Merge drm-upstream/drm-next into drm-intel-next-queued

Need MST sideband message transaction to power up/down nodes.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>

54 files changed, 14674 insertions, 165 deletions

diff --git a/lib/Kconfig b/lib/Kconfig
index 6762529ad9e4..b1445b22a6de 100644
--- a/lib/Kconfig
+++ b/lib/Kconfig
@@ -192,6 +192,9 @@ config CRC8
 	  when they need to do cyclic redundancy check according CRC8
 	  algorithm. Module will be called crc8.
 
+config XXHASH
+	tristate
+
 config AUDIT_GENERIC
 	bool
 	depends on AUDIT && !AUDIT_ARCH
@@ -246,6 +249,14 @@ config LZ4HC_COMPRESS
 config LZ4_DECOMPRESS
 	tristate
 
+config ZSTD_COMPRESS
+	select XXHASH
+	tristate
+
+config ZSTD_DECOMPRESS
+	select XXHASH
+	tristate
+
 source "lib/xz/Kconfig"
 
 #
@@ -559,9 +570,6 @@ config ARCH_HAS_PMEM_API
 config ARCH_HAS_UACCESS_FLUSHCACHE
 	bool
 
-config ARCH_HAS_MMIO_FLUSH
-	bool
-
 config STACKDEPOT
 	bool
 	select STACKTRACE
@@ -575,4 +583,7 @@ config PARMAN
 config PRIME_NUMBERS
 	tristate
 
+config STRING_SELFTEST
+	bool "Test string functions"
+
 endmenu
diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
index 98fe715522e8..2689b7c50c52 100644
--- a/lib/Kconfig.debug
+++ b/lib/Kconfig.debug
@@ -219,7 +219,8 @@ config FRAME_WARN
 	range 0 8192
 	default 0 if KASAN
 	default 2048 if GCC_PLUGIN_LATENT_ENTROPY
-	default 1024 if !64BIT
+	default 1280 if (!64BIT && PARISC)
+	default 1024 if (!64BIT && !PARISC)
 	default 2048 if 64BIT
 	help
 	  Tell gcc to warn at build time for stack frames larger than this.
@@ -374,6 +375,9 @@ config STACK_VALIDATION
 	  pointers (if CONFIG_FRAME_POINTER is enabled).  This helps ensure
 	  that runtime stack traces are more reliable.
 
+	  This is also a prerequisite for generation of ORC unwind data, which
+	  is needed for CONFIG_ORC_UNWINDER.
+
 	  For more information, see
 	  tools/objtool/Documentation/stack-validation.txt.
 
@@ -798,6 +802,13 @@ config HARDLOCKUP_DETECTOR_PERF
 	select SOFTLOCKUP_DETECTOR
 
 #
+# Enables a timestamp based low pass filter to compensate for perf based
+# hard lockup detection which runs too fast due to turbo modes.
+#
+config HARDLOCKUP_CHECK_TIMESTAMP
+	bool
+
+#
 # arch/ can define HAVE_HARDLOCKUP_DETECTOR_ARCH to provide their own hard
 # lockup detector rather than the perf based detector.
 #
@@ -1081,6 +1092,8 @@ config PROVE_LOCKING
 	select DEBUG_MUTEXES
 	select DEBUG_RT_MUTEXES if RT_MUTEXES
 	select DEBUG_LOCK_ALLOC
+	select LOCKDEP_CROSSRELEASE
+	select LOCKDEP_COMPLETIONS
 	select TRACE_IRQFLAGS
 	default n
 	help
@@ -1121,7 +1134,7 @@ config LOCKDEP
 	bool
 	depends on DEBUG_KERNEL && TRACE_IRQFLAGS_SUPPORT && STACKTRACE_SUPPORT && LOCKDEP_SUPPORT
 	select STACKTRACE
-	select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE && !ARC && !SCORE
+	select FRAME_POINTER if !MIPS && !PPC && !ARM_UNWIND && !S390 && !MICROBLAZE && !ARC && !SCORE && !X86
 	select KALLSYMS
 	select KALLSYMS_ALL
 
@@ -1150,6 +1163,22 @@ config LOCK_STAT
 	 CONFIG_LOCK_STAT defines "contended" and "acquired" lock events.
 	 (CONFIG_LOCKDEP defines "acquire" and "release" events.)
 
+config LOCKDEP_CROSSRELEASE
+	bool
+	help
+	 This makes lockdep work for crosslock which is a lock allowed to
+	 be released in a different context from the acquisition context.
+	 Normally a lock must be released in the context acquiring the lock.
+	 However, relexing this constraint helps synchronization primitives
+	 such as page locks or completions can use the lock correctness
+	 detector, lockdep.
+
+config LOCKDEP_COMPLETIONS
+	bool
+	help
+	 A deadlock caused by wait_for_completion() and complete() can be
+	 detected by lockdep using crossrelease feature.
+
 config DEBUG_LOCKDEP
 	bool "Lock dependency engine debugging"
 	depends on DEBUG_KERNEL && LOCKDEP
@@ -1540,7 +1569,7 @@ config FAULT_INJECTION_STACKTRACE_FILTER
 	depends on FAULT_INJECTION_DEBUG_FS && STACKTRACE_SUPPORT
 	depends on !X86_64
 	select STACKTRACE
-	select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC && !SCORE
+	select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC && !SCORE && !X86
 	help
 	  Provide stacktrace filter for fault-injection capabilities
 
@@ -1549,7 +1578,7 @@ config LATENCYTOP
 	depends on DEBUG_KERNEL
 	depends on STACKTRACE_SUPPORT
 	depends on PROC_FS
-	select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC
+	select FRAME_POINTER if !MIPS && !PPC && !S390 && !MICROBLAZE && !ARM_UNWIND && !ARC && !X86
 	select KALLSYMS
 	select KALLSYMS_ALL
 	select STACKTRACE
@@ -1902,6 +1931,17 @@ config TEST_KMOD
 
 	  If unsure, say N.
 
+config TEST_DEBUG_VIRTUAL
+	tristate "Test CONFIG_DEBUG_VIRTUAL feature"
+	depends on DEBUG_VIRTUAL
+	help
+	  Test the kernel's ability to detect incorrect calls to
+	  virt_to_phys() done against the non-linear part of the
+	  kernel's virtual address map.
+
+	  If unsure, say N.
+
+
 source "samples/Kconfig"
 
 source "lib/Kconfig.kgdb"
diff --git a/lib/Makefile b/lib/Makefile
index 40c18372b301..dafa79613fb4 100644
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -62,6 +62,7 @@ obj-$(CONFIG_TEST_BITMAP) += test_bitmap.o
 obj-$(CONFIG_TEST_UUID) += test_uuid.o
 obj-$(CONFIG_TEST_PARMAN) += test_parman.o
 obj-$(CONFIG_TEST_KMOD) += test_kmod.o
+obj-$(CONFIG_TEST_DEBUG_VIRTUAL) += test_debug_virtual.o
 
 ifeq ($(CONFIG_DEBUG_KOBJECT),y)
 CFLAGS_kobject.o += -DDEBUG
@@ -102,6 +103,7 @@ obj-$(CONFIG_CRC4)	+= crc4.o
 obj-$(CONFIG_CRC7)	+= crc7.o
 obj-$(CONFIG_LIBCRC32C)	+= libcrc32c.o
 obj-$(CONFIG_CRC8)	+= crc8.o
+obj-$(CONFIG_XXHASH)	+= xxhash.o
 obj-$(CONFIG_GENERIC_ALLOCATOR) += genalloc.o
 
 obj-$(CONFIG_842_COMPRESS) += 842/
@@ -115,6 +117,8 @@ obj-$(CONFIG_LZO_DECOMPRESS) += lzo/
 obj-$(CONFIG_LZ4_COMPRESS) += lz4/
 obj-$(CONFIG_LZ4HC_COMPRESS) += lz4/
 obj-$(CONFIG_LZ4_DECOMPRESS) += lz4/
+obj-$(CONFIG_ZSTD_COMPRESS) += zstd/
+obj-$(CONFIG_ZSTD_DECOMPRESS) += zstd/
 obj-$(CONFIG_XZ_DEC) += xz/
 obj-$(CONFIG_RAID6_PQ) += raid6/
 
diff --git a/lib/assoc_array.c b/lib/assoc_array.c
index 59fd7c0b119c..155c55d8db5f 100644
--- a/lib/assoc_array.c
+++ b/lib/assoc_array.c
@@ -1,6 +1,6 @@
 /* Generic associative array implementation.
  *
- * See Documentation/assoc_array.txt for information.
+ * See Documentation/core-api/assoc_array.rst for information.
  *
  * Copyright (C) 2013 Red Hat, Inc. All Rights Reserved.
  * Written by David Howells (dhowells@redhat.com)
diff --git a/lib/bitmap.c b/lib/bitmap.c
index 9a532805364b..c82c61b66e16 100644
--- a/lib/bitmap.c
+++ b/lib/bitmap.c
@@ -513,7 +513,7 @@ static int __bitmap_parselist(const char *buf, unsigned int buflen,
 		int nmaskbits)
 {
 	unsigned int a, b, old_a, old_b;
-	unsigned int group_size, used_size;
+	unsigned int group_size, used_size, off;
 	int c, old_c, totaldigits, ndigits;
 	const char __user __force *ubuf = (const char __user __force *)buf;
 	int at_start, in_range, in_partial_range;
@@ -599,6 +599,8 @@ static int __bitmap_parselist(const char *buf, unsigned int buflen,
 			a = old_a;
 			b = old_b;
 			old_a = old_b = 0;
+		} else {
+			used_size = group_size = b - a + 1;
 		}
 		/* if no digit is after '-', it's wrong*/
 		if (at_start && in_range)
@@ -608,17 +610,9 @@ static int __bitmap_parselist(const char *buf, unsigned int buflen,
 		if (b >= nmaskbits)
 			return -ERANGE;
 		while (a <= b) {
-			if (in_partial_range) {
-				static int pos_in_group = 1;
-
-				if (pos_in_group <= used_size)
-					set_bit(a, maskp);
-
-				if (a == b || ++pos_in_group > group_size)
-					pos_in_group = 1;
-			} else
-				set_bit(a, maskp);
-			a++;
+			off = min(b - a + 1, used_size);
+			bitmap_set(maskp, a, off);
+			a += group_size;
 		}
 	} while (buflen && c == ',');
 	return 0;
diff --git a/lib/cmdline.c b/lib/cmdline.c
index 4c0888c4a68d..171c19b6888e 100644
--- a/lib/cmdline.c
+++ b/lib/cmdline.c
@@ -244,5 +244,4 @@ char *next_arg(char *args, char **param, char **val)
 
 	/* Chew up trailing spaces. */
 	return skip_spaces(next);
-	//return next;
 }
diff --git a/lib/cpumask.c b/lib/cpumask.c
index 4731a0895760..8b1a1bd77539 100644
--- a/lib/cpumask.c
+++ b/lib/cpumask.c
@@ -6,6 +6,22 @@
 #include <linux/bootmem.h>
 
 /**
+ * cpumask_next - get the next cpu in a cpumask
+ * @n: the cpu prior to the place to search (ie. return will be > @n)
+ * @srcp: the cpumask pointer
+ *
+ * Returns >= nr_cpu_ids if no further cpus set.
+ */
+unsigned int cpumask_next(int n, const struct cpumask *srcp)
+{
+	/* -1 is a legal arg here. */
+	if (n != -1)
+		cpumask_check(n);
+	return find_next_bit(cpumask_bits(srcp), nr_cpumask_bits, n + 1);
+}
+EXPORT_SYMBOL(cpumask_next);
+
+/**
  * cpumask_next_and - get the next cpu in *src1p & *src2p
  * @n: the cpu prior to the place to search (ie. return will be > @n)
  * @src1p: the first cpumask pointer
diff --git a/lib/debugobjects.c b/lib/debugobjects.c
index 17afb0430161..2f5349c6e81a 100644
--- a/lib/debugobjects.c
+++ b/lib/debugobjects.c
@@ -18,6 +18,7 @@
 #include <linux/debugfs.h>
 #include <linux/slab.h>
 #include <linux/hash.h>
+#include <linux/kmemleak.h>
 
 #define ODEBUG_HASH_BITS	14
 #define ODEBUG_HASH_SIZE	(1 << ODEBUG_HASH_BITS)
@@ -110,6 +111,7 @@ static void fill_pool(void)
 		if (!new)
 			return;
 
+		kmemleak_ignore(new);
 		raw_spin_lock_irqsave(&pool_lock, flags);
 		hlist_add_head(&new->node, &obj_pool);
 		debug_objects_allocated++;
@@ -1080,6 +1082,7 @@ static int __init debug_objects_replace_static_objects(void)
 		obj = kmem_cache_zalloc(obj_cache, GFP_KERNEL);
 		if (!obj)
 			goto free;
+		kmemleak_ignore(obj);
 		hlist_add_head(&obj->node, &objects);
 	}
 
diff --git a/lib/errseq.c b/lib/errseq.c
index 841fa24e6e00..7b900c2a277a 100644
--- a/lib/errseq.c
+++ b/lib/errseq.c
@@ -41,23 +41,20 @@
 #define ERRSEQ_CTR_INC		(1 << (ERRSEQ_SHIFT + 1))
 
 /**
- * __errseq_set - set a errseq_t for later reporting
+ * errseq_set - set a errseq_t for later reporting
  * @eseq: errseq_t field that should be set
- * @err: error to set
+ * @err: error to set (must be between -1 and -MAX_ERRNO)
  *
  * This function sets the error in *eseq, and increments the sequence counter
  * if the last sequence was sampled at some point in the past.
  *
  * Any error set will always overwrite an existing error.
  *
- * Most callers will want to use the errseq_set inline wrapper to efficiently
- * handle the common case where err is 0.
- *
- * We do return an errseq_t here, primarily for debugging purposes. The return
- * value should not be used as a previously sampled value in later calls as it
- * will not have the SEEN flag set.
+ * We do return the latest value here, primarily for debugging purposes. The
+ * return value should not be used as a previously sampled value in later calls
+ * as it will not have the SEEN flag set.
  */
-errseq_t __errseq_set(errseq_t *eseq, int err)
+errseq_t errseq_set(errseq_t *eseq, int err)
 {
 	errseq_t cur, old;
 
@@ -107,7 +104,7 @@ errseq_t __errseq_set(errseq_t *eseq, int err)
 	}
 	return cur;
 }
-EXPORT_SYMBOL(__errseq_set);
+EXPORT_SYMBOL(errseq_set);
 
 /**
  * errseq_sample - grab current errseq_t value
diff --git a/lib/hexdump.c b/lib/hexdump.c
index 992457b1284c..81b70ed37209 100644
--- a/lib/hexdump.c
+++ b/lib/hexdump.c
@@ -9,6 +9,7 @@
 
 #include <linux/types.h>
 #include <linux/ctype.h>
+#include <linux/errno.h>
 #include <linux/kernel.h>
 #include <linux/export.h>
 #include <asm/unaligned.h>
@@ -42,7 +43,7 @@ EXPORT_SYMBOL(hex_to_bin);
  * @src: ascii hexadecimal string
  * @count: result length
  *
- * Return 0 on success, -1 in case of bad input.
+ * Return 0 on success, -EINVAL in case of bad input.
  */
 int hex2bin(u8 *dst, const char *src, size_t count)
 {
@@ -51,7 +52,7 @@ int hex2bin(u8 *dst, const char *src, size_t count)
 		int lo = hex_to_bin(*src++);
 
 		if ((hi < 0) || (lo < 0))
-			return -1;
+			return -EINVAL;
 
 		*dst++ = (hi << 4) | lo;
 	}
diff --git a/lib/idr.c b/lib/idr.c
index b13682bb0a1c..f9adf4805fd7 100644
--- a/lib/idr.c
+++ b/lib/idr.c
@@ -7,45 +7,32 @@
 DEFINE_PER_CPU(struct ida_bitmap *, ida_bitmap);
 static DEFINE_SPINLOCK(simple_ida_lock);
 
-/**
- * idr_alloc - allocate an id
- * @idr: idr handle
- * @ptr: pointer to be associated with the new id
- * @start: the minimum id (inclusive)
- * @end: the maximum id (exclusive)
- * @gfp: memory allocation flags
- *
- * Allocates an unused ID in the range [start, end).  Returns -ENOSPC
- * if there are no unused IDs in that range.
- *
- * Note that @end is treated as max when <= 0.  This is to always allow
- * using @start + N as @end as long as N is inside integer range.
- *
- * Simultaneous modifications to the @idr are not allowed and should be
- * prevented by the user, usually with a lock.  idr_alloc() may be called
- * concurrently with read-only accesses to the @idr, such as idr_find() and
- * idr_for_each_entry().
- */
-int idr_alloc(struct idr *idr, void *ptr, int start, int end, gfp_t gfp)
+int idr_alloc_cmn(struct idr *idr, void *ptr, unsigned long *index,
+		  unsigned long start, unsigned long end, gfp_t gfp,
+		  bool ext)
 {
-	void __rcu **slot;
 	struct radix_tree_iter iter;
+	void __rcu **slot;
 
-	if (WARN_ON_ONCE(start < 0))
-		return -EINVAL;
 	if (WARN_ON_ONCE(radix_tree_is_internal_node(ptr)))
 		return -EINVAL;
 
 	radix_tree_iter_init(&iter, start);
-	slot = idr_get_free(&idr->idr_rt, &iter, gfp, end);
+	if (ext)
+		slot = idr_get_free_ext(&idr->idr_rt, &iter, gfp, end);
+	else
+		slot = idr_get_free(&idr->idr_rt, &iter, gfp, end);
 	if (IS_ERR(slot))
 		return PTR_ERR(slot);
 
 	radix_tree_iter_replace(&idr->idr_rt, &iter, slot, ptr);
 	radix_tree_iter_tag_clear(&idr->idr_rt, &iter, IDR_FREE);
-	return iter.index;
+
+	if (index)
+		*index = iter.index;
+	return 0;
 }
-EXPORT_SYMBOL_GPL(idr_alloc);
+EXPORT_SYMBOL_GPL(idr_alloc_cmn);
 
 /**
  * idr_alloc_cyclic - allocate new idr entry in a cyclical fashion
@@ -134,6 +121,20 @@ void *idr_get_next(struct idr *idr, int *nextid)
 }
 EXPORT_SYMBOL(idr_get_next);
 
+void *idr_get_next_ext(struct idr *idr, unsigned long *nextid)
+{
+	struct radix_tree_iter iter;
+	void __rcu **slot;
+
+	slot = radix_tree_iter_find(&idr->idr_rt, &iter, *nextid);
+	if (!slot)
+		return NULL;
+
+	*nextid = iter.index;
+	return rcu_dereference_raw(*slot);
+}
+EXPORT_SYMBOL(idr_get_next_ext);
+
 /**
  * idr_replace - replace pointer for given id
  * @idr: idr handle
@@ -150,12 +151,19 @@ EXPORT_SYMBOL(idr_get_next);
  */
 void *idr_replace(struct idr *idr, void *ptr, int id)
 {
+	if (id < 0)
+		return ERR_PTR(-EINVAL);
+
+	return idr_replace_ext(idr, ptr, id);
+}
+EXPORT_SYMBOL(idr_replace);
+
+void *idr_replace_ext(struct idr *idr, void *ptr, unsigned long id)
+{
 	struct radix_tree_node *node;
 	void __rcu **slot = NULL;
 	void *entry;
 
-	if (WARN_ON_ONCE(id < 0))
-		return ERR_PTR(-EINVAL);
 	if (WARN_ON_ONCE(radix_tree_is_internal_node(ptr)))
 		return ERR_PTR(-EINVAL);
 
@@ -167,7 +175,7 @@ void *idr_replace(struct idr *idr, void *ptr, int id)
 
 	return entry;
 }
-EXPORT_SYMBOL(idr_replace);
+EXPORT_SYMBOL(idr_replace_ext);
 
 /**
  * DOC: IDA description
diff --git a/lib/interval_tree_test.c b/lib/interval_tree_test.c
index df495fe81421..0e343fd29570 100644
--- a/lib/interval_tree_test.c
+++ b/lib/interval_tree_test.c
@@ -19,14 +19,14 @@ __param(bool, search_all, false, "Searches will iterate all nodes in the tree");
 
 __param(uint, max_endpoint, ~0, "Largest value for the interval's endpoint");
 
-static struct rb_root root = RB_ROOT;
+static struct rb_root_cached root = RB_ROOT_CACHED;
 static struct interval_tree_node *nodes = NULL;
 static u32 *queries = NULL;
 
 static struct rnd_state rnd;
 
 static inline unsigned long
-search(struct rb_root *root, unsigned long start, unsigned long last)
+search(struct rb_root_cached *root, unsigned long start, unsigned long last)
 {
 	struct interval_tree_node *node;
 	unsigned long results = 0;
diff --git a/lib/iov_iter.c b/lib/iov_iter.c
index 52c8dd6d8e82..1c1c06ddc20a 100644
--- a/lib/iov_iter.c
+++ b/lib/iov_iter.c
@@ -687,8 +687,10 @@ EXPORT_SYMBOL(_copy_from_iter_full_nocache);
 
 static inline bool page_copy_sane(struct page *page, size_t offset, size_t n)
 {
-	size_t v = n + offset;
-	if (likely(n <= v && v <= (PAGE_SIZE << compound_order(page))))
+	struct page *head = compound_head(page);
+	size_t v = n + offset + page_address(page) - page_address(head);
+
+	if (likely(n <= v && v <= (PAGE_SIZE << compound_order(head))))
 		return true;
 	WARN_ON(1);
 	return false;
diff --git a/lib/kobject_uevent.c b/lib/kobject_uevent.c
index 719c155fce20..e590523ea476 100644
--- a/lib/kobject_uevent.c
+++ b/lib/kobject_uevent.c
@@ -52,6 +52,8 @@ static const char *kobject_actions[] = {
 	[KOBJ_MOVE] =		"move",
 	[KOBJ_ONLINE] =		"online",
 	[KOBJ_OFFLINE] =	"offline",
+	[KOBJ_BIND] =		"bind",
+	[KOBJ_UNBIND] =		"unbind",
 };
 
 static int kobject_action_type(const char *buf, size_t count,
diff --git a/lib/locking-selftest.c b/lib/locking-selftest.c
index 6f2b135dc5e8..cd0b5c964bd0 100644
--- a/lib/locking-selftest.c
+++ b/lib/locking-selftest.c
@@ -363,6 +363,103 @@ static void rsem_AA3(void)
 }
 
 /*
+ * read_lock(A)
+ * spin_lock(B)
+ *		spin_lock(B)
+ *		write_lock(A)
+ */
+static void rlock_ABBA1(void)
+{
+	RL(X1);
+	L(Y1);
+	U(Y1);
+	RU(X1);
+
+	L(Y1);
+	WL(X1);
+	WU(X1);
+	U(Y1); // should fail
+}
+
+static void rwsem_ABBA1(void)
+{
+	RSL(X1);
+	ML(Y1);
+	MU(Y1);
+	RSU(X1);
+
+	ML(Y1);
+	WSL(X1);
+	WSU(X1);
+	MU(Y1); // should fail
+}
+
+/*
+ * read_lock(A)
+ * spin_lock(B)
+ *		spin_lock(B)
+ *		read_lock(A)
+ */
+static void rlock_ABBA2(void)
+{
+	RL(X1);
+	L(Y1);
+	U(Y1);
+	RU(X1);
+
+	L(Y1);
+	RL(X1);
+	RU(X1);
+	U(Y1); // should NOT fail
+}
+
+static void rwsem_ABBA2(void)
+{
+	RSL(X1);
+	ML(Y1);
+	MU(Y1);
+	RSU(X1);
+
+	ML(Y1);
+	RSL(X1);
+	RSU(X1);
+	MU(Y1); // should fail
+}
+
+
+/*
+ * write_lock(A)
+ * spin_lock(B)
+ *		spin_lock(B)
+ *		write_lock(A)
+ */
+static void rlock_ABBA3(void)
+{
+	WL(X1);
+	L(Y1);
+	U(Y1);
+	WU(X1);
+
+	L(Y1);
+	WL(X1);
+	WU(X1);
+	U(Y1); // should fail
+}
+
+static void rwsem_ABBA3(void)
+{
+	WSL(X1);
+	ML(Y1);
+	MU(Y1);
+	WSU(X1);
+
+	ML(Y1);
+	WSL(X1);
+	WSU(X1);
+	MU(Y1); // should fail
+}
+
+/*
  * ABBA deadlock:
  */
 
@@ -1056,8 +1153,6 @@ static void dotest(void (*testcase_fn)(void), int expected, int lockclass_mask)
 	if (debug_locks != expected) {
 		unexpected_testcase_failures++;
 		pr_cont("FAILED|");
-
-		dump_stack();
 	} else {
 		testcase_successes++;
 		pr_cont("  ok  |");
@@ -1933,6 +2028,30 @@ void locking_selftest(void)
 	dotest(rsem_AA3, FAILURE, LOCKTYPE_RWSEM);
 	pr_cont("\n");
 
+	print_testname("mixed read-lock/lock-write ABBA");
+	pr_cont("             |");
+	dotest(rlock_ABBA1, FAILURE, LOCKTYPE_RWLOCK);
+	/*
+	 * Lockdep does indeed fail here, but there's nothing we can do about
+	 * that now.  Don't kill lockdep for it.
+	 */
+	unexpected_testcase_failures--;
+
+	pr_cont("             |");
+	dotest(rwsem_ABBA1, FAILURE, LOCKTYPE_RWSEM);
+
+	print_testname("mixed read-lock/lock-read ABBA");
+	pr_cont("             |");
+	dotest(rlock_ABBA2, SUCCESS, LOCKTYPE_RWLOCK);
+	pr_cont("             |");
+	dotest(rwsem_ABBA2, FAILURE, LOCKTYPE_RWSEM);
+
+	print_testname("mixed write-lock/lock-write ABBA");
+	pr_cont("             |");
+	dotest(rlock_ABBA3, FAILURE, LOCKTYPE_RWLOCK);
+	pr_cont("             |");
+	dotest(rwsem_ABBA3, FAILURE, LOCKTYPE_RWSEM);
+
 	printk("  --------------------------------------------------------------------------\n");
 
 	/*
diff --git a/lib/mpi/longlong.h b/lib/mpi/longlong.h
index 93336502af08..57fd45ab7af1 100644
--- a/lib/mpi/longlong.h
+++ b/lib/mpi/longlong.h
@@ -176,8 +176,8 @@ extern UDItype __udiv_qrnnd(UDItype *, UDItype, UDItype, UDItype);
 #define add_ssaaaa(sh, sl, ah, al, bh, bl) \
 	__asm__ ("adds %1, %4, %5\n" \
 		"adc  %0, %2, %3" \
-	: "=r" ((USItype)(sh)), \
-		"=&r" ((USItype)(sl)) \
+	: "=r" (sh), \
+		"=&r" (sl) \
 	: "%r" ((USItype)(ah)), \
 		"rI" ((USItype)(bh)), \
 		"%r" ((USItype)(al)), \
@@ -185,15 +185,15 @@ extern UDItype __udiv_qrnnd(UDItype *, UDItype, UDItype, UDItype);
 #define sub_ddmmss(sh, sl, ah, al, bh, bl) \
 	__asm__ ("subs %1, %4, %5\n" \
 		"sbc  %0, %2, %3" \
-	: "=r" ((USItype)(sh)), \
-		"=&r" ((USItype)(sl)) \
+	: "=r" (sh), \
+		"=&r" (sl) \
 	: "r" ((USItype)(ah)), \
 		"rI" ((USItype)(bh)), \
 		"r" ((USItype)(al)), \
 		"rI" ((USItype)(bl)))
 #if defined __ARM_ARCH_2__ || defined __ARM_ARCH_3__
 #define umul_ppmm(xh, xl, a, b) \
-	__asm__ ("%@ Inlined umul_ppmm\n" \
+	__asm__ ("@ Inlined umul_ppmm\n" \
 		"mov	%|r0, %2, lsr #16		@ AAAA\n" \
 		"mov	%|r2, %3, lsr #16		@ BBBB\n" \
 		"bic	%|r1, %2, %|r0, lsl #16		@ aaaa\n" \
@@ -206,19 +206,19 @@ extern UDItype __udiv_qrnnd(UDItype *, UDItype, UDItype, UDItype);
 		"addcs	%|r2, %|r2, #65536\n" \
 		"adds	%1, %|r1, %|r0, lsl #16\n" \
 		"adc	%0, %|r2, %|r0, lsr #16" \
-	: "=&r" ((USItype)(xh)), \
-		"=r" ((USItype)(xl)) \
+	: "=&r" (xh), \
+		"=r" (xl) \
 	: "r" ((USItype)(a)), \
 		"r" ((USItype)(b)) \
 	: "r0", "r1", "r2")
 #else
 #define umul_ppmm(xh, xl, a, b) \
-	__asm__ ("%@ Inlined umul_ppmm\n" \
-		"umull %r1, %r0, %r2, %r3" \
-	: "=&r" ((USItype)(xh)), \
-			"=&r" ((USItype)(xl)) \
+	__asm__ ("@ Inlined umul_ppmm\n" \
+		"umull %1, %0, %2, %3" \
+	: "=&r" (xh), \
+		"=&r" (xl) \
 	: "r" ((USItype)(a)), \
-			"r" ((USItype)(b)) \
+		"r" ((USItype)(b)) \
 	: "r0", "r1")
 #endif
 #define UMUL_TIME 20
diff --git a/lib/mpi/mpicoder.c b/lib/mpi/mpicoder.c
index 5a0f75a3bf01..eead4b339466 100644
--- a/lib/mpi/mpicoder.c
+++ b/lib/mpi/mpicoder.c
@@ -364,11 +364,11 @@ MPI mpi_read_raw_from_sgl(struct scatterlist *sgl, unsigned int nbytes)
 	}
 
 	miter.consumed = lzeros;
-	sg_miter_stop(&miter);
 
 	nbytes -= lzeros;
 	nbits = nbytes * 8;
 	if (nbits > MAX_EXTERN_MPI_BITS) {
+		sg_miter_stop(&miter);
 		pr_info("MPI: mpi too large (%u bits)\n", nbits);
 		return NULL;
 	}
@@ -376,6 +376,8 @@ MPI mpi_read_raw_from_sgl(struct scatterlist *sgl, unsigned int nbytes)
 	if (nbytes > 0)
 		nbits -= count_leading_zeros(*buff) - (BITS_PER_LONG - 8);
 
+	sg_miter_stop(&miter);
+
 	nlimbs = DIV_ROUND_UP(nbytes, BYTES_PER_MPI_LIMB);
 	val = mpi_alloc(nlimbs);
 	if (!val)
diff --git a/lib/nlattr.c b/lib/nlattr.c
index fb52435be42d..927c2f19f119 100644
--- a/lib/nlattr.c
+++ b/lib/nlattr.c
@@ -27,6 +27,30 @@ static const u8 nla_attr_minlen[NLA_TYPE_MAX+1] = {
 	[NLA_S64]	= sizeof(s64),
 };
 
+static int validate_nla_bitfield32(const struct nlattr *nla,
+				   u32 *valid_flags_allowed)
+{
+	const struct nla_bitfield32 *bf = nla_data(nla);
+	u32 *valid_flags_mask = valid_flags_allowed;
+
+	if (!valid_flags_allowed)
+		return -EINVAL;
+
+	/*disallow invalid bit selector */
+	if (bf->selector & ~*valid_flags_mask)
+		return -EINVAL;
+
+	/*disallow invalid bit values */
+	if (bf->value & ~*valid_flags_mask)
+		return -EINVAL;
+
+	/*disallow valid bit values that are not selected*/
+	if (bf->value & ~bf->selector)
+		return -EINVAL;
+
+	return 0;
+}
+
 static int validate_nla(const struct nlattr *nla, int maxtype,
 			const struct nla_policy *policy)
 {
@@ -46,6 +70,12 @@ static int validate_nla(const struct nlattr *nla, int maxtype,
 			return -ERANGE;
 		break;
 
+	case NLA_BITFIELD32:
+		if (attrlen != sizeof(struct nla_bitfield32))
+			return -ERANGE;
+
+		return validate_nla_bitfield32(nla, pt->validation_data);
+
 	case NLA_NUL_STRING:
 		if (pt->len)
 			minlen = min_t(int, attrlen, pt->len + 1);
@@ -272,6 +302,30 @@ size_t nla_strlcpy(char *dst, const struct nlattr *nla, size_t dstsize)
 EXPORT_SYMBOL(nla_strlcpy);
 
 /**
+ * nla_strdup - Copy string attribute payload into a newly allocated buffer
+ * @nla: attribute to copy the string from
+ * @flags: the type of memory to allocate (see kmalloc).
+ *
+ * Returns a pointer to the allocated buffer or NULL on error.
+ */
+char *nla_strdup(const struct nlattr *nla, gfp_t flags)
+{
+	size_t srclen = nla_len(nla);
+	char *src = nla_data(nla), *dst;
+
+	if (srclen > 0 && src[srclen - 1] == '\0')
+		srclen--;
+
+	dst = kmalloc(srclen + 1, flags);
+	if (dst != NULL) {
+		memcpy(dst, src, srclen);
+		dst[srclen] = '\0';
+	}
+	return dst;
+}
+EXPORT_SYMBOL(nla_strdup);
+
+/**
  * nla_memcpy - Copy a netlink attribute into another memory area
  * @dest: where to copy to memcpy
  * @src: netlink attribute to copy from
diff --git a/lib/oid_registry.c b/lib/oid_registry.c
index 318f382a010d..41b9e50711a7 100644
--- a/lib/oid_registry.c
+++ b/lib/oid_registry.c
@@ -142,9 +142,9 @@ int sprint_oid(const void *data, size_t datasize, char *buffer, size_t bufsize)
 		}
 		ret += count = snprintf(buffer, bufsize, ".%lu", num);
 		buffer += count;
-		bufsize -= count;
-		if (bufsize == 0)
+		if (bufsize <= count)
 			return -ENOBUFS;
+		bufsize -= count;
 	}
 
 	return ret;
diff --git a/lib/radix-tree.c b/lib/radix-tree.c
index 3527eb364964..8b1feca1230a 100644
--- a/lib/radix-tree.c
+++ b/lib/radix-tree.c
@@ -463,7 +463,7 @@ radix_tree_node_free(struct radix_tree_node *node)
  * To make use of this facility, the radix tree must be initialised without
  * __GFP_DIRECT_RECLAIM being passed to INIT_RADIX_TREE().
  */
-static int __radix_tree_preload(gfp_t gfp_mask, unsigned nr)
+static __must_check int __radix_tree_preload(gfp_t gfp_mask, unsigned nr)
 {
 	struct radix_tree_preload *rtp;
 	struct radix_tree_node *node;
@@ -2104,7 +2104,8 @@ EXPORT_SYMBOL(radix_tree_tagged);
  */
 void idr_preload(gfp_t gfp_mask)
 {
-	__radix_tree_preload(gfp_mask, IDR_PRELOAD_SIZE);
+	if (__radix_tree_preload(gfp_mask, IDR_PRELOAD_SIZE))
+		preempt_disable();
 }
 EXPORT_SYMBOL(idr_preload);
 
@@ -2118,13 +2119,13 @@ EXPORT_SYMBOL(idr_preload);
  */
 int ida_pre_get(struct ida *ida, gfp_t gfp)
 {
-	__radix_tree_preload(gfp, IDA_PRELOAD_SIZE);
 	/*
 	 * The IDA API has no preload_end() equivalent.  Instead,
 	 * ida_get_new() can return -EAGAIN, prompting the caller
 	 * to return to the ida_pre_get() step.
 	 */
-	preempt_enable();
+	if (!__radix_tree_preload(gfp, IDA_PRELOAD_SIZE))
+		preempt_enable();
 
 	if (!this_cpu_read(ida_bitmap)) {
 		struct ida_bitmap *bitmap = kmalloc(sizeof(*bitmap), gfp);
@@ -2138,13 +2139,13 @@ int ida_pre_get(struct ida *ida, gfp_t gfp)
 }
 EXPORT_SYMBOL(ida_pre_get);
 
-void __rcu **idr_get_free(struct radix_tree_root *root,
-			struct radix_tree_iter *iter, gfp_t gfp, int end)
+void __rcu **idr_get_free_cmn(struct radix_tree_root *root,
+			      struct radix_tree_iter *iter, gfp_t gfp,
+			      unsigned long max)
 {
 	struct radix_tree_node *node = NULL, *child;
 	void __rcu **slot = (void __rcu **)&root->rnode;
 	unsigned long maxindex, start = iter->next_index;
-	unsigned long max = end > 0 ? end - 1 : INT_MAX;
 	unsigned int shift, offset = 0;
 
  grow:
diff --git a/lib/raid6/Makefile b/lib/raid6/Makefile
index 3057011f5599..a93adf6dcfb2 100644
--- a/lib/raid6/Makefile
+++ b/lib/raid6/Makefile
@@ -5,7 +5,7 @@ raid6_pq-y	+= algos.o recov.o tables.o int1.o int2.o int4.o \
 
 raid6_pq-$(CONFIG_X86) += recov_ssse3.o recov_avx2.o mmx.o sse1.o sse2.o avx2.o avx512.o recov_avx512.o
 raid6_pq-$(CONFIG_ALTIVEC) += altivec1.o altivec2.o altivec4.o altivec8.o
-raid6_pq-$(CONFIG_KERNEL_MODE_NEON) += neon.o neon1.o neon2.o neon4.o neon8.o
+raid6_pq-$(CONFIG_KERNEL_MODE_NEON) += neon.o neon1.o neon2.o neon4.o neon8.o recov_neon.o recov_neon_inner.o
 raid6_pq-$(CONFIG_TILEGX) += tilegx8.o
 raid6_pq-$(CONFIG_S390) += s390vx8.o recov_s390xc.o
 
@@ -26,7 +26,9 @@ NEON_FLAGS := -ffreestanding
 ifeq ($(ARCH),arm)
 NEON_FLAGS += -mfloat-abi=softfp -mfpu=neon
 endif
+CFLAGS_recov_neon_inner.o += $(NEON_FLAGS)
 ifeq ($(ARCH),arm64)
+CFLAGS_REMOVE_recov_neon_inner.o += -mgeneral-regs-only
 CFLAGS_REMOVE_neon1.o += -mgeneral-regs-only
 CFLAGS_REMOVE_neon2.o += -mgeneral-regs-only
 CFLAGS_REMOVE_neon4.o += -mgeneral-regs-only
diff --git a/lib/raid6/algos.c b/lib/raid6/algos.c
index 7857049fd7d3..476994723258 100644
--- a/lib/raid6/algos.c
+++ b/lib/raid6/algos.c
@@ -113,6 +113,9 @@ const struct raid6_recov_calls *const raid6_recov_algos[] = {
 #ifdef CONFIG_S390
 	&raid6_recov_s390xc,
 #endif
+#if defined(CONFIG_KERNEL_MODE_NEON)
+	&raid6_recov_neon,
+#endif
 	&raid6_recov_intx1,
 	NULL
 };
diff --git a/lib/raid6/avx512.c b/lib/raid6/avx512.c
index f524a7972006..46df7977b971 100644
--- a/lib/raid6/avx512.c
+++ b/lib/raid6/avx512.c
@@ -29,7 +29,7 @@
 
 static const struct raid6_avx512_constants {
 	u64 x1d[8];
-} raid6_avx512_constants __aligned(512) = {
+} raid6_avx512_constants __aligned(512/8) = {
 	{ 0x1d1d1d1d1d1d1d1dULL, 0x1d1d1d1d1d1d1d1dULL,
 	  0x1d1d1d1d1d1d1d1dULL, 0x1d1d1d1d1d1d1d1dULL,
 	  0x1d1d1d1d1d1d1d1dULL, 0x1d1d1d1d1d1d1d1dULL,
diff --git a/lib/raid6/neon.uc b/lib/raid6/neon.uc
index 4fa51b761dd0..d5242f544551 100644
--- a/lib/raid6/neon.uc
+++ b/lib/raid6/neon.uc
@@ -46,8 +46,12 @@ static inline unative_t SHLBYTE(unative_t v)
  */
 static inline unative_t MASK(unative_t v)
 {
-	const uint8x16_t temp = NBYTES(0);
-	return (unative_t)vcltq_s8((int8x16_t)v, (int8x16_t)temp);
+	return (unative_t)vshrq_n_s8((int8x16_t)v, 7);
+}
+
+static inline unative_t PMUL(unative_t v, unative_t u)
+{
+	return (unative_t)vmulq_p8((poly8x16_t)v, (poly8x16_t)u);
 }
 
 void raid6_neon$#_gen_syndrome_real(int disks, unsigned long bytes, void **ptrs)
@@ -110,7 +114,30 @@ void raid6_neon$#_xor_syndrome_real(int disks, int start, int stop,
 			wq$$ = veorq_u8(w1$$, wd$$);
 		}
 		/* P/Q left side optimization */
-		for ( z = start-1 ; z >= 0 ; z-- ) {
+		for ( z = start-1 ; z >= 3 ; z -= 4 ) {
+			w2$$ = vshrq_n_u8(wq$$, 4);
+			w1$$ = vshlq_n_u8(wq$$, 4);
+
+			w2$$ = PMUL(w2$$, x1d);
+			wq$$ = veorq_u8(w1$$, w2$$);
+		}
+
+		switch (z) {
+		case 2:
+			w2$$ = vshrq_n_u8(wq$$, 5);
+			w1$$ = vshlq_n_u8(wq$$, 3);
+
+			w2$$ = PMUL(w2$$, x1d);
+			wq$$ = veorq_u8(w1$$, w2$$);
+			break;
+		case 1:
+			w2$$ = vshrq_n_u8(wq$$, 6);
+			w1$$ = vshlq_n_u8(wq$$, 2);
+
+			w2$$ = PMUL(w2$$, x1d);
+			wq$$ = veorq_u8(w1$$, w2$$);
+			break;
+		case 0:
 			w2$$ = MASK(wq$$);
 			w1$$ = SHLBYTE(wq$$);
 
diff --git a/lib/raid6/recov_neon.c b/lib/raid6/recov_neon.c
new file mode 100644
index 000000000000..eeb5c4065b92
--- /dev/null
+++ b/lib/raid6/recov_neon.c
@@ -0,0 +1,110 @@
+/*
+ * Copyright (C) 2012 Intel Corporation
+ * Copyright (C) 2017 Linaro Ltd. <ard.biesheuvel@linaro.org>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+
+#include <linux/raid/pq.h>
+
+#ifdef __KERNEL__
+#include <asm/neon.h>
+#else
+#define kernel_neon_begin()
+#define kernel_neon_end()
+#define cpu_has_neon()		(1)
+#endif
+
+static int raid6_has_neon(void)
+{
+	return cpu_has_neon();
+}
+
+void __raid6_2data_recov_neon(int bytes, uint8_t *p, uint8_t *q, uint8_t *dp,
+			      uint8_t *dq, const uint8_t *pbmul,
+			      const uint8_t *qmul);
+
+void __raid6_datap_recov_neon(int bytes, uint8_t *p, uint8_t *q, uint8_t *dq,
+			      const uint8_t *qmul);
+
+static void raid6_2data_recov_neon(int disks, size_t bytes, int faila,
+		int failb, void **ptrs)
+{
+	u8 *p, *q, *dp, *dq;
+	const u8 *pbmul;	/* P multiplier table for B data */
+	const u8 *qmul;		/* Q multiplier table (for both) */
+
+	p = (u8 *)ptrs[disks - 2];
+	q = (u8 *)ptrs[disks - 1];
+
+	/*
+	 * Compute syndrome with zero for the missing data pages
+	 * Use the dead data pages as temporary storage for
+	 * delta p and delta q
+	 */
+	dp = (u8 *)ptrs[faila];
+	ptrs[faila] = (void *)raid6_empty_zero_page;
+	ptrs[disks - 2] = dp;
+	dq = (u8 *)ptrs[failb];
+	ptrs[failb] = (void *)raid6_empty_zero_page;
+	ptrs[disks - 1] = dq;
+
+	raid6_call.gen_syndrome(disks, bytes, ptrs);
+
+	/* Restore pointer table */
+	ptrs[faila]     = dp;
+	ptrs[failb]     = dq;
+	ptrs[disks - 2] = p;
+	ptrs[disks - 1] = q;
+
+	/* Now, pick the proper data tables */
+	pbmul = raid6_vgfmul[raid6_gfexi[failb-faila]];
+	qmul  = raid6_vgfmul[raid6_gfinv[raid6_gfexp[faila] ^
+					 raid6_gfexp[failb]]];
+
+	kernel_neon_begin();
+	__raid6_2data_recov_neon(bytes, p, q, dp, dq, pbmul, qmul);
+	kernel_neon_end();
+}
+
+static void raid6_datap_recov_neon(int disks, size_t bytes, int faila,
+		void **ptrs)
+{
+	u8 *p, *q, *dq;
+	const u8 *qmul;		/* Q multiplier table */
+
+	p = (u8 *)ptrs[disks - 2];
+	q = (u8 *)ptrs[disks - 1];
+
+	/*
+	 * Compute syndrome with zero for the missing data page
+	 * Use the dead data page as temporary storage for delta q
+	 */
+	dq = (u8 *)ptrs[faila];
+	ptrs[faila] = (void *)raid6_empty_zero_page;
+	ptrs[disks - 1] = dq;
+
+	raid6_call.gen_syndrome(disks, bytes, ptrs);
+
+	/* Restore pointer table */
+	ptrs[faila]     = dq;
+	ptrs[disks - 1] = q;
+
+	/* Now, pick the proper data tables */
+	qmul = raid6_vgfmul[raid6_gfinv[raid6_gfexp[faila]]];
+
+	kernel_neon_begin();
+	__raid6_datap_recov_neon(bytes, p, q, dq, qmul);
+	kernel_neon_end();
+}
+
+const struct raid6_recov_calls raid6_recov_neon = {
+	.data2		= raid6_2data_recov_neon,
+	.datap		= raid6_datap_recov_neon,
+	.valid		= raid6_has_neon,
+	.name		= "neon",
+	.priority	= 10,
+};
diff --git a/lib/raid6/recov_neon_inner.c b/lib/raid6/recov_neon_inner.c
new file mode 100644
index 000000000000..8cd20c9f834a
--- /dev/null
+++ b/lib/raid6/recov_neon_inner.c
@@ -0,0 +1,117 @@
+/*
+ * Copyright (C) 2012 Intel Corporation
+ * Copyright (C) 2017 Linaro Ltd. <ard.biesheuvel@linaro.org>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License
+ * as published by the Free Software Foundation; version 2
+ * of the License.
+ */
+
+#include <arm_neon.h>
+
+static const uint8x16_t x0f = {
+	0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
+	0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f, 0x0f,
+};
+
+#ifdef CONFIG_ARM
+/*
+ * AArch32 does not provide this intrinsic natively because it does not
+ * implement the underlying instruction. AArch32 only provides a 64-bit
+ * wide vtbl.8 instruction, so use that instead.
+ */
+static uint8x16_t vqtbl1q_u8(uint8x16_t a, uint8x16_t b)
+{
+	union {
+		uint8x16_t	val;
+		uint8x8x2_t	pair;
+	} __a = { a };
+
+	return vcombine_u8(vtbl2_u8(__a.pair, vget_low_u8(b)),
+			   vtbl2_u8(__a.pair, vget_high_u8(b)));
+}
+#endif
+
+void __raid6_2data_recov_neon(int bytes, uint8_t *p, uint8_t *q, uint8_t *dp,
+			      uint8_t *dq, const uint8_t *pbmul,
+			      const uint8_t *qmul)
+{
+	uint8x16_t pm0 = vld1q_u8(pbmul);
+	uint8x16_t pm1 = vld1q_u8(pbmul + 16);
+	uint8x16_t qm0 = vld1q_u8(qmul);
+	uint8x16_t qm1 = vld1q_u8(qmul + 16);
+
+	/*
+	 * while ( bytes-- ) {
+	 *	uint8_t px, qx, db;
+	 *
+	 *	px    = *p ^ *dp;
+	 *	qx    = qmul[*q ^ *dq];
+	 *	*dq++ = db = pbmul[px] ^ qx;
+	 *	*dp++ = db ^ px;
+	 *	p++; q++;
+	 * }
+	 */
+
+	while (bytes) {
+		uint8x16_t vx, vy, px, qx, db;
+
+		px = veorq_u8(vld1q_u8(p), vld1q_u8(dp));
+		vx = veorq_u8(vld1q_u8(q), vld1q_u8(dq));
+
+		vy = (uint8x16_t)vshrq_n_s16((int16x8_t)vx, 4);
+		vx = vqtbl1q_u8(qm0, vandq_u8(vx, x0f));
+		vy = vqtbl1q_u8(qm1, vandq_u8(vy, x0f));
+		qx = veorq_u8(vx, vy);
+
+		vy = (uint8x16_t)vshrq_n_s16((int16x8_t)px, 4);
+		vx = vqtbl1q_u8(pm0, vandq_u8(px, x0f));
+		vy = vqtbl1q_u8(pm1, vandq_u8(vy, x0f));
+		vx = veorq_u8(vx, vy);
+		db = veorq_u8(vx, qx);
+
+		vst1q_u8(dq, db);
+		vst1q_u8(dp, veorq_u8(db, px));
+
+		bytes -= 16;
+		p += 16;
+		q += 16;
+		dp += 16;
+		dq += 16;
+	}
+}
+
+void __raid6_datap_recov_neon(int bytes, uint8_t *p, uint8_t *q, uint8_t *dq,
+			      const uint8_t *qmul)
+{
+	uint8x16_t qm0 = vld1q_u8(qmul);
+	uint8x16_t qm1 = vld1q_u8(qmul + 16);
+
+	/*
+	 * while (bytes--) {
+	 *	*p++ ^= *dq = qmul[*q ^ *dq];
+	 *	q++; dq++;
+	 * }
+	 */
+
+	while (bytes) {
+		uint8x16_t vx, vy;
+
+		vx = veorq_u8(vld1q_u8(q), vld1q_u8(dq));
+
+		vy = (uint8x16_t)vshrq_n_s16((int16x8_t)vx, 4);
+		vx = vqtbl1q_u8(qm0, vandq_u8(vx, x0f));
+		vy = vqtbl1q_u8(qm1, vandq_u8(vy, x0f));
+		vx = veorq_u8(vx, vy);
+		vy = veorq_u8(vx, vld1q_u8(p));
+
+		vst1q_u8(dq, vx);
+		vst1q_u8(p, vy);
+
+		bytes -= 16;
+		p += 16;
+		q += 16;
+		dq += 16;
+	}
+}
diff --git a/lib/rbtree.c b/lib/rbtree.c
index 4ba2828a67c0..ba4a9d165f1b 100644
--- a/lib/rbtree.c
+++ b/lib/rbtree.c
@@ -95,22 +95,35 @@ __rb_rotate_set_parents(struct rb_node *old, struct rb_node *new,
 
 static __always_inline void
 __rb_insert(struct rb_node *node, struct rb_root *root,
+	    bool newleft, struct rb_node **leftmost,
 	    void (*augment_rotate)(struct rb_node *old, struct rb_node *new))
 {
 	struct rb_node *parent = rb_red_parent(node), *gparent, *tmp;
 
+	if (newleft)
+		*leftmost = node;
+
 	while (true) {
 		/*
-		 * Loop invariant: node is red
-		 *
-		 * If there is a black parent, we are done.
-		 * Otherwise, take some corrective action as we don't
-		 * want a red root or two consecutive red nodes.
+		 * Loop invariant: node is red.
 		 */
-		if (!parent) {
+		if (unlikely(!parent)) {
+			/*
+			 * The inserted node is root. Either this is the
+			 * first node, or we recursed at Case 1 below and
+			 * are no longer violating 4).
+			 */
 			rb_set_parent_color(node, NULL, RB_BLACK);
 			break;
-		} else if (rb_is_black(parent))
+		}
+
+		/*
+		 * If there is a black parent, we are done.
+		 * Otherwise, take some corrective action as,
+		 * per 4), we don't want a red root or two
+		 * consecutive red nodes.
+		 */
+		if(rb_is_black(parent))
 			break;
 
 		gparent = rb_red_parent(parent);
@@ -119,7 +132,7 @@ __rb_insert(struct rb_node *node, struct rb_root *root,
 		if (parent != tmp) {	/* parent == gparent->rb_left */
 			if (tmp && rb_is_red(tmp)) {
 				/*
-				 * Case 1 - color flips
+				 * Case 1 - node's uncle is red (color flips).
 				 *
 				 *       G            g
 				 *      / \          / \
@@ -142,7 +155,8 @@ __rb_insert(struct rb_node *node, struct rb_root *root,
 			tmp = parent->rb_right;
 			if (node == tmp) {
 				/*
-				 * Case 2 - left rotate at parent
+				 * Case 2 - node's uncle is black and node is
+				 * the parent's right child (left rotate at parent).
 				 *
 				 *      G             G
 				 *     / \           / \
@@ -166,7 +180,8 @@ __rb_insert(struct rb_node *node, struct rb_root *root,
 			}
 
 			/*
-			 * Case 3 - right rotate at gparent
+			 * Case 3 - node's uncle is black and node is
+			 * the parent's left child (right rotate at gparent).
 			 *
 			 *        G           P
 			 *       / \         / \
@@ -434,19 +449,38 @@ static const struct rb_augment_callbacks dummy_callbacks = {
 
 void rb_insert_color(struct rb_node *node, struct rb_root *root)
 {
-	__rb_insert(node, root, dummy_rotate);
+	__rb_insert(node, root, false, NULL, dummy_rotate);
 }
 EXPORT_SYMBOL(rb_insert_color);
 
 void rb_erase(struct rb_node *node, struct rb_root *root)
 {
 	struct rb_node *rebalance;
-	rebalance = __rb_erase_augmented(node, root, &dummy_callbacks);
+	rebalance = __rb_erase_augmented(node, root,
+					 NULL, &dummy_callbacks);
 	if (rebalance)
 		____rb_erase_color(rebalance, root, dummy_rotate);
 }
 EXPORT_SYMBOL(rb_erase);
 
+void rb_insert_color_cached(struct rb_node *node,
+			    struct rb_root_cached *root, bool leftmost)
+{
+	__rb_insert(node, &root->rb_root, leftmost,
+		    &root->rb_leftmost, dummy_rotate);
+}
+EXPORT_SYMBOL(rb_insert_color_cached);
+
+void rb_erase_cached(struct rb_node *node, struct rb_root_cached *root)
+{
+	struct rb_node *rebalance;
+	rebalance = __rb_erase_augmented(node, &root->rb_root,
+					 &root->rb_leftmost, &dummy_callbacks);
+	if (rebalance)
+		____rb_erase_color(rebalance, &root->rb_root, dummy_rotate);
+}
+EXPORT_SYMBOL(rb_erase_cached);
+
 /*
  * Augmented rbtree manipulation functions.
  *
@@ -455,9 +489,10 @@ EXPORT_SYMBOL(rb_erase);
  */
 
 void __rb_insert_augmented(struct rb_node *node, struct rb_root *root,
+			   bool newleft, struct rb_node **leftmost,
 	void (*augment_rotate)(struct rb_node *old, struct rb_node *new))
 {
-	__rb_insert(node, root, augment_rotate);
+	__rb_insert(node, root, newleft, leftmost, augment_rotate);
 }
 EXPORT_SYMBOL(__rb_insert_augmented);
 
@@ -502,7 +537,7 @@ struct rb_node *rb_next(const struct rb_node *node)
 	 * as we can.
 	 */
 	if (node->rb_right) {
-		node = node->rb_right; 
+		node = node->rb_right;
 		while (node->rb_left)
 			node=node->rb_left;
 		return (struct rb_node *)node;
@@ -534,7 +569,7 @@ struct rb_node *rb_prev(const struct rb_node *node)
 	 * as we can.
 	 */
 	if (node->rb_left) {
-		node = node->rb_left; 
+		node = node->rb_left;
 		while (node->rb_right)
 			node=node->rb_right;
 		return (struct rb_node *)node;
diff --git a/lib/rbtree_test.c b/lib/rbtree_test.c
index 8b3c9dc88262..191a238e5a9d 100644
--- a/lib/rbtree_test.c
+++ b/lib/rbtree_test.c
@@ -1,11 +1,18 @@
 #include <linux/module.h>
+#include <linux/moduleparam.h>
 #include <linux/rbtree_augmented.h>
 #include <linux/random.h>
+#include <linux/slab.h>
 #include <asm/timex.h>
 
-#define NODES       100
-#define PERF_LOOPS  100000
-#define CHECK_LOOPS 100
+#define __param(type, name, init, msg)		\
+	static type name = init;		\
+	module_param(name, type, 0444);		\
+	MODULE_PARM_DESC(name, msg);
+
+__param(int, nnodes, 100, "Number of nodes in the rb-tree");
+__param(int, perf_loops, 100000, "Number of iterations modifying the rb-tree");
+__param(int, check_loops, 100, "Number of iterations modifying and verifying the rb-tree");
 
 struct test_node {
 	u32 key;
@@ -16,14 +23,14 @@ struct test_node {
 	u32 augmented;
 };
 
-static struct rb_root root = RB_ROOT;
-static struct test_node nodes[NODES];
+static struct rb_root_cached root = RB_ROOT_CACHED;
+static struct test_node *nodes = NULL;
 
 static struct rnd_state rnd;
 
-static void insert(struct test_node *node, struct rb_root *root)
+static void insert(struct test_node *node, struct rb_root_cached *root)
 {
-	struct rb_node **new = &root->rb_node, *parent = NULL;
+	struct rb_node **new = &root->rb_root.rb_node, *parent = NULL;
 	u32 key = node->key;
 
 	while (*new) {
@@ -35,14 +42,40 @@ static void insert(struct test_node *node, struct rb_root *root)
 	}
 
 	rb_link_node(&node->rb, parent, new);
-	rb_insert_color(&node->rb, root);
+	rb_insert_color(&node->rb, &root->rb_root);
+}
+
+static void insert_cached(struct test_node *node, struct rb_root_cached *root)
+{
+	struct rb_node **new = &root->rb_root.rb_node, *parent = NULL;
+	u32 key = node->key;
+	bool leftmost = true;
+
+	while (*new) {
+		parent = *new;
+		if (key < rb_entry(parent, struct test_node, rb)->key)
+			new = &parent->rb_left;
+		else {
+			new = &parent->rb_right;
+			leftmost = false;
+		}
+	}
+
+	rb_link_node(&node->rb, parent, new);
+	rb_insert_color_cached(&node->rb, root, leftmost);
 }
 
-static inline void erase(struct test_node *node, struct rb_root *root)
+static inline void erase(struct test_node *node, struct rb_root_cached *root)
 {
-	rb_erase(&node->rb, root);
+	rb_erase(&node->rb, &root->rb_root);
 }
 
+static inline void erase_cached(struct test_node *node, struct rb_root_cached *root)
+{
+	rb_erase_cached(&node->rb, root);
+}
+
+
 static inline u32 augment_recompute(struct test_node *node)
 {
 	u32 max = node->val, child_augmented;
@@ -64,9 +97,10 @@ static inline u32 augment_recompute(struct test_node *node)
 RB_DECLARE_CALLBACKS(static, augment_callbacks, struct test_node, rb,
 		     u32, augmented, augment_recompute)
 
-static void insert_augmented(struct test_node *node, struct rb_root *root)
+static void insert_augmented(struct test_node *node,
+			     struct rb_root_cached *root)
 {
-	struct rb_node **new = &root->rb_node, *rb_parent = NULL;
+	struct rb_node **new = &root->rb_root.rb_node, *rb_parent = NULL;
 	u32 key = node->key;
 	u32 val = node->val;
 	struct test_node *parent;
@@ -84,18 +118,53 @@ static void insert_augmented(struct test_node *node, struct rb_root *root)
 
 	node->augmented = val;
 	rb_link_node(&node->rb, rb_parent, new);
-	rb_insert_augmented(&node->rb, root, &augment_callbacks);
+	rb_insert_augmented(&node->rb, &root->rb_root, &augment_callbacks);
+}
+
+static void insert_augmented_cached(struct test_node *node,
+				    struct rb_root_cached *root)
+{
+	struct rb_node **new = &root->rb_root.rb_node, *rb_parent = NULL;
+	u32 key = node->key;
+	u32 val = node->val;
+	struct test_node *parent;
+	bool leftmost = true;
+
+	while (*new) {
+		rb_parent = *new;
+		parent = rb_entry(rb_parent, struct test_node, rb);
+		if (parent->augmented < val)
+			parent->augmented = val;
+		if (key < parent->key)
+			new = &parent->rb.rb_left;
+		else {
+			new = &parent->rb.rb_right;
+			leftmost = false;
+		}
+	}
+
+	node->augmented = val;
+	rb_link_node(&node->rb, rb_parent, new);
+	rb_insert_augmented_cached(&node->rb, root,
+				   leftmost, &augment_callbacks);
+}
+
+
+static void erase_augmented(struct test_node *node, struct rb_root_cached *root)
+{
+	rb_erase_augmented(&node->rb, &root->rb_root, &augment_callbacks);
 }
 
-static void erase_augmented(struct test_node *node, struct rb_root *root)
+static void erase_augmented_cached(struct test_node *node,
+				   struct rb_root_cached *root)
 {
-	rb_erase_augmented(&node->rb, root, &augment_callbacks);
+	rb_erase_augmented_cached(&node->rb, root, &augment_callbacks);
 }
 
 static void init(void)
 {
 	int i;
-	for (i = 0; i < NODES; i++) {
+	for (i = 0; i < nnodes; i++) {
 		nodes[i].key = prandom_u32_state(&rnd);
 		nodes[i].val = prandom_u32_state(&rnd);
 	}
@@ -118,7 +187,7 @@ static void check_postorder_foreach(int nr_nodes)
 {
 	struct test_node *cur, *n;
 	int count = 0;
-	rbtree_postorder_for_each_entry_safe(cur, n, &root, rb)
+	rbtree_postorder_for_each_entry_safe(cur, n, &root.rb_root, rb)
 		count++;
 
 	WARN_ON_ONCE(count != nr_nodes);
@@ -128,7 +197,7 @@ static void check_postorder(int nr_nodes)
 {
 	struct rb_node *rb;
 	int count = 0;
-	for (rb = rb_first_postorder(&root); rb; rb = rb_next_postorder(rb))
+	for (rb = rb_first_postorder(&root.rb_root); rb; rb = rb_next_postorder(rb))
 		count++;
 
 	WARN_ON_ONCE(count != nr_nodes);
@@ -140,7 +209,7 @@ static void check(int nr_nodes)
 	int count = 0, blacks = 0;
 	u32 prev_key = 0;
 
-	for (rb = rb_first(&root); rb; rb = rb_next(rb)) {
+	for (rb = rb_first(&root.rb_root); rb; rb = rb_next(rb)) {
 		struct test_node *node = rb_entry(rb, struct test_node, rb);
 		WARN_ON_ONCE(node->key < prev_key);
 		WARN_ON_ONCE(is_red(rb) &&
@@ -155,7 +224,7 @@ static void check(int nr_nodes)
 	}
 
 	WARN_ON_ONCE(count != nr_nodes);
-	WARN_ON_ONCE(count < (1 << black_path_count(rb_last(&root))) - 1);
+	WARN_ON_ONCE(count < (1 << black_path_count(rb_last(&root.rb_root))) - 1);
 
 	check_postorder(nr_nodes);
 	check_postorder_foreach(nr_nodes);
@@ -166,7 +235,7 @@ static void check_augmented(int nr_nodes)
 	struct rb_node *rb;
 
 	check(nr_nodes);
-	for (rb = rb_first(&root); rb; rb = rb_next(rb)) {
+	for (rb = rb_first(&root.rb_root); rb; rb = rb_next(rb)) {
 		struct test_node *node = rb_entry(rb, struct test_node, rb);
 		WARN_ON_ONCE(node->augmented != augment_recompute(node));
 	}
@@ -176,6 +245,11 @@ static int __init rbtree_test_init(void)
 {
 	int i, j;
 	cycles_t time1, time2, time;
+	struct rb_node *node;
+
+	nodes = kmalloc(nnodes * sizeof(*nodes), GFP_KERNEL);
+	if (!nodes)
+		return -ENOMEM;
 
 	printk(KERN_ALERT "rbtree testing");
 
@@ -184,27 +258,88 @@ static int __init rbtree_test_init(void)
 
 	time1 = get_cycles();
 
-	for (i = 0; i < PERF_LOOPS; i++) {
-		for (j = 0; j < NODES; j++)
+	for (i = 0; i < perf_loops; i++) {
+		for (j = 0; j < nnodes; j++)
 			insert(nodes + j, &root);
-		for (j = 0; j < NODES; j++)
+		for (j = 0; j < nnodes; j++)
 			erase(nodes + j, &root);
 	}
 
 	time2 = get_cycles();
 	time = time2 - time1;
 
-	time = div_u64(time, PERF_LOOPS);
-	printk(" -> %llu cycles\n", (unsigned long long)time);
+	time = div_u64(time, perf_loops);
+	printk(" -> test 1 (latency of nnodes insert+delete): %llu cycles\n",
+	       (unsigned long long)time);
+
+	time1 = get_cycles();
+
+	for (i = 0; i < perf_loops; i++) {
+		for (j = 0; j < nnodes; j++)
+			insert_cached(nodes + j, &root);
+		for (j = 0; j < nnodes; j++)
+			erase_cached(nodes + j, &root);
+	}
+
+	time2 = get_cycles();
+	time = time2 - time1;
+
+	time = div_u64(time, perf_loops);
+	printk(" -> test 2 (latency of nnodes cached insert+delete): %llu cycles\n",
+	       (unsigned long long)time);
+
+	for (i = 0; i < nnodes; i++)
+		insert(nodes + i, &root);
+
+	time1 = get_cycles();
+
+	for (i = 0; i < perf_loops; i++) {
+		for (node = rb_first(&root.rb_root); node; node = rb_next(node))
+			;
+	}
+
+	time2 = get_cycles();
+	time = time2 - time1;
+
+	time = div_u64(time, perf_loops);
+	printk(" -> test 3 (latency of inorder traversal): %llu cycles\n",
+	       (unsigned long long)time);
+
+	time1 = get_cycles();
+
+	for (i = 0; i < perf_loops; i++)
+		node = rb_first(&root.rb_root);
+
+	time2 = get_cycles();
+	time = time2 - time1;
+
+	time = div_u64(time, perf_loops);
+	printk(" -> test 4 (latency to fetch first node)\n");
+	printk("        non-cached: %llu cycles\n", (unsigned long long)time);
+
+	time1 = get_cycles();
+
+	for (i = 0; i < perf_loops; i++)
+		node = rb_first_cached(&root);
+
+	time2 = get_cycles();
+	time = time2 - time1;
+
+	time = div_u64(time, perf_loops);
+	printk("        cached: %llu cycles\n", (unsigned long long)time);
 
-	for (i = 0; i < CHECK_LOOPS; i++) {
+	for (i = 0; i < nnodes; i++)
+		erase(nodes + i, &root);
+
+	/* run checks */
+	for (i = 0; i < check_loops; i++) {
 		init();
-		for (j = 0; j < NODES; j++) {
+		for (j = 0; j < nnodes; j++) {
 			check(j);
 			insert(nodes + j, &root);
 		}
-		for (j = 0; j < NODES; j++) {
-			check(NODES - j);
+		for (j = 0; j < nnodes; j++) {
+			check(nnodes - j);
 			erase(nodes + j, &root);
 		}
 		check(0);
@@ -216,32 +351,49 @@ static int __init rbtree_test_init(void)
 
 	time1 = get_cycles();
 
-	for (i = 0; i < PERF_LOOPS; i++) {
-		for (j = 0; j < NODES; j++)
+	for (i = 0; i < perf_loops; i++) {
+		for (j = 0; j < nnodes; j++)
 			insert_augmented(nodes + j, &root);
-		for (j = 0; j < NODES; j++)
+		for (j = 0; j < nnodes; j++)
 			erase_augmented(nodes + j, &root);
 	}
 
 	time2 = get_cycles();
 	time = time2 - time1;
 
-	time = div_u64(time, PERF_LOOPS);
-	printk(" -> %llu cycles\n", (unsigned long long)time);
+	time = div_u64(time, perf_loops);
+	printk(" -> test 1 (latency of nnodes insert+delete): %llu cycles\n", (unsigned long long)time);
+
+	time1 = get_cycles();
+
+	for (i = 0; i < perf_loops; i++) {
+		for (j = 0; j < nnodes; j++)
+			insert_augmented_cached(nodes + j, &root);
+		for (j = 0; j < nnodes; j++)
+			erase_augmented_cached(nodes + j, &root);
+	}
+
+	time2 = get_cycles();
+	time = time2 - time1;
+
+	time = div_u64(time, perf_loops);
+	printk(" -> test 2 (latency of nnodes cached insert+delete): %llu cycles\n", (unsigned long long)time);
 
-	for (i = 0; i < CHECK_LOOPS; i++) {
+	for (i = 0; i < check_loops; i++) {
 		init();
-		for (j = 0; j < NODES; j++) {
+		for (j = 0; j < nnodes; j++) {
 			check_augmented(j);
 			insert_augmented(nodes + j, &root);
 		}
-		for (j = 0; j < NODES; j++) {
-			check_augmented(NODES - j);
+		for (j = 0; j < nnodes; j++) {
+			check_augmented(nnodes - j);
 			erase_augmented(nodes + j, &root);
 		}
 		check_augmented(0);
 	}
 
+	kfree(nodes);
+
 	return -EAGAIN; /* Fail will directly unload the module */
 }
 
diff --git a/lib/rhashtable.c b/lib/rhashtable.c
index 707ca5d677c6..ddd7dde87c3c 100644
--- a/lib/rhashtable.c
+++ b/lib/rhashtable.c
@@ -735,9 +735,9 @@ EXPORT_SYMBOL_GPL(rhashtable_walk_exit);
  * rhashtable_walk_start - Start a hash table walk
  * @iter:	Hash table iterator
  *
- * Start a hash table walk.  Note that we take the RCU lock in all
- * cases including when we return an error.  So you must always call
- * rhashtable_walk_stop to clean up.
+ * Start a hash table walk at the current iterator position.  Note that we take
+ * the RCU lock in all cases including when we return an error.  So you must
+ * always call rhashtable_walk_stop to clean up.
  *
  * Returns zero if successful.
  *
@@ -846,7 +846,8 @@ EXPORT_SYMBOL_GPL(rhashtable_walk_next);
  * rhashtable_walk_stop - Finish a hash table walk
  * @iter:	Hash table iterator
  *
- * Finish a hash table walk.
+ * Finish a hash table walk.  Does not reset the iterator to the start of the
+ * hash table.
  */
 void rhashtable_walk_stop(struct rhashtable_iter *iter)
 	__releases(RCU)
diff --git a/lib/string.c b/lib/string.c
index ebbb99c775bd..9921dc202db4 100644
--- a/lib/string.c
+++ b/lib/string.c
@@ -723,6 +723,72 @@ void memzero_explicit(void *s, size_t count)
 }
 EXPORT_SYMBOL(memzero_explicit);
 
+#ifndef __HAVE_ARCH_MEMSET16
+/**
+ * memset16() - Fill a memory area with a uint16_t
+ * @s: Pointer to the start of the area.
+ * @v: The value to fill the area with
+ * @count: The number of values to store
+ *
+ * Differs from memset() in that it fills with a uint16_t instead
+ * of a byte.  Remember that @count is the number of uint16_ts to
+ * store, not the number of bytes.
+ */
+void *memset16(uint16_t *s, uint16_t v, size_t count)
+{
+	uint16_t *xs = s;
+
+	while (count--)
+		*xs++ = v;
+	return s;
+}
+EXPORT_SYMBOL(memset16);
+#endif
+
+#ifndef __HAVE_ARCH_MEMSET32
+/**
+ * memset32() - Fill a memory area with a uint32_t
+ * @s: Pointer to the start of the area.
+ * @v: The value to fill the area with
+ * @count: The number of values to store
+ *
+ * Differs from memset() in that it fills with a uint32_t instead
+ * of a byte.  Remember that @count is the number of uint32_ts to
+ * store, not the number of bytes.
+ */
+void *memset32(uint32_t *s, uint32_t v, size_t count)
+{
+	uint32_t *xs = s;
+
+	while (count--)
+		*xs++ = v;
+	return s;
+}
+EXPORT_SYMBOL(memset32);
+#endif
+
+#ifndef __HAVE_ARCH_MEMSET64
+/**
+ * memset64() - Fill a memory area with a uint64_t
+ * @s: Pointer to the start of the area.
+ * @v: The value to fill the area with
+ * @count: The number of values to store
+ *
+ * Differs from memset() in that it fills with a uint64_t instead
+ * of a byte.  Remember that @count is the number of uint64_ts to
+ * store, not the number of bytes.
+ */
+void *memset64(uint64_t *s, uint64_t v, size_t count)
+{
+	uint64_t *xs = s;
+
+	while (count--)
+		*xs++ = v;
+	return s;
+}
+EXPORT_SYMBOL(memset64);
+#endif
+
 #ifndef __HAVE_ARCH_MEMCPY
 /**
  * memcpy - Copy one area of memory to another
@@ -985,3 +1051,144 @@ void fortify_panic(const char *name)
 	BUG();
 }
 EXPORT_SYMBOL(fortify_panic);
+
+#ifdef CONFIG_STRING_SELFTEST
+#include <linux/slab.h>
+#include <linux/module.h>
+
+static __init int memset16_selftest(void)
+{
+	unsigned i, j, k;
+	u16 v, *p;
+
+	p = kmalloc(256 * 2 * 2, GFP_KERNEL);
+	if (!p)
+		return -1;
+
+	for (i = 0; i < 256; i++) {
+		for (j = 0; j < 256; j++) {
+			memset(p, 0xa1, 256 * 2 * sizeof(v));
+			memset16(p + i, 0xb1b2, j);
+			for (k = 0; k < 512; k++) {
+				v = p[k];
+				if (k < i) {
+					if (v != 0xa1a1)
+						goto fail;
+				} else if (k < i + j) {
+					if (v != 0xb1b2)
+						goto fail;
+				} else {
+					if (v != 0xa1a1)
+						goto fail;
+				}
+			}
+		}
+	}
+
+fail:
+	kfree(p);
+	if (i < 256)
+		return (i << 24) | (j << 16) | k;
+	return 0;
+}
+
+static __init int memset32_selftest(void)
+{
+	unsigned i, j, k;
+	u32 v, *p;
+
+	p = kmalloc(256 * 2 * 4, GFP_KERNEL);
+	if (!p)
+		return -1;
+
+	for (i = 0; i < 256; i++) {
+		for (j = 0; j < 256; j++) {
+			memset(p, 0xa1, 256 * 2 * sizeof(v));
+			memset32(p + i, 0xb1b2b3b4, j);
+			for (k = 0; k < 512; k++) {
+				v = p[k];
+				if (k < i) {
+					if (v != 0xa1a1a1a1)
+						goto fail;
+				} else if (k < i + j) {
+					if (v != 0xb1b2b3b4)
+						goto fail;
+				} else {
+					if (v != 0xa1a1a1a1)
+						goto fail;
+				}
+			}
+		}
+	}
+
+fail:
+	kfree(p);
+	if (i < 256)
+		return (i << 24) | (j << 16) | k;
+	return 0;
+}
+
+static __init int memset64_selftest(void)
+{
+	unsigned i, j, k;
+	u64 v, *p;
+
+	p = kmalloc(256 * 2 * 8, GFP_KERNEL);
+	if (!p)
+		return -1;
+
+	for (i = 0; i < 256; i++) {
+		for (j = 0; j < 256; j++) {
+			memset(p, 0xa1, 256 * 2 * sizeof(v));
+			memset64(p + i, 0xb1b2b3b4b5b6b7b8ULL, j);
+			for (k = 0; k < 512; k++) {
+				v = p[k];
+				if (k < i) {
+					if (v != 0xa1a1a1a1a1a1a1a1ULL)
+						goto fail;
+				} else if (k < i + j) {
+					if (v != 0xb1b2b3b4b5b6b7b8ULL)
+						goto fail;
+				} else {
+					if (v != 0xa1a1a1a1a1a1a1a1ULL)
+						goto fail;
+				}
+			}
+		}
+	}
+
+fail:
+	kfree(p);
+	if (i < 256)
+		return (i << 24) | (j << 16) | k;
+	return 0;
+}
+
+static __init int string_selftest_init(void)
+{
+	int test, subtest;
+
+	test = 1;
+	subtest = memset16_selftest();
+	if (subtest)
+		goto fail;
+
+	test = 2;
+	subtest = memset32_selftest();
+	if (subtest)
+		goto fail;
+
+	test = 3;
+	subtest = memset64_selftest();
+	if (subtest)
+		goto fail;
+
+	pr_info("String selftests succeeded\n");
+	return 0;
+fail:
+	pr_crit("String selftest failure %d.%08x\n", test, subtest);
+	return 0;
+}
+
+module_init(string_selftest_init);
+#endif	/* CONFIG_STRING_SELFTEST */
diff --git a/lib/string_helpers.c b/lib/string_helpers.c
index ecaac2c0526f..29c490e5d478 100644
--- a/lib/string_helpers.c
+++ b/lib/string_helpers.c
@@ -576,7 +576,7 @@ char *kstrdup_quotable_cmdline(struct task_struct *task, gfp_t gfp)
 	char *buffer, *quoted;
 	int i, res;
 
-	buffer = kmalloc(PAGE_SIZE, GFP_TEMPORARY);
+	buffer = kmalloc(PAGE_SIZE, GFP_KERNEL);
 	if (!buffer)
 		return NULL;
 
@@ -612,7 +612,7 @@ char *kstrdup_quotable_file(struct file *file, gfp_t gfp)
 		return kstrdup("<unknown>", gfp);
 
 	/* We add 11 spaces for ' (deleted)' to be appended */
-	temp = kmalloc(PATH_MAX + 11, GFP_TEMPORARY);
+	temp = kmalloc(PATH_MAX + 11, GFP_KERNEL);
 	if (!temp)
 		return kstrdup("<no_memory>", gfp);
 
diff --git a/lib/swiotlb.c b/lib/swiotlb.c
index a8d74a733a38..8c6c83ef57a4 100644
--- a/lib/swiotlb.c
+++ b/lib/swiotlb.c
@@ -30,6 +30,7 @@
 #include <linux/highmem.h>
 #include <linux/gfp.h>
 #include <linux/scatterlist.h>
+#include <linux/mem_encrypt.h>
 
 #include <asm/io.h>
 #include <asm/dma.h>
@@ -155,6 +156,15 @@ unsigned long swiotlb_size_or_default(void)
 	return size ? size : (IO_TLB_DEFAULT_SIZE);
 }
 
+void __weak swiotlb_set_mem_attributes(void *vaddr, unsigned long size) { }
+
+/* For swiotlb, clear memory encryption mask from dma addresses */
+static dma_addr_t swiotlb_phys_to_dma(struct device *hwdev,
+				      phys_addr_t address)
+{
+	return __sme_clr(phys_to_dma(hwdev, address));
+}
+
 /* Note that this doesn't work with highmem page */
 static dma_addr_t swiotlb_virt_to_bus(struct device *hwdev,
 				      volatile void *address)
@@ -183,6 +193,31 @@ void swiotlb_print_info(void)
 	       bytes >> 20, vstart, vend - 1);
 }
 
+/*
+ * Early SWIOTLB allocation may be too early to allow an architecture to
+ * perform the desired operations.  This function allows the architecture to
+ * call SWIOTLB when the operations are possible.  It needs to be called
+ * before the SWIOTLB memory is used.
+ */
+void __init swiotlb_update_mem_attributes(void)
+{
+	void *vaddr;
+	unsigned long bytes;
+
+	if (no_iotlb_memory || late_alloc)
+		return;
+
+	vaddr = phys_to_virt(io_tlb_start);
+	bytes = PAGE_ALIGN(io_tlb_nslabs << IO_TLB_SHIFT);
+	swiotlb_set_mem_attributes(vaddr, bytes);
+	memset(vaddr, 0, bytes);
+
+	vaddr = phys_to_virt(io_tlb_overflow_buffer);
+	bytes = PAGE_ALIGN(io_tlb_overflow);
+	swiotlb_set_mem_attributes(vaddr, bytes);
+	memset(vaddr, 0, bytes);
+}
+
 int __init swiotlb_init_with_tbl(char *tlb, unsigned long nslabs, int verbose)
 {
 	void *v_overflow_buffer;
@@ -320,6 +355,7 @@ swiotlb_late_init_with_tbl(char *tlb, unsigned long nslabs)
 	io_tlb_start = virt_to_phys(tlb);
 	io_tlb_end = io_tlb_start + bytes;
 
+	swiotlb_set_mem_attributes(tlb, bytes);
 	memset(tlb, 0, bytes);
 
 	/*
@@ -330,6 +366,8 @@ swiotlb_late_init_with_tbl(char *tlb, unsigned long nslabs)
 	if (!v_overflow_buffer)
 		goto cleanup2;
 
+	swiotlb_set_mem_attributes(v_overflow_buffer, io_tlb_overflow);
+	memset(v_overflow_buffer, 0, io_tlb_overflow);
 	io_tlb_overflow_buffer = virt_to_phys(v_overflow_buffer);
 
 	/*
@@ -469,6 +507,9 @@ phys_addr_t swiotlb_tbl_map_single(struct device *hwdev,
 	if (no_iotlb_memory)
 		panic("Can not allocate SWIOTLB buffer earlier and can't now provide you with the DMA bounce buffer");
 
+	if (sme_active())
+		pr_warn_once("SME is active and system is using DMA bounce buffers\n");
+
 	mask = dma_get_seg_boundary(hwdev);
 
 	tbl_dma_addr &= mask;
@@ -581,7 +622,7 @@ map_single(struct device *hwdev, phys_addr_t phys, size_t size,
 		return SWIOTLB_MAP_ERROR;
 	}
 
-	start_dma_addr = phys_to_dma(hwdev, io_tlb_start);
+	start_dma_addr = swiotlb_phys_to_dma(hwdev, io_tlb_start);
 	return swiotlb_tbl_map_single(hwdev, start_dma_addr, phys, size,
 				      dir, attrs);
 }
@@ -702,7 +743,7 @@ swiotlb_alloc_coherent(struct device *hwdev, size_t size,
 			goto err_warn;
 
 		ret = phys_to_virt(paddr);
-		dev_addr = phys_to_dma(hwdev, paddr);
+		dev_addr = swiotlb_phys_to_dma(hwdev, paddr);
 
 		/* Confirm address can be DMA'd by device */
 		if (dev_addr + size - 1 > dma_mask) {
@@ -812,10 +853,10 @@ dma_addr_t swiotlb_map_page(struct device *dev, struct page *page,
 	map = map_single(dev, phys, size, dir, attrs);
 	if (map == SWIOTLB_MAP_ERROR) {
 		swiotlb_full(dev, size, dir, 1);
-		return phys_to_dma(dev, io_tlb_overflow_buffer);
+		return swiotlb_phys_to_dma(dev, io_tlb_overflow_buffer);
 	}
 
-	dev_addr = phys_to_dma(dev, map);
+	dev_addr = swiotlb_phys_to_dma(dev, map);
 
 	/* Ensure that the address returned is DMA'ble */
 	if (dma_capable(dev, dev_addr, size))
@@ -824,7 +865,7 @@ dma_addr_t swiotlb_map_page(struct device *dev, struct page *page,
 	attrs |= DMA_ATTR_SKIP_CPU_SYNC;
 	swiotlb_tbl_unmap_single(dev, map, size, dir, attrs);
 
-	return phys_to_dma(dev, io_tlb_overflow_buffer);
+	return swiotlb_phys_to_dma(dev, io_tlb_overflow_buffer);
 }
 EXPORT_SYMBOL_GPL(swiotlb_map_page);
 
@@ -958,7 +999,7 @@ swiotlb_map_sg_attrs(struct device *hwdev, struct scatterlist *sgl, int nelems,
 				sg_dma_len(sgl) = 0;
 				return 0;
 			}
-			sg->dma_address = phys_to_dma(hwdev, map);
+			sg->dma_address = swiotlb_phys_to_dma(hwdev, map);
 		} else
 			sg->dma_address = dev_addr;
 		sg_dma_len(sg) = sg->length;
@@ -1026,7 +1067,7 @@ EXPORT_SYMBOL(swiotlb_sync_sg_for_device);
 int
 swiotlb_dma_mapping_error(struct device *hwdev, dma_addr_t dma_addr)
 {
-	return (dma_addr == phys_to_dma(hwdev, io_tlb_overflow_buffer));
+	return (dma_addr == swiotlb_phys_to_dma(hwdev, io_tlb_overflow_buffer));
 }
 EXPORT_SYMBOL(swiotlb_dma_mapping_error);
 
@@ -1039,6 +1080,6 @@ EXPORT_SYMBOL(swiotlb_dma_mapping_error);
 int
 swiotlb_dma_supported(struct device *hwdev, u64 mask)
 {
-	return phys_to_dma(hwdev, io_tlb_end - 1) <= mask;
+	return swiotlb_phys_to_dma(hwdev, io_tlb_end - 1) <= mask;
 }
 EXPORT_SYMBOL(swiotlb_dma_supported);
diff --git a/lib/test_bitmap.c b/lib/test_bitmap.c
index 2526a2975c51..aa1f2669bdd5 100644
--- a/lib/test_bitmap.c
+++ b/lib/test_bitmap.c
@@ -165,6 +165,96 @@ static void __init test_zero_fill_copy(void)
 	expect_eq_pbl("128-1023", bmap2, 1024);
 }
 
+#define PARSE_TIME 0x1
+
+struct test_bitmap_parselist{
+	const int errno;
+	const char *in;
+	const unsigned long *expected;
+	const int nbits;
+	const int flags;
+};
+
+static const unsigned long exp[] __initconst = {
+	BITMAP_FROM_U64(1),
+	BITMAP_FROM_U64(2),
+	BITMAP_FROM_U64(0x0000ffff),
+	BITMAP_FROM_U64(0xffff0000),
+	BITMAP_FROM_U64(0x55555555),
+	BITMAP_FROM_U64(0xaaaaaaaa),
+	BITMAP_FROM_U64(0x11111111),
+	BITMAP_FROM_U64(0x22222222),
+	BITMAP_FROM_U64(0xffffffff),
+	BITMAP_FROM_U64(0xfffffffe),
+	BITMAP_FROM_U64(0x3333333311111111ULL),
+	BITMAP_FROM_U64(0xffffffff77777777ULL)
+};
+
+static const unsigned long exp2[] __initconst = {
+	BITMAP_FROM_U64(0x3333333311111111ULL),
+	BITMAP_FROM_U64(0xffffffff77777777ULL)
+};
+
+static const struct test_bitmap_parselist parselist_tests[] __initconst = {
+#define step (sizeof(u64) / sizeof(unsigned long))
+
+	{0, "0",			&exp[0], 8, 0},
+	{0, "1",			&exp[1 * step], 8, 0},
+	{0, "0-15",			&exp[2 * step], 32, 0},
+	{0, "16-31",			&exp[3 * step], 32, 0},
+	{0, "0-31:1/2",			&exp[4 * step], 32, 0},
+	{0, "1-31:1/2",			&exp[5 * step], 32, 0},
+	{0, "0-31:1/4",			&exp[6 * step], 32, 0},
+	{0, "1-31:1/4",			&exp[7 * step], 32, 0},
+	{0, "0-31:4/4",			&exp[8 * step], 32, 0},
+	{0, "1-31:4/4",			&exp[9 * step], 32, 0},
+	{0, "0-31:1/4,32-63:2/4",	&exp[10 * step], 64, 0},
+	{0, "0-31:3/4,32-63:4/4",	&exp[11 * step], 64, 0},
+
+	{0, "0-31:1/4,32-63:2/4,64-95:3/4,96-127:4/4",	exp2, 128, 0},
+
+	{0, "0-2047:128/256", NULL, 2048, PARSE_TIME},
+
+	{-EINVAL, "-1",	NULL, 8, 0},
+	{-EINVAL, "-0",	NULL, 8, 0},
+	{-EINVAL, "10-1", NULL, 8, 0},
+	{-EINVAL, "0-31:10/1", NULL, 8, 0},
+};
+
+static void __init test_bitmap_parselist(void)
+{
+	int i;
+	int err;
+	cycles_t cycles;
+	DECLARE_BITMAP(bmap, 2048);
+
+	for (i = 0; i < ARRAY_SIZE(parselist_tests); i++) {
+#define ptest parselist_tests[i]
+
+		cycles = get_cycles();
+		err = bitmap_parselist(ptest.in, bmap, ptest.nbits);
+		cycles = get_cycles() - cycles;
+
+		if (err != ptest.errno) {
+			pr_err("test %d: input is %s, errno is %d, expected %d\n",
+					i, ptest.in, err, ptest.errno);
+			continue;
+		}
+
+		if (!err && ptest.expected
+			 && !__bitmap_equal(bmap, ptest.expected, ptest.nbits)) {
+			pr_err("test %d: input is %s, result is 0x%lx, expected 0x%lx\n",
+					i, ptest.in, bmap[0], *ptest.expected);
+			continue;
+		}
+
+		if (ptest.flags & PARSE_TIME)
+			pr_err("test %d: input is '%s' OK, Time: %llu\n",
+					i, ptest.in,
+					(unsigned long long)cycles);
+	}
+}
+
 static void __init test_bitmap_u32_array_conversions(void)
 {
 	DECLARE_BITMAP(bmap1, 1024);
@@ -365,6 +455,7 @@ static int __init test_bitmap_init(void)
 {
 	test_zero_fill_copy();
 	test_bitmap_u32_array_conversions();
+	test_bitmap_parselist();
 	test_mem_optimisations();
 
 	if (failed_tests == 0)
diff --git a/lib/test_bpf.c b/lib/test_bpf.c
index d9d5a410955c..aa8812ae6776 100644
--- a/lib/test_bpf.c
+++ b/lib/test_bpf.c
@@ -952,6 +952,32 @@ static struct bpf_test tests[] = {
 		{ { 2, 0 }, { 3, 1 }, { 4, MAX_K } },
 	},
 	{
+		"JGE (jt 0), test 1",
+		.u.insns = {
+			BPF_STMT(BPF_LDX | BPF_LEN, 0),
+			BPF_STMT(BPF_LD | BPF_B | BPF_ABS, 2),
+			BPF_JUMP(BPF_JMP | BPF_JGE | BPF_X, 0, 0, 1),
+			BPF_STMT(BPF_RET | BPF_K, 1),
+			BPF_STMT(BPF_RET | BPF_K, MAX_K)
+		},
+		CLASSIC,
+		{ 4, 4, 4, 3, 3 },
+		{ { 2, 0 }, { 3, 1 }, { 4, 1 } },
+	},
+	{
+		"JGE (jt 0), test 2",
+		.u.insns = {
+			BPF_STMT(BPF_LDX | BPF_LEN, 0),
+			BPF_STMT(BPF_LD | BPF_B | BPF_ABS, 2),
+			BPF_JUMP(BPF_JMP | BPF_JGE | BPF_X, 0, 0, 1),
+			BPF_STMT(BPF_RET | BPF_K, 1),
+			BPF_STMT(BPF_RET | BPF_K, MAX_K)
+		},
+		CLASSIC,
+		{ 4, 4, 5, 3, 3 },
+		{ { 4, 1 }, { 5, 1 }, { 6, MAX_K } },
+	},
+	{
 		"JGE",
 		.u.insns = {
 			BPF_STMT(BPF_LDX | BPF_LEN, 0),
@@ -4492,6 +4518,35 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JSLT | BPF_K */
+	{
+		"JMP_JSLT_K: Signed jump: if (-2 < -1) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 0xfffffffffffffffeLL),
+			BPF_JMP_IMM(BPF_JSLT, R1, -1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JSLT_K: Signed jump: if (-1 < -1) return 0",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_LD_IMM64(R1, 0xffffffffffffffffLL),
+			BPF_JMP_IMM(BPF_JSLT, R1, -1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JSGT | BPF_K */
 	{
 		"JMP_JSGT_K: Signed jump: if (-1 > -2) return 1",
@@ -4521,6 +4576,73 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JSLE | BPF_K */
+	{
+		"JMP_JSLE_K: Signed jump: if (-2 <= -1) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 0xfffffffffffffffeLL),
+			BPF_JMP_IMM(BPF_JSLE, R1, -1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JSLE_K: Signed jump: if (-1 <= -1) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 0xffffffffffffffffLL),
+			BPF_JMP_IMM(BPF_JSLE, R1, -1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JSLE_K: Signed jump: value walk 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 3),
+			BPF_JMP_IMM(BPF_JSLE, R1, 0, 6),
+			BPF_ALU64_IMM(BPF_SUB, R1, 1),
+			BPF_JMP_IMM(BPF_JSLE, R1, 0, 4),
+			BPF_ALU64_IMM(BPF_SUB, R1, 1),
+			BPF_JMP_IMM(BPF_JSLE, R1, 0, 2),
+			BPF_ALU64_IMM(BPF_SUB, R1, 1),
+			BPF_JMP_IMM(BPF_JSLE, R1, 0, 1),
+			BPF_EXIT_INSN(),		/* bad exit */
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),	/* good exit */
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JSLE_K: Signed jump: value walk 2",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 3),
+			BPF_JMP_IMM(BPF_JSLE, R1, 0, 4),
+			BPF_ALU64_IMM(BPF_SUB, R1, 2),
+			BPF_JMP_IMM(BPF_JSLE, R1, 0, 2),
+			BPF_ALU64_IMM(BPF_SUB, R1, 2),
+			BPF_JMP_IMM(BPF_JSLE, R1, 0, 1),
+			BPF_EXIT_INSN(),		/* bad exit */
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),	/* good exit */
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JSGE | BPF_K */
 	{
 		"JMP_JSGE_K: Signed jump: if (-1 >= -2) return 1",
@@ -4617,6 +4739,35 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JLT | BPF_K */
+	{
+		"JMP_JLT_K: if (2 < 3) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 2),
+			BPF_JMP_IMM(BPF_JLT, R1, 3, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JGT_K: Unsigned jump: if (1 < -1) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 1),
+			BPF_JMP_IMM(BPF_JLT, R1, -1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JGE | BPF_K */
 	{
 		"JMP_JGE_K: if (3 >= 2) return 1",
@@ -4632,6 +4783,21 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JLE | BPF_K */
+	{
+		"JMP_JLE_K: if (2 <= 3) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 2),
+			BPF_JMP_IMM(BPF_JLE, R1, 3, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JGT | BPF_K jump backwards */
 	{
 		"JMP_JGT_K: if (3 > 2) return 1 (jump backwards)",
@@ -4662,6 +4828,36 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JLT | BPF_K jump backwards */
+	{
+		"JMP_JGT_K: if (2 < 3) return 1 (jump backwards)",
+		.u.insns_int = {
+			BPF_JMP_IMM(BPF_JA, 0, 0, 2), /* goto start */
+			BPF_ALU32_IMM(BPF_MOV, R0, 1), /* out: */
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 0), /* start: */
+			BPF_LD_IMM64(R1, 2), /* note: this takes 2 insns */
+			BPF_JMP_IMM(BPF_JLT, R1, 3, -6), /* goto out */
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JLE_K: if (3 <= 3) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 3),
+			BPF_JMP_IMM(BPF_JLE, R1, 3, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JNE | BPF_K */
 	{
 		"JMP_JNE_K: if (3 != 2) return 1",
@@ -4752,6 +4948,37 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JSLT | BPF_X */
+	{
+		"JMP_JSLT_X: Signed jump: if (-2 < -1) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, -1),
+			BPF_LD_IMM64(R2, -2),
+			BPF_JMP_REG(BPF_JSLT, R2, R1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JSLT_X: Signed jump: if (-1 < -1) return 0",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_LD_IMM64(R1, -1),
+			BPF_LD_IMM64(R2, -1),
+			BPF_JMP_REG(BPF_JSLT, R1, R2, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JSGE | BPF_X */
 	{
 		"JMP_JSGE_X: Signed jump: if (-1 >= -2) return 1",
@@ -4783,6 +5010,37 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JSLE | BPF_X */
+	{
+		"JMP_JSLE_X: Signed jump: if (-2 <= -1) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, -1),
+			BPF_LD_IMM64(R2, -2),
+			BPF_JMP_REG(BPF_JSLE, R2, R1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JSLE_X: Signed jump: if (-1 <= -1) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, -1),
+			BPF_LD_IMM64(R2, -1),
+			BPF_JMP_REG(BPF_JSLE, R1, R2, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JGT | BPF_X */
 	{
 		"JMP_JGT_X: if (3 > 2) return 1",
@@ -4814,6 +5072,37 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JLT | BPF_X */
+	{
+		"JMP_JLT_X: if (2 < 3) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 3),
+			BPF_LD_IMM64(R2, 2),
+			BPF_JMP_REG(BPF_JLT, R2, R1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JLT_X: Unsigned jump: if (1 < -1) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, -1),
+			BPF_LD_IMM64(R2, 1),
+			BPF_JMP_REG(BPF_JLT, R2, R1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JGE | BPF_X */
 	{
 		"JMP_JGE_X: if (3 >= 2) return 1",
@@ -4845,6 +5134,37 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	/* BPF_JMP | BPF_JLE | BPF_X */
+	{
+		"JMP_JLE_X: if (2 <= 3) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 3),
+			BPF_LD_IMM64(R2, 2),
+			BPF_JMP_REG(BPF_JLE, R2, R1, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
+	{
+		"JMP_JLE_X: if (3 <= 3) return 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 3),
+			BPF_LD_IMM64(R2, 3),
+			BPF_JMP_REG(BPF_JLE, R1, R2, 1),
+			BPF_EXIT_INSN(),
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	{
 		/* Mainly testing JIT + imm64 here. */
 		"JMP_JGE_X: ldimm64 test 1",
@@ -4890,6 +5210,50 @@ static struct bpf_test tests[] = {
 		{ },
 		{ { 0, 1 } },
 	},
+	{
+		"JMP_JLE_X: ldimm64 test 1",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 3),
+			BPF_LD_IMM64(R2, 2),
+			BPF_JMP_REG(BPF_JLE, R2, R1, 2),
+			BPF_LD_IMM64(R0, 0xffffffffffffffffULL),
+			BPF_LD_IMM64(R0, 0xeeeeeeeeeeeeeeeeULL),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 0xeeeeeeeeU } },
+	},
+	{
+		"JMP_JLE_X: ldimm64 test 2",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 0),
+			BPF_LD_IMM64(R1, 3),
+			BPF_LD_IMM64(R2, 2),
+			BPF_JMP_REG(BPF_JLE, R2, R1, 0),
+			BPF_LD_IMM64(R0, 0xffffffffffffffffULL),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 0xffffffffU } },
+	},
+	{
+		"JMP_JLE_X: ldimm64 test 3",
+		.u.insns_int = {
+			BPF_ALU32_IMM(BPF_MOV, R0, 1),
+			BPF_LD_IMM64(R1, 3),
+			BPF_LD_IMM64(R2, 2),
+			BPF_JMP_REG(BPF_JLE, R2, R1, 4),
+			BPF_LD_IMM64(R0, 0xffffffffffffffffULL),
+			BPF_LD_IMM64(R0, 0xeeeeeeeeeeeeeeeeULL),
+			BPF_EXIT_INSN(),
+		},
+		INTERNAL,
+		{ },
+		{ { 0, 1 } },
+	},
 	/* BPF_JMP | BPF_JNE | BPF_X */
 	{
 		"JMP_JNE_X: if (3 != 2) return 1",
diff --git a/lib/test_debug_virtual.c b/lib/test_debug_virtual.c
new file mode 100644
index 000000000000..b9cdeecc19dc
--- /dev/null
+++ b/lib/test_debug_virtual.c
@@ -0,0 +1,49 @@
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/export.h>
+#include <linux/mm.h>
+#include <linux/vmalloc.h>
+#include <linux/slab.h>
+#include <linux/sizes.h>
+
+#include <asm/page.h>
+#ifdef CONFIG_MIPS
+#include <asm/bootinfo.h>
+#endif
+
+struct foo {
+	unsigned int bar;
+};
+
+struct foo *foo;
+
+static int __init test_debug_virtual_init(void)
+{
+	phys_addr_t pa;
+	void *va;
+
+	va = (void *)VMALLOC_START;
+	pa = virt_to_phys(va);
+
+	pr_info("PA: %pa for VA: 0x%lx\n", &pa, (unsigned long)va);
+
+	foo = kzalloc(sizeof(*foo), GFP_KERNEL);
+	if (!foo)
+		return -ENOMEM;
+
+	pa = virt_to_phys(foo);
+	va = foo;
+	pr_info("PA: %pa for VA: 0x%lx\n", &pa, (unsigned long)va);
+
+	return 0;
+}
+module_init(test_debug_virtual_init);
+
+static void __exit test_debug_virtual_exit(void)
+{
+	kfree(foo);
+}
+module_exit(test_debug_virtual_exit);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Test module for CONFIG_DEBUG_VIRTUAL");
diff --git a/lib/test_firmware.c b/lib/test_firmware.c
index 09371b0a9baf..64a4c76cba2b 100644
--- a/lib/test_firmware.c
+++ b/lib/test_firmware.c
@@ -19,10 +19,85 @@
 #include <linux/miscdevice.h>
 #include <linux/slab.h>
 #include <linux/uaccess.h>
+#include <linux/delay.h>
+#include <linux/kthread.h>
+
+#define TEST_FIRMWARE_NAME	"test-firmware.bin"
+#define TEST_FIRMWARE_NUM_REQS	4
 
 static DEFINE_MUTEX(test_fw_mutex);
 static const struct firmware *test_firmware;
 
+struct test_batched_req {
+	u8 idx;
+	int rc;
+	bool sent;
+	const struct firmware *fw;
+	const char *name;
+	struct completion completion;
+	struct task_struct *task;
+	struct device *dev;
+};
+
+/**
+ * test_config - represents configuration for the test for different triggers
+ *
+ * @name: the name of the firmware file to look for
+ * @sync_direct: when the sync trigger is used if this is true
+ *	request_firmware_direct() will be used instead.
+ * @send_uevent: whether or not to send a uevent for async requests
+ * @num_requests: number of requests to try per test case. This is trigger
+ *	specific.
+ * @reqs: stores all requests information
+ * @read_fw_idx: index of thread from which we want to read firmware results
+ *	from through the read_fw trigger.
+ * @test_result: a test may use this to collect the result from the call
+ *	of the request_firmware*() calls used in their tests. In order of
+ *	priority we always keep first any setup error. If no setup errors were
+ *	found then we move on to the first error encountered while running the
+ *	API. Note that for async calls this typically will be a successful
+ *	result (0) unless of course you've used bogus parameters, or the system
+ *	is out of memory.  In the async case the callback is expected to do a
+ *	bit more homework to figure out what happened, unfortunately the only
+ *	information passed today on error is the fact that no firmware was
+ *	found so we can only assume -ENOENT on async calls if the firmware is
+ *	NULL.
+ *
+ *	Errors you can expect:
+ *
+ *	API specific:
+ *
+ *	0:		success for sync, for async it means request was sent
+ *	-EINVAL:	invalid parameters or request
+ *	-ENOENT:	files not found
+ *
+ *	System environment:
+ *
+ *	-ENOMEM:	memory pressure on system
+ *	-ENODEV:	out of number of devices to test
+ *	-EINVAL:	an unexpected error has occurred
+ * @req_firmware: if @sync_direct is true this is set to
+ *	request_firmware_direct(), otherwise request_firmware()
+ */
+struct test_config {
+	char *name;
+	bool sync_direct;
+	bool send_uevent;
+	u8 num_requests;
+	u8 read_fw_idx;
+
+	/*
+	 * These below don't belong her but we'll move them once we create
+	 * a struct fw_test_device and stuff the misc_dev under there later.
+	 */
+	struct test_batched_req *reqs;
+	int test_result;
+	int (*req_firmware)(const struct firmware **fw, const char *name,
+			    struct device *device);
+};
+
+struct test_config *test_fw_config;
+
 static ssize_t test_fw_misc_read(struct file *f, char __user *buf,
 				 size_t size, loff_t *offset)
 {
@@ -42,6 +117,338 @@ static const struct file_operations test_fw_fops = {
 	.read           = test_fw_misc_read,
 };
 
+static void __test_release_all_firmware(void)
+{
+	struct test_batched_req *req;
+	u8 i;
+
+	if (!test_fw_config->reqs)
+		return;
+
+	for (i = 0; i < test_fw_config->num_requests; i++) {
+		req = &test_fw_config->reqs[i];
+		if (req->fw)
+			release_firmware(req->fw);
+	}
+
+	vfree(test_fw_config->reqs);
+	test_fw_config->reqs = NULL;
+}
+
+static void test_release_all_firmware(void)
+{
+	mutex_lock(&test_fw_mutex);
+	__test_release_all_firmware();
+	mutex_unlock(&test_fw_mutex);
+}
+
+
+static void __test_firmware_config_free(void)
+{
+	__test_release_all_firmware();
+	kfree_const(test_fw_config->name);
+	test_fw_config->name = NULL;
+}
+
+/*
+ * XXX: move to kstrncpy() once merged.
+ *
+ * Users should use kfree_const() when freeing these.
+ */
+static int __kstrncpy(char **dst, const char *name, size_t count, gfp_t gfp)
+{
+	*dst = kstrndup(name, count, gfp);
+	if (!*dst)
+		return -ENOSPC;
+	return count;
+}
+
+static int __test_firmware_config_init(void)
+{
+	int ret;
+
+	ret = __kstrncpy(&test_fw_config->name, TEST_FIRMWARE_NAME,
+			 strlen(TEST_FIRMWARE_NAME), GFP_KERNEL);
+	if (ret < 0)
+		goto out;
+
+	test_fw_config->num_requests = TEST_FIRMWARE_NUM_REQS;
+	test_fw_config->send_uevent = true;
+	test_fw_config->sync_direct = false;
+	test_fw_config->req_firmware = request_firmware;
+	test_fw_config->test_result = 0;
+	test_fw_config->reqs = NULL;
+
+	return 0;
+
+out:
+	__test_firmware_config_free();
+	return ret;
+}
+
+static ssize_t reset_store(struct device *dev,
+			   struct device_attribute *attr,
+			   const char *buf, size_t count)
+{
+	int ret;
+
+	mutex_lock(&test_fw_mutex);
+
+	__test_firmware_config_free();
+
+	ret = __test_firmware_config_init();
+	if (ret < 0) {
+		ret = -ENOMEM;
+		pr_err("could not alloc settings for config trigger: %d\n",
+		       ret);
+		goto out;
+	}
+
+	pr_info("reset\n");
+	ret = count;
+
+out:
+	mutex_unlock(&test_fw_mutex);
+
+	return ret;
+}
+static DEVICE_ATTR_WO(reset);
+
+static ssize_t config_show(struct device *dev,
+			   struct device_attribute *attr,
+			   char *buf)
+{
+	int len = 0;
+
+	mutex_lock(&test_fw_mutex);
+
+	len += snprintf(buf, PAGE_SIZE,
+			"Custom trigger configuration for: %s\n",
+			dev_name(dev));
+
+	if (test_fw_config->name)
+		len += snprintf(buf+len, PAGE_SIZE,
+				"name:\t%s\n",
+				test_fw_config->name);
+	else
+		len += snprintf(buf+len, PAGE_SIZE,
+				"name:\tEMTPY\n");
+
+	len += snprintf(buf+len, PAGE_SIZE,
+			"num_requests:\t%u\n", test_fw_config->num_requests);
+
+	len += snprintf(buf+len, PAGE_SIZE,
+			"send_uevent:\t\t%s\n",
+			test_fw_config->send_uevent ?
+			"FW_ACTION_HOTPLUG" :
+			"FW_ACTION_NOHOTPLUG");
+	len += snprintf(buf+len, PAGE_SIZE,
+			"sync_direct:\t\t%s\n",
+			test_fw_config->sync_direct ? "true" : "false");
+	len += snprintf(buf+len, PAGE_SIZE,
+			"read_fw_idx:\t%u\n", test_fw_config->read_fw_idx);
+
+	mutex_unlock(&test_fw_mutex);
+
+	return len;
+}
+static DEVICE_ATTR_RO(config);
+
+static ssize_t config_name_store(struct device *dev,
+				 struct device_attribute *attr,
+				 const char *buf, size_t count)
+{
+	int ret;
+
+	mutex_lock(&test_fw_mutex);
+	kfree_const(test_fw_config->name);
+	ret = __kstrncpy(&test_fw_config->name, buf, count, GFP_KERNEL);
+	mutex_unlock(&test_fw_mutex);
+
+	return ret;
+}
+
+/*
+ * As per sysfs_kf_seq_show() the buf is max PAGE_SIZE.
+ */
+static ssize_t config_test_show_str(char *dst,
+				    char *src)
+{
+	int len;
+
+	mutex_lock(&test_fw_mutex);
+	len = snprintf(dst, PAGE_SIZE, "%s\n", src);
+	mutex_unlock(&test_fw_mutex);
+
+	return len;
+}
+
+static int test_dev_config_update_bool(const char *buf, size_t size,
+				       bool *cfg)
+{
+	int ret;
+
+	mutex_lock(&test_fw_mutex);
+	if (strtobool(buf, cfg) < 0)
+		ret = -EINVAL;
+	else
+		ret = size;
+	mutex_unlock(&test_fw_mutex);
+
+	return ret;
+}
+
+static ssize_t
+test_dev_config_show_bool(char *buf,
+			  bool config)
+{
+	bool val;
+
+	mutex_lock(&test_fw_mutex);
+	val = config;
+	mutex_unlock(&test_fw_mutex);
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", val);
+}
+
+static ssize_t test_dev_config_show_int(char *buf, int cfg)
+{
+	int val;
+
+	mutex_lock(&test_fw_mutex);
+	val = cfg;
+	mutex_unlock(&test_fw_mutex);
+
+	return snprintf(buf, PAGE_SIZE, "%d\n", val);
+}
+
+static int test_dev_config_update_u8(const char *buf, size_t size, u8 *cfg)
+{
+	int ret;
+	long new;
+
+	ret = kstrtol(buf, 10, &new);
+	if (ret)
+		return ret;
+
+	if (new > U8_MAX)
+		return -EINVAL;
+
+	mutex_lock(&test_fw_mutex);
+	*(u8 *)cfg = new;
+	mutex_unlock(&test_fw_mutex);
+
+	/* Always return full write size even if we didn't consume all */
+	return size;
+}
+
+static ssize_t test_dev_config_show_u8(char *buf, u8 cfg)
+{
+	u8 val;
+
+	mutex_lock(&test_fw_mutex);
+	val = cfg;
+	mutex_unlock(&test_fw_mutex);
+
+	return snprintf(buf, PAGE_SIZE, "%u\n", val);
+}
+
+static ssize_t config_name_show(struct device *dev,
+				struct device_attribute *attr,
+				char *buf)
+{
+	return config_test_show_str(buf, test_fw_config->name);
+}
+static DEVICE_ATTR(config_name, 0644, config_name_show, config_name_store);
+
+static ssize_t config_num_requests_store(struct device *dev,
+					 struct device_attribute *attr,
+					 const char *buf, size_t count)
+{
+	int rc;
+
+	mutex_lock(&test_fw_mutex);
+	if (test_fw_config->reqs) {
+		pr_err("Must call release_all_firmware prior to changing config\n");
+		rc = -EINVAL;
+		goto out;
+	}
+	mutex_unlock(&test_fw_mutex);
+
+	rc = test_dev_config_update_u8(buf, count,
+				       &test_fw_config->num_requests);
+
+out:
+	return rc;
+}
+
+static ssize_t config_num_requests_show(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	return test_dev_config_show_u8(buf, test_fw_config->num_requests);
+}
+static DEVICE_ATTR(config_num_requests, 0644, config_num_requests_show,
+		   config_num_requests_store);
+
+static ssize_t config_sync_direct_store(struct device *dev,
+					struct device_attribute *attr,
+					const char *buf, size_t count)
+{
+	int rc = test_dev_config_update_bool(buf, count,
+					     &test_fw_config->sync_direct);
+
+	if (rc == count)
+		test_fw_config->req_firmware = test_fw_config->sync_direct ?
+				       request_firmware_direct :
+				       request_firmware;
+	return rc;
+}
+
+static ssize_t config_sync_direct_show(struct device *dev,
+				       struct device_attribute *attr,
+				       char *buf)
+{
+	return test_dev_config_show_bool(buf, test_fw_config->sync_direct);
+}
+static DEVICE_ATTR(config_sync_direct, 0644, config_sync_direct_show,
+		   config_sync_direct_store);
+
+static ssize_t config_send_uevent_store(struct device *dev,
+					struct device_attribute *attr,
+					const char *buf, size_t count)
+{
+	return test_dev_config_update_bool(buf, count,
+					   &test_fw_config->send_uevent);
+}
+
+static ssize_t config_send_uevent_show(struct device *dev,
+				       struct device_attribute *attr,
+				       char *buf)
+{
+	return test_dev_config_show_bool(buf, test_fw_config->send_uevent);
+}
+static DEVICE_ATTR(config_send_uevent, 0644, config_send_uevent_show,
+		   config_send_uevent_store);
+
+static ssize_t config_read_fw_idx_store(struct device *dev,
+					struct device_attribute *attr,
+					const char *buf, size_t count)
+{
+	return test_dev_config_update_u8(buf, count,
+					 &test_fw_config->read_fw_idx);
+}
+
+static ssize_t config_read_fw_idx_show(struct device *dev,
+				       struct device_attribute *attr,
+				       char *buf)
+{
+	return test_dev_config_show_u8(buf, test_fw_config->read_fw_idx);
+}
+static DEVICE_ATTR(config_read_fw_idx, 0644, config_read_fw_idx_show,
+		   config_read_fw_idx_store);
+
+
 static ssize_t trigger_request_store(struct device *dev,
 				     struct device_attribute *attr,
 				     const char *buf, size_t count)
@@ -170,12 +577,301 @@ out:
 }
 static DEVICE_ATTR_WO(trigger_custom_fallback);
 
+static int test_fw_run_batch_request(void *data)
+{
+	struct test_batched_req *req = data;
+
+	if (!req) {
+		test_fw_config->test_result = -EINVAL;
+		return -EINVAL;
+	}
+
+	req->rc = test_fw_config->req_firmware(&req->fw, req->name, req->dev);
+	if (req->rc) {
+		pr_info("#%u: batched sync load failed: %d\n",
+			req->idx, req->rc);
+		if (!test_fw_config->test_result)
+			test_fw_config->test_result = req->rc;
+	} else if (req->fw) {
+		req->sent = true;
+		pr_info("#%u: batched sync loaded %zu\n",
+			req->idx, req->fw->size);
+	}
+	complete(&req->completion);
+
+	req->task = NULL;
+
+	return 0;
+}
+
+/*
+ * We use a kthread as otherwise the kernel serializes all our sync requests
+ * and we would not be able to mimic batched requests on a sync call. Batched
+ * requests on a sync call can for instance happen on a device driver when
+ * multiple cards are used and firmware loading happens outside of probe.
+ */
+static ssize_t trigger_batched_requests_store(struct device *dev,
+					      struct device_attribute *attr,
+					      const char *buf, size_t count)
+{
+	struct test_batched_req *req;
+	int rc;
+	u8 i;
+
+	mutex_lock(&test_fw_mutex);
+
+	test_fw_config->reqs = vzalloc(sizeof(struct test_batched_req) *
+				       test_fw_config->num_requests * 2);
+	if (!test_fw_config->reqs) {
+		rc = -ENOMEM;
+		goto out_unlock;
+	}
+
+	pr_info("batched sync firmware loading '%s' %u times\n",
+		test_fw_config->name, test_fw_config->num_requests);
+
+	for (i = 0; i < test_fw_config->num_requests; i++) {
+		req = &test_fw_config->reqs[i];
+		if (!req) {
+			WARN_ON(1);
+			rc = -ENOMEM;
+			goto out_bail;
+		}
+		req->fw = NULL;
+		req->idx = i;
+		req->name = test_fw_config->name;
+		req->dev = dev;
+		init_completion(&req->completion);
+		req->task = kthread_run(test_fw_run_batch_request, req,
+					     "%s-%u", KBUILD_MODNAME, req->idx);
+		if (!req->task || IS_ERR(req->task)) {
+			pr_err("Setting up thread %u failed\n", req->idx);
+			req->task = NULL;
+			rc = -ENOMEM;
+			goto out_bail;
+		}
+	}
+
+	rc = count;
+
+	/*
+	 * We require an explicit release to enable more time and delay of
+	 * calling release_firmware() to improve our chances of forcing a
+	 * batched request. If we instead called release_firmware() right away
+	 * then we might miss on an opportunity of having a successful firmware
+	 * request pass on the opportunity to be come a batched request.
+	 */
+
+out_bail:
+	for (i = 0; i < test_fw_config->num_requests; i++) {
+		req = &test_fw_config->reqs[i];
+		if (req->task || req->sent)
+			wait_for_completion(&req->completion);
+	}
+
+	/* Override any worker error if we had a general setup error */
+	if (rc < 0)
+		test_fw_config->test_result = rc;
+
+out_unlock:
+	mutex_unlock(&test_fw_mutex);
+
+	return rc;
+}
+static DEVICE_ATTR_WO(trigger_batched_requests);
+
+/*
+ * We wait for each callback to return with the lock held, no need to lock here
+ */
+static void trigger_batched_cb(const struct firmware *fw, void *context)
+{
+	struct test_batched_req *req = context;
+
+	if (!req) {
+		test_fw_config->test_result = -EINVAL;
+		return;
+	}
+
+	/* forces *some* batched requests to queue up */
+	if (!req->idx)
+		ssleep(2);
+
+	req->fw = fw;
+
+	/*
+	 * Unfortunately the firmware API gives us nothing other than a null FW
+	 * if the firmware was not found on async requests.  Best we can do is
+	 * just assume -ENOENT. A better API would pass the actual return
+	 * value to the callback.
+	 */
+	if (!fw && !test_fw_config->test_result)
+		test_fw_config->test_result = -ENOENT;
+
+	complete(&req->completion);
+}
+
+static
+ssize_t trigger_batched_requests_async_store(struct device *dev,
+					     struct device_attribute *attr,
+					     const char *buf, size_t count)
+{
+	struct test_batched_req *req;
+	bool send_uevent;
+	int rc;
+	u8 i;
+
+	mutex_lock(&test_fw_mutex);
+
+	test_fw_config->reqs = vzalloc(sizeof(struct test_batched_req) *
+				       test_fw_config->num_requests * 2);
+	if (!test_fw_config->reqs) {
+		rc = -ENOMEM;
+		goto out;
+	}
+
+	pr_info("batched loading '%s' custom fallback mechanism %u times\n",
+		test_fw_config->name, test_fw_config->num_requests);
+
+	send_uevent = test_fw_config->send_uevent ? FW_ACTION_HOTPLUG :
+		FW_ACTION_NOHOTPLUG;
+
+	for (i = 0; i < test_fw_config->num_requests; i++) {
+		req = &test_fw_config->reqs[i];
+		if (!req) {
+			WARN_ON(1);
+			goto out_bail;
+		}
+		req->name = test_fw_config->name;
+		req->fw = NULL;
+		req->idx = i;
+		init_completion(&req->completion);
+		rc = request_firmware_nowait(THIS_MODULE, send_uevent,
+					     req->name,
+					     dev, GFP_KERNEL, req,
+					     trigger_batched_cb);
+		if (rc) {
+			pr_info("#%u: batched async load failed setup: %d\n",
+				i, rc);
+			req->rc = rc;
+			goto out_bail;
+		} else
+			req->sent = true;
+	}
+
+	rc = count;
+
+out_bail:
+
+	/*
+	 * We require an explicit release to enable more time and delay of
+	 * calling release_firmware() to improve our chances of forcing a
+	 * batched request. If we instead called release_firmware() right away
+	 * then we might miss on an opportunity of having a successful firmware
+	 * request pass on the opportunity to be come a batched request.
+	 */
+
+	for (i = 0; i < test_fw_config->num_requests; i++) {
+		req = &test_fw_config->reqs[i];
+		if (req->sent)
+			wait_for_completion(&req->completion);
+	}
+
+	/* Override any worker error if we had a general setup error */
+	if (rc < 0)
+		test_fw_config->test_result = rc;
+
+out:
+	mutex_unlock(&test_fw_mutex);
+
+	return rc;
+}
+static DEVICE_ATTR_WO(trigger_batched_requests_async);
+
+static ssize_t test_result_show(struct device *dev,
+				struct device_attribute *attr,
+				char *buf)
+{
+	return test_dev_config_show_int(buf, test_fw_config->test_result);
+}
+static DEVICE_ATTR_RO(test_result);
+
+static ssize_t release_all_firmware_store(struct device *dev,
+					  struct device_attribute *attr,
+					  const char *buf, size_t count)
+{
+	test_release_all_firmware();
+	return count;
+}
+static DEVICE_ATTR_WO(release_all_firmware);
+
+static ssize_t read_firmware_show(struct device *dev,
+				  struct device_attribute *attr,
+				  char *buf)
+{
+	struct test_batched_req *req;
+	u8 idx;
+	ssize_t rc = 0;
+
+	mutex_lock(&test_fw_mutex);
+
+	idx = test_fw_config->read_fw_idx;
+	if (idx >= test_fw_config->num_requests) {
+		rc = -ERANGE;
+		goto out;
+	}
+
+	if (!test_fw_config->reqs) {
+		rc = -EINVAL;
+		goto out;
+	}
+
+	req = &test_fw_config->reqs[idx];
+	if (!req->fw) {
+		pr_err("#%u: failed to async load firmware\n", idx);
+		rc = -ENOENT;
+		goto out;
+	}
+
+	pr_info("#%u: loaded %zu\n", idx, req->fw->size);
+
+	if (req->fw->size > PAGE_SIZE) {
+		pr_err("Testing interface must use PAGE_SIZE firmware for now\n");
+		rc = -EINVAL;
+	}
+	memcpy(buf, req->fw->data, req->fw->size);
+
+	rc = req->fw->size;
+out:
+	mutex_unlock(&test_fw_mutex);
+
+	return rc;
+}
+static DEVICE_ATTR_RO(read_firmware);
+
 #define TEST_FW_DEV_ATTR(name)          &dev_attr_##name.attr
 
 static struct attribute *test_dev_attrs[] = {
+	TEST_FW_DEV_ATTR(reset),
+
+	TEST_FW_DEV_ATTR(config),
+	TEST_FW_DEV_ATTR(config_name),
+	TEST_FW_DEV_ATTR(config_num_requests),
+	TEST_FW_DEV_ATTR(config_sync_direct),
+	TEST_FW_DEV_ATTR(config_send_uevent),
+	TEST_FW_DEV_ATTR(config_read_fw_idx),
+
+	/* These don't use the config at all - they could be ported! */
 	TEST_FW_DEV_ATTR(trigger_request),
 	TEST_FW_DEV_ATTR(trigger_async_request),
 	TEST_FW_DEV_ATTR(trigger_custom_fallback),
+
+	/* These use the config and can use the test_result */
+	TEST_FW_DEV_ATTR(trigger_batched_requests),
+	TEST_FW_DEV_ATTR(trigger_batched_requests_async),
+
+	TEST_FW_DEV_ATTR(release_all_firmware),
+	TEST_FW_DEV_ATTR(test_result),
+	TEST_FW_DEV_ATTR(read_firmware),
 	NULL,
 };
 
@@ -192,8 +888,17 @@ static int __init test_firmware_init(void)
 {
 	int rc;
 
+	test_fw_config = kzalloc(sizeof(struct test_config), GFP_KERNEL);
+	if (!test_fw_config)
+		return -ENOMEM;
+
+	rc = __test_firmware_config_init();
+	if (rc)
+		return rc;
+
 	rc = misc_register(&test_fw_misc_device);
 	if (rc) {
+		kfree(test_fw_config);
 		pr_err("could not register misc device: %d\n", rc);
 		return rc;
 	}
@@ -207,8 +912,13 @@ module_init(test_firmware_init);
 
 static void __exit test_firmware_exit(void)
 {
+	mutex_lock(&test_fw_mutex);
 	release_firmware(test_firmware);
 	misc_deregister(&test_fw_misc_device);
+	__test_firmware_config_free();
+	kfree(test_fw_config);
+	mutex_unlock(&test_fw_mutex);
+
 	pr_warn("removed interface\n");
 }
 
diff --git a/lib/test_kmod.c b/lib/test_kmod.c
index ff9148969b92..fba78d25e825 100644
--- a/lib/test_kmod.c
+++ b/lib/test_kmod.c
@@ -924,7 +924,7 @@ static int test_dev_config_update_uint_range(struct kmod_test_device *test_dev,
 	if (ret)
 		return ret;
 
-	if (new < min || new >  max || new > UINT_MAX)
+	if (new < min || new > max)
 		return -EINVAL;
 
 	mutex_lock(&test_dev->config_mutex);
@@ -946,7 +946,7 @@ static int test_dev_config_update_int(struct kmod_test_device *test_dev,
 	if (ret)
 		return ret;
 
-	if (new > INT_MAX || new < INT_MIN)
+	if (new < INT_MIN || new > INT_MAX)
 		return -EINVAL;
 
 	mutex_lock(&test_dev->config_mutex);
diff --git a/lib/xxhash.c b/lib/xxhash.c
new file mode 100644
index 000000000000..aa61e2a3802f
--- /dev/null
+++ b/lib/xxhash.c
@@ -0,0 +1,500 @@
+/*
+ * xxHash - Extremely Fast Hash algorithm
+ * Copyright (C) 2012-2016, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ *     notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ *     copyright notice, this list of conditions and the following disclaimer
+ *     in the documentation and/or other materials provided with the
+ *     distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at:
+ * - xxHash homepage: http://cyan4973.github.io/xxHash/
+ * - xxHash source repository: https://github.com/Cyan4973/xxHash
+ */
+
+#include <asm/unaligned.h>
+#include <linux/errno.h>
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/string.h>
+#include <linux/xxhash.h>
+
+/*-*************************************
+ * Macros
+ **************************************/
+#define xxh_rotl32(x, r) ((x << r) | (x >> (32 - r)))
+#define xxh_rotl64(x, r) ((x << r) | (x >> (64 - r)))
+
+#ifdef __LITTLE_ENDIAN
+# define XXH_CPU_LITTLE_ENDIAN 1
+#else
+# define XXH_CPU_LITTLE_ENDIAN 0
+#endif
+
+/*-*************************************
+ * Constants
+ **************************************/
+static const uint32_t PRIME32_1 = 2654435761U;
+static const uint32_t PRIME32_2 = 2246822519U;
+static const uint32_t PRIME32_3 = 3266489917U;
+static const uint32_t PRIME32_4 =  668265263U;
+static const uint32_t PRIME32_5 =  374761393U;
+
+static const uint64_t PRIME64_1 = 11400714785074694791ULL;
+static const uint64_t PRIME64_2 = 14029467366897019727ULL;
+static const uint64_t PRIME64_3 =  1609587929392839161ULL;
+static const uint64_t PRIME64_4 =  9650029242287828579ULL;
+static const uint64_t PRIME64_5 =  2870177450012600261ULL;
+
+/*-**************************
+ *  Utils
+ ***************************/
+void xxh32_copy_state(struct xxh32_state *dst, const struct xxh32_state *src)
+{
+	memcpy(dst, src, sizeof(*dst));
+}
+EXPORT_SYMBOL(xxh32_copy_state);
+
+void xxh64_copy_state(struct xxh64_state *dst, const struct xxh64_state *src)
+{
+	memcpy(dst, src, sizeof(*dst));
+}
+EXPORT_SYMBOL(xxh64_copy_state);
+
+/*-***************************
+ * Simple Hash Functions
+ ****************************/
+static uint32_t xxh32_round(uint32_t seed, const uint32_t input)
+{
+	seed += input * PRIME32_2;
+	seed = xxh_rotl32(seed, 13);
+	seed *= PRIME32_1;
+	return seed;
+}
+
+uint32_t xxh32(const void *input, const size_t len, const uint32_t seed)
+{
+	const uint8_t *p = (const uint8_t *)input;
+	const uint8_t *b_end = p + len;
+	uint32_t h32;
+
+	if (len >= 16) {
+		const uint8_t *const limit = b_end - 16;
+		uint32_t v1 = seed + PRIME32_1 + PRIME32_2;
+		uint32_t v2 = seed + PRIME32_2;
+		uint32_t v3 = seed + 0;
+		uint32_t v4 = seed - PRIME32_1;
+
+		do {
+			v1 = xxh32_round(v1, get_unaligned_le32(p));
+			p += 4;
+			v2 = xxh32_round(v2, get_unaligned_le32(p));
+			p += 4;
+			v3 = xxh32_round(v3, get_unaligned_le32(p));
+			p += 4;
+			v4 = xxh32_round(v4, get_unaligned_le32(p));
+			p += 4;
+		} while (p <= limit);
+
+		h32 = xxh_rotl32(v1, 1) + xxh_rotl32(v2, 7) +
+			xxh_rotl32(v3, 12) + xxh_rotl32(v4, 18);
+	} else {
+		h32 = seed + PRIME32_5;
+	}
+
+	h32 += (uint32_t)len;
+
+	while (p + 4 <= b_end) {
+		h32 += get_unaligned_le32(p) * PRIME32_3;
+		h32 = xxh_rotl32(h32, 17) * PRIME32_4;
+		p += 4;
+	}
+
+	while (p < b_end) {
+		h32 += (*p) * PRIME32_5;
+		h32 = xxh_rotl32(h32, 11) * PRIME32_1;
+		p++;
+	}
+
+	h32 ^= h32 >> 15;
+	h32 *= PRIME32_2;
+	h32 ^= h32 >> 13;
+	h32 *= PRIME32_3;
+	h32 ^= h32 >> 16;
+
+	return h32;
+}
+EXPORT_SYMBOL(xxh32);
+
+static uint64_t xxh64_round(uint64_t acc, const uint64_t input)
+{
+	acc += input * PRIME64_2;
+	acc = xxh_rotl64(acc, 31);
+	acc *= PRIME64_1;
+	return acc;
+}
+
+static uint64_t xxh64_merge_round(uint64_t acc, uint64_t val)
+{
+	val = xxh64_round(0, val);
+	acc ^= val;
+	acc = acc * PRIME64_1 + PRIME64_4;
+	return acc;
+}
+
+uint64_t xxh64(const void *input, const size_t len, const uint64_t seed)
+{
+	const uint8_t *p = (const uint8_t *)input;
+	const uint8_t *const b_end = p + len;
+	uint64_t h64;
+
+	if (len >= 32) {
+		const uint8_t *const limit = b_end - 32;
+		uint64_t v1 = seed + PRIME64_1 + PRIME64_2;
+		uint64_t v2 = seed + PRIME64_2;
+		uint64_t v3 = seed + 0;
+		uint64_t v4 = seed - PRIME64_1;
+
+		do {
+			v1 = xxh64_round(v1, get_unaligned_le64(p));
+			p += 8;
+			v2 = xxh64_round(v2, get_unaligned_le64(p));
+			p += 8;
+			v3 = xxh64_round(v3, get_unaligned_le64(p));
+			p += 8;
+			v4 = xxh64_round(v4, get_unaligned_le64(p));
+			p += 8;
+		} while (p <= limit);
+
+		h64 = xxh_rotl64(v1, 1) + xxh_rotl64(v2, 7) +
+			xxh_rotl64(v3, 12) + xxh_rotl64(v4, 18);
+		h64 = xxh64_merge_round(h64, v1);
+		h64 = xxh64_merge_round(h64, v2);
+		h64 = xxh64_merge_round(h64, v3);
+		h64 = xxh64_merge_round(h64, v4);
+
+	} else {
+		h64  = seed + PRIME64_5;
+	}
+
+	h64 += (uint64_t)len;
+
+	while (p + 8 <= b_end) {
+		const uint64_t k1 = xxh64_round(0, get_unaligned_le64(p));
+
+		h64 ^= k1;
+		h64 = xxh_rotl64(h64, 27) * PRIME64_1 + PRIME64_4;
+		p += 8;
+	}
+
+	if (p + 4 <= b_end) {
+		h64 ^= (uint64_t)(get_unaligned_le32(p)) * PRIME64_1;
+		h64 = xxh_rotl64(h64, 23) * PRIME64_2 + PRIME64_3;
+		p += 4;
+	}
+
+	while (p < b_end) {
+		h64 ^= (*p) * PRIME64_5;
+		h64 = xxh_rotl64(h64, 11) * PRIME64_1;
+		p++;
+	}
+
+	h64 ^= h64 >> 33;
+	h64 *= PRIME64_2;
+	h64 ^= h64 >> 29;
+	h64 *= PRIME64_3;
+	h64 ^= h64 >> 32;
+
+	return h64;
+}
+EXPORT_SYMBOL(xxh64);
+
+/*-**************************************************
+ * Advanced Hash Functions
+ ***************************************************/
+void xxh32_reset(struct xxh32_state *statePtr, const uint32_t seed)
+{
+	/* use a local state for memcpy() to avoid strict-aliasing warnings */
+	struct xxh32_state state;
+
+	memset(&state, 0, sizeof(state));
+	state.v1 = seed + PRIME32_1 + PRIME32_2;
+	state.v2 = seed + PRIME32_2;
+	state.v3 = seed + 0;
+	state.v4 = seed - PRIME32_1;
+	memcpy(statePtr, &state, sizeof(state));
+}
+EXPORT_SYMBOL(xxh32_reset);
+
+void xxh64_reset(struct xxh64_state *statePtr, const uint64_t seed)
+{
+	/* use a local state for memcpy() to avoid strict-aliasing warnings */
+	struct xxh64_state state;
+
+	memset(&state, 0, sizeof(state));
+	state.v1 = seed + PRIME64_1 + PRIME64_2;
+	state.v2 = seed + PRIME64_2;
+	state.v3 = seed + 0;
+	state.v4 = seed - PRIME64_1;
+	memcpy(statePtr, &state, sizeof(state));
+}
+EXPORT_SYMBOL(xxh64_reset);
+
+int xxh32_update(struct xxh32_state *state, const void *input, const size_t len)
+{
+	const uint8_t *p = (const uint8_t *)input;
+	const uint8_t *const b_end = p + len;
+
+	if (input == NULL)
+		return -EINVAL;
+
+	state->total_len_32 += (uint32_t)len;
+	state->large_len |= (len >= 16) | (state->total_len_32 >= 16);
+
+	if (state->memsize + len < 16) { /* fill in tmp buffer */
+		memcpy((uint8_t *)(state->mem32) + state->memsize, input, len);
+		state->memsize += (uint32_t)len;
+		return 0;
+	}
+
+	if (state->memsize) { /* some data left from previous update */
+		const uint32_t *p32 = state->mem32;
+
+		memcpy((uint8_t *)(state->mem32) + state->memsize, input,
+			16 - state->memsize);
+
+		state->v1 = xxh32_round(state->v1, get_unaligned_le32(p32));
+		p32++;
+		state->v2 = xxh32_round(state->v2, get_unaligned_le32(p32));
+		p32++;
+		state->v3 = xxh32_round(state->v3, get_unaligned_le32(p32));
+		p32++;
+		state->v4 = xxh32_round(state->v4, get_unaligned_le32(p32));
+		p32++;
+
+		p += 16-state->memsize;
+		state->memsize = 0;
+	}
+
+	if (p <= b_end - 16) {
+		const uint8_t *const limit = b_end - 16;
+		uint32_t v1 = state->v1;
+		uint32_t v2 = state->v2;
+		uint32_t v3 = state->v3;
+		uint32_t v4 = state->v4;
+
+		do {
+			v1 = xxh32_round(v1, get_unaligned_le32(p));
+			p += 4;
+			v2 = xxh32_round(v2, get_unaligned_le32(p));
+			p += 4;
+			v3 = xxh32_round(v3, get_unaligned_le32(p));
+			p += 4;
+			v4 = xxh32_round(v4, get_unaligned_le32(p));
+			p += 4;
+		} while (p <= limit);
+
+		state->v1 = v1;
+		state->v2 = v2;
+		state->v3 = v3;
+		state->v4 = v4;
+	}
+
+	if (p < b_end) {
+		memcpy(state->mem32, p, (size_t)(b_end-p));
+		state->memsize = (uint32_t)(b_end-p);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(xxh32_update);
+
+uint32_t xxh32_digest(const struct xxh32_state *state)
+{
+	const uint8_t *p = (const uint8_t *)state->mem32;
+	const uint8_t *const b_end = (const uint8_t *)(state->mem32) +
+		state->memsize;
+	uint32_t h32;
+
+	if (state->large_len) {
+		h32 = xxh_rotl32(state->v1, 1) + xxh_rotl32(state->v2, 7) +
+			xxh_rotl32(state->v3, 12) + xxh_rotl32(state->v4, 18);
+	} else {
+		h32 = state->v3 /* == seed */ + PRIME32_5;
+	}
+
+	h32 += state->total_len_32;
+
+	while (p + 4 <= b_end) {
+		h32 += get_unaligned_le32(p) * PRIME32_3;
+		h32 = xxh_rotl32(h32, 17) * PRIME32_4;
+		p += 4;
+	}
+
+	while (p < b_end) {
+		h32 += (*p) * PRIME32_5;
+		h32 = xxh_rotl32(h32, 11) * PRIME32_1;
+		p++;
+	}
+
+	h32 ^= h32 >> 15;
+	h32 *= PRIME32_2;
+	h32 ^= h32 >> 13;
+	h32 *= PRIME32_3;
+	h32 ^= h32 >> 16;
+
+	return h32;
+}
+EXPORT_SYMBOL(xxh32_digest);
+
+int xxh64_update(struct xxh64_state *state, const void *input, const size_t len)
+{
+	const uint8_t *p = (const uint8_t *)input;
+	const uint8_t *const b_end = p + len;
+
+	if (input == NULL)
+		return -EINVAL;
+
+	state->total_len += len;
+
+	if (state->memsize + len < 32) { /* fill in tmp buffer */
+		memcpy(((uint8_t *)state->mem64) + state->memsize, input, len);
+		state->memsize += (uint32_t)len;
+		return 0;
+	}
+
+	if (state->memsize) { /* tmp buffer is full */
+		uint64_t *p64 = state->mem64;
+
+		memcpy(((uint8_t *)p64) + state->memsize, input,
+			32 - state->memsize);
+
+		state->v1 = xxh64_round(state->v1, get_unaligned_le64(p64));
+		p64++;
+		state->v2 = xxh64_round(state->v2, get_unaligned_le64(p64));
+		p64++;
+		state->v3 = xxh64_round(state->v3, get_unaligned_le64(p64));
+		p64++;
+		state->v4 = xxh64_round(state->v4, get_unaligned_le64(p64));
+
+		p += 32 - state->memsize;
+		state->memsize = 0;
+	}
+
+	if (p + 32 <= b_end) {
+		const uint8_t *const limit = b_end - 32;
+		uint64_t v1 = state->v1;
+		uint64_t v2 = state->v2;
+		uint64_t v3 = state->v3;
+		uint64_t v4 = state->v4;
+
+		do {
+			v1 = xxh64_round(v1, get_unaligned_le64(p));
+			p += 8;
+			v2 = xxh64_round(v2, get_unaligned_le64(p));
+			p += 8;
+			v3 = xxh64_round(v3, get_unaligned_le64(p));
+			p += 8;
+			v4 = xxh64_round(v4, get_unaligned_le64(p));
+			p += 8;
+		} while (p <= limit);
+
+		state->v1 = v1;
+		state->v2 = v2;
+		state->v3 = v3;
+		state->v4 = v4;
+	}
+
+	if (p < b_end) {
+		memcpy(state->mem64, p, (size_t)(b_end-p));
+		state->memsize = (uint32_t)(b_end - p);
+	}
+
+	return 0;
+}
+EXPORT_SYMBOL(xxh64_update);
+
+uint64_t xxh64_digest(const struct xxh64_state *state)
+{
+	const uint8_t *p = (const uint8_t *)state->mem64;
+	const uint8_t *const b_end = (const uint8_t *)state->mem64 +
+		state->memsize;
+	uint64_t h64;
+
+	if (state->total_len >= 32) {
+		const uint64_t v1 = state->v1;
+		const uint64_t v2 = state->v2;
+		const uint64_t v3 = state->v3;
+		const uint64_t v4 = state->v4;
+
+		h64 = xxh_rotl64(v1, 1) + xxh_rotl64(v2, 7) +
+			xxh_rotl64(v3, 12) + xxh_rotl64(v4, 18);
+		h64 = xxh64_merge_round(h64, v1);
+		h64 = xxh64_merge_round(h64, v2);
+		h64 = xxh64_merge_round(h64, v3);
+		h64 = xxh64_merge_round(h64, v4);
+	} else {
+		h64  = state->v3 + PRIME64_5;
+	}
+
+	h64 += (uint64_t)state->total_len;
+
+	while (p + 8 <= b_end) {
+		const uint64_t k1 = xxh64_round(0, get_unaligned_le64(p));
+
+		h64 ^= k1;
+		h64 = xxh_rotl64(h64, 27) * PRIME64_1 + PRIME64_4;
+		p += 8;
+	}
+
+	if (p + 4 <= b_end) {
+		h64 ^= (uint64_t)(get_unaligned_le32(p)) * PRIME64_1;
+		h64 = xxh_rotl64(h64, 23) * PRIME64_2 + PRIME64_3;
+		p += 4;
+	}
+
+	while (p < b_end) {
+		h64 ^= (*p) * PRIME64_5;
+		h64 = xxh_rotl64(h64, 11) * PRIME64_1;
+		p++;
+	}
+
+	h64 ^= h64 >> 33;
+	h64 *= PRIME64_2;
+	h64 ^= h64 >> 29;
+	h64 *= PRIME64_3;
+	h64 ^= h64 >> 32;
+
+	return h64;
+}
+EXPORT_SYMBOL(xxh64_digest);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("xxHash");
diff --git a/lib/zstd/Makefile b/lib/zstd/Makefile
new file mode 100644
index 000000000000..dd0a359c135b
--- /dev/null
+++ b/lib/zstd/Makefile
@@ -0,0 +1,18 @@
+obj-$(CONFIG_ZSTD_COMPRESS) += zstd_compress.o
+obj-$(CONFIG_ZSTD_DECOMPRESS) += zstd_decompress.o
+
+ccflags-y += -O3
+
+# Object files unique to zstd_compress and zstd_decompress
+zstd_compress-y := fse_compress.o huf_compress.o compress.o
+zstd_decompress-y := huf_decompress.o decompress.o
+
+# These object files are shared between the modules.
+# Always add them to zstd_compress.
+# Unless both zstd_compress and zstd_decompress are built in
+# then also add them to zstd_decompress.
+zstd_compress-y += entropy_common.o fse_decompress.o zstd_common.o
+
+ifneq ($(CONFIG_ZSTD_COMPRESS)$(CONFIG_ZSTD_DECOMPRESS),yy)
+	zstd_decompress-y += entropy_common.o fse_decompress.o zstd_common.o
+endif
diff --git a/lib/zstd/bitstream.h b/lib/zstd/bitstream.h
new file mode 100644
index 000000000000..a826b99e1d63
--- /dev/null
+++ b/lib/zstd/bitstream.h
@@ -0,0 +1,374 @@
+/*
+ * bitstream
+ * Part of FSE library
+ * header file (to include)
+ * Copyright (C) 2013-2016, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ */
+#ifndef BITSTREAM_H_MODULE
+#define BITSTREAM_H_MODULE
+
+/*
+*  This API consists of small unitary functions, which must be inlined for best performance.
+*  Since link-time-optimization is not available for all compilers,
+*  these functions are defined into a .h to be included.
+*/
+
+/*-****************************************
+*  Dependencies
+******************************************/
+#include "error_private.h" /* error codes and messages */
+#include "mem.h"	   /* unaligned access routines */
+
+/*=========================================
+*  Target specific
+=========================================*/
+#define STREAM_ACCUMULATOR_MIN_32 25
+#define STREAM_ACCUMULATOR_MIN_64 57
+#define STREAM_ACCUMULATOR_MIN ((U32)(ZSTD_32bits() ? STREAM_ACCUMULATOR_MIN_32 : STREAM_ACCUMULATOR_MIN_64))
+
+/*-******************************************
+*  bitStream encoding API (write forward)
+********************************************/
+/* bitStream can mix input from multiple sources.
+*  A critical property of these streams is that they encode and decode in **reverse** direction.
+*  So the first bit sequence you add will be the last to be read, like a LIFO stack.
+*/
+typedef struct {
+	size_t bitContainer;
+	int bitPos;
+	char *startPtr;
+	char *ptr;
+	char *endPtr;
+} BIT_CStream_t;
+
+ZSTD_STATIC size_t BIT_initCStream(BIT_CStream_t *bitC, void *dstBuffer, size_t dstCapacity);
+ZSTD_STATIC void BIT_addBits(BIT_CStream_t *bitC, size_t value, unsigned nbBits);
+ZSTD_STATIC void BIT_flushBits(BIT_CStream_t *bitC);
+ZSTD_STATIC size_t BIT_closeCStream(BIT_CStream_t *bitC);
+
+/* Start with initCStream, providing the size of buffer to write into.
+*  bitStream will never write outside of this buffer.
+*  `dstCapacity` must be >= sizeof(bitD->bitContainer), otherwise @return will be an error code.
+*
+*  bits are first added to a local register.
+*  Local register is size_t, hence 64-bits on 64-bits systems, or 32-bits on 32-bits systems.
+*  Writing data into memory is an explicit operation, performed by the flushBits function.
+*  Hence keep track how many bits are potentially stored into local register to avoid register overflow.
+*  After a flushBits, a maximum of 7 bits might still be stored into local register.
+*
+*  Avoid storing elements of more than 24 bits if you want compatibility with 32-bits bitstream readers.
+*
+*  Last operation is to close the bitStream.
+*  The function returns the final size of CStream in bytes.
+*  If data couldn't fit into `dstBuffer`, it will return a 0 ( == not storable)
+*/
+
+/*-********************************************
+*  bitStream decoding API (read backward)
+**********************************************/
+typedef struct {
+	size_t bitContainer;
+	unsigned bitsConsumed;
+	const char *ptr;
+	const char *start;
+} BIT_DStream_t;
+
+typedef enum {
+	BIT_DStream_unfinished = 0,
+	BIT_DStream_endOfBuffer = 1,
+	BIT_DStream_completed = 2,
+	BIT_DStream_overflow = 3
+} BIT_DStream_status; /* result of BIT_reloadDStream() */
+/* 1,2,4,8 would be better for bitmap combinations, but slows down performance a bit ... :( */
+
+ZSTD_STATIC size_t BIT_initDStream(BIT_DStream_t *bitD, const void *srcBuffer, size_t srcSize);
+ZSTD_STATIC size_t BIT_readBits(BIT_DStream_t *bitD, unsigned nbBits);
+ZSTD_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t *bitD);
+ZSTD_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t *bitD);
+
+/* Start by invoking BIT_initDStream().
+*  A chunk of the bitStream is then stored into a local register.
+*  Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
+*  You can then retrieve bitFields stored into the local register, **in reverse order**.
+*  Local register is explicitly reloaded from memory by the BIT_reloadDStream() method.
+*  A reload guarantee a minimum of ((8*sizeof(bitD->bitContainer))-7) bits when its result is BIT_DStream_unfinished.
+*  Otherwise, it can be less than that, so proceed accordingly.
+*  Checking if DStream has reached its end can be performed with BIT_endOfDStream().
+*/
+
+/*-****************************************
+*  unsafe API
+******************************************/
+ZSTD_STATIC void BIT_addBitsFast(BIT_CStream_t *bitC, size_t value, unsigned nbBits);
+/* faster, but works only if value is "clean", meaning all high bits above nbBits are 0 */
+
+ZSTD_STATIC void BIT_flushBitsFast(BIT_CStream_t *bitC);
+/* unsafe version; does not check buffer overflow */
+
+ZSTD_STATIC size_t BIT_readBitsFast(BIT_DStream_t *bitD, unsigned nbBits);
+/* faster, but works only if nbBits >= 1 */
+
+/*-**************************************************************
+*  Internal functions
+****************************************************************/
+ZSTD_STATIC unsigned BIT_highbit32(register U32 val) { return 31 - __builtin_clz(val); }
+
+/*=====    Local Constants   =====*/
+static const unsigned BIT_mask[] = {0,       1,       3,       7,	0xF,      0x1F,     0x3F,     0x7F,      0xFF,
+				    0x1FF,   0x3FF,   0x7FF,   0xFFF,    0x1FFF,   0x3FFF,   0x7FFF,   0xFFFF,    0x1FFFF,
+				    0x3FFFF, 0x7FFFF, 0xFFFFF, 0x1FFFFF, 0x3FFFFF, 0x7FFFFF, 0xFFFFFF, 0x1FFFFFF, 0x3FFFFFF}; /* up to 26 bits */
+
+/*-**************************************************************
+*  bitStream encoding
+****************************************************************/
+/*! BIT_initCStream() :
+ *  `dstCapacity` must be > sizeof(void*)
+ *  @return : 0 if success,
+			  otherwise an error code (can be tested using ERR_isError() ) */
+ZSTD_STATIC size_t BIT_initCStream(BIT_CStream_t *bitC, void *startPtr, size_t dstCapacity)
+{
+	bitC->bitContainer = 0;
+	bitC->bitPos = 0;
+	bitC->startPtr = (char *)startPtr;
+	bitC->ptr = bitC->startPtr;
+	bitC->endPtr = bitC->startPtr + dstCapacity - sizeof(bitC->ptr);
+	if (dstCapacity <= sizeof(bitC->ptr))
+		return ERROR(dstSize_tooSmall);
+	return 0;
+}
+
+/*! BIT_addBits() :
+	can add up to 26 bits into `bitC`.
+	Does not check for register overflow ! */
+ZSTD_STATIC void BIT_addBits(BIT_CStream_t *bitC, size_t value, unsigned nbBits)
+{
+	bitC->bitContainer |= (value & BIT_mask[nbBits]) << bitC->bitPos;
+	bitC->bitPos += nbBits;
+}
+
+/*! BIT_addBitsFast() :
+ *  works only if `value` is _clean_, meaning all high bits above nbBits are 0 */
+ZSTD_STATIC void BIT_addBitsFast(BIT_CStream_t *bitC, size_t value, unsigned nbBits)
+{
+	bitC->bitContainer |= value << bitC->bitPos;
+	bitC->bitPos += nbBits;
+}
+
+/*! BIT_flushBitsFast() :
+ *  unsafe version; does not check buffer overflow */
+ZSTD_STATIC void BIT_flushBitsFast(BIT_CStream_t *bitC)
+{
+	size_t const nbBytes = bitC->bitPos >> 3;
+	ZSTD_writeLEST(bitC->ptr, bitC->bitContainer);
+	bitC->ptr += nbBytes;
+	bitC->bitPos &= 7;
+	bitC->bitContainer >>= nbBytes * 8; /* if bitPos >= sizeof(bitContainer)*8 --> undefined behavior */
+}
+
+/*! BIT_flushBits() :
+ *  safe version; check for buffer overflow, and prevents it.
+ *  note : does not signal buffer overflow. This will be revealed later on using BIT_closeCStream() */
+ZSTD_STATIC void BIT_flushBits(BIT_CStream_t *bitC)
+{
+	size_t const nbBytes = bitC->bitPos >> 3;
+	ZSTD_writeLEST(bitC->ptr, bitC->bitContainer);
+	bitC->ptr += nbBytes;
+	if (bitC->ptr > bitC->endPtr)
+		bitC->ptr = bitC->endPtr;
+	bitC->bitPos &= 7;
+	bitC->bitContainer >>= nbBytes * 8; /* if bitPos >= sizeof(bitContainer)*8 --> undefined behavior */
+}
+
+/*! BIT_closeCStream() :
+ *  @return : size of CStream, in bytes,
+			  or 0 if it could not fit into dstBuffer */
+ZSTD_STATIC size_t BIT_closeCStream(BIT_CStream_t *bitC)
+{
+	BIT_addBitsFast(bitC, 1, 1); /* endMark */
+	BIT_flushBits(bitC);
+
+	if (bitC->ptr >= bitC->endPtr)
+		return 0; /* doesn't fit within authorized budget : cancel */
+
+	return (bitC->ptr - bitC->startPtr) + (bitC->bitPos > 0);
+}
+
+/*-********************************************************
+* bitStream decoding
+**********************************************************/
+/*! BIT_initDStream() :
+*   Initialize a BIT_DStream_t.
+*   `bitD` : a pointer to an already allocated BIT_DStream_t structure.
+*   `srcSize` must be the *exact* size of the bitStream, in bytes.
+*   @return : size of stream (== srcSize) or an errorCode if a problem is detected
+*/
+ZSTD_STATIC size_t BIT_initDStream(BIT_DStream_t *bitD, const void *srcBuffer, size_t srcSize)
+{
+	if (srcSize < 1) {
+		memset(bitD, 0, sizeof(*bitD));
+		return ERROR(srcSize_wrong);
+	}
+
+	if (srcSize >= sizeof(bitD->bitContainer)) { /* normal case */
+		bitD->start = (const char *)srcBuffer;
+		bitD->ptr = (const char *)srcBuffer + srcSize - sizeof(bitD->bitContainer);
+		bitD->bitContainer = ZSTD_readLEST(bitD->ptr);
+		{
+			BYTE const lastByte = ((const BYTE *)srcBuffer)[srcSize - 1];
+			bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0; /* ensures bitsConsumed is always set */
+			if (lastByte == 0)
+				return ERROR(GENERIC); /* endMark not present */
+		}
+	} else {
+		bitD->start = (const char *)srcBuffer;
+		bitD->ptr = bitD->start;
+		bitD->bitContainer = *(const BYTE *)(bitD->start);
+		switch (srcSize) {
+		case 7: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[6]) << (sizeof(bitD->bitContainer) * 8 - 16);
+		case 6: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[5]) << (sizeof(bitD->bitContainer) * 8 - 24);
+		case 5: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[4]) << (sizeof(bitD->bitContainer) * 8 - 32);
+		case 4: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[3]) << 24;
+		case 3: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[2]) << 16;
+		case 2: bitD->bitContainer += (size_t)(((const BYTE *)(srcBuffer))[1]) << 8;
+		default:;
+		}
+		{
+			BYTE const lastByte = ((const BYTE *)srcBuffer)[srcSize - 1];
+			bitD->bitsConsumed = lastByte ? 8 - BIT_highbit32(lastByte) : 0;
+			if (lastByte == 0)
+				return ERROR(GENERIC); /* endMark not present */
+		}
+		bitD->bitsConsumed += (U32)(sizeof(bitD->bitContainer) - srcSize) * 8;
+	}
+
+	return srcSize;
+}
+
+ZSTD_STATIC size_t BIT_getUpperBits(size_t bitContainer, U32 const start) { return bitContainer >> start; }
+
+ZSTD_STATIC size_t BIT_getMiddleBits(size_t bitContainer, U32 const start, U32 const nbBits) { return (bitContainer >> start) & BIT_mask[nbBits]; }
+
+ZSTD_STATIC size_t BIT_getLowerBits(size_t bitContainer, U32 const nbBits) { return bitContainer & BIT_mask[nbBits]; }
+
+/*! BIT_lookBits() :
+ *  Provides next n bits from local register.
+ *  local register is not modified.
+ *  On 32-bits, maxNbBits==24.
+ *  On 64-bits, maxNbBits==56.
+ *  @return : value extracted
+ */
+ZSTD_STATIC size_t BIT_lookBits(const BIT_DStream_t *bitD, U32 nbBits)
+{
+	U32 const bitMask = sizeof(bitD->bitContainer) * 8 - 1;
+	return ((bitD->bitContainer << (bitD->bitsConsumed & bitMask)) >> 1) >> ((bitMask - nbBits) & bitMask);
+}
+
+/*! BIT_lookBitsFast() :
+*   unsafe version; only works only if nbBits >= 1 */
+ZSTD_STATIC size_t BIT_lookBitsFast(const BIT_DStream_t *bitD, U32 nbBits)
+{
+	U32 const bitMask = sizeof(bitD->bitContainer) * 8 - 1;
+	return (bitD->bitContainer << (bitD->bitsConsumed & bitMask)) >> (((bitMask + 1) - nbBits) & bitMask);
+}
+
+ZSTD_STATIC void BIT_skipBits(BIT_DStream_t *bitD, U32 nbBits) { bitD->bitsConsumed += nbBits; }
+
+/*! BIT_readBits() :
+ *  Read (consume) next n bits from local register and update.
+ *  Pay attention to not read more than nbBits contained into local register.
+ *  @return : extracted value.
+ */
+ZSTD_STATIC size_t BIT_readBits(BIT_DStream_t *bitD, U32 nbBits)
+{
+	size_t const value = BIT_lookBits(bitD, nbBits);
+	BIT_skipBits(bitD, nbBits);
+	return value;
+}
+
+/*! BIT_readBitsFast() :
+*   unsafe version; only works only if nbBits >= 1 */
+ZSTD_STATIC size_t BIT_readBitsFast(BIT_DStream_t *bitD, U32 nbBits)
+{
+	size_t const value = BIT_lookBitsFast(bitD, nbBits);
+	BIT_skipBits(bitD, nbBits);
+	return value;
+}
+
+/*! BIT_reloadDStream() :
+*   Refill `bitD` from buffer previously set in BIT_initDStream() .
+*   This function is safe, it guarantees it will not read beyond src buffer.
+*   @return : status of `BIT_DStream_t` internal register.
+			  if status == BIT_DStream_unfinished, internal register is filled with >= (sizeof(bitD->bitContainer)*8 - 7) bits */
+ZSTD_STATIC BIT_DStream_status BIT_reloadDStream(BIT_DStream_t *bitD)
+{
+	if (bitD->bitsConsumed > (sizeof(bitD->bitContainer) * 8)) /* should not happen => corruption detected */
+		return BIT_DStream_overflow;
+
+	if (bitD->ptr >= bitD->start + sizeof(bitD->bitContainer)) {
+		bitD->ptr -= bitD->bitsConsumed >> 3;
+		bitD->bitsConsumed &= 7;
+		bitD->bitContainer = ZSTD_readLEST(bitD->ptr);
+		return BIT_DStream_unfinished;
+	}
+	if (bitD->ptr == bitD->start) {
+		if (bitD->bitsConsumed < sizeof(bitD->bitContainer) * 8)
+			return BIT_DStream_endOfBuffer;
+		return BIT_DStream_completed;
+	}
+	{
+		U32 nbBytes = bitD->bitsConsumed >> 3;
+		BIT_DStream_status result = BIT_DStream_unfinished;
+		if (bitD->ptr - nbBytes < bitD->start) {
+			nbBytes = (U32)(bitD->ptr - bitD->start); /* ptr > start */
+			result = BIT_DStream_endOfBuffer;
+		}
+		bitD->ptr -= nbBytes;
+		bitD->bitsConsumed -= nbBytes * 8;
+		bitD->bitContainer = ZSTD_readLEST(bitD->ptr); /* reminder : srcSize > sizeof(bitD) */
+		return result;
+	}
+}
+
+/*! BIT_endOfDStream() :
+*   @return Tells if DStream has exactly reached its end (all bits consumed).
+*/
+ZSTD_STATIC unsigned BIT_endOfDStream(const BIT_DStream_t *DStream)
+{
+	return ((DStream->ptr == DStream->start) && (DStream->bitsConsumed == sizeof(DStream->bitContainer) * 8));
+}
+
+#endif /* BITSTREAM_H_MODULE */
diff --git a/lib/zstd/compress.c b/lib/zstd/compress.c
new file mode 100644
index 000000000000..f9166cf4f7a9
--- /dev/null
+++ b/lib/zstd/compress.c
@@ -0,0 +1,3484 @@
+/**
+ * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd.
+ * An additional grant of patent rights can be found in the PATENTS file in the
+ * same directory.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ */
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#include "fse.h"
+#include "huf.h"
+#include "mem.h"
+#include "zstd_internal.h" /* includes zstd.h */
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/string.h> /* memset */
+
+/*-*************************************
+*  Constants
+***************************************/
+static const U32 g_searchStrength = 8; /* control skip over incompressible data */
+#define HASH_READ_SIZE 8
+typedef enum { ZSTDcs_created = 0, ZSTDcs_init, ZSTDcs_ongoing, ZSTDcs_ending } ZSTD_compressionStage_e;
+
+/*-*************************************
+*  Helper functions
+***************************************/
+size_t ZSTD_compressBound(size_t srcSize) { return FSE_compressBound(srcSize) + 12; }
+
+/*-*************************************
+*  Sequence storage
+***************************************/
+static void ZSTD_resetSeqStore(seqStore_t *ssPtr)
+{
+	ssPtr->lit = ssPtr->litStart;
+	ssPtr->sequences = ssPtr->sequencesStart;
+	ssPtr->longLengthID = 0;
+}
+
+/*-*************************************
+*  Context memory management
+***************************************/
+struct ZSTD_CCtx_s {
+	const BYTE *nextSrc;  /* next block here to continue on curr prefix */
+	const BYTE *base;     /* All regular indexes relative to this position */
+	const BYTE *dictBase; /* extDict indexes relative to this position */
+	U32 dictLimit;	/* below that point, need extDict */
+	U32 lowLimit;	 /* below that point, no more data */
+	U32 nextToUpdate;     /* index from which to continue dictionary update */
+	U32 nextToUpdate3;    /* index from which to continue dictionary update */
+	U32 hashLog3;	 /* dispatch table : larger == faster, more memory */
+	U32 loadedDictEnd;    /* index of end of dictionary */
+	U32 forceWindow;      /* force back-references to respect limit of 1<<wLog, even for dictionary */
+	U32 forceRawDict;     /* Force loading dictionary in "content-only" mode (no header analysis) */
+	ZSTD_compressionStage_e stage;
+	U32 rep[ZSTD_REP_NUM];
+	U32 repToConfirm[ZSTD_REP_NUM];
+	U32 dictID;
+	ZSTD_parameters params;
+	void *workSpace;
+	size_t workSpaceSize;
+	size_t blockSize;
+	U64 frameContentSize;
+	struct xxh64_state xxhState;
+	ZSTD_customMem customMem;
+
+	seqStore_t seqStore; /* sequences storage ptrs */
+	U32 *hashTable;
+	U32 *hashTable3;
+	U32 *chainTable;
+	HUF_CElt *hufTable;
+	U32 flagStaticTables;
+	HUF_repeat flagStaticHufTable;
+	FSE_CTable offcodeCTable[FSE_CTABLE_SIZE_U32(OffFSELog, MaxOff)];
+	FSE_CTable matchlengthCTable[FSE_CTABLE_SIZE_U32(MLFSELog, MaxML)];
+	FSE_CTable litlengthCTable[FSE_CTABLE_SIZE_U32(LLFSELog, MaxLL)];
+	unsigned tmpCounters[HUF_COMPRESS_WORKSPACE_SIZE_U32];
+};
+
+size_t ZSTD_CCtxWorkspaceBound(ZSTD_compressionParameters cParams)
+{
+	size_t const blockSize = MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, (size_t)1 << cParams.windowLog);
+	U32 const divider = (cParams.searchLength == 3) ? 3 : 4;
+	size_t const maxNbSeq = blockSize / divider;
+	size_t const tokenSpace = blockSize + 11 * maxNbSeq;
+	size_t const chainSize = (cParams.strategy == ZSTD_fast) ? 0 : (1 << cParams.chainLog);
+	size_t const hSize = ((size_t)1) << cParams.hashLog;
+	U32 const hashLog3 = (cParams.searchLength > 3) ? 0 : MIN(ZSTD_HASHLOG3_MAX, cParams.windowLog);
+	size_t const h3Size = ((size_t)1) << hashLog3;
+	size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
+	size_t const optSpace =
+	    ((MaxML + 1) + (MaxLL + 1) + (MaxOff + 1) + (1 << Litbits)) * sizeof(U32) + (ZSTD_OPT_NUM + 1) * (sizeof(ZSTD_match_t) + sizeof(ZSTD_optimal_t));
+	size_t const workspaceSize = tableSpace + (256 * sizeof(U32)) /* huffTable */ + tokenSpace +
+				     (((cParams.strategy == ZSTD_btopt) || (cParams.strategy == ZSTD_btopt2)) ? optSpace : 0);
+
+	return ZSTD_ALIGN(sizeof(ZSTD_stack)) + ZSTD_ALIGN(sizeof(ZSTD_CCtx)) + ZSTD_ALIGN(workspaceSize);
+}
+
+static ZSTD_CCtx *ZSTD_createCCtx_advanced(ZSTD_customMem customMem)
+{
+	ZSTD_CCtx *cctx;
+	if (!customMem.customAlloc || !customMem.customFree)
+		return NULL;
+	cctx = (ZSTD_CCtx *)ZSTD_malloc(sizeof(ZSTD_CCtx), customMem);
+	if (!cctx)
+		return NULL;
+	memset(cctx, 0, sizeof(ZSTD_CCtx));
+	cctx->customMem = customMem;
+	return cctx;
+}
+
+ZSTD_CCtx *ZSTD_initCCtx(void *workspace, size_t workspaceSize)
+{
+	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
+	ZSTD_CCtx *cctx = ZSTD_createCCtx_advanced(stackMem);
+	if (cctx) {
+		cctx->workSpace = ZSTD_stackAllocAll(cctx->customMem.opaque, &cctx->workSpaceSize);
+	}
+	return cctx;
+}
+
+size_t ZSTD_freeCCtx(ZSTD_CCtx *cctx)
+{
+	if (cctx == NULL)
+		return 0; /* support free on NULL */
+	ZSTD_free(cctx->workSpace, cctx->customMem);
+	ZSTD_free(cctx, cctx->customMem);
+	return 0; /* reserved as a potential error code in the future */
+}
+
+const seqStore_t *ZSTD_getSeqStore(const ZSTD_CCtx *ctx) /* hidden interface */ { return &(ctx->seqStore); }
+
+static ZSTD_parameters ZSTD_getParamsFromCCtx(const ZSTD_CCtx *cctx) { return cctx->params; }
+
+/** ZSTD_checkParams() :
+	ensure param values remain within authorized range.
+	@return : 0, or an error code if one value is beyond authorized range */
+size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams)
+{
+#define CLAMPCHECK(val, min, max)                                       \
+	{                                                               \
+		if ((val < min) | (val > max))                          \
+			return ERROR(compressionParameter_unsupported); \
+	}
+	CLAMPCHECK(cParams.windowLog, ZSTD_WINDOWLOG_MIN, ZSTD_WINDOWLOG_MAX);
+	CLAMPCHECK(cParams.chainLog, ZSTD_CHAINLOG_MIN, ZSTD_CHAINLOG_MAX);
+	CLAMPCHECK(cParams.hashLog, ZSTD_HASHLOG_MIN, ZSTD_HASHLOG_MAX);
+	CLAMPCHECK(cParams.searchLog, ZSTD_SEARCHLOG_MIN, ZSTD_SEARCHLOG_MAX);
+	CLAMPCHECK(cParams.searchLength, ZSTD_SEARCHLENGTH_MIN, ZSTD_SEARCHLENGTH_MAX);
+	CLAMPCHECK(cParams.targetLength, ZSTD_TARGETLENGTH_MIN, ZSTD_TARGETLENGTH_MAX);
+	if ((U32)(cParams.strategy) > (U32)ZSTD_btopt2)
+		return ERROR(compressionParameter_unsupported);
+	return 0;
+}
+
+/** ZSTD_cycleLog() :
+ *  condition for correct operation : hashLog > 1 */
+static U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat)
+{
+	U32 const btScale = ((U32)strat >= (U32)ZSTD_btlazy2);
+	return hashLog - btScale;
+}
+
+/** ZSTD_adjustCParams() :
+	optimize `cPar` for a given input (`srcSize` and `dictSize`).
+	mostly downsizing to reduce memory consumption and initialization.
+	Both `srcSize` and `dictSize` are optional (use 0 if unknown),
+	but if both are 0, no optimization can be done.
+	Note : cPar is considered validated at this stage. Use ZSTD_checkParams() to ensure that. */
+ZSTD_compressionParameters ZSTD_adjustCParams(ZSTD_compressionParameters cPar, unsigned long long srcSize, size_t dictSize)
+{
+	if (srcSize + dictSize == 0)
+		return cPar; /* no size information available : no adjustment */
+
+	/* resize params, to use less memory when necessary */
+	{
+		U32 const minSrcSize = (srcSize == 0) ? 500 : 0;
+		U64 const rSize = srcSize + dictSize + minSrcSize;
+		if (rSize < ((U64)1 << ZSTD_WINDOWLOG_MAX)) {
+			U32 const srcLog = MAX(ZSTD_HASHLOG_MIN, ZSTD_highbit32((U32)(rSize)-1) + 1);
+			if (cPar.windowLog > srcLog)
+				cPar.windowLog = srcLog;
+		}
+	}
+	if (cPar.hashLog > cPar.windowLog)
+		cPar.hashLog = cPar.windowLog;
+	{
+		U32 const cycleLog = ZSTD_cycleLog(cPar.chainLog, cPar.strategy);
+		if (cycleLog > cPar.windowLog)
+			cPar.chainLog -= (cycleLog - cPar.windowLog);
+	}
+
+	if (cPar.windowLog < ZSTD_WINDOWLOG_ABSOLUTEMIN)
+		cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN; /* required for frame header */
+
+	return cPar;
+}
+
+static U32 ZSTD_equivalentParams(ZSTD_parameters param1, ZSTD_parameters param2)
+{
+	return (param1.cParams.hashLog == param2.cParams.hashLog) & (param1.cParams.chainLog == param2.cParams.chainLog) &
+	       (param1.cParams.strategy == param2.cParams.strategy) & ((param1.cParams.searchLength == 3) == (param2.cParams.searchLength == 3));
+}
+
+/*! ZSTD_continueCCtx() :
+	reuse CCtx without reset (note : requires no dictionary) */
+static size_t ZSTD_continueCCtx(ZSTD_CCtx *cctx, ZSTD_parameters params, U64 frameContentSize)
+{
+	U32 const end = (U32)(cctx->nextSrc - cctx->base);
+	cctx->params = params;
+	cctx->frameContentSize = frameContentSize;
+	cctx->lowLimit = end;
+	cctx->dictLimit = end;
+	cctx->nextToUpdate = end + 1;
+	cctx->stage = ZSTDcs_init;
+	cctx->dictID = 0;
+	cctx->loadedDictEnd = 0;
+	{
+		int i;
+		for (i = 0; i < ZSTD_REP_NUM; i++)
+			cctx->rep[i] = repStartValue[i];
+	}
+	cctx->seqStore.litLengthSum = 0; /* force reset of btopt stats */
+	xxh64_reset(&cctx->xxhState, 0);
+	return 0;
+}
+
+typedef enum { ZSTDcrp_continue, ZSTDcrp_noMemset, ZSTDcrp_fullReset } ZSTD_compResetPolicy_e;
+
+/*! ZSTD_resetCCtx_advanced() :
+	note : `params` must be validated */
+static size_t ZSTD_resetCCtx_advanced(ZSTD_CCtx *zc, ZSTD_parameters params, U64 frameContentSize, ZSTD_compResetPolicy_e const crp)
+{
+	if (crp == ZSTDcrp_continue)
+		if (ZSTD_equivalentParams(params, zc->params)) {
+			zc->flagStaticTables = 0;
+			zc->flagStaticHufTable = HUF_repeat_none;
+			return ZSTD_continueCCtx(zc, params, frameContentSize);
+		}
+
+	{
+		size_t const blockSize = MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, (size_t)1 << params.cParams.windowLog);
+		U32 const divider = (params.cParams.searchLength == 3) ? 3 : 4;
+		size_t const maxNbSeq = blockSize / divider;
+		size_t const tokenSpace = blockSize + 11 * maxNbSeq;
+		size_t const chainSize = (params.cParams.strategy == ZSTD_fast) ? 0 : (1 << params.cParams.chainLog);
+		size_t const hSize = ((size_t)1) << params.cParams.hashLog;
+		U32 const hashLog3 = (params.cParams.searchLength > 3) ? 0 : MIN(ZSTD_HASHLOG3_MAX, params.cParams.windowLog);
+		size_t const h3Size = ((size_t)1) << hashLog3;
+		size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
+		void *ptr;
+
+		/* Check if workSpace is large enough, alloc a new one if needed */
+		{
+			size_t const optSpace = ((MaxML + 1) + (MaxLL + 1) + (MaxOff + 1) + (1 << Litbits)) * sizeof(U32) +
+						(ZSTD_OPT_NUM + 1) * (sizeof(ZSTD_match_t) + sizeof(ZSTD_optimal_t));
+			size_t const neededSpace = tableSpace + (256 * sizeof(U32)) /* huffTable */ + tokenSpace +
+						   (((params.cParams.strategy == ZSTD_btopt) || (params.cParams.strategy == ZSTD_btopt2)) ? optSpace : 0);
+			if (zc->workSpaceSize < neededSpace) {
+				ZSTD_free(zc->workSpace, zc->customMem);
+				zc->workSpace = ZSTD_malloc(neededSpace, zc->customMem);
+				if (zc->workSpace == NULL)
+					return ERROR(memory_allocation);
+				zc->workSpaceSize = neededSpace;
+			}
+		}
+
+		if (crp != ZSTDcrp_noMemset)
+			memset(zc->workSpace, 0, tableSpace); /* reset tables only */
+		xxh64_reset(&zc->xxhState, 0);
+		zc->hashLog3 = hashLog3;
+		zc->hashTable = (U32 *)(zc->workSpace);
+		zc->chainTable = zc->hashTable + hSize;
+		zc->hashTable3 = zc->chainTable + chainSize;
+		ptr = zc->hashTable3 + h3Size;
+		zc->hufTable = (HUF_CElt *)ptr;
+		zc->flagStaticTables = 0;
+		zc->flagStaticHufTable = HUF_repeat_none;
+		ptr = ((U32 *)ptr) + 256; /* note : HUF_CElt* is incomplete type, size is simulated using U32 */
+
+		zc->nextToUpdate = 1;
+		zc->nextSrc = NULL;
+		zc->base = NULL;
+		zc->dictBase = NULL;
+		zc->dictLimit = 0;
+		zc->lowLimit = 0;
+		zc->params = params;
+		zc->blockSize = blockSize;
+		zc->frameContentSize = frameContentSize;
+		{
+			int i;
+			for (i = 0; i < ZSTD_REP_NUM; i++)
+				zc->rep[i] = repStartValue[i];
+		}
+
+		if ((params.cParams.strategy == ZSTD_btopt) || (params.cParams.strategy == ZSTD_btopt2)) {
+			zc->seqStore.litFreq = (U32 *)ptr;
+			zc->seqStore.litLengthFreq = zc->seqStore.litFreq + (1 << Litbits);
+			zc->seqStore.matchLengthFreq = zc->seqStore.litLengthFreq + (MaxLL + 1);
+			zc->seqStore.offCodeFreq = zc->seqStore.matchLengthFreq + (MaxML + 1);
+			ptr = zc->seqStore.offCodeFreq + (MaxOff + 1);
+			zc->seqStore.matchTable = (ZSTD_match_t *)ptr;
+			ptr = zc->seqStore.matchTable + ZSTD_OPT_NUM + 1;
+			zc->seqStore.priceTable = (ZSTD_optimal_t *)ptr;
+			ptr = zc->seqStore.priceTable + ZSTD_OPT_NUM + 1;
+			zc->seqStore.litLengthSum = 0;
+		}
+		zc->seqStore.sequencesStart = (seqDef *)ptr;
+		ptr = zc->seqStore.sequencesStart + maxNbSeq;
+		zc->seqStore.llCode = (BYTE *)ptr;
+		zc->seqStore.mlCode = zc->seqStore.llCode + maxNbSeq;
+		zc->seqStore.ofCode = zc->seqStore.mlCode + maxNbSeq;
+		zc->seqStore.litStart = zc->seqStore.ofCode + maxNbSeq;
+
+		zc->stage = ZSTDcs_init;
+		zc->dictID = 0;
+		zc->loadedDictEnd = 0;
+
+		return 0;
+	}
+}
+
+/* ZSTD_invalidateRepCodes() :
+ * ensures next compression will not use repcodes from previous block.
+ * Note : only works with regular variant;
+ *        do not use with extDict variant ! */
+void ZSTD_invalidateRepCodes(ZSTD_CCtx *cctx)
+{
+	int i;
+	for (i = 0; i < ZSTD_REP_NUM; i++)
+		cctx->rep[i] = 0;
+}
+
+/*! ZSTD_copyCCtx() :
+*   Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
+*   Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
+*   @return : 0, or an error code */
+size_t ZSTD_copyCCtx(ZSTD_CCtx *dstCCtx, const ZSTD_CCtx *srcCCtx, unsigned long long pledgedSrcSize)
+{
+	if (srcCCtx->stage != ZSTDcs_init)
+		return ERROR(stage_wrong);
+
+	memcpy(&dstCCtx->customMem, &srcCCtx->customMem, sizeof(ZSTD_customMem));
+	{
+		ZSTD_parameters params = srcCCtx->params;
+		params.fParams.contentSizeFlag = (pledgedSrcSize > 0);
+		ZSTD_resetCCtx_advanced(dstCCtx, params, pledgedSrcSize, ZSTDcrp_noMemset);
+	}
+
+	/* copy tables */
+	{
+		size_t const chainSize = (srcCCtx->params.cParams.strategy == ZSTD_fast) ? 0 : (1 << srcCCtx->params.cParams.chainLog);
+		size_t const hSize = ((size_t)1) << srcCCtx->params.cParams.hashLog;
+		size_t const h3Size = (size_t)1 << srcCCtx->hashLog3;
+		size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
+		memcpy(dstCCtx->workSpace, srcCCtx->workSpace, tableSpace);
+	}
+
+	/* copy dictionary offsets */
+	dstCCtx->nextToUpdate = srcCCtx->nextToUpdate;
+	dstCCtx->nextToUpdate3 = srcCCtx->nextToUpdate3;
+	dstCCtx->nextSrc = srcCCtx->nextSrc;
+	dstCCtx->base = srcCCtx->base;
+	dstCCtx->dictBase = srcCCtx->dictBase;
+	dstCCtx->dictLimit = srcCCtx->dictLimit;
+	dstCCtx->lowLimit = srcCCtx->lowLimit;
+	dstCCtx->loadedDictEnd = srcCCtx->loadedDictEnd;
+	dstCCtx->dictID = srcCCtx->dictID;
+
+	/* copy entropy tables */
+	dstCCtx->flagStaticTables = srcCCtx->flagStaticTables;
+	dstCCtx->flagStaticHufTable = srcCCtx->flagStaticHufTable;
+	if (srcCCtx->flagStaticTables) {
+		memcpy(dstCCtx->litlengthCTable, srcCCtx->litlengthCTable, sizeof(dstCCtx->litlengthCTable));
+		memcpy(dstCCtx->matchlengthCTable, srcCCtx->matchlengthCTable, sizeof(dstCCtx->matchlengthCTable));
+		memcpy(dstCCtx->offcodeCTable, srcCCtx->offcodeCTable, sizeof(dstCCtx->offcodeCTable));
+	}
+	if (srcCCtx->flagStaticHufTable) {
+		memcpy(dstCCtx->hufTable, srcCCtx->hufTable, 256 * 4);
+	}
+
+	return 0;
+}
+
+/*! ZSTD_reduceTable() :
+*   reduce table indexes by `reducerValue` */
+static void ZSTD_reduceTable(U32 *const table, U32 const size, U32 const reducerValue)
+{
+	U32 u;
+	for (u = 0; u < size; u++) {
+		if (table[u] < reducerValue)
+			table[u] = 0;
+		else
+			table[u] -= reducerValue;
+	}
+}
+
+/*! ZSTD_reduceIndex() :
+*   rescale all indexes to avoid future overflow (indexes are U32) */
+static void ZSTD_reduceIndex(ZSTD_CCtx *zc, const U32 reducerValue)
+{
+	{
+		U32 const hSize = 1 << zc->params.cParams.hashLog;
+		ZSTD_reduceTable(zc->hashTable, hSize, reducerValue);
+	}
+
+	{
+		U32 const chainSize = (zc->params.cParams.strategy == ZSTD_fast) ? 0 : (1 << zc->params.cParams.chainLog);
+		ZSTD_reduceTable(zc->chainTable, chainSize, reducerValue);
+	}
+
+	{
+		U32 const h3Size = (zc->hashLog3) ? 1 << zc->hashLog3 : 0;
+		ZSTD_reduceTable(zc->hashTable3, h3Size, reducerValue);
+	}
+}
+
+/*-*******************************************************
+*  Block entropic compression
+*********************************************************/
+
+/* See doc/zstd_compression_format.md for detailed format description */
+
+size_t ZSTD_noCompressBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	if (srcSize + ZSTD_blockHeaderSize > dstCapacity)
+		return ERROR(dstSize_tooSmall);
+	memcpy((BYTE *)dst + ZSTD_blockHeaderSize, src, srcSize);
+	ZSTD_writeLE24(dst, (U32)(srcSize << 2) + (U32)bt_raw);
+	return ZSTD_blockHeaderSize + srcSize;
+}
+
+static size_t ZSTD_noCompressLiterals(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	BYTE *const ostart = (BYTE * const)dst;
+	U32 const flSize = 1 + (srcSize > 31) + (srcSize > 4095);
+
+	if (srcSize + flSize > dstCapacity)
+		return ERROR(dstSize_tooSmall);
+
+	switch (flSize) {
+	case 1: /* 2 - 1 - 5 */ ostart[0] = (BYTE)((U32)set_basic + (srcSize << 3)); break;
+	case 2: /* 2 - 2 - 12 */ ZSTD_writeLE16(ostart, (U16)((U32)set_basic + (1 << 2) + (srcSize << 4))); break;
+	default: /*note : should not be necessary : flSize is within {1,2,3} */
+	case 3: /* 2 - 2 - 20 */ ZSTD_writeLE32(ostart, (U32)((U32)set_basic + (3 << 2) + (srcSize << 4))); break;
+	}
+
+	memcpy(ostart + flSize, src, srcSize);
+	return srcSize + flSize;
+}
+
+static size_t ZSTD_compressRleLiteralsBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	BYTE *const ostart = (BYTE * const)dst;
+	U32 const flSize = 1 + (srcSize > 31) + (srcSize > 4095);
+
+	(void)dstCapacity; /* dstCapacity already guaranteed to be >=4, hence large enough */
+
+	switch (flSize) {
+	case 1: /* 2 - 1 - 5 */ ostart[0] = (BYTE)((U32)set_rle + (srcSize << 3)); break;
+	case 2: /* 2 - 2 - 12 */ ZSTD_writeLE16(ostart, (U16)((U32)set_rle + (1 << 2) + (srcSize << 4))); break;
+	default: /*note : should not be necessary : flSize is necessarily within {1,2,3} */
+	case 3: /* 2 - 2 - 20 */ ZSTD_writeLE32(ostart, (U32)((U32)set_rle + (3 << 2) + (srcSize << 4))); break;
+	}
+
+	ostart[flSize] = *(const BYTE *)src;
+	return flSize + 1;
+}
+
+static size_t ZSTD_minGain(size_t srcSize) { return (srcSize >> 6) + 2; }
+
+static size_t ZSTD_compressLiterals(ZSTD_CCtx *zc, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	size_t const minGain = ZSTD_minGain(srcSize);
+	size_t const lhSize = 3 + (srcSize >= 1 KB) + (srcSize >= 16 KB);
+	BYTE *const ostart = (BYTE *)dst;
+	U32 singleStream = srcSize < 256;
+	symbolEncodingType_e hType = set_compressed;
+	size_t cLitSize;
+
+/* small ? don't even attempt compression (speed opt) */
+#define LITERAL_NOENTROPY 63
+	{
+		size_t const minLitSize = zc->flagStaticHufTable == HUF_repeat_valid ? 6 : LITERAL_NOENTROPY;
+		if (srcSize <= minLitSize)
+			return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
+	}
+
+	if (dstCapacity < lhSize + 1)
+		return ERROR(dstSize_tooSmall); /* not enough space for compression */
+	{
+		HUF_repeat repeat = zc->flagStaticHufTable;
+		int const preferRepeat = zc->params.cParams.strategy < ZSTD_lazy ? srcSize <= 1024 : 0;
+		if (repeat == HUF_repeat_valid && lhSize == 3)
+			singleStream = 1;
+		cLitSize = singleStream ? HUF_compress1X_repeat(ostart + lhSize, dstCapacity - lhSize, src, srcSize, 255, 11, zc->tmpCounters,
+								sizeof(zc->tmpCounters), zc->hufTable, &repeat, preferRepeat)
+					: HUF_compress4X_repeat(ostart + lhSize, dstCapacity - lhSize, src, srcSize, 255, 11, zc->tmpCounters,
+								sizeof(zc->tmpCounters), zc->hufTable, &repeat, preferRepeat);
+		if (repeat != HUF_repeat_none) {
+			hType = set_repeat;
+		} /* reused the existing table */
+		else {
+			zc->flagStaticHufTable = HUF_repeat_check;
+		} /* now have a table to reuse */
+	}
+
+	if ((cLitSize == 0) | (cLitSize >= srcSize - minGain)) {
+		zc->flagStaticHufTable = HUF_repeat_none;
+		return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
+	}
+	if (cLitSize == 1) {
+		zc->flagStaticHufTable = HUF_repeat_none;
+		return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
+	}
+
+	/* Build header */
+	switch (lhSize) {
+	case 3: /* 2 - 2 - 10 - 10 */
+	{
+		U32 const lhc = hType + ((!singleStream) << 2) + ((U32)srcSize << 4) + ((U32)cLitSize << 14);
+		ZSTD_writeLE24(ostart, lhc);
+		break;
+	}
+	case 4: /* 2 - 2 - 14 - 14 */
+	{
+		U32 const lhc = hType + (2 << 2) + ((U32)srcSize << 4) + ((U32)cLitSize << 18);
+		ZSTD_writeLE32(ostart, lhc);
+		break;
+	}
+	default: /* should not be necessary, lhSize is only {3,4,5} */
+	case 5:  /* 2 - 2 - 18 - 18 */
+	{
+		U32 const lhc = hType + (3 << 2) + ((U32)srcSize << 4) + ((U32)cLitSize << 22);
+		ZSTD_writeLE32(ostart, lhc);
+		ostart[4] = (BYTE)(cLitSize >> 10);
+		break;
+	}
+	}
+	return lhSize + cLitSize;
+}
+
+static const BYTE LL_Code[64] = {0,  1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 16, 16, 17, 17, 18, 18,
+				 19, 19, 20, 20, 20, 20, 21, 21, 21, 21, 22, 22, 22, 22, 22, 22, 22, 22, 23, 23, 23, 23,
+				 23, 23, 23, 23, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 24};
+
+static const BYTE ML_Code[128] = {0,  1,  2,  3,  4,  5,  6,  7,  8,  9,  10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25,
+				  26, 27, 28, 29, 30, 31, 32, 32, 33, 33, 34, 34, 35, 35, 36, 36, 36, 36, 37, 37, 37, 37, 38, 38, 38, 38,
+				  38, 38, 38, 38, 39, 39, 39, 39, 39, 39, 39, 39, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40, 40,
+				  40, 40, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 41, 42, 42, 42, 42, 42, 42, 42, 42,
+				  42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42, 42};
+
+void ZSTD_seqToCodes(const seqStore_t *seqStorePtr)
+{
+	BYTE const LL_deltaCode = 19;
+	BYTE const ML_deltaCode = 36;
+	const seqDef *const sequences = seqStorePtr->sequencesStart;
+	BYTE *const llCodeTable = seqStorePtr->llCode;
+	BYTE *const ofCodeTable = seqStorePtr->ofCode;
+	BYTE *const mlCodeTable = seqStorePtr->mlCode;
+	U32 const nbSeq = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
+	U32 u;
+	for (u = 0; u < nbSeq; u++) {
+		U32 const llv = sequences[u].litLength;
+		U32 const mlv = sequences[u].matchLength;
+		llCodeTable[u] = (llv > 63) ? (BYTE)ZSTD_highbit32(llv) + LL_deltaCode : LL_Code[llv];
+		ofCodeTable[u] = (BYTE)ZSTD_highbit32(sequences[u].offset);
+		mlCodeTable[u] = (mlv > 127) ? (BYTE)ZSTD_highbit32(mlv) + ML_deltaCode : ML_Code[mlv];
+	}
+	if (seqStorePtr->longLengthID == 1)
+		llCodeTable[seqStorePtr->longLengthPos] = MaxLL;
+	if (seqStorePtr->longLengthID == 2)
+		mlCodeTable[seqStorePtr->longLengthPos] = MaxML;
+}
+
+ZSTD_STATIC size_t ZSTD_compressSequences_internal(ZSTD_CCtx *zc, void *dst, size_t dstCapacity)
+{
+	const int longOffsets = zc->params.cParams.windowLog > STREAM_ACCUMULATOR_MIN;
+	const seqStore_t *seqStorePtr = &(zc->seqStore);
+	FSE_CTable *CTable_LitLength = zc->litlengthCTable;
+	FSE_CTable *CTable_OffsetBits = zc->offcodeCTable;
+	FSE_CTable *CTable_MatchLength = zc->matchlengthCTable;
+	U32 LLtype, Offtype, MLtype; /* compressed, raw or rle */
+	const seqDef *const sequences = seqStorePtr->sequencesStart;
+	const BYTE *const ofCodeTable = seqStorePtr->ofCode;
+	const BYTE *const llCodeTable = seqStorePtr->llCode;
+	const BYTE *const mlCodeTable = seqStorePtr->mlCode;
+	BYTE *const ostart = (BYTE *)dst;
+	BYTE *const oend = ostart + dstCapacity;
+	BYTE *op = ostart;
+	size_t const nbSeq = seqStorePtr->sequences - seqStorePtr->sequencesStart;
+	BYTE *seqHead;
+
+	U32 *count;
+	S16 *norm;
+	U32 *workspace;
+	size_t workspaceSize = sizeof(zc->tmpCounters);
+	{
+		size_t spaceUsed32 = 0;
+		count = (U32 *)zc->tmpCounters + spaceUsed32;
+		spaceUsed32 += MaxSeq + 1;
+		norm = (S16 *)((U32 *)zc->tmpCounters + spaceUsed32);
+		spaceUsed32 += ALIGN(sizeof(S16) * (MaxSeq + 1), sizeof(U32)) >> 2;
+
+		workspace = (U32 *)zc->tmpCounters + spaceUsed32;
+		workspaceSize -= (spaceUsed32 << 2);
+	}
+
+	/* Compress literals */
+	{
+		const BYTE *const literals = seqStorePtr->litStart;
+		size_t const litSize = seqStorePtr->lit - literals;
+		size_t const cSize = ZSTD_compressLiterals(zc, op, dstCapacity, literals, litSize);
+		if (ZSTD_isError(cSize))
+			return cSize;
+		op += cSize;
+	}
+
+	/* Sequences Header */
+	if ((oend - op) < 3 /*max nbSeq Size*/ + 1 /*seqHead */)
+		return ERROR(dstSize_tooSmall);
+	if (nbSeq < 0x7F)
+		*op++ = (BYTE)nbSeq;
+	else if (nbSeq < LONGNBSEQ)
+		op[0] = (BYTE)((nbSeq >> 8) + 0x80), op[1] = (BYTE)nbSeq, op += 2;
+	else
+		op[0] = 0xFF, ZSTD_writeLE16(op + 1, (U16)(nbSeq - LONGNBSEQ)), op += 3;
+	if (nbSeq == 0)
+		return op - ostart;
+
+	/* seqHead : flags for FSE encoding type */
+	seqHead = op++;
+
+#define MIN_SEQ_FOR_DYNAMIC_FSE 64
+#define MAX_SEQ_FOR_STATIC_FSE 1000
+
+	/* convert length/distances into codes */
+	ZSTD_seqToCodes(seqStorePtr);
+
+	/* CTable for Literal Lengths */
+	{
+		U32 max = MaxLL;
+		size_t const mostFrequent = FSE_countFast_wksp(count, &max, llCodeTable, nbSeq, workspace);
+		if ((mostFrequent == nbSeq) && (nbSeq > 2)) {
+			*op++ = llCodeTable[0];
+			FSE_buildCTable_rle(CTable_LitLength, (BYTE)max);
+			LLtype = set_rle;
+		} else if ((zc->flagStaticTables) && (nbSeq < MAX_SEQ_FOR_STATIC_FSE)) {
+			LLtype = set_repeat;
+		} else if ((nbSeq < MIN_SEQ_FOR_DYNAMIC_FSE) || (mostFrequent < (nbSeq >> (LL_defaultNormLog - 1)))) {
+			FSE_buildCTable_wksp(CTable_LitLength, LL_defaultNorm, MaxLL, LL_defaultNormLog, workspace, workspaceSize);
+			LLtype = set_basic;
+		} else {
+			size_t nbSeq_1 = nbSeq;
+			const U32 tableLog = FSE_optimalTableLog(LLFSELog, nbSeq, max);
+			if (count[llCodeTable[nbSeq - 1]] > 1) {
+				count[llCodeTable[nbSeq - 1]]--;
+				nbSeq_1--;
+			}
+			FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max);
+			{
+				size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog); /* overflow protected */
+				if (FSE_isError(NCountSize))
+					return NCountSize;
+				op += NCountSize;
+			}
+			FSE_buildCTable_wksp(CTable_LitLength, norm, max, tableLog, workspace, workspaceSize);
+			LLtype = set_compressed;
+		}
+	}
+
+	/* CTable for Offsets */
+	{
+		U32 max = MaxOff;
+		size_t const mostFrequent = FSE_countFast_wksp(count, &max, ofCodeTable, nbSeq, workspace);
+		if ((mostFrequent == nbSeq) && (nbSeq > 2)) {
+			*op++ = ofCodeTable[0];
+			FSE_buildCTable_rle(CTable_OffsetBits, (BYTE)max);
+			Offtype = set_rle;
+		} else if ((zc->flagStaticTables) && (nbSeq < MAX_SEQ_FOR_STATIC_FSE)) {
+			Offtype = set_repeat;
+		} else if ((nbSeq < MIN_SEQ_FOR_DYNAMIC_FSE) || (mostFrequent < (nbSeq >> (OF_defaultNormLog - 1)))) {
+			FSE_buildCTable_wksp(CTable_OffsetBits, OF_defaultNorm, MaxOff, OF_defaultNormLog, workspace, workspaceSize);
+			Offtype = set_basic;
+		} else {
+			size_t nbSeq_1 = nbSeq;
+			const U32 tableLog = FSE_optimalTableLog(OffFSELog, nbSeq, max);
+			if (count[ofCodeTable[nbSeq - 1]] > 1) {
+				count[ofCodeTable[nbSeq - 1]]--;
+				nbSeq_1--;
+			}
+			FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max);
+			{
+				size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog); /* overflow protected */
+				if (FSE_isError(NCountSize))
+					return NCountSize;
+				op += NCountSize;
+			}
+			FSE_buildCTable_wksp(CTable_OffsetBits, norm, max, tableLog, workspace, workspaceSize);
+			Offtype = set_compressed;
+		}
+	}
+
+	/* CTable for MatchLengths */
+	{
+		U32 max = MaxML;
+		size_t const mostFrequent = FSE_countFast_wksp(count, &max, mlCodeTable, nbSeq, workspace);
+		if ((mostFrequent == nbSeq) && (nbSeq > 2)) {
+			*op++ = *mlCodeTable;
+			FSE_buildCTable_rle(CTable_MatchLength, (BYTE)max);
+			MLtype = set_rle;
+		} else if ((zc->flagStaticTables) && (nbSeq < MAX_SEQ_FOR_STATIC_FSE)) {
+			MLtype = set_repeat;
+		} else if ((nbSeq < MIN_SEQ_FOR_DYNAMIC_FSE) || (mostFrequent < (nbSeq >> (ML_defaultNormLog - 1)))) {
+			FSE_buildCTable_wksp(CTable_MatchLength, ML_defaultNorm, MaxML, ML_defaultNormLog, workspace, workspaceSize);
+			MLtype = set_basic;
+		} else {
+			size_t nbSeq_1 = nbSeq;
+			const U32 tableLog = FSE_optimalTableLog(MLFSELog, nbSeq, max);
+			if (count[mlCodeTable[nbSeq - 1]] > 1) {
+				count[mlCodeTable[nbSeq - 1]]--;
+				nbSeq_1--;
+			}
+			FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max);
+			{
+				size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog); /* overflow protected */
+				if (FSE_isError(NCountSize))
+					return NCountSize;
+				op += NCountSize;
+			}
+			FSE_buildCTable_wksp(CTable_MatchLength, norm, max, tableLog, workspace, workspaceSize);
+			MLtype = set_compressed;
+		}
+	}
+
+	*seqHead = (BYTE)((LLtype << 6) + (Offtype << 4) + (MLtype << 2));
+	zc->flagStaticTables = 0;
+
+	/* Encoding Sequences */
+	{
+		BIT_CStream_t blockStream;
+		FSE_CState_t stateMatchLength;
+		FSE_CState_t stateOffsetBits;
+		FSE_CState_t stateLitLength;
+
+		CHECK_E(BIT_initCStream(&blockStream, op, oend - op), dstSize_tooSmall); /* not enough space remaining */
+
+		/* first symbols */
+		FSE_initCState2(&stateMatchLength, CTable_MatchLength, mlCodeTable[nbSeq - 1]);
+		FSE_initCState2(&stateOffsetBits, CTable_OffsetBits, ofCodeTable[nbSeq - 1]);
+		FSE_initCState2(&stateLitLength, CTable_LitLength, llCodeTable[nbSeq - 1]);
+		BIT_addBits(&blockStream, sequences[nbSeq - 1].litLength, LL_bits[llCodeTable[nbSeq - 1]]);
+		if (ZSTD_32bits())
+			BIT_flushBits(&blockStream);
+		BIT_addBits(&blockStream, sequences[nbSeq - 1].matchLength, ML_bits[mlCodeTable[nbSeq - 1]]);
+		if (ZSTD_32bits())
+			BIT_flushBits(&blockStream);
+		if (longOffsets) {
+			U32 const ofBits = ofCodeTable[nbSeq - 1];
+			int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN - 1);
+			if (extraBits) {
+				BIT_addBits(&blockStream, sequences[nbSeq - 1].offset, extraBits);
+				BIT_flushBits(&blockStream);
+			}
+			BIT_addBits(&blockStream, sequences[nbSeq - 1].offset >> extraBits, ofBits - extraBits);
+		} else {
+			BIT_addBits(&blockStream, sequences[nbSeq - 1].offset, ofCodeTable[nbSeq - 1]);
+		}
+		BIT_flushBits(&blockStream);
+
+		{
+			size_t n;
+			for (n = nbSeq - 2; n < nbSeq; n--) { /* intentional underflow */
+				BYTE const llCode = llCodeTable[n];
+				BYTE const ofCode = ofCodeTable[n];
+				BYTE const mlCode = mlCodeTable[n];
+				U32 const llBits = LL_bits[llCode];
+				U32 const ofBits = ofCode; /* 32b*/ /* 64b*/
+				U32 const mlBits = ML_bits[mlCode];
+				/* (7)*/							    /* (7)*/
+				FSE_encodeSymbol(&blockStream, &stateOffsetBits, ofCode); /* 15 */  /* 15 */
+				FSE_encodeSymbol(&blockStream, &stateMatchLength, mlCode); /* 24 */ /* 24 */
+				if (ZSTD_32bits())
+					BIT_flushBits(&blockStream);				  /* (7)*/
+				FSE_encodeSymbol(&blockStream, &stateLitLength, llCode); /* 16 */ /* 33 */
+				if (ZSTD_32bits() || (ofBits + mlBits + llBits >= 64 - 7 - (LLFSELog + MLFSELog + OffFSELog)))
+					BIT_flushBits(&blockStream); /* (7)*/
+				BIT_addBits(&blockStream, sequences[n].litLength, llBits);
+				if (ZSTD_32bits() && ((llBits + mlBits) > 24))
+					BIT_flushBits(&blockStream);
+				BIT_addBits(&blockStream, sequences[n].matchLength, mlBits);
+				if (ZSTD_32bits())
+					BIT_flushBits(&blockStream); /* (7)*/
+				if (longOffsets) {
+					int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN - 1);
+					if (extraBits) {
+						BIT_addBits(&blockStream, sequences[n].offset, extraBits);
+						BIT_flushBits(&blockStream); /* (7)*/
+					}
+					BIT_addBits(&blockStream, sequences[n].offset >> extraBits, ofBits - extraBits); /* 31 */
+				} else {
+					BIT_addBits(&blockStream, sequences[n].offset, ofBits); /* 31 */
+				}
+				BIT_flushBits(&blockStream); /* (7)*/
+			}
+		}
+
+		FSE_flushCState(&blockStream, &stateMatchLength);
+		FSE_flushCState(&blockStream, &stateOffsetBits);
+		FSE_flushCState(&blockStream, &stateLitLength);
+
+		{
+			size_t const streamSize = BIT_closeCStream(&blockStream);
+			if (streamSize == 0)
+				return ERROR(dstSize_tooSmall); /* not enough space */
+			op += streamSize;
+		}
+	}
+	return op - ostart;
+}
+
+ZSTD_STATIC size_t ZSTD_compressSequences(ZSTD_CCtx *zc, void *dst, size_t dstCapacity, size_t srcSize)
+{
+	size_t const cSize = ZSTD_compressSequences_internal(zc, dst, dstCapacity);
+	size_t const minGain = ZSTD_minGain(srcSize);
+	size_t const maxCSize = srcSize - minGain;
+	/* If the srcSize <= dstCapacity, then there is enough space to write a
+	 * raw uncompressed block. Since we ran out of space, the block must not
+	 * be compressible, so fall back to a raw uncompressed block.
+	 */
+	int const uncompressibleError = cSize == ERROR(dstSize_tooSmall) && srcSize <= dstCapacity;
+	int i;
+
+	if (ZSTD_isError(cSize) && !uncompressibleError)
+		return cSize;
+	if (cSize >= maxCSize || uncompressibleError) {
+		zc->flagStaticHufTable = HUF_repeat_none;
+		return 0;
+	}
+	/* confirm repcodes */
+	for (i = 0; i < ZSTD_REP_NUM; i++)
+		zc->rep[i] = zc->repToConfirm[i];
+	return cSize;
+}
+
+/*! ZSTD_storeSeq() :
+	Store a sequence (literal length, literals, offset code and match length code) into seqStore_t.
+	`offsetCode` : distance to match, or 0 == repCode.
+	`matchCode` : matchLength - MINMATCH
+*/
+ZSTD_STATIC void ZSTD_storeSeq(seqStore_t *seqStorePtr, size_t litLength, const void *literals, U32 offsetCode, size_t matchCode)
+{
+	/* copy Literals */
+	ZSTD_wildcopy(seqStorePtr->lit, literals, litLength);
+	seqStorePtr->lit += litLength;
+
+	/* literal Length */
+	if (litLength > 0xFFFF) {
+		seqStorePtr->longLengthID = 1;
+		seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
+	}
+	seqStorePtr->sequences[0].litLength = (U16)litLength;
+
+	/* match offset */
+	seqStorePtr->sequences[0].offset = offsetCode + 1;
+
+	/* match Length */
+	if (matchCode > 0xFFFF) {
+		seqStorePtr->longLengthID = 2;
+		seqStorePtr->longLengthPos = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
+	}
+	seqStorePtr->sequences[0].matchLength = (U16)matchCode;
+
+	seqStorePtr->sequences++;
+}
+
+/*-*************************************
+*  Match length counter
+***************************************/
+static unsigned ZSTD_NbCommonBytes(register size_t val)
+{
+	if (ZSTD_isLittleEndian()) {
+		if (ZSTD_64bits()) {
+			return (__builtin_ctzll((U64)val) >> 3);
+		} else { /* 32 bits */
+			return (__builtin_ctz((U32)val) >> 3);
+		}
+	} else { /* Big Endian CPU */
+		if (ZSTD_64bits()) {
+			return (__builtin_clzll(val) >> 3);
+		} else { /* 32 bits */
+			return (__builtin_clz((U32)val) >> 3);
+		}
+	}
+}
+
+static size_t ZSTD_count(const BYTE *pIn, const BYTE *pMatch, const BYTE *const pInLimit)
+{
+	const BYTE *const pStart = pIn;
+	const BYTE *const pInLoopLimit = pInLimit - (sizeof(size_t) - 1);
+
+	while (pIn < pInLoopLimit) {
+		size_t const diff = ZSTD_readST(pMatch) ^ ZSTD_readST(pIn);
+		if (!diff) {
+			pIn += sizeof(size_t);
+			pMatch += sizeof(size_t);
+			continue;
+		}
+		pIn += ZSTD_NbCommonBytes(diff);
+		return (size_t)(pIn - pStart);
+	}
+	if (ZSTD_64bits())
+		if ((pIn < (pInLimit - 3)) && (ZSTD_read32(pMatch) == ZSTD_read32(pIn))) {
+			pIn += 4;
+			pMatch += 4;
+		}
+	if ((pIn < (pInLimit - 1)) && (ZSTD_read16(pMatch) == ZSTD_read16(pIn))) {
+		pIn += 2;
+		pMatch += 2;
+	}
+	if ((pIn < pInLimit) && (*pMatch == *pIn))
+		pIn++;
+	return (size_t)(pIn - pStart);
+}
+
+/** ZSTD_count_2segments() :
+*   can count match length with `ip` & `match` in 2 different segments.
+*   convention : on reaching mEnd, match count continue starting from iStart
+*/
+static size_t ZSTD_count_2segments(const BYTE *ip, const BYTE *match, const BYTE *iEnd, const BYTE *mEnd, const BYTE *iStart)
+{
+	const BYTE *const vEnd = MIN(ip + (mEnd - match), iEnd);
+	size_t const matchLength = ZSTD_count(ip, match, vEnd);
+	if (match + matchLength != mEnd)
+		return matchLength;
+	return matchLength + ZSTD_count(ip + matchLength, iStart, iEnd);
+}
+
+/*-*************************************
+*  Hashes
+***************************************/
+static const U32 prime3bytes = 506832829U;
+static U32 ZSTD_hash3(U32 u, U32 h) { return ((u << (32 - 24)) * prime3bytes) >> (32 - h); }
+ZSTD_STATIC size_t ZSTD_hash3Ptr(const void *ptr, U32 h) { return ZSTD_hash3(ZSTD_readLE32(ptr), h); } /* only in zstd_opt.h */
+
+static const U32 prime4bytes = 2654435761U;
+static U32 ZSTD_hash4(U32 u, U32 h) { return (u * prime4bytes) >> (32 - h); }
+static size_t ZSTD_hash4Ptr(const void *ptr, U32 h) { return ZSTD_hash4(ZSTD_read32(ptr), h); }
+
+static const U64 prime5bytes = 889523592379ULL;
+static size_t ZSTD_hash5(U64 u, U32 h) { return (size_t)(((u << (64 - 40)) * prime5bytes) >> (64 - h)); }
+static size_t ZSTD_hash5Ptr(const void *p, U32 h) { return ZSTD_hash5(ZSTD_readLE64(p), h); }
+
+static const U64 prime6bytes = 227718039650203ULL;
+static size_t ZSTD_hash6(U64 u, U32 h) { return (size_t)(((u << (64 - 48)) * prime6bytes) >> (64 - h)); }
+static size_t ZSTD_hash6Ptr(const void *p, U32 h) { return ZSTD_hash6(ZSTD_readLE64(p), h); }
+
+static const U64 prime7bytes = 58295818150454627ULL;
+static size_t ZSTD_hash7(U64 u, U32 h) { return (size_t)(((u << (64 - 56)) * prime7bytes) >> (64 - h)); }
+static size_t ZSTD_hash7Ptr(const void *p, U32 h) { return ZSTD_hash7(ZSTD_readLE64(p), h); }
+
+static const U64 prime8bytes = 0xCF1BBCDCB7A56463ULL;
+static size_t ZSTD_hash8(U64 u, U32 h) { return (size_t)(((u)*prime8bytes) >> (64 - h)); }
+static size_t ZSTD_hash8Ptr(const void *p, U32 h) { return ZSTD_hash8(ZSTD_readLE64(p), h); }
+
+static size_t ZSTD_hashPtr(const void *p, U32 hBits, U32 mls)
+{
+	switch (mls) {
+	// case 3: return ZSTD_hash3Ptr(p, hBits);
+	default:
+	case 4: return ZSTD_hash4Ptr(p, hBits);
+	case 5: return ZSTD_hash5Ptr(p, hBits);
+	case 6: return ZSTD_hash6Ptr(p, hBits);
+	case 7: return ZSTD_hash7Ptr(p, hBits);
+	case 8: return ZSTD_hash8Ptr(p, hBits);
+	}
+}
+
+/*-*************************************
+*  Fast Scan
+***************************************/
+static void ZSTD_fillHashTable(ZSTD_CCtx *zc, const void *end, const U32 mls)
+{
+	U32 *const hashTable = zc->hashTable;
+	U32 const hBits = zc->params.cParams.hashLog;
+	const BYTE *const base = zc->base;
+	const BYTE *ip = base + zc->nextToUpdate;
+	const BYTE *const iend = ((const BYTE *)end) - HASH_READ_SIZE;
+	const size_t fastHashFillStep = 3;
+
+	while (ip <= iend) {
+		hashTable[ZSTD_hashPtr(ip, hBits, mls)] = (U32)(ip - base);
+		ip += fastHashFillStep;
+	}
+}
+
+FORCE_INLINE
+void ZSTD_compressBlock_fast_generic(ZSTD_CCtx *cctx, const void *src, size_t srcSize, const U32 mls)
+{
+	U32 *const hashTable = cctx->hashTable;
+	U32 const hBits = cctx->params.cParams.hashLog;
+	seqStore_t *seqStorePtr = &(cctx->seqStore);
+	const BYTE *const base = cctx->base;
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *ip = istart;
+	const BYTE *anchor = istart;
+	const U32 lowestIndex = cctx->dictLimit;
+	const BYTE *const lowest = base + lowestIndex;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *const ilimit = iend - HASH_READ_SIZE;
+	U32 offset_1 = cctx->rep[0], offset_2 = cctx->rep[1];
+	U32 offsetSaved = 0;
+
+	/* init */
+	ip += (ip == lowest);
+	{
+		U32 const maxRep = (U32)(ip - lowest);
+		if (offset_2 > maxRep)
+			offsetSaved = offset_2, offset_2 = 0;
+		if (offset_1 > maxRep)
+			offsetSaved = offset_1, offset_1 = 0;
+	}
+
+	/* Main Search Loop */
+	while (ip < ilimit) { /* < instead of <=, because repcode check at (ip+1) */
+		size_t mLength;
+		size_t const h = ZSTD_hashPtr(ip, hBits, mls);
+		U32 const curr = (U32)(ip - base);
+		U32 const matchIndex = hashTable[h];
+		const BYTE *match = base + matchIndex;
+		hashTable[h] = curr; /* update hash table */
+
+		if ((offset_1 > 0) & (ZSTD_read32(ip + 1 - offset_1) == ZSTD_read32(ip + 1))) {
+			mLength = ZSTD_count(ip + 1 + 4, ip + 1 + 4 - offset_1, iend) + 4;
+			ip++;
+			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, 0, mLength - MINMATCH);
+		} else {
+			U32 offset;
+			if ((matchIndex <= lowestIndex) || (ZSTD_read32(match) != ZSTD_read32(ip))) {
+				ip += ((ip - anchor) >> g_searchStrength) + 1;
+				continue;
+			}
+			mLength = ZSTD_count(ip + 4, match + 4, iend) + 4;
+			offset = (U32)(ip - match);
+			while (((ip > anchor) & (match > lowest)) && (ip[-1] == match[-1])) {
+				ip--;
+				match--;
+				mLength++;
+			} /* catch up */
+			offset_2 = offset_1;
+			offset_1 = offset;
+
+			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
+		}
+
+		/* match found */
+		ip += mLength;
+		anchor = ip;
+
+		if (ip <= ilimit) {
+			/* Fill Table */
+			hashTable[ZSTD_hashPtr(base + curr + 2, hBits, mls)] = curr + 2; /* here because curr+2 could be > iend-8 */
+			hashTable[ZSTD_hashPtr(ip - 2, hBits, mls)] = (U32)(ip - 2 - base);
+			/* check immediate repcode */
+			while ((ip <= ilimit) && ((offset_2 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_2)))) {
+				/* store sequence */
+				size_t const rLength = ZSTD_count(ip + 4, ip + 4 - offset_2, iend) + 4;
+				{
+					U32 const tmpOff = offset_2;
+					offset_2 = offset_1;
+					offset_1 = tmpOff;
+				} /* swap offset_2 <=> offset_1 */
+				hashTable[ZSTD_hashPtr(ip, hBits, mls)] = (U32)(ip - base);
+				ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, rLength - MINMATCH);
+				ip += rLength;
+				anchor = ip;
+				continue; /* faster when present ... (?) */
+			}
+		}
+	}
+
+	/* save reps for next block */
+	cctx->repToConfirm[0] = offset_1 ? offset_1 : offsetSaved;
+	cctx->repToConfirm[1] = offset_2 ? offset_2 : offsetSaved;
+
+	/* Last Literals */
+	{
+		size_t const lastLLSize = iend - anchor;
+		memcpy(seqStorePtr->lit, anchor, lastLLSize);
+		seqStorePtr->lit += lastLLSize;
+	}
+}
+
+static void ZSTD_compressBlock_fast(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+	const U32 mls = ctx->params.cParams.searchLength;
+	switch (mls) {
+	default: /* includes case 3 */
+	case 4: ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 4); return;
+	case 5: ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 5); return;
+	case 6: ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 6); return;
+	case 7: ZSTD_compressBlock_fast_generic(ctx, src, srcSize, 7); return;
+	}
+}
+
+static void ZSTD_compressBlock_fast_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const U32 mls)
+{
+	U32 *hashTable = ctx->hashTable;
+	const U32 hBits = ctx->params.cParams.hashLog;
+	seqStore_t *seqStorePtr = &(ctx->seqStore);
+	const BYTE *const base = ctx->base;
+	const BYTE *const dictBase = ctx->dictBase;
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *ip = istart;
+	const BYTE *anchor = istart;
+	const U32 lowestIndex = ctx->lowLimit;
+	const BYTE *const dictStart = dictBase + lowestIndex;
+	const U32 dictLimit = ctx->dictLimit;
+	const BYTE *const lowPrefixPtr = base + dictLimit;
+	const BYTE *const dictEnd = dictBase + dictLimit;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *const ilimit = iend - 8;
+	U32 offset_1 = ctx->rep[0], offset_2 = ctx->rep[1];
+
+	/* Search Loop */
+	while (ip < ilimit) { /* < instead of <=, because (ip+1) */
+		const size_t h = ZSTD_hashPtr(ip, hBits, mls);
+		const U32 matchIndex = hashTable[h];
+		const BYTE *matchBase = matchIndex < dictLimit ? dictBase : base;
+		const BYTE *match = matchBase + matchIndex;
+		const U32 curr = (U32)(ip - base);
+		const U32 repIndex = curr + 1 - offset_1; /* offset_1 expected <= curr +1 */
+		const BYTE *repBase = repIndex < dictLimit ? dictBase : base;
+		const BYTE *repMatch = repBase + repIndex;
+		size_t mLength;
+		hashTable[h] = curr; /* update hash table */
+
+		if ((((U32)((dictLimit - 1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > lowestIndex)) &&
+		    (ZSTD_read32(repMatch) == ZSTD_read32(ip + 1))) {
+			const BYTE *repMatchEnd = repIndex < dictLimit ? dictEnd : iend;
+			mLength = ZSTD_count_2segments(ip + 1 + EQUAL_READ32, repMatch + EQUAL_READ32, iend, repMatchEnd, lowPrefixPtr) + EQUAL_READ32;
+			ip++;
+			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, 0, mLength - MINMATCH);
+		} else {
+			if ((matchIndex < lowestIndex) || (ZSTD_read32(match) != ZSTD_read32(ip))) {
+				ip += ((ip - anchor) >> g_searchStrength) + 1;
+				continue;
+			}
+			{
+				const BYTE *matchEnd = matchIndex < dictLimit ? dictEnd : iend;
+				const BYTE *lowMatchPtr = matchIndex < dictLimit ? dictStart : lowPrefixPtr;
+				U32 offset;
+				mLength = ZSTD_count_2segments(ip + EQUAL_READ32, match + EQUAL_READ32, iend, matchEnd, lowPrefixPtr) + EQUAL_READ32;
+				while (((ip > anchor) & (match > lowMatchPtr)) && (ip[-1] == match[-1])) {
+					ip--;
+					match--;
+					mLength++;
+				} /* catch up */
+				offset = curr - matchIndex;
+				offset_2 = offset_1;
+				offset_1 = offset;
+				ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
+			}
+		}
+
+		/* found a match : store it */
+		ip += mLength;
+		anchor = ip;
+
+		if (ip <= ilimit) {
+			/* Fill Table */
+			hashTable[ZSTD_hashPtr(base + curr + 2, hBits, mls)] = curr + 2;
+			hashTable[ZSTD_hashPtr(ip - 2, hBits, mls)] = (U32)(ip - 2 - base);
+			/* check immediate repcode */
+			while (ip <= ilimit) {
+				U32 const curr2 = (U32)(ip - base);
+				U32 const repIndex2 = curr2 - offset_2;
+				const BYTE *repMatch2 = repIndex2 < dictLimit ? dictBase + repIndex2 : base + repIndex2;
+				if ((((U32)((dictLimit - 1) - repIndex2) >= 3) & (repIndex2 > lowestIndex)) /* intentional overflow */
+				    && (ZSTD_read32(repMatch2) == ZSTD_read32(ip))) {
+					const BYTE *const repEnd2 = repIndex2 < dictLimit ? dictEnd : iend;
+					size_t repLength2 =
+					    ZSTD_count_2segments(ip + EQUAL_READ32, repMatch2 + EQUAL_READ32, iend, repEnd2, lowPrefixPtr) + EQUAL_READ32;
+					U32 tmpOffset = offset_2;
+					offset_2 = offset_1;
+					offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
+					ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, repLength2 - MINMATCH);
+					hashTable[ZSTD_hashPtr(ip, hBits, mls)] = curr2;
+					ip += repLength2;
+					anchor = ip;
+					continue;
+				}
+				break;
+			}
+		}
+	}
+
+	/* save reps for next block */
+	ctx->repToConfirm[0] = offset_1;
+	ctx->repToConfirm[1] = offset_2;
+
+	/* Last Literals */
+	{
+		size_t const lastLLSize = iend - anchor;
+		memcpy(seqStorePtr->lit, anchor, lastLLSize);
+		seqStorePtr->lit += lastLLSize;
+	}
+}
+
+static void ZSTD_compressBlock_fast_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+	U32 const mls = ctx->params.cParams.searchLength;
+	switch (mls) {
+	default: /* includes case 3 */
+	case 4: ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 4); return;
+	case 5: ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 5); return;
+	case 6: ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 6); return;
+	case 7: ZSTD_compressBlock_fast_extDict_generic(ctx, src, srcSize, 7); return;
+	}
+}
+
+/*-*************************************
+*  Double Fast
+***************************************/
+static void ZSTD_fillDoubleHashTable(ZSTD_CCtx *cctx, const void *end, const U32 mls)
+{
+	U32 *const hashLarge = cctx->hashTable;
+	U32 const hBitsL = cctx->params.cParams.hashLog;
+	U32 *const hashSmall = cctx->chainTable;
+	U32 const hBitsS = cctx->params.cParams.chainLog;
+	const BYTE *const base = cctx->base;
+	const BYTE *ip = base + cctx->nextToUpdate;
+	const BYTE *const iend = ((const BYTE *)end) - HASH_READ_SIZE;
+	const size_t fastHashFillStep = 3;
+
+	while (ip <= iend) {
+		hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip - base);
+		hashLarge[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip - base);
+		ip += fastHashFillStep;
+	}
+}
+
+FORCE_INLINE
+void ZSTD_compressBlock_doubleFast_generic(ZSTD_CCtx *cctx, const void *src, size_t srcSize, const U32 mls)
+{
+	U32 *const hashLong = cctx->hashTable;
+	const U32 hBitsL = cctx->params.cParams.hashLog;
+	U32 *const hashSmall = cctx->chainTable;
+	const U32 hBitsS = cctx->params.cParams.chainLog;
+	seqStore_t *seqStorePtr = &(cctx->seqStore);
+	const BYTE *const base = cctx->base;
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *ip = istart;
+	const BYTE *anchor = istart;
+	const U32 lowestIndex = cctx->dictLimit;
+	const BYTE *const lowest = base + lowestIndex;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *const ilimit = iend - HASH_READ_SIZE;
+	U32 offset_1 = cctx->rep[0], offset_2 = cctx->rep[1];
+	U32 offsetSaved = 0;
+
+	/* init */
+	ip += (ip == lowest);
+	{
+		U32 const maxRep = (U32)(ip - lowest);
+		if (offset_2 > maxRep)
+			offsetSaved = offset_2, offset_2 = 0;
+		if (offset_1 > maxRep)
+			offsetSaved = offset_1, offset_1 = 0;
+	}
+
+	/* Main Search Loop */
+	while (ip < ilimit) { /* < instead of <=, because repcode check at (ip+1) */
+		size_t mLength;
+		size_t const h2 = ZSTD_hashPtr(ip, hBitsL, 8);
+		size_t const h = ZSTD_hashPtr(ip, hBitsS, mls);
+		U32 const curr = (U32)(ip - base);
+		U32 const matchIndexL = hashLong[h2];
+		U32 const matchIndexS = hashSmall[h];
+		const BYTE *matchLong = base + matchIndexL;
+		const BYTE *match = base + matchIndexS;
+		hashLong[h2] = hashSmall[h] = curr; /* update hash tables */
+
+		if ((offset_1 > 0) & (ZSTD_read32(ip + 1 - offset_1) == ZSTD_read32(ip + 1))) { /* note : by construction, offset_1 <= curr */
+			mLength = ZSTD_count(ip + 1 + 4, ip + 1 + 4 - offset_1, iend) + 4;
+			ip++;
+			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, 0, mLength - MINMATCH);
+		} else {
+			U32 offset;
+			if ((matchIndexL > lowestIndex) && (ZSTD_read64(matchLong) == ZSTD_read64(ip))) {
+				mLength = ZSTD_count(ip + 8, matchLong + 8, iend) + 8;
+				offset = (U32)(ip - matchLong);
+				while (((ip > anchor) & (matchLong > lowest)) && (ip[-1] == matchLong[-1])) {
+					ip--;
+					matchLong--;
+					mLength++;
+				} /* catch up */
+			} else if ((matchIndexS > lowestIndex) && (ZSTD_read32(match) == ZSTD_read32(ip))) {
+				size_t const h3 = ZSTD_hashPtr(ip + 1, hBitsL, 8);
+				U32 const matchIndex3 = hashLong[h3];
+				const BYTE *match3 = base + matchIndex3;
+				hashLong[h3] = curr + 1;
+				if ((matchIndex3 > lowestIndex) && (ZSTD_read64(match3) == ZSTD_read64(ip + 1))) {
+					mLength = ZSTD_count(ip + 9, match3 + 8, iend) + 8;
+					ip++;
+					offset = (U32)(ip - match3);
+					while (((ip > anchor) & (match3 > lowest)) && (ip[-1] == match3[-1])) {
+						ip--;
+						match3--;
+						mLength++;
+					} /* catch up */
+				} else {
+					mLength = ZSTD_count(ip + 4, match + 4, iend) + 4;
+					offset = (U32)(ip - match);
+					while (((ip > anchor) & (match > lowest)) && (ip[-1] == match[-1])) {
+						ip--;
+						match--;
+						mLength++;
+					} /* catch up */
+				}
+			} else {
+				ip += ((ip - anchor) >> g_searchStrength) + 1;
+				continue;
+			}
+
+			offset_2 = offset_1;
+			offset_1 = offset;
+
+			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
+		}
+
+		/* match found */
+		ip += mLength;
+		anchor = ip;
+
+		if (ip <= ilimit) {
+			/* Fill Table */
+			hashLong[ZSTD_hashPtr(base + curr + 2, hBitsL, 8)] = hashSmall[ZSTD_hashPtr(base + curr + 2, hBitsS, mls)] =
+			    curr + 2; /* here because curr+2 could be > iend-8 */
+			hashLong[ZSTD_hashPtr(ip - 2, hBitsL, 8)] = hashSmall[ZSTD_hashPtr(ip - 2, hBitsS, mls)] = (U32)(ip - 2 - base);
+
+			/* check immediate repcode */
+			while ((ip <= ilimit) && ((offset_2 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_2)))) {
+				/* store sequence */
+				size_t const rLength = ZSTD_count(ip + 4, ip + 4 - offset_2, iend) + 4;
+				{
+					U32 const tmpOff = offset_2;
+					offset_2 = offset_1;
+					offset_1 = tmpOff;
+				} /* swap offset_2 <=> offset_1 */
+				hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip - base);
+				hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip - base);
+				ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, rLength - MINMATCH);
+				ip += rLength;
+				anchor = ip;
+				continue; /* faster when present ... (?) */
+			}
+		}
+	}
+
+	/* save reps for next block */
+	cctx->repToConfirm[0] = offset_1 ? offset_1 : offsetSaved;
+	cctx->repToConfirm[1] = offset_2 ? offset_2 : offsetSaved;
+
+	/* Last Literals */
+	{
+		size_t const lastLLSize = iend - anchor;
+		memcpy(seqStorePtr->lit, anchor, lastLLSize);
+		seqStorePtr->lit += lastLLSize;
+	}
+}
+
+static void ZSTD_compressBlock_doubleFast(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+	const U32 mls = ctx->params.cParams.searchLength;
+	switch (mls) {
+	default: /* includes case 3 */
+	case 4: ZSTD_compressBlock_doubleFast_generic(ctx, src, srcSize, 4); return;
+	case 5: ZSTD_compressBlock_doubleFast_generic(ctx, src, srcSize, 5); return;
+	case 6: ZSTD_compressBlock_doubleFast_generic(ctx, src, srcSize, 6); return;
+	case 7: ZSTD_compressBlock_doubleFast_generic(ctx, src, srcSize, 7); return;
+	}
+}
+
+static void ZSTD_compressBlock_doubleFast_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const U32 mls)
+{
+	U32 *const hashLong = ctx->hashTable;
+	U32 const hBitsL = ctx->params.cParams.hashLog;
+	U32 *const hashSmall = ctx->chainTable;
+	U32 const hBitsS = ctx->params.cParams.chainLog;
+	seqStore_t *seqStorePtr = &(ctx->seqStore);
+	const BYTE *const base = ctx->base;
+	const BYTE *const dictBase = ctx->dictBase;
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *ip = istart;
+	const BYTE *anchor = istart;
+	const U32 lowestIndex = ctx->lowLimit;
+	const BYTE *const dictStart = dictBase + lowestIndex;
+	const U32 dictLimit = ctx->dictLimit;
+	const BYTE *const lowPrefixPtr = base + dictLimit;
+	const BYTE *const dictEnd = dictBase + dictLimit;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *const ilimit = iend - 8;
+	U32 offset_1 = ctx->rep[0], offset_2 = ctx->rep[1];
+
+	/* Search Loop */
+	while (ip < ilimit) { /* < instead of <=, because (ip+1) */
+		const size_t hSmall = ZSTD_hashPtr(ip, hBitsS, mls);
+		const U32 matchIndex = hashSmall[hSmall];
+		const BYTE *matchBase = matchIndex < dictLimit ? dictBase : base;
+		const BYTE *match = matchBase + matchIndex;
+
+		const size_t hLong = ZSTD_hashPtr(ip, hBitsL, 8);
+		const U32 matchLongIndex = hashLong[hLong];
+		const BYTE *matchLongBase = matchLongIndex < dictLimit ? dictBase : base;
+		const BYTE *matchLong = matchLongBase + matchLongIndex;
+
+		const U32 curr = (U32)(ip - base);
+		const U32 repIndex = curr + 1 - offset_1; /* offset_1 expected <= curr +1 */
+		const BYTE *repBase = repIndex < dictLimit ? dictBase : base;
+		const BYTE *repMatch = repBase + repIndex;
+		size_t mLength;
+		hashSmall[hSmall] = hashLong[hLong] = curr; /* update hash table */
+
+		if ((((U32)((dictLimit - 1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > lowestIndex)) &&
+		    (ZSTD_read32(repMatch) == ZSTD_read32(ip + 1))) {
+			const BYTE *repMatchEnd = repIndex < dictLimit ? dictEnd : iend;
+			mLength = ZSTD_count_2segments(ip + 1 + 4, repMatch + 4, iend, repMatchEnd, lowPrefixPtr) + 4;
+			ip++;
+			ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, 0, mLength - MINMATCH);
+		} else {
+			if ((matchLongIndex > lowestIndex) && (ZSTD_read64(matchLong) == ZSTD_read64(ip))) {
+				const BYTE *matchEnd = matchLongIndex < dictLimit ? dictEnd : iend;
+				const BYTE *lowMatchPtr = matchLongIndex < dictLimit ? dictStart : lowPrefixPtr;
+				U32 offset;
+				mLength = ZSTD_count_2segments(ip + 8, matchLong + 8, iend, matchEnd, lowPrefixPtr) + 8;
+				offset = curr - matchLongIndex;
+				while (((ip > anchor) & (matchLong > lowMatchPtr)) && (ip[-1] == matchLong[-1])) {
+					ip--;
+					matchLong--;
+					mLength++;
+				} /* catch up */
+				offset_2 = offset_1;
+				offset_1 = offset;
+				ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
+
+			} else if ((matchIndex > lowestIndex) && (ZSTD_read32(match) == ZSTD_read32(ip))) {
+				size_t const h3 = ZSTD_hashPtr(ip + 1, hBitsL, 8);
+				U32 const matchIndex3 = hashLong[h3];
+				const BYTE *const match3Base = matchIndex3 < dictLimit ? dictBase : base;
+				const BYTE *match3 = match3Base + matchIndex3;
+				U32 offset;
+				hashLong[h3] = curr + 1;
+				if ((matchIndex3 > lowestIndex) && (ZSTD_read64(match3) == ZSTD_read64(ip + 1))) {
+					const BYTE *matchEnd = matchIndex3 < dictLimit ? dictEnd : iend;
+					const BYTE *lowMatchPtr = matchIndex3 < dictLimit ? dictStart : lowPrefixPtr;
+					mLength = ZSTD_count_2segments(ip + 9, match3 + 8, iend, matchEnd, lowPrefixPtr) + 8;
+					ip++;
+					offset = curr + 1 - matchIndex3;
+					while (((ip > anchor) & (match3 > lowMatchPtr)) && (ip[-1] == match3[-1])) {
+						ip--;
+						match3--;
+						mLength++;
+					} /* catch up */
+				} else {
+					const BYTE *matchEnd = matchIndex < dictLimit ? dictEnd : iend;
+					const BYTE *lowMatchPtr = matchIndex < dictLimit ? dictStart : lowPrefixPtr;
+					mLength = ZSTD_count_2segments(ip + 4, match + 4, iend, matchEnd, lowPrefixPtr) + 4;
+					offset = curr - matchIndex;
+					while (((ip > anchor) & (match > lowMatchPtr)) && (ip[-1] == match[-1])) {
+						ip--;
+						match--;
+						mLength++;
+					} /* catch up */
+				}
+				offset_2 = offset_1;
+				offset_1 = offset;
+				ZSTD_storeSeq(seqStorePtr, ip - anchor, anchor, offset + ZSTD_REP_MOVE, mLength - MINMATCH);
+
+			} else {
+				ip += ((ip - anchor) >> g_searchStrength) + 1;
+				continue;
+			}
+		}
+
+		/* found a match : store it */
+		ip += mLength;
+		anchor = ip;
+
+		if (ip <= ilimit) {
+			/* Fill Table */
+			hashSmall[ZSTD_hashPtr(base + curr + 2, hBitsS, mls)] = curr + 2;
+			hashLong[ZSTD_hashPtr(base + curr + 2, hBitsL, 8)] = curr + 2;
+			hashSmall[ZSTD_hashPtr(ip - 2, hBitsS, mls)] = (U32)(ip - 2 - base);
+			hashLong[ZSTD_hashPtr(ip - 2, hBitsL, 8)] = (U32)(ip - 2 - base);
+			/* check immediate repcode */
+			while (ip <= ilimit) {
+				U32 const curr2 = (U32)(ip - base);
+				U32 const repIndex2 = curr2 - offset_2;
+				const BYTE *repMatch2 = repIndex2 < dictLimit ? dictBase + repIndex2 : base + repIndex2;
+				if ((((U32)((dictLimit - 1) - repIndex2) >= 3) & (repIndex2 > lowestIndex)) /* intentional overflow */
+				    && (ZSTD_read32(repMatch2) == ZSTD_read32(ip))) {
+					const BYTE *const repEnd2 = repIndex2 < dictLimit ? dictEnd : iend;
+					size_t const repLength2 =
+					    ZSTD_count_2segments(ip + EQUAL_READ32, repMatch2 + EQUAL_READ32, iend, repEnd2, lowPrefixPtr) + EQUAL_READ32;
+					U32 tmpOffset = offset_2;
+					offset_2 = offset_1;
+					offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
+					ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, repLength2 - MINMATCH);
+					hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = curr2;
+					hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = curr2;
+					ip += repLength2;
+					anchor = ip;
+					continue;
+				}
+				break;
+			}
+		}
+	}
+
+	/* save reps for next block */
+	ctx->repToConfirm[0] = offset_1;
+	ctx->repToConfirm[1] = offset_2;
+
+	/* Last Literals */
+	{
+		size_t const lastLLSize = iend - anchor;
+		memcpy(seqStorePtr->lit, anchor, lastLLSize);
+		seqStorePtr->lit += lastLLSize;
+	}
+}
+
+static void ZSTD_compressBlock_doubleFast_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+	U32 const mls = ctx->params.cParams.searchLength;
+	switch (mls) {
+	default: /* includes case 3 */
+	case 4: ZSTD_compressBlock_doubleFast_extDict_generic(ctx, src, srcSize, 4); return;
+	case 5: ZSTD_compressBlock_doubleFast_extDict_generic(ctx, src, srcSize, 5); return;
+	case 6: ZSTD_compressBlock_doubleFast_extDict_generic(ctx, src, srcSize, 6); return;
+	case 7: ZSTD_compressBlock_doubleFast_extDict_generic(ctx, src, srcSize, 7); return;
+	}
+}
+
+/*-*************************************
+*  Binary Tree search
+***************************************/
+/** ZSTD_insertBt1() : add one or multiple positions to tree.
+*   ip : assumed <= iend-8 .
+*   @return : nb of positions added */
+static U32 ZSTD_insertBt1(ZSTD_CCtx *zc, const BYTE *const ip, const U32 mls, const BYTE *const iend, U32 nbCompares, U32 extDict)
+{
+	U32 *const hashTable = zc->hashTable;
+	U32 const hashLog = zc->params.cParams.hashLog;
+	size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
+	U32 *const bt = zc->chainTable;
+	U32 const btLog = zc->params.cParams.chainLog - 1;
+	U32 const btMask = (1 << btLog) - 1;
+	U32 matchIndex = hashTable[h];
+	size_t commonLengthSmaller = 0, commonLengthLarger = 0;
+	const BYTE *const base = zc->base;
+	const BYTE *const dictBase = zc->dictBase;
+	const U32 dictLimit = zc->dictLimit;
+	const BYTE *const dictEnd = dictBase + dictLimit;
+	const BYTE *const prefixStart = base + dictLimit;
+	const BYTE *match;
+	const U32 curr = (U32)(ip - base);
+	const U32 btLow = btMask >= curr ? 0 : curr - btMask;
+	U32 *smallerPtr = bt + 2 * (curr & btMask);
+	U32 *largerPtr = smallerPtr + 1;
+	U32 dummy32; /* to be nullified at the end */
+	U32 const windowLow = zc->lowLimit;
+	U32 matchEndIdx = curr + 8;
+	size_t bestLength = 8;
+
+	hashTable[h] = curr; /* Update Hash Table */
+
+	while (nbCompares-- && (matchIndex > windowLow)) {
+		U32 *const nextPtr = bt + 2 * (matchIndex & btMask);
+		size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
+
+		if ((!extDict) || (matchIndex + matchLength >= dictLimit)) {
+			match = base + matchIndex;
+			if (match[matchLength] == ip[matchLength])
+				matchLength += ZSTD_count(ip + matchLength + 1, match + matchLength + 1, iend) + 1;
+		} else {
+			match = dictBase + matchIndex;
+			matchLength += ZSTD_count_2segments(ip + matchLength, match + matchLength, iend, dictEnd, prefixStart);
+			if (matchIndex + matchLength >= dictLimit)
+				match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
+		}
+
+		if (matchLength > bestLength) {
+			bestLength = matchLength;
+			if (matchLength > matchEndIdx - matchIndex)
+				matchEndIdx = matchIndex + (U32)matchLength;
+		}
+
+		if (ip + matchLength == iend) /* equal : no way to know if inf or sup */
+			break;		      /* drop , to guarantee consistency ; miss a bit of compression, but other solutions can corrupt the tree */
+
+		if (match[matchLength] < ip[matchLength]) { /* necessarily within correct buffer */
+			/* match is smaller than curr */
+			*smallerPtr = matchIndex;	  /* update smaller idx */
+			commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
+			if (matchIndex <= btLow) {
+				smallerPtr = &dummy32;
+				break;
+			}			  /* beyond tree size, stop the search */
+			smallerPtr = nextPtr + 1; /* new "smaller" => larger of match */
+			matchIndex = nextPtr[1];  /* new matchIndex larger than previous (closer to curr) */
+		} else {
+			/* match is larger than curr */
+			*largerPtr = matchIndex;
+			commonLengthLarger = matchLength;
+			if (matchIndex <= btLow) {
+				largerPtr = &dummy32;
+				break;
+			} /* beyond tree size, stop the search */
+			largerPtr = nextPtr;
+			matchIndex = nextPtr[0];
+		}
+	}
+
+	*smallerPtr = *largerPtr = 0;
+	if (bestLength > 384)
+		return MIN(192, (U32)(bestLength - 384)); /* speed optimization */
+	if (matchEndIdx > curr + 8)
+		return matchEndIdx - curr - 8;
+	return 1;
+}
+
+static size_t ZSTD_insertBtAndFindBestMatch(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iend, size_t *offsetPtr, U32 nbCompares, const U32 mls,
+					    U32 extDict)
+{
+	U32 *const hashTable = zc->hashTable;
+	U32 const hashLog = zc->params.cParams.hashLog;
+	size_t const h = ZSTD_hashPtr(ip, hashLog, mls);
+	U32 *const bt = zc->chainTable;
+	U32 const btLog = zc->params.cParams.chainLog - 1;
+	U32 const btMask = (1 << btLog) - 1;
+	U32 matchIndex = hashTable[h];
+	size_t commonLengthSmaller = 0, commonLengthLarger = 0;
+	const BYTE *const base = zc->base;
+	const BYTE *const dictBase = zc->dictBase;
+	const U32 dictLimit = zc->dictLimit;
+	const BYTE *const dictEnd = dictBase + dictLimit;
+	const BYTE *const prefixStart = base + dictLimit;
+	const U32 curr = (U32)(ip - base);
+	const U32 btLow = btMask >= curr ? 0 : curr - btMask;
+	const U32 windowLow = zc->lowLimit;
+	U32 *smallerPtr = bt + 2 * (curr & btMask);
+	U32 *largerPtr = bt + 2 * (curr & btMask) + 1;
+	U32 matchEndIdx = curr + 8;
+	U32 dummy32; /* to be nullified at the end */
+	size_t bestLength = 0;
+
+	hashTable[h] = curr; /* Update Hash Table */
+
+	while (nbCompares-- && (matchIndex > windowLow)) {
+		U32 *const nextPtr = bt + 2 * (matchIndex & btMask);
+		size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
+		const BYTE *match;
+
+		if ((!extDict) || (matchIndex + matchLength >= dictLimit)) {
+			match = base + matchIndex;
+			if (match[matchLength] == ip[matchLength])
+				matchLength += ZSTD_count(ip + matchLength + 1, match + matchLength + 1, iend) + 1;
+		} else {
+			match = dictBase + matchIndex;
+			matchLength += ZSTD_count_2segments(ip + matchLength, match + matchLength, iend, dictEnd, prefixStart);
+			if (matchIndex + matchLength >= dictLimit)
+				match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
+		}
+
+		if (matchLength > bestLength) {
+			if (matchLength > matchEndIdx - matchIndex)
+				matchEndIdx = matchIndex + (U32)matchLength;
+			if ((4 * (int)(matchLength - bestLength)) > (int)(ZSTD_highbit32(curr - matchIndex + 1) - ZSTD_highbit32((U32)offsetPtr[0] + 1)))
+				bestLength = matchLength, *offsetPtr = ZSTD_REP_MOVE + curr - matchIndex;
+			if (ip + matchLength == iend) /* equal : no way to know if inf or sup */
+				break;		      /* drop, to guarantee consistency (miss a little bit of compression) */
+		}
+
+		if (match[matchLength] < ip[matchLength]) {
+			/* match is smaller than curr */
+			*smallerPtr = matchIndex;	  /* update smaller idx */
+			commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
+			if (matchIndex <= btLow) {
+				smallerPtr = &dummy32;
+				break;
+			}			  /* beyond tree size, stop the search */
+			smallerPtr = nextPtr + 1; /* new "smaller" => larger of match */
+			matchIndex = nextPtr[1];  /* new matchIndex larger than previous (closer to curr) */
+		} else {
+			/* match is larger than curr */
+			*largerPtr = matchIndex;
+			commonLengthLarger = matchLength;
+			if (matchIndex <= btLow) {
+				largerPtr = &dummy32;
+				break;
+			} /* beyond tree size, stop the search */
+			largerPtr = nextPtr;
+			matchIndex = nextPtr[0];
+		}
+	}
+
+	*smallerPtr = *largerPtr = 0;
+
+	zc->nextToUpdate = (matchEndIdx > curr + 8) ? matchEndIdx - 8 : curr + 1;
+	return bestLength;
+}
+
+static void ZSTD_updateTree(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iend, const U32 nbCompares, const U32 mls)
+{
+	const BYTE *const base = zc->base;
+	const U32 target = (U32)(ip - base);
+	U32 idx = zc->nextToUpdate;
+
+	while (idx < target)
+		idx += ZSTD_insertBt1(zc, base + idx, mls, iend, nbCompares, 0);
+}
+
+/** ZSTD_BtFindBestMatch() : Tree updater, providing best match */
+static size_t ZSTD_BtFindBestMatch(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts, const U32 mls)
+{
+	if (ip < zc->base + zc->nextToUpdate)
+		return 0; /* skipped area */
+	ZSTD_updateTree(zc, ip, iLimit, maxNbAttempts, mls);
+	return ZSTD_insertBtAndFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, mls, 0);
+}
+
+static size_t ZSTD_BtFindBestMatch_selectMLS(ZSTD_CCtx *zc, /* Index table will be updated */
+					     const BYTE *ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts, const U32 matchLengthSearch)
+{
+	switch (matchLengthSearch) {
+	default: /* includes case 3 */
+	case 4: return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4);
+	case 5: return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5);
+	case 7:
+	case 6: return ZSTD_BtFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6);
+	}
+}
+
+static void ZSTD_updateTree_extDict(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iend, const U32 nbCompares, const U32 mls)
+{
+	const BYTE *const base = zc->base;
+	const U32 target = (U32)(ip - base);
+	U32 idx = zc->nextToUpdate;
+
+	while (idx < target)
+		idx += ZSTD_insertBt1(zc, base + idx, mls, iend, nbCompares, 1);
+}
+
+/** Tree updater, providing best match */
+static size_t ZSTD_BtFindBestMatch_extDict(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts,
+					   const U32 mls)
+{
+	if (ip < zc->base + zc->nextToUpdate)
+		return 0; /* skipped area */
+	ZSTD_updateTree_extDict(zc, ip, iLimit, maxNbAttempts, mls);
+	return ZSTD_insertBtAndFindBestMatch(zc, ip, iLimit, offsetPtr, maxNbAttempts, mls, 1);
+}
+
+static size_t ZSTD_BtFindBestMatch_selectMLS_extDict(ZSTD_CCtx *zc, /* Index table will be updated */
+						     const BYTE *ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts,
+						     const U32 matchLengthSearch)
+{
+	switch (matchLengthSearch) {
+	default: /* includes case 3 */
+	case 4: return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4);
+	case 5: return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5);
+	case 7:
+	case 6: return ZSTD_BtFindBestMatch_extDict(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6);
+	}
+}
+
+/* *********************************
+*  Hash Chain
+***********************************/
+#define NEXT_IN_CHAIN(d, mask) chainTable[(d)&mask]
+
+/* Update chains up to ip (excluded)
+   Assumption : always within prefix (i.e. not within extDict) */
+FORCE_INLINE
+U32 ZSTD_insertAndFindFirstIndex(ZSTD_CCtx *zc, const BYTE *ip, U32 mls)
+{
+	U32 *const hashTable = zc->hashTable;
+	const U32 hashLog = zc->params.cParams.hashLog;
+	U32 *const chainTable = zc->chainTable;
+	const U32 chainMask = (1 << zc->params.cParams.chainLog) - 1;
+	const BYTE *const base = zc->base;
+	const U32 target = (U32)(ip - base);
+	U32 idx = zc->nextToUpdate;
+
+	while (idx < target) { /* catch up */
+		size_t const h = ZSTD_hashPtr(base + idx, hashLog, mls);
+		NEXT_IN_CHAIN(idx, chainMask) = hashTable[h];
+		hashTable[h] = idx;
+		idx++;
+	}
+
+	zc->nextToUpdate = target;
+	return hashTable[ZSTD_hashPtr(ip, hashLog, mls)];
+}
+
+/* inlining is important to hardwire a hot branch (template emulation) */
+FORCE_INLINE
+size_t ZSTD_HcFindBestMatch_generic(ZSTD_CCtx *zc, /* Index table will be updated */
+				    const BYTE *const ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts, const U32 mls,
+				    const U32 extDict)
+{
+	U32 *const chainTable = zc->chainTable;
+	const U32 chainSize = (1 << zc->params.cParams.chainLog);
+	const U32 chainMask = chainSize - 1;
+	const BYTE *const base = zc->base;
+	const BYTE *const dictBase = zc->dictBase;
+	const U32 dictLimit = zc->dictLimit;
+	const BYTE *const prefixStart = base + dictLimit;
+	const BYTE *const dictEnd = dictBase + dictLimit;
+	const U32 lowLimit = zc->lowLimit;
+	const U32 curr = (U32)(ip - base);
+	const U32 minChain = curr > chainSize ? curr - chainSize : 0;
+	int nbAttempts = maxNbAttempts;
+	size_t ml = EQUAL_READ32 - 1;
+
+	/* HC4 match finder */
+	U32 matchIndex = ZSTD_insertAndFindFirstIndex(zc, ip, mls);
+
+	for (; (matchIndex > lowLimit) & (nbAttempts > 0); nbAttempts--) {
+		const BYTE *match;
+		size_t currMl = 0;
+		if ((!extDict) || matchIndex >= dictLimit) {
+			match = base + matchIndex;
+			if (match[ml] == ip[ml]) /* potentially better */
+				currMl = ZSTD_count(ip, match, iLimit);
+		} else {
+			match = dictBase + matchIndex;
+			if (ZSTD_read32(match) == ZSTD_read32(ip)) /* assumption : matchIndex <= dictLimit-4 (by table construction) */
+				currMl = ZSTD_count_2segments(ip + EQUAL_READ32, match + EQUAL_READ32, iLimit, dictEnd, prefixStart) + EQUAL_READ32;
+		}
+
+		/* save best solution */
+		if (currMl > ml) {
+			ml = currMl;
+			*offsetPtr = curr - matchIndex + ZSTD_REP_MOVE;
+			if (ip + currMl == iLimit)
+				break; /* best possible, and avoid read overflow*/
+		}
+
+		if (matchIndex <= minChain)
+			break;
+		matchIndex = NEXT_IN_CHAIN(matchIndex, chainMask);
+	}
+
+	return ml;
+}
+
+FORCE_INLINE size_t ZSTD_HcFindBestMatch_selectMLS(ZSTD_CCtx *zc, const BYTE *ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts,
+						   const U32 matchLengthSearch)
+{
+	switch (matchLengthSearch) {
+	default: /* includes case 3 */
+	case 4: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4, 0);
+	case 5: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5, 0);
+	case 7:
+	case 6: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6, 0);
+	}
+}
+
+FORCE_INLINE size_t ZSTD_HcFindBestMatch_extDict_selectMLS(ZSTD_CCtx *zc, const BYTE *ip, const BYTE *const iLimit, size_t *offsetPtr, const U32 maxNbAttempts,
+							   const U32 matchLengthSearch)
+{
+	switch (matchLengthSearch) {
+	default: /* includes case 3 */
+	case 4: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 4, 1);
+	case 5: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 5, 1);
+	case 7:
+	case 6: return ZSTD_HcFindBestMatch_generic(zc, ip, iLimit, offsetPtr, maxNbAttempts, 6, 1);
+	}
+}
+
+/* *******************************
+*  Common parser - lazy strategy
+*********************************/
+FORCE_INLINE
+void ZSTD_compressBlock_lazy_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const U32 searchMethod, const U32 depth)
+{
+	seqStore_t *seqStorePtr = &(ctx->seqStore);
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *ip = istart;
+	const BYTE *anchor = istart;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *const ilimit = iend - 8;
+	const BYTE *const base = ctx->base + ctx->dictLimit;
+
+	U32 const maxSearches = 1 << ctx->params.cParams.searchLog;
+	U32 const mls = ctx->params.cParams.searchLength;
+
+	typedef size_t (*searchMax_f)(ZSTD_CCtx * zc, const BYTE *ip, const BYTE *iLimit, size_t *offsetPtr, U32 maxNbAttempts, U32 matchLengthSearch);
+	searchMax_f const searchMax = searchMethod ? ZSTD_BtFindBestMatch_selectMLS : ZSTD_HcFindBestMatch_selectMLS;
+	U32 offset_1 = ctx->rep[0], offset_2 = ctx->rep[1], savedOffset = 0;
+
+	/* init */
+	ip += (ip == base);
+	ctx->nextToUpdate3 = ctx->nextToUpdate;
+	{
+		U32 const maxRep = (U32)(ip - base);
+		if (offset_2 > maxRep)
+			savedOffset = offset_2, offset_2 = 0;
+		if (offset_1 > maxRep)
+			savedOffset = offset_1, offset_1 = 0;
+	}
+
+	/* Match Loop */
+	while (ip < ilimit) {
+		size_t matchLength = 0;
+		size_t offset = 0;
+		const BYTE *start = ip + 1;
+
+		/* check repCode */
+		if ((offset_1 > 0) & (ZSTD_read32(ip + 1) == ZSTD_read32(ip + 1 - offset_1))) {
+			/* repcode : we take it */
+			matchLength = ZSTD_count(ip + 1 + EQUAL_READ32, ip + 1 + EQUAL_READ32 - offset_1, iend) + EQUAL_READ32;
+			if (depth == 0)
+				goto _storeSequence;
+		}
+
+		/* first search (depth 0) */
+		{
+			size_t offsetFound = 99999999;
+			size_t const ml2 = searchMax(ctx, ip, iend, &offsetFound, maxSearches, mls);
+			if (ml2 > matchLength)
+				matchLength = ml2, start = ip, offset = offsetFound;
+		}
+
+		if (matchLength < EQUAL_READ32) {
+			ip += ((ip - anchor) >> g_searchStrength) + 1; /* jump faster over incompressible sections */
+			continue;
+		}
+
+		/* let's try to find a better solution */
+		if (depth >= 1)
+			while (ip < ilimit) {
+				ip++;
+				if ((offset) && ((offset_1 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_1)))) {
+					size_t const mlRep = ZSTD_count(ip + EQUAL_READ32, ip + EQUAL_READ32 - offset_1, iend) + EQUAL_READ32;
+					int const gain2 = (int)(mlRep * 3);
+					int const gain1 = (int)(matchLength * 3 - ZSTD_highbit32((U32)offset + 1) + 1);
+					if ((mlRep >= EQUAL_READ32) && (gain2 > gain1))
+						matchLength = mlRep, offset = 0, start = ip;
+				}
+				{
+					size_t offset2 = 99999999;
+					size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
+					int const gain2 = (int)(ml2 * 4 - ZSTD_highbit32((U32)offset2 + 1)); /* raw approx */
+					int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 4);
+					if ((ml2 >= EQUAL_READ32) && (gain2 > gain1)) {
+						matchLength = ml2, offset = offset2, start = ip;
+						continue; /* search a better one */
+					}
+				}
+
+				/* let's find an even better one */
+				if ((depth == 2) && (ip < ilimit)) {
+					ip++;
+					if ((offset) && ((offset_1 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_1)))) {
+						size_t const ml2 = ZSTD_count(ip + EQUAL_READ32, ip + EQUAL_READ32 - offset_1, iend) + EQUAL_READ32;
+						int const gain2 = (int)(ml2 * 4);
+						int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 1);
+						if ((ml2 >= EQUAL_READ32) && (gain2 > gain1))
+							matchLength = ml2, offset = 0, start = ip;
+					}
+					{
+						size_t offset2 = 99999999;
+						size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
+						int const gain2 = (int)(ml2 * 4 - ZSTD_highbit32((U32)offset2 + 1)); /* raw approx */
+						int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 7);
+						if ((ml2 >= EQUAL_READ32) && (gain2 > gain1)) {
+							matchLength = ml2, offset = offset2, start = ip;
+							continue;
+						}
+					}
+				}
+				break; /* nothing found : store previous solution */
+			}
+
+		/* NOTE:
+		 * start[-offset+ZSTD_REP_MOVE-1] is undefined behavior.
+		 * (-offset+ZSTD_REP_MOVE-1) is unsigned, and is added to start, which
+		 * overflows the pointer, which is undefined behavior.
+		 */
+		/* catch up */
+		if (offset) {
+			while ((start > anchor) && (start > base + offset - ZSTD_REP_MOVE) &&
+			       (start[-1] == (start-offset+ZSTD_REP_MOVE)[-1])) /* only search for offset within prefix */
+			{
+				start--;
+				matchLength++;
+			}
+			offset_2 = offset_1;
+			offset_1 = (U32)(offset - ZSTD_REP_MOVE);
+		}
+
+	/* store sequence */
+_storeSequence:
+		{
+			size_t const litLength = start - anchor;
+			ZSTD_storeSeq(seqStorePtr, litLength, anchor, (U32)offset, matchLength - MINMATCH);
+			anchor = ip = start + matchLength;
+		}
+
+		/* check immediate repcode */
+		while ((ip <= ilimit) && ((offset_2 > 0) & (ZSTD_read32(ip) == ZSTD_read32(ip - offset_2)))) {
+			/* store sequence */
+			matchLength = ZSTD_count(ip + EQUAL_READ32, ip + EQUAL_READ32 - offset_2, iend) + EQUAL_READ32;
+			offset = offset_2;
+			offset_2 = offset_1;
+			offset_1 = (U32)offset; /* swap repcodes */
+			ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, matchLength - MINMATCH);
+			ip += matchLength;
+			anchor = ip;
+			continue; /* faster when present ... (?) */
+		}
+	}
+
+	/* Save reps for next block */
+	ctx->repToConfirm[0] = offset_1 ? offset_1 : savedOffset;
+	ctx->repToConfirm[1] = offset_2 ? offset_2 : savedOffset;
+
+	/* Last Literals */
+	{
+		size_t const lastLLSize = iend - anchor;
+		memcpy(seqStorePtr->lit, anchor, lastLLSize);
+		seqStorePtr->lit += lastLLSize;
+	}
+}
+
+static void ZSTD_compressBlock_btlazy2(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 1, 2); }
+
+static void ZSTD_compressBlock_lazy2(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 2); }
+
+static void ZSTD_compressBlock_lazy(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 1); }
+
+static void ZSTD_compressBlock_greedy(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_generic(ctx, src, srcSize, 0, 0); }
+
+FORCE_INLINE
+void ZSTD_compressBlock_lazy_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const U32 searchMethod, const U32 depth)
+{
+	seqStore_t *seqStorePtr = &(ctx->seqStore);
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *ip = istart;
+	const BYTE *anchor = istart;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *const ilimit = iend - 8;
+	const BYTE *const base = ctx->base;
+	const U32 dictLimit = ctx->dictLimit;
+	const U32 lowestIndex = ctx->lowLimit;
+	const BYTE *const prefixStart = base + dictLimit;
+	const BYTE *const dictBase = ctx->dictBase;
+	const BYTE *const dictEnd = dictBase + dictLimit;
+	const BYTE *const dictStart = dictBase + ctx->lowLimit;
+
+	const U32 maxSearches = 1 << ctx->params.cParams.searchLog;
+	const U32 mls = ctx->params.cParams.searchLength;
+
+	typedef size_t (*searchMax_f)(ZSTD_CCtx * zc, const BYTE *ip, const BYTE *iLimit, size_t *offsetPtr, U32 maxNbAttempts, U32 matchLengthSearch);
+	searchMax_f searchMax = searchMethod ? ZSTD_BtFindBestMatch_selectMLS_extDict : ZSTD_HcFindBestMatch_extDict_selectMLS;
+
+	U32 offset_1 = ctx->rep[0], offset_2 = ctx->rep[1];
+
+	/* init */
+	ctx->nextToUpdate3 = ctx->nextToUpdate;
+	ip += (ip == prefixStart);
+
+	/* Match Loop */
+	while (ip < ilimit) {
+		size_t matchLength = 0;
+		size_t offset = 0;
+		const BYTE *start = ip + 1;
+		U32 curr = (U32)(ip - base);
+
+		/* check repCode */
+		{
+			const U32 repIndex = (U32)(curr + 1 - offset_1);
+			const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
+			const BYTE *const repMatch = repBase + repIndex;
+			if (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
+				if (ZSTD_read32(ip + 1) == ZSTD_read32(repMatch)) {
+					/* repcode detected we should take it */
+					const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
+					matchLength =
+					    ZSTD_count_2segments(ip + 1 + EQUAL_READ32, repMatch + EQUAL_READ32, iend, repEnd, prefixStart) + EQUAL_READ32;
+					if (depth == 0)
+						goto _storeSequence;
+				}
+		}
+
+		/* first search (depth 0) */
+		{
+			size_t offsetFound = 99999999;
+			size_t const ml2 = searchMax(ctx, ip, iend, &offsetFound, maxSearches, mls);
+			if (ml2 > matchLength)
+				matchLength = ml2, start = ip, offset = offsetFound;
+		}
+
+		if (matchLength < EQUAL_READ32) {
+			ip += ((ip - anchor) >> g_searchStrength) + 1; /* jump faster over incompressible sections */
+			continue;
+		}
+
+		/* let's try to find a better solution */
+		if (depth >= 1)
+			while (ip < ilimit) {
+				ip++;
+				curr++;
+				/* check repCode */
+				if (offset) {
+					const U32 repIndex = (U32)(curr - offset_1);
+					const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
+					const BYTE *const repMatch = repBase + repIndex;
+					if (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
+						if (ZSTD_read32(ip) == ZSTD_read32(repMatch)) {
+							/* repcode detected */
+							const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
+							size_t const repLength =
+							    ZSTD_count_2segments(ip + EQUAL_READ32, repMatch + EQUAL_READ32, iend, repEnd, prefixStart) +
+							    EQUAL_READ32;
+							int const gain2 = (int)(repLength * 3);
+							int const gain1 = (int)(matchLength * 3 - ZSTD_highbit32((U32)offset + 1) + 1);
+							if ((repLength >= EQUAL_READ32) && (gain2 > gain1))
+								matchLength = repLength, offset = 0, start = ip;
+						}
+				}
+
+				/* search match, depth 1 */
+				{
+					size_t offset2 = 99999999;
+					size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
+					int const gain2 = (int)(ml2 * 4 - ZSTD_highbit32((U32)offset2 + 1)); /* raw approx */
+					int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 4);
+					if ((ml2 >= EQUAL_READ32) && (gain2 > gain1)) {
+						matchLength = ml2, offset = offset2, start = ip;
+						continue; /* search a better one */
+					}
+				}
+
+				/* let's find an even better one */
+				if ((depth == 2) && (ip < ilimit)) {
+					ip++;
+					curr++;
+					/* check repCode */
+					if (offset) {
+						const U32 repIndex = (U32)(curr - offset_1);
+						const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
+						const BYTE *const repMatch = repBase + repIndex;
+						if (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
+							if (ZSTD_read32(ip) == ZSTD_read32(repMatch)) {
+								/* repcode detected */
+								const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
+								size_t repLength = ZSTD_count_2segments(ip + EQUAL_READ32, repMatch + EQUAL_READ32, iend,
+													repEnd, prefixStart) +
+										   EQUAL_READ32;
+								int gain2 = (int)(repLength * 4);
+								int gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 1);
+								if ((repLength >= EQUAL_READ32) && (gain2 > gain1))
+									matchLength = repLength, offset = 0, start = ip;
+							}
+					}
+
+					/* search match, depth 2 */
+					{
+						size_t offset2 = 99999999;
+						size_t const ml2 = searchMax(ctx, ip, iend, &offset2, maxSearches, mls);
+						int const gain2 = (int)(ml2 * 4 - ZSTD_highbit32((U32)offset2 + 1)); /* raw approx */
+						int const gain1 = (int)(matchLength * 4 - ZSTD_highbit32((U32)offset + 1) + 7);
+						if ((ml2 >= EQUAL_READ32) && (gain2 > gain1)) {
+							matchLength = ml2, offset = offset2, start = ip;
+							continue;
+						}
+					}
+				}
+				break; /* nothing found : store previous solution */
+			}
+
+		/* catch up */
+		if (offset) {
+			U32 const matchIndex = (U32)((start - base) - (offset - ZSTD_REP_MOVE));
+			const BYTE *match = (matchIndex < dictLimit) ? dictBase + matchIndex : base + matchIndex;
+			const BYTE *const mStart = (matchIndex < dictLimit) ? dictStart : prefixStart;
+			while ((start > anchor) && (match > mStart) && (start[-1] == match[-1])) {
+				start--;
+				match--;
+				matchLength++;
+			} /* catch up */
+			offset_2 = offset_1;
+			offset_1 = (U32)(offset - ZSTD_REP_MOVE);
+		}
+
+	/* store sequence */
+	_storeSequence : {
+		size_t const litLength = start - anchor;
+		ZSTD_storeSeq(seqStorePtr, litLength, anchor, (U32)offset, matchLength - MINMATCH);
+		anchor = ip = start + matchLength;
+	}
+
+		/* check immediate repcode */
+		while (ip <= ilimit) {
+			const U32 repIndex = (U32)((ip - base) - offset_2);
+			const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
+			const BYTE *const repMatch = repBase + repIndex;
+			if (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
+				if (ZSTD_read32(ip) == ZSTD_read32(repMatch)) {
+					/* repcode detected we should take it */
+					const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
+					matchLength =
+					    ZSTD_count_2segments(ip + EQUAL_READ32, repMatch + EQUAL_READ32, iend, repEnd, prefixStart) + EQUAL_READ32;
+					offset = offset_2;
+					offset_2 = offset_1;
+					offset_1 = (U32)offset; /* swap offset history */
+					ZSTD_storeSeq(seqStorePtr, 0, anchor, 0, matchLength - MINMATCH);
+					ip += matchLength;
+					anchor = ip;
+					continue; /* faster when present ... (?) */
+				}
+			break;
+		}
+	}
+
+	/* Save reps for next block */
+	ctx->repToConfirm[0] = offset_1;
+	ctx->repToConfirm[1] = offset_2;
+
+	/* Last Literals */
+	{
+		size_t const lastLLSize = iend - anchor;
+		memcpy(seqStorePtr->lit, anchor, lastLLSize);
+		seqStorePtr->lit += lastLLSize;
+	}
+}
+
+void ZSTD_compressBlock_greedy_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize) { ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 0); }
+
+static void ZSTD_compressBlock_lazy_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+	ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 1);
+}
+
+static void ZSTD_compressBlock_lazy2_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+	ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 0, 2);
+}
+
+static void ZSTD_compressBlock_btlazy2_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+	ZSTD_compressBlock_lazy_extDict_generic(ctx, src, srcSize, 1, 2);
+}
+
+/* The optimal parser */
+#include "zstd_opt.h"
+
+static void ZSTD_compressBlock_btopt(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+#ifdef ZSTD_OPT_H_91842398743
+	ZSTD_compressBlock_opt_generic(ctx, src, srcSize, 0);
+#else
+	(void)ctx;
+	(void)src;
+	(void)srcSize;
+	return;
+#endif
+}
+
+static void ZSTD_compressBlock_btopt2(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+#ifdef ZSTD_OPT_H_91842398743
+	ZSTD_compressBlock_opt_generic(ctx, src, srcSize, 1);
+#else
+	(void)ctx;
+	(void)src;
+	(void)srcSize;
+	return;
+#endif
+}
+
+static void ZSTD_compressBlock_btopt_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+#ifdef ZSTD_OPT_H_91842398743
+	ZSTD_compressBlock_opt_extDict_generic(ctx, src, srcSize, 0);
+#else
+	(void)ctx;
+	(void)src;
+	(void)srcSize;
+	return;
+#endif
+}
+
+static void ZSTD_compressBlock_btopt2_extDict(ZSTD_CCtx *ctx, const void *src, size_t srcSize)
+{
+#ifdef ZSTD_OPT_H_91842398743
+	ZSTD_compressBlock_opt_extDict_generic(ctx, src, srcSize, 1);
+#else
+	(void)ctx;
+	(void)src;
+	(void)srcSize;
+	return;
+#endif
+}
+
+typedef void (*ZSTD_blockCompressor)(ZSTD_CCtx *ctx, const void *src, size_t srcSize);
+
+static ZSTD_blockCompressor ZSTD_selectBlockCompressor(ZSTD_strategy strat, int extDict)
+{
+	static const ZSTD_blockCompressor blockCompressor[2][8] = {
+	    {ZSTD_compressBlock_fast, ZSTD_compressBlock_doubleFast, ZSTD_compressBlock_greedy, ZSTD_compressBlock_lazy, ZSTD_compressBlock_lazy2,
+	     ZSTD_compressBlock_btlazy2, ZSTD_compressBlock_btopt, ZSTD_compressBlock_btopt2},
+	    {ZSTD_compressBlock_fast_extDict, ZSTD_compressBlock_doubleFast_extDict, ZSTD_compressBlock_greedy_extDict, ZSTD_compressBlock_lazy_extDict,
+	     ZSTD_compressBlock_lazy2_extDict, ZSTD_compressBlock_btlazy2_extDict, ZSTD_compressBlock_btopt_extDict, ZSTD_compressBlock_btopt2_extDict}};
+
+	return blockCompressor[extDict][(U32)strat];
+}
+
+static size_t ZSTD_compressBlock_internal(ZSTD_CCtx *zc, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	ZSTD_blockCompressor const blockCompressor = ZSTD_selectBlockCompressor(zc->params.cParams.strategy, zc->lowLimit < zc->dictLimit);
+	const BYTE *const base = zc->base;
+	const BYTE *const istart = (const BYTE *)src;
+	const U32 curr = (U32)(istart - base);
+	if (srcSize < MIN_CBLOCK_SIZE + ZSTD_blockHeaderSize + 1)
+		return 0; /* don't even attempt compression below a certain srcSize */
+	ZSTD_resetSeqStore(&(zc->seqStore));
+	if (curr > zc->nextToUpdate + 384)
+		zc->nextToUpdate = curr - MIN(192, (U32)(curr - zc->nextToUpdate - 384)); /* update tree not updated after finding very long rep matches */
+	blockCompressor(zc, src, srcSize);
+	return ZSTD_compressSequences(zc, dst, dstCapacity, srcSize);
+}
+
+/*! ZSTD_compress_generic() :
+*   Compress a chunk of data into one or multiple blocks.
+*   All blocks will be terminated, all input will be consumed.
+*   Function will issue an error if there is not enough `dstCapacity` to hold the compressed content.
+*   Frame is supposed already started (header already produced)
+*   @return : compressed size, or an error code
+*/
+static size_t ZSTD_compress_generic(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, U32 lastFrameChunk)
+{
+	size_t blockSize = cctx->blockSize;
+	size_t remaining = srcSize;
+	const BYTE *ip = (const BYTE *)src;
+	BYTE *const ostart = (BYTE *)dst;
+	BYTE *op = ostart;
+	U32 const maxDist = 1 << cctx->params.cParams.windowLog;
+
+	if (cctx->params.fParams.checksumFlag && srcSize)
+		xxh64_update(&cctx->xxhState, src, srcSize);
+
+	while (remaining) {
+		U32 const lastBlock = lastFrameChunk & (blockSize >= remaining);
+		size_t cSize;
+
+		if (dstCapacity < ZSTD_blockHeaderSize + MIN_CBLOCK_SIZE)
+			return ERROR(dstSize_tooSmall); /* not enough space to store compressed block */
+		if (remaining < blockSize)
+			blockSize = remaining;
+
+		/* preemptive overflow correction */
+		if (cctx->lowLimit > (3U << 29)) {
+			U32 const cycleMask = (1 << ZSTD_cycleLog(cctx->params.cParams.hashLog, cctx->params.cParams.strategy)) - 1;
+			U32 const curr = (U32)(ip - cctx->base);
+			U32 const newCurr = (curr & cycleMask) + (1 << cctx->params.cParams.windowLog);
+			U32 const correction = curr - newCurr;
+			ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX_64 <= 30);
+			ZSTD_reduceIndex(cctx, correction);
+			cctx->base += correction;
+			cctx->dictBase += correction;
+			cctx->lowLimit -= correction;
+			cctx->dictLimit -= correction;
+			if (cctx->nextToUpdate < correction)
+				cctx->nextToUpdate = 0;
+			else
+				cctx->nextToUpdate -= correction;
+		}
+
+		if ((U32)(ip + blockSize - cctx->base) > cctx->loadedDictEnd + maxDist) {
+			/* enforce maxDist */
+			U32 const newLowLimit = (U32)(ip + blockSize - cctx->base) - maxDist;
+			if (cctx->lowLimit < newLowLimit)
+				cctx->lowLimit = newLowLimit;
+			if (cctx->dictLimit < cctx->lowLimit)
+				cctx->dictLimit = cctx->lowLimit;
+		}
+
+		cSize = ZSTD_compressBlock_internal(cctx, op + ZSTD_blockHeaderSize, dstCapacity - ZSTD_blockHeaderSize, ip, blockSize);
+		if (ZSTD_isError(cSize))
+			return cSize;
+
+		if (cSize == 0) { /* block is not compressible */
+			U32 const cBlockHeader24 = lastBlock + (((U32)bt_raw) << 1) + (U32)(blockSize << 3);
+			if (blockSize + ZSTD_blockHeaderSize > dstCapacity)
+				return ERROR(dstSize_tooSmall);
+			ZSTD_writeLE32(op, cBlockHeader24); /* no pb, 4th byte will be overwritten */
+			memcpy(op + ZSTD_blockHeaderSize, ip, blockSize);
+			cSize = ZSTD_blockHeaderSize + blockSize;
+		} else {
+			U32 const cBlockHeader24 = lastBlock + (((U32)bt_compressed) << 1) + (U32)(cSize << 3);
+			ZSTD_writeLE24(op, cBlockHeader24);
+			cSize += ZSTD_blockHeaderSize;
+		}
+
+		remaining -= blockSize;
+		dstCapacity -= cSize;
+		ip += blockSize;
+		op += cSize;
+	}
+
+	if (lastFrameChunk && (op > ostart))
+		cctx->stage = ZSTDcs_ending;
+	return op - ostart;
+}
+
+static size_t ZSTD_writeFrameHeader(void *dst, size_t dstCapacity, ZSTD_parameters params, U64 pledgedSrcSize, U32 dictID)
+{
+	BYTE *const op = (BYTE *)dst;
+	U32 const dictIDSizeCode = (dictID > 0) + (dictID >= 256) + (dictID >= 65536); /* 0-3 */
+	U32 const checksumFlag = params.fParams.checksumFlag > 0;
+	U32 const windowSize = 1U << params.cParams.windowLog;
+	U32 const singleSegment = params.fParams.contentSizeFlag && (windowSize >= pledgedSrcSize);
+	BYTE const windowLogByte = (BYTE)((params.cParams.windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN) << 3);
+	U32 const fcsCode =
+	    params.fParams.contentSizeFlag ? (pledgedSrcSize >= 256) + (pledgedSrcSize >= 65536 + 256) + (pledgedSrcSize >= 0xFFFFFFFFU) : 0; /* 0-3 */
+	BYTE const frameHeaderDecriptionByte = (BYTE)(dictIDSizeCode + (checksumFlag << 2) + (singleSegment << 5) + (fcsCode << 6));
+	size_t pos;
+
+	if (dstCapacity < ZSTD_frameHeaderSize_max)
+		return ERROR(dstSize_tooSmall);
+
+	ZSTD_writeLE32(dst, ZSTD_MAGICNUMBER);
+	op[4] = frameHeaderDecriptionByte;
+	pos = 5;
+	if (!singleSegment)
+		op[pos++] = windowLogByte;
+	switch (dictIDSizeCode) {
+	default: /* impossible */
+	case 0: break;
+	case 1:
+		op[pos] = (BYTE)(dictID);
+		pos++;
+		break;
+	case 2:
+		ZSTD_writeLE16(op + pos, (U16)dictID);
+		pos += 2;
+		break;
+	case 3:
+		ZSTD_writeLE32(op + pos, dictID);
+		pos += 4;
+		break;
+	}
+	switch (fcsCode) {
+	default: /* impossible */
+	case 0:
+		if (singleSegment)
+			op[pos++] = (BYTE)(pledgedSrcSize);
+		break;
+	case 1:
+		ZSTD_writeLE16(op + pos, (U16)(pledgedSrcSize - 256));
+		pos += 2;
+		break;
+	case 2:
+		ZSTD_writeLE32(op + pos, (U32)(pledgedSrcSize));
+		pos += 4;
+		break;
+	case 3:
+		ZSTD_writeLE64(op + pos, (U64)(pledgedSrcSize));
+		pos += 8;
+		break;
+	}
+	return pos;
+}
+
+static size_t ZSTD_compressContinue_internal(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, U32 frame, U32 lastFrameChunk)
+{
+	const BYTE *const ip = (const BYTE *)src;
+	size_t fhSize = 0;
+
+	if (cctx->stage == ZSTDcs_created)
+		return ERROR(stage_wrong); /* missing init (ZSTD_compressBegin) */
+
+	if (frame && (cctx->stage == ZSTDcs_init)) {
+		fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, cctx->params, cctx->frameContentSize, cctx->dictID);
+		if (ZSTD_isError(fhSize))
+			return fhSize;
+		dstCapacity -= fhSize;
+		dst = (char *)dst + fhSize;
+		cctx->stage = ZSTDcs_ongoing;
+	}
+
+	/* Check if blocks follow each other */
+	if (src != cctx->nextSrc) {
+		/* not contiguous */
+		ptrdiff_t const delta = cctx->nextSrc - ip;
+		cctx->lowLimit = cctx->dictLimit;
+		cctx->dictLimit = (U32)(cctx->nextSrc - cctx->base);
+		cctx->dictBase = cctx->base;
+		cctx->base -= delta;
+		cctx->nextToUpdate = cctx->dictLimit;
+		if (cctx->dictLimit - cctx->lowLimit < HASH_READ_SIZE)
+			cctx->lowLimit = cctx->dictLimit; /* too small extDict */
+	}
+
+	/* if input and dictionary overlap : reduce dictionary (area presumed modified by input) */
+	if ((ip + srcSize > cctx->dictBase + cctx->lowLimit) & (ip < cctx->dictBase + cctx->dictLimit)) {
+		ptrdiff_t const highInputIdx = (ip + srcSize) - cctx->dictBase;
+		U32 const lowLimitMax = (highInputIdx > (ptrdiff_t)cctx->dictLimit) ? cctx->dictLimit : (U32)highInputIdx;
+		cctx->lowLimit = lowLimitMax;
+	}
+
+	cctx->nextSrc = ip + srcSize;
+
+	if (srcSize) {
+		size_t const cSize = frame ? ZSTD_compress_generic(cctx, dst, dstCapacity, src, srcSize, lastFrameChunk)
+					   : ZSTD_compressBlock_internal(cctx, dst, dstCapacity, src, srcSize);
+		if (ZSTD_isError(cSize))
+			return cSize;
+		return cSize + fhSize;
+	} else
+		return fhSize;
+}
+
+size_t ZSTD_compressContinue(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 1, 0);
+}
+
+size_t ZSTD_getBlockSizeMax(ZSTD_CCtx *cctx) { return MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, 1 << cctx->params.cParams.windowLog); }
+
+size_t ZSTD_compressBlock(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	size_t const blockSizeMax = ZSTD_getBlockSizeMax(cctx);
+	if (srcSize > blockSizeMax)
+		return ERROR(srcSize_wrong);
+	return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 0, 0);
+}
+
+/*! ZSTD_loadDictionaryContent() :
+ *  @return : 0, or an error code
+ */
+static size_t ZSTD_loadDictionaryContent(ZSTD_CCtx *zc, const void *src, size_t srcSize)
+{
+	const BYTE *const ip = (const BYTE *)src;
+	const BYTE *const iend = ip + srcSize;
+
+	/* input becomes curr prefix */
+	zc->lowLimit = zc->dictLimit;
+	zc->dictLimit = (U32)(zc->nextSrc - zc->base);
+	zc->dictBase = zc->base;
+	zc->base += ip - zc->nextSrc;
+	zc->nextToUpdate = zc->dictLimit;
+	zc->loadedDictEnd = zc->forceWindow ? 0 : (U32)(iend - zc->base);
+
+	zc->nextSrc = iend;
+	if (srcSize <= HASH_READ_SIZE)
+		return 0;
+
+	switch (zc->params.cParams.strategy) {
+	case ZSTD_fast: ZSTD_fillHashTable(zc, iend, zc->params.cParams.searchLength); break;
+
+	case ZSTD_dfast: ZSTD_fillDoubleHashTable(zc, iend, zc->params.cParams.searchLength); break;
+
+	case ZSTD_greedy:
+	case ZSTD_lazy:
+	case ZSTD_lazy2:
+		if (srcSize >= HASH_READ_SIZE)
+			ZSTD_insertAndFindFirstIndex(zc, iend - HASH_READ_SIZE, zc->params.cParams.searchLength);
+		break;
+
+	case ZSTD_btlazy2:
+	case ZSTD_btopt:
+	case ZSTD_btopt2:
+		if (srcSize >= HASH_READ_SIZE)
+			ZSTD_updateTree(zc, iend - HASH_READ_SIZE, iend, 1 << zc->params.cParams.searchLog, zc->params.cParams.searchLength);
+		break;
+
+	default:
+		return ERROR(GENERIC); /* strategy doesn't exist; impossible */
+	}
+
+	zc->nextToUpdate = (U32)(iend - zc->base);
+	return 0;
+}
+
+/* Dictionaries that assign zero probability to symbols that show up causes problems
+   when FSE encoding.  Refuse dictionaries that assign zero probability to symbols
+   that we may encounter during compression.
+   NOTE: This behavior is not standard and could be improved in the future. */
+static size_t ZSTD_checkDictNCount(short *normalizedCounter, unsigned dictMaxSymbolValue, unsigned maxSymbolValue)
+{
+	U32 s;
+	if (dictMaxSymbolValue < maxSymbolValue)
+		return ERROR(dictionary_corrupted);
+	for (s = 0; s <= maxSymbolValue; ++s) {
+		if (normalizedCounter[s] == 0)
+			return ERROR(dictionary_corrupted);
+	}
+	return 0;
+}
+
+/* Dictionary format :
+ * See :
+ * https://github.com/facebook/zstd/blob/master/doc/zstd_compression_format.md#dictionary-format
+ */
+/*! ZSTD_loadZstdDictionary() :
+ * @return : 0, or an error code
+ *  assumptions : magic number supposed already checked
+ *                dictSize supposed > 8
+ */
+static size_t ZSTD_loadZstdDictionary(ZSTD_CCtx *cctx, const void *dict, size_t dictSize)
+{
+	const BYTE *dictPtr = (const BYTE *)dict;
+	const BYTE *const dictEnd = dictPtr + dictSize;
+	short offcodeNCount[MaxOff + 1];
+	unsigned offcodeMaxValue = MaxOff;
+
+	dictPtr += 4; /* skip magic number */
+	cctx->dictID = cctx->params.fParams.noDictIDFlag ? 0 : ZSTD_readLE32(dictPtr);
+	dictPtr += 4;
+
+	{
+		size_t const hufHeaderSize = HUF_readCTable_wksp(cctx->hufTable, 255, dictPtr, dictEnd - dictPtr, cctx->tmpCounters, sizeof(cctx->tmpCounters));
+		if (HUF_isError(hufHeaderSize))
+			return ERROR(dictionary_corrupted);
+		dictPtr += hufHeaderSize;
+	}
+
+	{
+		unsigned offcodeLog;
+		size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, dictEnd - dictPtr);
+		if (FSE_isError(offcodeHeaderSize))
+			return ERROR(dictionary_corrupted);
+		if (offcodeLog > OffFSELog)
+			return ERROR(dictionary_corrupted);
+		/* Defer checking offcodeMaxValue because we need to know the size of the dictionary content */
+		CHECK_E(FSE_buildCTable_wksp(cctx->offcodeCTable, offcodeNCount, offcodeMaxValue, offcodeLog, cctx->tmpCounters, sizeof(cctx->tmpCounters)),
+			dictionary_corrupted);
+		dictPtr += offcodeHeaderSize;
+	}
+
+	{
+		short matchlengthNCount[MaxML + 1];
+		unsigned matchlengthMaxValue = MaxML, matchlengthLog;
+		size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd - dictPtr);
+		if (FSE_isError(matchlengthHeaderSize))
+			return ERROR(dictionary_corrupted);
+		if (matchlengthLog > MLFSELog)
+			return ERROR(dictionary_corrupted);
+		/* Every match length code must have non-zero probability */
+		CHECK_F(ZSTD_checkDictNCount(matchlengthNCount, matchlengthMaxValue, MaxML));
+		CHECK_E(
+		    FSE_buildCTable_wksp(cctx->matchlengthCTable, matchlengthNCount, matchlengthMaxValue, matchlengthLog, cctx->tmpCounters, sizeof(cctx->tmpCounters)),
+		    dictionary_corrupted);
+		dictPtr += matchlengthHeaderSize;
+	}
+
+	{
+		short litlengthNCount[MaxLL + 1];
+		unsigned litlengthMaxValue = MaxLL, litlengthLog;
+		size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd - dictPtr);
+		if (FSE_isError(litlengthHeaderSize))
+			return ERROR(dictionary_corrupted);
+		if (litlengthLog > LLFSELog)
+			return ERROR(dictionary_corrupted);
+		/* Every literal length code must have non-zero probability */
+		CHECK_F(ZSTD_checkDictNCount(litlengthNCount, litlengthMaxValue, MaxLL));
+		CHECK_E(FSE_buildCTable_wksp(cctx->litlengthCTable, litlengthNCount, litlengthMaxValue, litlengthLog, cctx->tmpCounters, sizeof(cctx->tmpCounters)),
+			dictionary_corrupted);
+		dictPtr += litlengthHeaderSize;
+	}
+
+	if (dictPtr + 12 > dictEnd)
+		return ERROR(dictionary_corrupted);
+	cctx->rep[0] = ZSTD_readLE32(dictPtr + 0);
+	cctx->rep[1] = ZSTD_readLE32(dictPtr + 4);
+	cctx->rep[2] = ZSTD_readLE32(dictPtr + 8);
+	dictPtr += 12;
+
+	{
+		size_t const dictContentSize = (size_t)(dictEnd - dictPtr);
+		U32 offcodeMax = MaxOff;
+		if (dictContentSize <= ((U32)-1) - 128 KB) {
+			U32 const maxOffset = (U32)dictContentSize + 128 KB; /* The maximum offset that must be supported */
+			offcodeMax = ZSTD_highbit32(maxOffset);		     /* Calculate minimum offset code required to represent maxOffset */
+		}
+		/* All offset values <= dictContentSize + 128 KB must be representable */
+		CHECK_F(ZSTD_checkDictNCount(offcodeNCount, offcodeMaxValue, MIN(offcodeMax, MaxOff)));
+		/* All repCodes must be <= dictContentSize and != 0*/
+		{
+			U32 u;
+			for (u = 0; u < 3; u++) {
+				if (cctx->rep[u] == 0)
+					return ERROR(dictionary_corrupted);
+				if (cctx->rep[u] > dictContentSize)
+					return ERROR(dictionary_corrupted);
+			}
+		}
+
+		cctx->flagStaticTables = 1;
+		cctx->flagStaticHufTable = HUF_repeat_valid;
+		return ZSTD_loadDictionaryContent(cctx, dictPtr, dictContentSize);
+	}
+}
+
+/** ZSTD_compress_insertDictionary() :
+*   @return : 0, or an error code */
+static size_t ZSTD_compress_insertDictionary(ZSTD_CCtx *cctx, const void *dict, size_t dictSize)
+{
+	if ((dict == NULL) || (dictSize <= 8))
+		return 0;
+
+	/* dict as pure content */
+	if ((ZSTD_readLE32(dict) != ZSTD_DICT_MAGIC) || (cctx->forceRawDict))
+		return ZSTD_loadDictionaryContent(cctx, dict, dictSize);
+
+	/* dict as zstd dictionary */
+	return ZSTD_loadZstdDictionary(cctx, dict, dictSize);
+}
+
+/*! ZSTD_compressBegin_internal() :
+*   @return : 0, or an error code */
+static size_t ZSTD_compressBegin_internal(ZSTD_CCtx *cctx, const void *dict, size_t dictSize, ZSTD_parameters params, U64 pledgedSrcSize)
+{
+	ZSTD_compResetPolicy_e const crp = dictSize ? ZSTDcrp_fullReset : ZSTDcrp_continue;
+	CHECK_F(ZSTD_resetCCtx_advanced(cctx, params, pledgedSrcSize, crp));
+	return ZSTD_compress_insertDictionary(cctx, dict, dictSize);
+}
+
+/*! ZSTD_compressBegin_advanced() :
+*   @return : 0, or an error code */
+size_t ZSTD_compressBegin_advanced(ZSTD_CCtx *cctx, const void *dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize)
+{
+	/* compression parameters verification and optimization */
+	CHECK_F(ZSTD_checkCParams(params.cParams));
+	return ZSTD_compressBegin_internal(cctx, dict, dictSize, params, pledgedSrcSize);
+}
+
+size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx *cctx, const void *dict, size_t dictSize, int compressionLevel)
+{
+	ZSTD_parameters const params = ZSTD_getParams(compressionLevel, 0, dictSize);
+	return ZSTD_compressBegin_internal(cctx, dict, dictSize, params, 0);
+}
+
+size_t ZSTD_compressBegin(ZSTD_CCtx *cctx, int compressionLevel) { return ZSTD_compressBegin_usingDict(cctx, NULL, 0, compressionLevel); }
+
+/*! ZSTD_writeEpilogue() :
+*   Ends a frame.
+*   @return : nb of bytes written into dst (or an error code) */
+static size_t ZSTD_writeEpilogue(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity)
+{
+	BYTE *const ostart = (BYTE *)dst;
+	BYTE *op = ostart;
+	size_t fhSize = 0;
+
+	if (cctx->stage == ZSTDcs_created)
+		return ERROR(stage_wrong); /* init missing */
+
+	/* special case : empty frame */
+	if (cctx->stage == ZSTDcs_init) {
+		fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, cctx->params, 0, 0);
+		if (ZSTD_isError(fhSize))
+			return fhSize;
+		dstCapacity -= fhSize;
+		op += fhSize;
+		cctx->stage = ZSTDcs_ongoing;
+	}
+
+	if (cctx->stage != ZSTDcs_ending) {
+		/* write one last empty block, make it the "last" block */
+		U32 const cBlockHeader24 = 1 /* last block */ + (((U32)bt_raw) << 1) + 0;
+		if (dstCapacity < 4)
+			return ERROR(dstSize_tooSmall);
+		ZSTD_writeLE32(op, cBlockHeader24);
+		op += ZSTD_blockHeaderSize;
+		dstCapacity -= ZSTD_blockHeaderSize;
+	}
+
+	if (cctx->params.fParams.checksumFlag) {
+		U32 const checksum = (U32)xxh64_digest(&cctx->xxhState);
+		if (dstCapacity < 4)
+			return ERROR(dstSize_tooSmall);
+		ZSTD_writeLE32(op, checksum);
+		op += 4;
+	}
+
+	cctx->stage = ZSTDcs_created; /* return to "created but no init" status */
+	return op - ostart;
+}
+
+size_t ZSTD_compressEnd(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	size_t endResult;
+	size_t const cSize = ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 1, 1);
+	if (ZSTD_isError(cSize))
+		return cSize;
+	endResult = ZSTD_writeEpilogue(cctx, (char *)dst + cSize, dstCapacity - cSize);
+	if (ZSTD_isError(endResult))
+		return endResult;
+	return cSize + endResult;
+}
+
+static size_t ZSTD_compress_internal(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize,
+				     ZSTD_parameters params)
+{
+	CHECK_F(ZSTD_compressBegin_internal(cctx, dict, dictSize, params, srcSize));
+	return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
+}
+
+size_t ZSTD_compress_usingDict(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize,
+			       ZSTD_parameters params)
+{
+	return ZSTD_compress_internal(ctx, dst, dstCapacity, src, srcSize, dict, dictSize, params);
+}
+
+size_t ZSTD_compressCCtx(ZSTD_CCtx *ctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, ZSTD_parameters params)
+{
+	return ZSTD_compress_internal(ctx, dst, dstCapacity, src, srcSize, NULL, 0, params);
+}
+
+/* =====  Dictionary API  ===== */
+
+struct ZSTD_CDict_s {
+	void *dictBuffer;
+	const void *dictContent;
+	size_t dictContentSize;
+	ZSTD_CCtx *refContext;
+}; /* typedef'd tp ZSTD_CDict within "zstd.h" */
+
+size_t ZSTD_CDictWorkspaceBound(ZSTD_compressionParameters cParams) { return ZSTD_CCtxWorkspaceBound(cParams) + ZSTD_ALIGN(sizeof(ZSTD_CDict)); }
+
+static ZSTD_CDict *ZSTD_createCDict_advanced(const void *dictBuffer, size_t dictSize, unsigned byReference, ZSTD_parameters params, ZSTD_customMem customMem)
+{
+	if (!customMem.customAlloc || !customMem.customFree)
+		return NULL;
+
+	{
+		ZSTD_CDict *const cdict = (ZSTD_CDict *)ZSTD_malloc(sizeof(ZSTD_CDict), customMem);
+		ZSTD_CCtx *const cctx = ZSTD_createCCtx_advanced(customMem);
+
+		if (!cdict || !cctx) {
+			ZSTD_free(cdict, customMem);
+			ZSTD_freeCCtx(cctx);
+			return NULL;
+		}
+
+		if ((byReference) || (!dictBuffer) || (!dictSize)) {
+			cdict->dictBuffer = NULL;
+			cdict->dictContent = dictBuffer;
+		} else {
+			void *const internalBuffer = ZSTD_malloc(dictSize, customMem);
+			if (!internalBuffer) {
+				ZSTD_free(cctx, customMem);
+				ZSTD_free(cdict, customMem);
+				return NULL;
+			}
+			memcpy(internalBuffer, dictBuffer, dictSize);
+			cdict->dictBuffer = internalBuffer;
+			cdict->dictContent = internalBuffer;
+		}
+
+		{
+			size_t const errorCode = ZSTD_compressBegin_advanced(cctx, cdict->dictContent, dictSize, params, 0);
+			if (ZSTD_isError(errorCode)) {
+				ZSTD_free(cdict->dictBuffer, customMem);
+				ZSTD_free(cdict, customMem);
+				ZSTD_freeCCtx(cctx);
+				return NULL;
+			}
+		}
+
+		cdict->refContext = cctx;
+		cdict->dictContentSize = dictSize;
+		return cdict;
+	}
+}
+
+ZSTD_CDict *ZSTD_initCDict(const void *dict, size_t dictSize, ZSTD_parameters params, void *workspace, size_t workspaceSize)
+{
+	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
+	return ZSTD_createCDict_advanced(dict, dictSize, 1, params, stackMem);
+}
+
+size_t ZSTD_freeCDict(ZSTD_CDict *cdict)
+{
+	if (cdict == NULL)
+		return 0; /* support free on NULL */
+	{
+		ZSTD_customMem const cMem = cdict->refContext->customMem;
+		ZSTD_freeCCtx(cdict->refContext);
+		ZSTD_free(cdict->dictBuffer, cMem);
+		ZSTD_free(cdict, cMem);
+		return 0;
+	}
+}
+
+static ZSTD_parameters ZSTD_getParamsFromCDict(const ZSTD_CDict *cdict) { return ZSTD_getParamsFromCCtx(cdict->refContext); }
+
+size_t ZSTD_compressBegin_usingCDict(ZSTD_CCtx *cctx, const ZSTD_CDict *cdict, unsigned long long pledgedSrcSize)
+{
+	if (cdict->dictContentSize)
+		CHECK_F(ZSTD_copyCCtx(cctx, cdict->refContext, pledgedSrcSize))
+	else {
+		ZSTD_parameters params = cdict->refContext->params;
+		params.fParams.contentSizeFlag = (pledgedSrcSize > 0);
+		CHECK_F(ZSTD_compressBegin_advanced(cctx, NULL, 0, params, pledgedSrcSize));
+	}
+	return 0;
+}
+
+/*! ZSTD_compress_usingCDict() :
+*   Compression using a digested Dictionary.
+*   Faster startup than ZSTD_compress_usingDict(), recommended when same dictionary is used multiple times.
+*   Note that compression level is decided during dictionary creation */
+size_t ZSTD_compress_usingCDict(ZSTD_CCtx *cctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const ZSTD_CDict *cdict)
+{
+	CHECK_F(ZSTD_compressBegin_usingCDict(cctx, cdict, srcSize));
+
+	if (cdict->refContext->params.fParams.contentSizeFlag == 1) {
+		cctx->params.fParams.contentSizeFlag = 1;
+		cctx->frameContentSize = srcSize;
+	} else {
+		cctx->params.fParams.contentSizeFlag = 0;
+	}
+
+	return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
+}
+
+/* ******************************************************************
+*  Streaming
+********************************************************************/
+
+typedef enum { zcss_init, zcss_load, zcss_flush, zcss_final } ZSTD_cStreamStage;
+
+struct ZSTD_CStream_s {
+	ZSTD_CCtx *cctx;
+	ZSTD_CDict *cdictLocal;
+	const ZSTD_CDict *cdict;
+	char *inBuff;
+	size_t inBuffSize;
+	size_t inToCompress;
+	size_t inBuffPos;
+	size_t inBuffTarget;
+	size_t blockSize;
+	char *outBuff;
+	size_t outBuffSize;
+	size_t outBuffContentSize;
+	size_t outBuffFlushedSize;
+	ZSTD_cStreamStage stage;
+	U32 checksum;
+	U32 frameEnded;
+	U64 pledgedSrcSize;
+	U64 inputProcessed;
+	ZSTD_parameters params;
+	ZSTD_customMem customMem;
+}; /* typedef'd to ZSTD_CStream within "zstd.h" */
+
+size_t ZSTD_CStreamWorkspaceBound(ZSTD_compressionParameters cParams)
+{
+	size_t const inBuffSize = (size_t)1 << cParams.windowLog;
+	size_t const blockSize = MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, inBuffSize);
+	size_t const outBuffSize = ZSTD_compressBound(blockSize) + 1;
+
+	return ZSTD_CCtxWorkspaceBound(cParams) + ZSTD_ALIGN(sizeof(ZSTD_CStream)) + ZSTD_ALIGN(inBuffSize) + ZSTD_ALIGN(outBuffSize);
+}
+
+ZSTD_CStream *ZSTD_createCStream_advanced(ZSTD_customMem customMem)
+{
+	ZSTD_CStream *zcs;
+
+	if (!customMem.customAlloc || !customMem.customFree)
+		return NULL;
+
+	zcs = (ZSTD_CStream *)ZSTD_malloc(sizeof(ZSTD_CStream), customMem);
+	if (zcs == NULL)
+		return NULL;
+	memset(zcs, 0, sizeof(ZSTD_CStream));
+	memcpy(&zcs->customMem, &customMem, sizeof(ZSTD_customMem));
+	zcs->cctx = ZSTD_createCCtx_advanced(customMem);
+	if (zcs->cctx == NULL) {
+		ZSTD_freeCStream(zcs);
+		return NULL;
+	}
+	return zcs;
+}
+
+size_t ZSTD_freeCStream(ZSTD_CStream *zcs)
+{
+	if (zcs == NULL)
+		return 0; /* support free on NULL */
+	{
+		ZSTD_customMem const cMem = zcs->customMem;
+		ZSTD_freeCCtx(zcs->cctx);
+		zcs->cctx = NULL;
+		ZSTD_freeCDict(zcs->cdictLocal);
+		zcs->cdictLocal = NULL;
+		ZSTD_free(zcs->inBuff, cMem);
+		zcs->inBuff = NULL;
+		ZSTD_free(zcs->outBuff, cMem);
+		zcs->outBuff = NULL;
+		ZSTD_free(zcs, cMem);
+		return 0;
+	}
+}
+
+/*======   Initialization   ======*/
+
+size_t ZSTD_CStreamInSize(void) { return ZSTD_BLOCKSIZE_ABSOLUTEMAX; }
+size_t ZSTD_CStreamOutSize(void) { return ZSTD_compressBound(ZSTD_BLOCKSIZE_ABSOLUTEMAX) + ZSTD_blockHeaderSize + 4 /* 32-bits hash */; }
+
+static size_t ZSTD_resetCStream_internal(ZSTD_CStream *zcs, unsigned long long pledgedSrcSize)
+{
+	if (zcs->inBuffSize == 0)
+		return ERROR(stage_wrong); /* zcs has not been init at least once => can't reset */
+
+	if (zcs->cdict)
+		CHECK_F(ZSTD_compressBegin_usingCDict(zcs->cctx, zcs->cdict, pledgedSrcSize))
+	else
+		CHECK_F(ZSTD_compressBegin_advanced(zcs->cctx, NULL, 0, zcs->params, pledgedSrcSize));
+
+	zcs->inToCompress = 0;
+	zcs->inBuffPos = 0;
+	zcs->inBuffTarget = zcs->blockSize;
+	zcs->outBuffContentSize = zcs->outBuffFlushedSize = 0;
+	zcs->stage = zcss_load;
+	zcs->frameEnded = 0;
+	zcs->pledgedSrcSize = pledgedSrcSize;
+	zcs->inputProcessed = 0;
+	return 0; /* ready to go */
+}
+
+size_t ZSTD_resetCStream(ZSTD_CStream *zcs, unsigned long long pledgedSrcSize)
+{
+
+	zcs->params.fParams.contentSizeFlag = (pledgedSrcSize > 0);
+
+	return ZSTD_resetCStream_internal(zcs, pledgedSrcSize);
+}
+
+static size_t ZSTD_initCStream_advanced(ZSTD_CStream *zcs, const void *dict, size_t dictSize, ZSTD_parameters params, unsigned long long pledgedSrcSize)
+{
+	/* allocate buffers */
+	{
+		size_t const neededInBuffSize = (size_t)1 << params.cParams.windowLog;
+		if (zcs->inBuffSize < neededInBuffSize) {
+			zcs->inBuffSize = neededInBuffSize;
+			ZSTD_free(zcs->inBuff, zcs->customMem);
+			zcs->inBuff = (char *)ZSTD_malloc(neededInBuffSize, zcs->customMem);
+			if (zcs->inBuff == NULL)
+				return ERROR(memory_allocation);
+		}
+		zcs->blockSize = MIN(ZSTD_BLOCKSIZE_ABSOLUTEMAX, neededInBuffSize);
+	}
+	if (zcs->outBuffSize < ZSTD_compressBound(zcs->blockSize) + 1) {
+		zcs->outBuffSize = ZSTD_compressBound(zcs->blockSize) + 1;
+		ZSTD_free(zcs->outBuff, zcs->customMem);
+		zcs->outBuff = (char *)ZSTD_malloc(zcs->outBuffSize, zcs->customMem);
+		if (zcs->outBuff == NULL)
+			return ERROR(memory_allocation);
+	}
+
+	if (dict && dictSize >= 8) {
+		ZSTD_freeCDict(zcs->cdictLocal);
+		zcs->cdictLocal = ZSTD_createCDict_advanced(dict, dictSize, 0, params, zcs->customMem);
+		if (zcs->cdictLocal == NULL)
+			return ERROR(memory_allocation);
+		zcs->cdict = zcs->cdictLocal;
+	} else
+		zcs->cdict = NULL;
+
+	zcs->checksum = params.fParams.checksumFlag > 0;
+	zcs->params = params;
+
+	return ZSTD_resetCStream_internal(zcs, pledgedSrcSize);
+}
+
+ZSTD_CStream *ZSTD_initCStream(ZSTD_parameters params, unsigned long long pledgedSrcSize, void *workspace, size_t workspaceSize)
+{
+	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
+	ZSTD_CStream *const zcs = ZSTD_createCStream_advanced(stackMem);
+	if (zcs) {
+		size_t const code = ZSTD_initCStream_advanced(zcs, NULL, 0, params, pledgedSrcSize);
+		if (ZSTD_isError(code)) {
+			return NULL;
+		}
+	}
+	return zcs;
+}
+
+ZSTD_CStream *ZSTD_initCStream_usingCDict(const ZSTD_CDict *cdict, unsigned long long pledgedSrcSize, void *workspace, size_t workspaceSize)
+{
+	ZSTD_parameters const params = ZSTD_getParamsFromCDict(cdict);
+	ZSTD_CStream *const zcs = ZSTD_initCStream(params, pledgedSrcSize, workspace, workspaceSize);
+	if (zcs) {
+		zcs->cdict = cdict;
+		if (ZSTD_isError(ZSTD_resetCStream_internal(zcs, pledgedSrcSize))) {
+			return NULL;
+		}
+	}
+	return zcs;
+}
+
+/*======   Compression   ======*/
+
+typedef enum { zsf_gather, zsf_flush, zsf_end } ZSTD_flush_e;
+
+ZSTD_STATIC size_t ZSTD_limitCopy(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	size_t const length = MIN(dstCapacity, srcSize);
+	memcpy(dst, src, length);
+	return length;
+}
+
+static size_t ZSTD_compressStream_generic(ZSTD_CStream *zcs, void *dst, size_t *dstCapacityPtr, const void *src, size_t *srcSizePtr, ZSTD_flush_e const flush)
+{
+	U32 someMoreWork = 1;
+	const char *const istart = (const char *)src;
+	const char *const iend = istart + *srcSizePtr;
+	const char *ip = istart;
+	char *const ostart = (char *)dst;
+	char *const oend = ostart + *dstCapacityPtr;
+	char *op = ostart;
+
+	while (someMoreWork) {
+		switch (zcs->stage) {
+		case zcss_init:
+			return ERROR(init_missing); /* call ZBUFF_compressInit() first ! */
+
+		case zcss_load:
+			/* complete inBuffer */
+			{
+				size_t const toLoad = zcs->inBuffTarget - zcs->inBuffPos;
+				size_t const loaded = ZSTD_limitCopy(zcs->inBuff + zcs->inBuffPos, toLoad, ip, iend - ip);
+				zcs->inBuffPos += loaded;
+				ip += loaded;
+				if ((zcs->inBuffPos == zcs->inToCompress) || (!flush && (toLoad != loaded))) {
+					someMoreWork = 0;
+					break; /* not enough input to get a full block : stop there, wait for more */
+				}
+			}
+			/* compress curr block (note : this stage cannot be stopped in the middle) */
+			{
+				void *cDst;
+				size_t cSize;
+				size_t const iSize = zcs->inBuffPos - zcs->inToCompress;
+				size_t oSize = oend - op;
+				if (oSize >= ZSTD_compressBound(iSize))
+					cDst = op; /* compress directly into output buffer (avoid flush stage) */
+				else
+					cDst = zcs->outBuff, oSize = zcs->outBuffSize;
+				cSize = (flush == zsf_end) ? ZSTD_compressEnd(zcs->cctx, cDst, oSize, zcs->inBuff + zcs->inToCompress, iSize)
+							   : ZSTD_compressContinue(zcs->cctx, cDst, oSize, zcs->inBuff + zcs->inToCompress, iSize);
+				if (ZSTD_isError(cSize))
+					return cSize;
+				if (flush == zsf_end)
+					zcs->frameEnded = 1;
+				/* prepare next block */
+				zcs->inBuffTarget = zcs->inBuffPos + zcs->blockSize;
+				if (zcs->inBuffTarget > zcs->inBuffSize)
+					zcs->inBuffPos = 0, zcs->inBuffTarget = zcs->blockSize; /* note : inBuffSize >= blockSize */
+				zcs->inToCompress = zcs->inBuffPos;
+				if (cDst == op) {
+					op += cSize;
+					break;
+				} /* no need to flush */
+				zcs->outBuffContentSize = cSize;
+				zcs->outBuffFlushedSize = 0;
+				zcs->stage = zcss_flush; /* pass-through to flush stage */
+			}
+
+		case zcss_flush: {
+			size_t const toFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
+			size_t const flushed = ZSTD_limitCopy(op, oend - op, zcs->outBuff + zcs->outBuffFlushedSize, toFlush);
+			op += flushed;
+			zcs->outBuffFlushedSize += flushed;
+			if (toFlush != flushed) {
+				someMoreWork = 0;
+				break;
+			} /* dst too small to store flushed data : stop there */
+			zcs->outBuffContentSize = zcs->outBuffFlushedSize = 0;
+			zcs->stage = zcss_load;
+			break;
+		}
+
+		case zcss_final:
+			someMoreWork = 0; /* do nothing */
+			break;
+
+		default:
+			return ERROR(GENERIC); /* impossible */
+		}
+	}
+
+	*srcSizePtr = ip - istart;
+	*dstCapacityPtr = op - ostart;
+	zcs->inputProcessed += *srcSizePtr;
+	if (zcs->frameEnded)
+		return 0;
+	{
+		size_t hintInSize = zcs->inBuffTarget - zcs->inBuffPos;
+		if (hintInSize == 0)
+			hintInSize = zcs->blockSize;
+		return hintInSize;
+	}
+}
+
+size_t ZSTD_compressStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output, ZSTD_inBuffer *input)
+{
+	size_t sizeRead = input->size - input->pos;
+	size_t sizeWritten = output->size - output->pos;
+	size_t const result =
+	    ZSTD_compressStream_generic(zcs, (char *)(output->dst) + output->pos, &sizeWritten, (const char *)(input->src) + input->pos, &sizeRead, zsf_gather);
+	input->pos += sizeRead;
+	output->pos += sizeWritten;
+	return result;
+}
+
+/*======   Finalize   ======*/
+
+/*! ZSTD_flushStream() :
+*   @return : amount of data remaining to flush */
+size_t ZSTD_flushStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output)
+{
+	size_t srcSize = 0;
+	size_t sizeWritten = output->size - output->pos;
+	size_t const result = ZSTD_compressStream_generic(zcs, (char *)(output->dst) + output->pos, &sizeWritten, &srcSize,
+							  &srcSize, /* use a valid src address instead of NULL */
+							  zsf_flush);
+	output->pos += sizeWritten;
+	if (ZSTD_isError(result))
+		return result;
+	return zcs->outBuffContentSize - zcs->outBuffFlushedSize; /* remaining to flush */
+}
+
+size_t ZSTD_endStream(ZSTD_CStream *zcs, ZSTD_outBuffer *output)
+{
+	BYTE *const ostart = (BYTE *)(output->dst) + output->pos;
+	BYTE *const oend = (BYTE *)(output->dst) + output->size;
+	BYTE *op = ostart;
+
+	if ((zcs->pledgedSrcSize) && (zcs->inputProcessed != zcs->pledgedSrcSize))
+		return ERROR(srcSize_wrong); /* pledgedSrcSize not respected */
+
+	if (zcs->stage != zcss_final) {
+		/* flush whatever remains */
+		size_t srcSize = 0;
+		size_t sizeWritten = output->size - output->pos;
+		size_t const notEnded =
+		    ZSTD_compressStream_generic(zcs, ostart, &sizeWritten, &srcSize, &srcSize, zsf_end); /* use a valid src address instead of NULL */
+		size_t const remainingToFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
+		op += sizeWritten;
+		if (remainingToFlush) {
+			output->pos += sizeWritten;
+			return remainingToFlush + ZSTD_BLOCKHEADERSIZE /* final empty block */ + (zcs->checksum * 4);
+		}
+		/* create epilogue */
+		zcs->stage = zcss_final;
+		zcs->outBuffContentSize = !notEnded ? 0 : ZSTD_compressEnd(zcs->cctx, zcs->outBuff, zcs->outBuffSize, NULL,
+									   0); /* write epilogue, including final empty block, into outBuff */
+	}
+
+	/* flush epilogue */
+	{
+		size_t const toFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
+		size_t const flushed = ZSTD_limitCopy(op, oend - op, zcs->outBuff + zcs->outBuffFlushedSize, toFlush);
+		op += flushed;
+		zcs->outBuffFlushedSize += flushed;
+		output->pos += op - ostart;
+		if (toFlush == flushed)
+			zcs->stage = zcss_init; /* end reached */
+		return toFlush - flushed;
+	}
+}
+
+/*-=====  Pre-defined compression levels  =====-*/
+
+#define ZSTD_DEFAULT_CLEVEL 1
+#define ZSTD_MAX_CLEVEL 22
+int ZSTD_maxCLevel(void) { return ZSTD_MAX_CLEVEL; }
+
+static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEVEL + 1] = {
+    {
+	/* "default" */
+	/* W,  C,  H,  S,  L, TL, strat */
+	{18, 12, 12, 1, 7, 16, ZSTD_fast},    /* level  0 - never used */
+	{19, 13, 14, 1, 7, 16, ZSTD_fast},    /* level  1 */
+	{19, 15, 16, 1, 6, 16, ZSTD_fast},    /* level  2 */
+	{20, 16, 17, 1, 5, 16, ZSTD_dfast},   /* level  3.*/
+	{20, 18, 18, 1, 5, 16, ZSTD_dfast},   /* level  4.*/
+	{20, 15, 18, 3, 5, 16, ZSTD_greedy},  /* level  5 */
+	{21, 16, 19, 2, 5, 16, ZSTD_lazy},    /* level  6 */
+	{21, 17, 20, 3, 5, 16, ZSTD_lazy},    /* level  7 */
+	{21, 18, 20, 3, 5, 16, ZSTD_lazy2},   /* level  8 */
+	{21, 20, 20, 3, 5, 16, ZSTD_lazy2},   /* level  9 */
+	{21, 19, 21, 4, 5, 16, ZSTD_lazy2},   /* level 10 */
+	{22, 20, 22, 4, 5, 16, ZSTD_lazy2},   /* level 11 */
+	{22, 20, 22, 5, 5, 16, ZSTD_lazy2},   /* level 12 */
+	{22, 21, 22, 5, 5, 16, ZSTD_lazy2},   /* level 13 */
+	{22, 21, 22, 6, 5, 16, ZSTD_lazy2},   /* level 14 */
+	{22, 21, 21, 5, 5, 16, ZSTD_btlazy2}, /* level 15 */
+	{23, 22, 22, 5, 5, 16, ZSTD_btlazy2}, /* level 16 */
+	{23, 21, 22, 4, 5, 24, ZSTD_btopt},   /* level 17 */
+	{23, 23, 22, 6, 5, 32, ZSTD_btopt},   /* level 18 */
+	{23, 23, 22, 6, 3, 48, ZSTD_btopt},   /* level 19 */
+	{25, 25, 23, 7, 3, 64, ZSTD_btopt2},  /* level 20 */
+	{26, 26, 23, 7, 3, 256, ZSTD_btopt2}, /* level 21 */
+	{27, 27, 25, 9, 3, 512, ZSTD_btopt2}, /* level 22 */
+    },
+    {
+	/* for srcSize <= 256 KB */
+	/* W,  C,  H,  S,  L,  T, strat */
+	{0, 0, 0, 0, 0, 0, ZSTD_fast},	 /* level  0 - not used */
+	{18, 13, 14, 1, 6, 8, ZSTD_fast},      /* level  1 */
+	{18, 14, 13, 1, 5, 8, ZSTD_dfast},     /* level  2 */
+	{18, 16, 15, 1, 5, 8, ZSTD_dfast},     /* level  3 */
+	{18, 15, 17, 1, 5, 8, ZSTD_greedy},    /* level  4.*/
+	{18, 16, 17, 4, 5, 8, ZSTD_greedy},    /* level  5.*/
+	{18, 16, 17, 3, 5, 8, ZSTD_lazy},      /* level  6.*/
+	{18, 17, 17, 4, 4, 8, ZSTD_lazy},      /* level  7 */
+	{18, 17, 17, 4, 4, 8, ZSTD_lazy2},     /* level  8 */
+	{18, 17, 17, 5, 4, 8, ZSTD_lazy2},     /* level  9 */
+	{18, 17, 17, 6, 4, 8, ZSTD_lazy2},     /* level 10 */
+	{18, 18, 17, 6, 4, 8, ZSTD_lazy2},     /* level 11.*/
+	{18, 18, 17, 7, 4, 8, ZSTD_lazy2},     /* level 12.*/
+	{18, 19, 17, 6, 4, 8, ZSTD_btlazy2},   /* level 13 */
+	{18, 18, 18, 4, 4, 16, ZSTD_btopt},    /* level 14.*/
+	{18, 18, 18, 4, 3, 16, ZSTD_btopt},    /* level 15.*/
+	{18, 19, 18, 6, 3, 32, ZSTD_btopt},    /* level 16.*/
+	{18, 19, 18, 8, 3, 64, ZSTD_btopt},    /* level 17.*/
+	{18, 19, 18, 9, 3, 128, ZSTD_btopt},   /* level 18.*/
+	{18, 19, 18, 10, 3, 256, ZSTD_btopt},  /* level 19.*/
+	{18, 19, 18, 11, 3, 512, ZSTD_btopt2}, /* level 20.*/
+	{18, 19, 18, 12, 3, 512, ZSTD_btopt2}, /* level 21.*/
+	{18, 19, 18, 13, 3, 512, ZSTD_btopt2}, /* level 22.*/
+    },
+    {
+	/* for srcSize <= 128 KB */
+	/* W,  C,  H,  S,  L,  T, strat */
+	{17, 12, 12, 1, 7, 8, ZSTD_fast},      /* level  0 - not used */
+	{17, 12, 13, 1, 6, 8, ZSTD_fast},      /* level  1 */
+	{17, 13, 16, 1, 5, 8, ZSTD_fast},      /* level  2 */
+	{17, 16, 16, 2, 5, 8, ZSTD_dfast},     /* level  3 */
+	{17, 13, 15, 3, 4, 8, ZSTD_greedy},    /* level  4 */
+	{17, 15, 17, 4, 4, 8, ZSTD_greedy},    /* level  5 */
+	{17, 16, 17, 3, 4, 8, ZSTD_lazy},      /* level  6 */
+	{17, 15, 17, 4, 4, 8, ZSTD_lazy2},     /* level  7 */
+	{17, 17, 17, 4, 4, 8, ZSTD_lazy2},     /* level  8 */
+	{17, 17, 17, 5, 4, 8, ZSTD_lazy2},     /* level  9 */
+	{17, 17, 17, 6, 4, 8, ZSTD_lazy2},     /* level 10 */
+	{17, 17, 17, 7, 4, 8, ZSTD_lazy2},     /* level 11 */
+	{17, 17, 17, 8, 4, 8, ZSTD_lazy2},     /* level 12 */
+	{17, 18, 17, 6, 4, 8, ZSTD_btlazy2},   /* level 13.*/
+	{17, 17, 17, 7, 3, 8, ZSTD_btopt},     /* level 14.*/
+	{17, 17, 17, 7, 3, 16, ZSTD_btopt},    /* level 15.*/
+	{17, 18, 17, 7, 3, 32, ZSTD_btopt},    /* level 16.*/
+	{17, 18, 17, 7, 3, 64, ZSTD_btopt},    /* level 17.*/
+	{17, 18, 17, 7, 3, 256, ZSTD_btopt},   /* level 18.*/
+	{17, 18, 17, 8, 3, 256, ZSTD_btopt},   /* level 19.*/
+	{17, 18, 17, 9, 3, 256, ZSTD_btopt2},  /* level 20.*/
+	{17, 18, 17, 10, 3, 256, ZSTD_btopt2}, /* level 21.*/
+	{17, 18, 17, 11, 3, 512, ZSTD_btopt2}, /* level 22.*/
+    },
+    {
+	/* for srcSize <= 16 KB */
+	/* W,  C,  H,  S,  L,  T, strat */
+	{14, 12, 12, 1, 7, 6, ZSTD_fast},      /* level  0 - not used */
+	{14, 14, 14, 1, 6, 6, ZSTD_fast},      /* level  1 */
+	{14, 14, 14, 1, 4, 6, ZSTD_fast},      /* level  2 */
+	{14, 14, 14, 1, 4, 6, ZSTD_dfast},     /* level  3.*/
+	{14, 14, 14, 4, 4, 6, ZSTD_greedy},    /* level  4.*/
+	{14, 14, 14, 3, 4, 6, ZSTD_lazy},      /* level  5.*/
+	{14, 14, 14, 4, 4, 6, ZSTD_lazy2},     /* level  6 */
+	{14, 14, 14, 5, 4, 6, ZSTD_lazy2},     /* level  7 */
+	{14, 14, 14, 6, 4, 6, ZSTD_lazy2},     /* level  8.*/
+	{14, 15, 14, 6, 4, 6, ZSTD_btlazy2},   /* level  9.*/
+	{14, 15, 14, 3, 3, 6, ZSTD_btopt},     /* level 10.*/
+	{14, 15, 14, 6, 3, 8, ZSTD_btopt},     /* level 11.*/
+	{14, 15, 14, 6, 3, 16, ZSTD_btopt},    /* level 12.*/
+	{14, 15, 14, 6, 3, 24, ZSTD_btopt},    /* level 13.*/
+	{14, 15, 15, 6, 3, 48, ZSTD_btopt},    /* level 14.*/
+	{14, 15, 15, 6, 3, 64, ZSTD_btopt},    /* level 15.*/
+	{14, 15, 15, 6, 3, 96, ZSTD_btopt},    /* level 16.*/
+	{14, 15, 15, 6, 3, 128, ZSTD_btopt},   /* level 17.*/
+	{14, 15, 15, 6, 3, 256, ZSTD_btopt},   /* level 18.*/
+	{14, 15, 15, 7, 3, 256, ZSTD_btopt},   /* level 19.*/
+	{14, 15, 15, 8, 3, 256, ZSTD_btopt2},  /* level 20.*/
+	{14, 15, 15, 9, 3, 256, ZSTD_btopt2},  /* level 21.*/
+	{14, 15, 15, 10, 3, 256, ZSTD_btopt2}, /* level 22.*/
+    },
+};
+
+/*! ZSTD_getCParams() :
+*   @return ZSTD_compressionParameters structure for a selected compression level, `srcSize` and `dictSize`.
+*   Size values are optional, provide 0 if not known or unused */
+ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long srcSize, size_t dictSize)
+{
+	ZSTD_compressionParameters cp;
+	size_t const addedSize = srcSize ? 0 : 500;
+	U64 const rSize = srcSize + dictSize ? srcSize + dictSize + addedSize : (U64)-1;
+	U32 const tableID = (rSize <= 256 KB) + (rSize <= 128 KB) + (rSize <= 16 KB); /* intentional underflow for srcSizeHint == 0 */
+	if (compressionLevel <= 0)
+		compressionLevel = ZSTD_DEFAULT_CLEVEL; /* 0 == default; no negative compressionLevel yet */
+	if (compressionLevel > ZSTD_MAX_CLEVEL)
+		compressionLevel = ZSTD_MAX_CLEVEL;
+	cp = ZSTD_defaultCParameters[tableID][compressionLevel];
+	if (ZSTD_32bits()) { /* auto-correction, for 32-bits mode */
+		if (cp.windowLog > ZSTD_WINDOWLOG_MAX)
+			cp.windowLog = ZSTD_WINDOWLOG_MAX;
+		if (cp.chainLog > ZSTD_CHAINLOG_MAX)
+			cp.chainLog = ZSTD_CHAINLOG_MAX;
+		if (cp.hashLog > ZSTD_HASHLOG_MAX)
+			cp.hashLog = ZSTD_HASHLOG_MAX;
+	}
+	cp = ZSTD_adjustCParams(cp, srcSize, dictSize);
+	return cp;
+}
+
+/*! ZSTD_getParams() :
+*   same as ZSTD_getCParams(), but @return a `ZSTD_parameters` object (instead of `ZSTD_compressionParameters`).
+*   All fields of `ZSTD_frameParameters` are set to default (0) */
+ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long srcSize, size_t dictSize)
+{
+	ZSTD_parameters params;
+	ZSTD_compressionParameters const cParams = ZSTD_getCParams(compressionLevel, srcSize, dictSize);
+	memset(&params, 0, sizeof(params));
+	params.cParams = cParams;
+	return params;
+}
+
+EXPORT_SYMBOL(ZSTD_maxCLevel);
+EXPORT_SYMBOL(ZSTD_compressBound);
+
+EXPORT_SYMBOL(ZSTD_CCtxWorkspaceBound);
+EXPORT_SYMBOL(ZSTD_initCCtx);
+EXPORT_SYMBOL(ZSTD_compressCCtx);
+EXPORT_SYMBOL(ZSTD_compress_usingDict);
+
+EXPORT_SYMBOL(ZSTD_CDictWorkspaceBound);
+EXPORT_SYMBOL(ZSTD_initCDict);
+EXPORT_SYMBOL(ZSTD_compress_usingCDict);
+
+EXPORT_SYMBOL(ZSTD_CStreamWorkspaceBound);
+EXPORT_SYMBOL(ZSTD_initCStream);
+EXPORT_SYMBOL(ZSTD_initCStream_usingCDict);
+EXPORT_SYMBOL(ZSTD_resetCStream);
+EXPORT_SYMBOL(ZSTD_compressStream);
+EXPORT_SYMBOL(ZSTD_flushStream);
+EXPORT_SYMBOL(ZSTD_endStream);
+EXPORT_SYMBOL(ZSTD_CStreamInSize);
+EXPORT_SYMBOL(ZSTD_CStreamOutSize);
+
+EXPORT_SYMBOL(ZSTD_getCParams);
+EXPORT_SYMBOL(ZSTD_getParams);
+EXPORT_SYMBOL(ZSTD_checkCParams);
+EXPORT_SYMBOL(ZSTD_adjustCParams);
+
+EXPORT_SYMBOL(ZSTD_compressBegin);
+EXPORT_SYMBOL(ZSTD_compressBegin_usingDict);
+EXPORT_SYMBOL(ZSTD_compressBegin_advanced);
+EXPORT_SYMBOL(ZSTD_copyCCtx);
+EXPORT_SYMBOL(ZSTD_compressBegin_usingCDict);
+EXPORT_SYMBOL(ZSTD_compressContinue);
+EXPORT_SYMBOL(ZSTD_compressEnd);
+
+EXPORT_SYMBOL(ZSTD_getBlockSizeMax);
+EXPORT_SYMBOL(ZSTD_compressBlock);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("Zstd Compressor");
diff --git a/lib/zstd/decompress.c b/lib/zstd/decompress.c
new file mode 100644
index 000000000000..b17846725ca0
--- /dev/null
+++ b/lib/zstd/decompress.c
@@ -0,0 +1,2528 @@
+/**
+ * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd.
+ * An additional grant of patent rights can be found in the PATENTS file in the
+ * same directory.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ */
+
+/* ***************************************************************
+*  Tuning parameters
+*****************************************************************/
+/*!
+*  MAXWINDOWSIZE_DEFAULT :
+*  maximum window size accepted by DStream, by default.
+*  Frames requiring more memory will be rejected.
+*/
+#ifndef ZSTD_MAXWINDOWSIZE_DEFAULT
+#define ZSTD_MAXWINDOWSIZE_DEFAULT ((1 << ZSTD_WINDOWLOG_MAX) + 1) /* defined within zstd.h */
+#endif
+
+/*-*******************************************************
+*  Dependencies
+*********************************************************/
+#include "fse.h"
+#include "huf.h"
+#include "mem.h" /* low level memory routines */
+#include "zstd_internal.h"
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/string.h> /* memcpy, memmove, memset */
+
+#define ZSTD_PREFETCH(ptr) __builtin_prefetch(ptr, 0, 0)
+
+/*-*************************************
+*  Macros
+***************************************/
+#define ZSTD_isError ERR_isError /* for inlining */
+#define FSE_isError ERR_isError
+#define HUF_isError ERR_isError
+
+/*_*******************************************************
+*  Memory operations
+**********************************************************/
+static void ZSTD_copy4(void *dst, const void *src) { memcpy(dst, src, 4); }
+
+/*-*************************************************************
+*   Context management
+***************************************************************/
+typedef enum {
+	ZSTDds_getFrameHeaderSize,
+	ZSTDds_decodeFrameHeader,
+	ZSTDds_decodeBlockHeader,
+	ZSTDds_decompressBlock,
+	ZSTDds_decompressLastBlock,
+	ZSTDds_checkChecksum,
+	ZSTDds_decodeSkippableHeader,
+	ZSTDds_skipFrame
+} ZSTD_dStage;
+
+typedef struct {
+	FSE_DTable LLTable[FSE_DTABLE_SIZE_U32(LLFSELog)];
+	FSE_DTable OFTable[FSE_DTABLE_SIZE_U32(OffFSELog)];
+	FSE_DTable MLTable[FSE_DTABLE_SIZE_U32(MLFSELog)];
+	HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)]; /* can accommodate HUF_decompress4X */
+	U64 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32 / 2];
+	U32 rep[ZSTD_REP_NUM];
+} ZSTD_entropyTables_t;
+
+struct ZSTD_DCtx_s {
+	const FSE_DTable *LLTptr;
+	const FSE_DTable *MLTptr;
+	const FSE_DTable *OFTptr;
+	const HUF_DTable *HUFptr;
+	ZSTD_entropyTables_t entropy;
+	const void *previousDstEnd; /* detect continuity */
+	const void *base;	   /* start of curr segment */
+	const void *vBase;	  /* virtual start of previous segment if it was just before curr one */
+	const void *dictEnd;	/* end of previous segment */
+	size_t expected;
+	ZSTD_frameParams fParams;
+	blockType_e bType; /* used in ZSTD_decompressContinue(), to transfer blockType between header decoding and block decoding stages */
+	ZSTD_dStage stage;
+	U32 litEntropy;
+	U32 fseEntropy;
+	struct xxh64_state xxhState;
+	size_t headerSize;
+	U32 dictID;
+	const BYTE *litPtr;
+	ZSTD_customMem customMem;
+	size_t litSize;
+	size_t rleSize;
+	BYTE litBuffer[ZSTD_BLOCKSIZE_ABSOLUTEMAX + WILDCOPY_OVERLENGTH];
+	BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];
+}; /* typedef'd to ZSTD_DCtx within "zstd.h" */
+
+size_t ZSTD_DCtxWorkspaceBound(void) { return ZSTD_ALIGN(sizeof(ZSTD_stack)) + ZSTD_ALIGN(sizeof(ZSTD_DCtx)); }
+
+size_t ZSTD_decompressBegin(ZSTD_DCtx *dctx)
+{
+	dctx->expected = ZSTD_frameHeaderSize_prefix;
+	dctx->stage = ZSTDds_getFrameHeaderSize;
+	dctx->previousDstEnd = NULL;
+	dctx->base = NULL;
+	dctx->vBase = NULL;
+	dctx->dictEnd = NULL;
+	dctx->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001); /* cover both little and big endian */
+	dctx->litEntropy = dctx->fseEntropy = 0;
+	dctx->dictID = 0;
+	ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.rep) == sizeof(repStartValue));
+	memcpy(dctx->entropy.rep, repStartValue, sizeof(repStartValue)); /* initial repcodes */
+	dctx->LLTptr = dctx->entropy.LLTable;
+	dctx->MLTptr = dctx->entropy.MLTable;
+	dctx->OFTptr = dctx->entropy.OFTable;
+	dctx->HUFptr = dctx->entropy.hufTable;
+	return 0;
+}
+
+ZSTD_DCtx *ZSTD_createDCtx_advanced(ZSTD_customMem customMem)
+{
+	ZSTD_DCtx *dctx;
+
+	if (!customMem.customAlloc || !customMem.customFree)
+		return NULL;
+
+	dctx = (ZSTD_DCtx *)ZSTD_malloc(sizeof(ZSTD_DCtx), customMem);
+	if (!dctx)
+		return NULL;
+	memcpy(&dctx->customMem, &customMem, sizeof(customMem));
+	ZSTD_decompressBegin(dctx);
+	return dctx;
+}
+
+ZSTD_DCtx *ZSTD_initDCtx(void *workspace, size_t workspaceSize)
+{
+	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
+	return ZSTD_createDCtx_advanced(stackMem);
+}
+
+size_t ZSTD_freeDCtx(ZSTD_DCtx *dctx)
+{
+	if (dctx == NULL)
+		return 0; /* support free on NULL */
+	ZSTD_free(dctx, dctx->customMem);
+	return 0; /* reserved as a potential error code in the future */
+}
+
+void ZSTD_copyDCtx(ZSTD_DCtx *dstDCtx, const ZSTD_DCtx *srcDCtx)
+{
+	size_t const workSpaceSize = (ZSTD_BLOCKSIZE_ABSOLUTEMAX + WILDCOPY_OVERLENGTH) + ZSTD_frameHeaderSize_max;
+	memcpy(dstDCtx, srcDCtx, sizeof(ZSTD_DCtx) - workSpaceSize); /* no need to copy workspace */
+}
+
+static void ZSTD_refDDict(ZSTD_DCtx *dstDCtx, const ZSTD_DDict *ddict);
+
+/*-*************************************************************
+*   Decompression section
+***************************************************************/
+
+/*! ZSTD_isFrame() :
+ *  Tells if the content of `buffer` starts with a valid Frame Identifier.
+ *  Note : Frame Identifier is 4 bytes. If `size < 4`, @return will always be 0.
+ *  Note 2 : Legacy Frame Identifiers are considered valid only if Legacy Support is enabled.
+ *  Note 3 : Skippable Frame Identifiers are considered valid. */
+unsigned ZSTD_isFrame(const void *buffer, size_t size)
+{
+	if (size < 4)
+		return 0;
+	{
+		U32 const magic = ZSTD_readLE32(buffer);
+		if (magic == ZSTD_MAGICNUMBER)
+			return 1;
+		if ((magic & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START)
+			return 1;
+	}
+	return 0;
+}
+
+/** ZSTD_frameHeaderSize() :
+*   srcSize must be >= ZSTD_frameHeaderSize_prefix.
+*   @return : size of the Frame Header */
+static size_t ZSTD_frameHeaderSize(const void *src, size_t srcSize)
+{
+	if (srcSize < ZSTD_frameHeaderSize_prefix)
+		return ERROR(srcSize_wrong);
+	{
+		BYTE const fhd = ((const BYTE *)src)[4];
+		U32 const dictID = fhd & 3;
+		U32 const singleSegment = (fhd >> 5) & 1;
+		U32 const fcsId = fhd >> 6;
+		return ZSTD_frameHeaderSize_prefix + !singleSegment + ZSTD_did_fieldSize[dictID] + ZSTD_fcs_fieldSize[fcsId] + (singleSegment && !fcsId);
+	}
+}
+
+/** ZSTD_getFrameParams() :
+*   decode Frame Header, or require larger `srcSize`.
+*   @return : 0, `fparamsPtr` is correctly filled,
+*            >0, `srcSize` is too small, result is expected `srcSize`,
+*             or an error code, which can be tested using ZSTD_isError() */
+size_t ZSTD_getFrameParams(ZSTD_frameParams *fparamsPtr, const void *src, size_t srcSize)
+{
+	const BYTE *ip = (const BYTE *)src;
+
+	if (srcSize < ZSTD_frameHeaderSize_prefix)
+		return ZSTD_frameHeaderSize_prefix;
+	if (ZSTD_readLE32(src) != ZSTD_MAGICNUMBER) {
+		if ((ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
+			if (srcSize < ZSTD_skippableHeaderSize)
+				return ZSTD_skippableHeaderSize; /* magic number + skippable frame length */
+			memset(fparamsPtr, 0, sizeof(*fparamsPtr));
+			fparamsPtr->frameContentSize = ZSTD_readLE32((const char *)src + 4);
+			fparamsPtr->windowSize = 0; /* windowSize==0 means a frame is skippable */
+			return 0;
+		}
+		return ERROR(prefix_unknown);
+	}
+
+	/* ensure there is enough `srcSize` to fully read/decode frame header */
+	{
+		size_t const fhsize = ZSTD_frameHeaderSize(src, srcSize);
+		if (srcSize < fhsize)
+			return fhsize;
+	}
+
+	{
+		BYTE const fhdByte = ip[4];
+		size_t pos = 5;
+		U32 const dictIDSizeCode = fhdByte & 3;
+		U32 const checksumFlag = (fhdByte >> 2) & 1;
+		U32 const singleSegment = (fhdByte >> 5) & 1;
+		U32 const fcsID = fhdByte >> 6;
+		U32 const windowSizeMax = 1U << ZSTD_WINDOWLOG_MAX;
+		U32 windowSize = 0;
+		U32 dictID = 0;
+		U64 frameContentSize = 0;
+		if ((fhdByte & 0x08) != 0)
+			return ERROR(frameParameter_unsupported); /* reserved bits, which must be zero */
+		if (!singleSegment) {
+			BYTE const wlByte = ip[pos++];
+			U32 const windowLog = (wlByte >> 3) + ZSTD_WINDOWLOG_ABSOLUTEMIN;
+			if (windowLog > ZSTD_WINDOWLOG_MAX)
+				return ERROR(frameParameter_windowTooLarge); /* avoids issue with 1 << windowLog */
+			windowSize = (1U << windowLog);
+			windowSize += (windowSize >> 3) * (wlByte & 7);
+		}
+
+		switch (dictIDSizeCode) {
+		default: /* impossible */
+		case 0: break;
+		case 1:
+			dictID = ip[pos];
+			pos++;
+			break;
+		case 2:
+			dictID = ZSTD_readLE16(ip + pos);
+			pos += 2;
+			break;
+		case 3:
+			dictID = ZSTD_readLE32(ip + pos);
+			pos += 4;
+			break;
+		}
+		switch (fcsID) {
+		default: /* impossible */
+		case 0:
+			if (singleSegment)
+				frameContentSize = ip[pos];
+			break;
+		case 1: frameContentSize = ZSTD_readLE16(ip + pos) + 256; break;
+		case 2: frameContentSize = ZSTD_readLE32(ip + pos); break;
+		case 3: frameContentSize = ZSTD_readLE64(ip + pos); break;
+		}
+		if (!windowSize)
+			windowSize = (U32)frameContentSize;
+		if (windowSize > windowSizeMax)
+			return ERROR(frameParameter_windowTooLarge);
+		fparamsPtr->frameContentSize = frameContentSize;
+		fparamsPtr->windowSize = windowSize;
+		fparamsPtr->dictID = dictID;
+		fparamsPtr->checksumFlag = checksumFlag;
+	}
+	return 0;
+}
+
+/** ZSTD_getFrameContentSize() :
+*   compatible with legacy mode
+*   @return : decompressed size of the single frame pointed to be `src` if known, otherwise
+*             - ZSTD_CONTENTSIZE_UNKNOWN if the size cannot be determined
+*             - ZSTD_CONTENTSIZE_ERROR if an error occurred (e.g. invalid magic number, srcSize too small) */
+unsigned long long ZSTD_getFrameContentSize(const void *src, size_t srcSize)
+{
+	{
+		ZSTD_frameParams fParams;
+		if (ZSTD_getFrameParams(&fParams, src, srcSize) != 0)
+			return ZSTD_CONTENTSIZE_ERROR;
+		if (fParams.windowSize == 0) {
+			/* Either skippable or empty frame, size == 0 either way */
+			return 0;
+		} else if (fParams.frameContentSize != 0) {
+			return fParams.frameContentSize;
+		} else {
+			return ZSTD_CONTENTSIZE_UNKNOWN;
+		}
+	}
+}
+
+/** ZSTD_findDecompressedSize() :
+ *  compatible with legacy mode
+ *  `srcSize` must be the exact length of some number of ZSTD compressed and/or
+ *      skippable frames
+ *  @return : decompressed size of the frames contained */
+unsigned long long ZSTD_findDecompressedSize(const void *src, size_t srcSize)
+{
+	{
+		unsigned long long totalDstSize = 0;
+		while (srcSize >= ZSTD_frameHeaderSize_prefix) {
+			const U32 magicNumber = ZSTD_readLE32(src);
+
+			if ((magicNumber & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
+				size_t skippableSize;
+				if (srcSize < ZSTD_skippableHeaderSize)
+					return ERROR(srcSize_wrong);
+				skippableSize = ZSTD_readLE32((const BYTE *)src + 4) + ZSTD_skippableHeaderSize;
+				if (srcSize < skippableSize) {
+					return ZSTD_CONTENTSIZE_ERROR;
+				}
+
+				src = (const BYTE *)src + skippableSize;
+				srcSize -= skippableSize;
+				continue;
+			}
+
+			{
+				unsigned long long const ret = ZSTD_getFrameContentSize(src, srcSize);
+				if (ret >= ZSTD_CONTENTSIZE_ERROR)
+					return ret;
+
+				/* check for overflow */
+				if (totalDstSize + ret < totalDstSize)
+					return ZSTD_CONTENTSIZE_ERROR;
+				totalDstSize += ret;
+			}
+			{
+				size_t const frameSrcSize = ZSTD_findFrameCompressedSize(src, srcSize);
+				if (ZSTD_isError(frameSrcSize)) {
+					return ZSTD_CONTENTSIZE_ERROR;
+				}
+
+				src = (const BYTE *)src + frameSrcSize;
+				srcSize -= frameSrcSize;
+			}
+		}
+
+		if (srcSize) {
+			return ZSTD_CONTENTSIZE_ERROR;
+		}
+
+		return totalDstSize;
+	}
+}
+
+/** ZSTD_decodeFrameHeader() :
+*   `headerSize` must be the size provided by ZSTD_frameHeaderSize().
+*   @return : 0 if success, or an error code, which can be tested using ZSTD_isError() */
+static size_t ZSTD_decodeFrameHeader(ZSTD_DCtx *dctx, const void *src, size_t headerSize)
+{
+	size_t const result = ZSTD_getFrameParams(&(dctx->fParams), src, headerSize);
+	if (ZSTD_isError(result))
+		return result; /* invalid header */
+	if (result > 0)
+		return ERROR(srcSize_wrong); /* headerSize too small */
+	if (dctx->fParams.dictID && (dctx->dictID != dctx->fParams.dictID))
+		return ERROR(dictionary_wrong);
+	if (dctx->fParams.checksumFlag)
+		xxh64_reset(&dctx->xxhState, 0);
+	return 0;
+}
+
+typedef struct {
+	blockType_e blockType;
+	U32 lastBlock;
+	U32 origSize;
+} blockProperties_t;
+
+/*! ZSTD_getcBlockSize() :
+*   Provides the size of compressed block from block header `src` */
+size_t ZSTD_getcBlockSize(const void *src, size_t srcSize, blockProperties_t *bpPtr)
+{
+	if (srcSize < ZSTD_blockHeaderSize)
+		return ERROR(srcSize_wrong);
+	{
+		U32 const cBlockHeader = ZSTD_readLE24(src);
+		U32 const cSize = cBlockHeader >> 3;
+		bpPtr->lastBlock = cBlockHeader & 1;
+		bpPtr->blockType = (blockType_e)((cBlockHeader >> 1) & 3);
+		bpPtr->origSize = cSize; /* only useful for RLE */
+		if (bpPtr->blockType == bt_rle)
+			return 1;
+		if (bpPtr->blockType == bt_reserved)
+			return ERROR(corruption_detected);
+		return cSize;
+	}
+}
+
+static size_t ZSTD_copyRawBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	if (srcSize > dstCapacity)
+		return ERROR(dstSize_tooSmall);
+	memcpy(dst, src, srcSize);
+	return srcSize;
+}
+
+static size_t ZSTD_setRleBlock(void *dst, size_t dstCapacity, const void *src, size_t srcSize, size_t regenSize)
+{
+	if (srcSize != 1)
+		return ERROR(srcSize_wrong);
+	if (regenSize > dstCapacity)
+		return ERROR(dstSize_tooSmall);
+	memset(dst, *(const BYTE *)src, regenSize);
+	return regenSize;
+}
+
+/*! ZSTD_decodeLiteralsBlock() :
+	@return : nb of bytes read from src (< srcSize ) */
+size_t ZSTD_decodeLiteralsBlock(ZSTD_DCtx *dctx, const void *src, size_t srcSize) /* note : srcSize < BLOCKSIZE */
+{
+	if (srcSize < MIN_CBLOCK_SIZE)
+		return ERROR(corruption_detected);
+
+	{
+		const BYTE *const istart = (const BYTE *)src;
+		symbolEncodingType_e const litEncType = (symbolEncodingType_e)(istart[0] & 3);
+
+		switch (litEncType) {
+		case set_repeat:
+			if (dctx->litEntropy == 0)
+				return ERROR(dictionary_corrupted);
+		/* fall-through */
+		case set_compressed:
+			if (srcSize < 5)
+				return ERROR(corruption_detected); /* srcSize >= MIN_CBLOCK_SIZE == 3; here we need up to 5 for case 3 */
+			{
+				size_t lhSize, litSize, litCSize;
+				U32 singleStream = 0;
+				U32 const lhlCode = (istart[0] >> 2) & 3;
+				U32 const lhc = ZSTD_readLE32(istart);
+				switch (lhlCode) {
+				case 0:
+				case 1:
+				default: /* note : default is impossible, since lhlCode into [0..3] */
+					/* 2 - 2 - 10 - 10 */
+					singleStream = !lhlCode;
+					lhSize = 3;
+					litSize = (lhc >> 4) & 0x3FF;
+					litCSize = (lhc >> 14) & 0x3FF;
+					break;
+				case 2:
+					/* 2 - 2 - 14 - 14 */
+					lhSize = 4;
+					litSize = (lhc >> 4) & 0x3FFF;
+					litCSize = lhc >> 18;
+					break;
+				case 3:
+					/* 2 - 2 - 18 - 18 */
+					lhSize = 5;
+					litSize = (lhc >> 4) & 0x3FFFF;
+					litCSize = (lhc >> 22) + (istart[4] << 10);
+					break;
+				}
+				if (litSize > ZSTD_BLOCKSIZE_ABSOLUTEMAX)
+					return ERROR(corruption_detected);
+				if (litCSize + lhSize > srcSize)
+					return ERROR(corruption_detected);
+
+				if (HUF_isError(
+					(litEncType == set_repeat)
+					    ? (singleStream ? HUF_decompress1X_usingDTable(dctx->litBuffer, litSize, istart + lhSize, litCSize, dctx->HUFptr)
+							    : HUF_decompress4X_usingDTable(dctx->litBuffer, litSize, istart + lhSize, litCSize, dctx->HUFptr))
+					    : (singleStream
+						   ? HUF_decompress1X2_DCtx_wksp(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart + lhSize, litCSize,
+										 dctx->entropy.workspace, sizeof(dctx->entropy.workspace))
+						   : HUF_decompress4X_hufOnly_wksp(dctx->entropy.hufTable, dctx->litBuffer, litSize, istart + lhSize, litCSize,
+										   dctx->entropy.workspace, sizeof(dctx->entropy.workspace)))))
+					return ERROR(corruption_detected);
+
+				dctx->litPtr = dctx->litBuffer;
+				dctx->litSize = litSize;
+				dctx->litEntropy = 1;
+				if (litEncType == set_compressed)
+					dctx->HUFptr = dctx->entropy.hufTable;
+				memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
+				return litCSize + lhSize;
+			}
+
+		case set_basic: {
+			size_t litSize, lhSize;
+			U32 const lhlCode = ((istart[0]) >> 2) & 3;
+			switch (lhlCode) {
+			case 0:
+			case 2:
+			default: /* note : default is impossible, since lhlCode into [0..3] */
+				lhSize = 1;
+				litSize = istart[0] >> 3;
+				break;
+			case 1:
+				lhSize = 2;
+				litSize = ZSTD_readLE16(istart) >> 4;
+				break;
+			case 3:
+				lhSize = 3;
+				litSize = ZSTD_readLE24(istart) >> 4;
+				break;
+			}
+
+			if (lhSize + litSize + WILDCOPY_OVERLENGTH > srcSize) { /* risk reading beyond src buffer with wildcopy */
+				if (litSize + lhSize > srcSize)
+					return ERROR(corruption_detected);
+				memcpy(dctx->litBuffer, istart + lhSize, litSize);
+				dctx->litPtr = dctx->litBuffer;
+				dctx->litSize = litSize;
+				memset(dctx->litBuffer + dctx->litSize, 0, WILDCOPY_OVERLENGTH);
+				return lhSize + litSize;
+			}
+			/* direct reference into compressed stream */
+			dctx->litPtr = istart + lhSize;
+			dctx->litSize = litSize;
+			return lhSize + litSize;
+		}
+
+		case set_rle: {
+			U32 const lhlCode = ((istart[0]) >> 2) & 3;
+			size_t litSize, lhSize;
+			switch (lhlCode) {
+			case 0:
+			case 2:
+			default: /* note : default is impossible, since lhlCode into [0..3] */
+				lhSize = 1;
+				litSize = istart[0] >> 3;
+				break;
+			case 1:
+				lhSize = 2;
+				litSize = ZSTD_readLE16(istart) >> 4;
+				break;
+			case 3:
+				lhSize = 3;
+				litSize = ZSTD_readLE24(istart) >> 4;
+				if (srcSize < 4)
+					return ERROR(corruption_detected); /* srcSize >= MIN_CBLOCK_SIZE == 3; here we need lhSize+1 = 4 */
+				break;
+			}
+			if (litSize > ZSTD_BLOCKSIZE_ABSOLUTEMAX)
+				return ERROR(corruption_detected);
+			memset(dctx->litBuffer, istart[lhSize], litSize + WILDCOPY_OVERLENGTH);
+			dctx->litPtr = dctx->litBuffer;
+			dctx->litSize = litSize;
+			return lhSize + 1;
+		}
+		default:
+			return ERROR(corruption_detected); /* impossible */
+		}
+	}
+}
+
+typedef union {
+	FSE_decode_t realData;
+	U32 alignedBy4;
+} FSE_decode_t4;
+
+static const FSE_decode_t4 LL_defaultDTable[(1 << LL_DEFAULTNORMLOG) + 1] = {
+    {{LL_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
+    {{0, 0, 4}},		 /* 0 : base, symbol, bits */
+    {{16, 0, 4}},
+    {{32, 1, 5}},
+    {{0, 3, 5}},
+    {{0, 4, 5}},
+    {{0, 6, 5}},
+    {{0, 7, 5}},
+    {{0, 9, 5}},
+    {{0, 10, 5}},
+    {{0, 12, 5}},
+    {{0, 14, 6}},
+    {{0, 16, 5}},
+    {{0, 18, 5}},
+    {{0, 19, 5}},
+    {{0, 21, 5}},
+    {{0, 22, 5}},
+    {{0, 24, 5}},
+    {{32, 25, 5}},
+    {{0, 26, 5}},
+    {{0, 27, 6}},
+    {{0, 29, 6}},
+    {{0, 31, 6}},
+    {{32, 0, 4}},
+    {{0, 1, 4}},
+    {{0, 2, 5}},
+    {{32, 4, 5}},
+    {{0, 5, 5}},
+    {{32, 7, 5}},
+    {{0, 8, 5}},
+    {{32, 10, 5}},
+    {{0, 11, 5}},
+    {{0, 13, 6}},
+    {{32, 16, 5}},
+    {{0, 17, 5}},
+    {{32, 19, 5}},
+    {{0, 20, 5}},
+    {{32, 22, 5}},
+    {{0, 23, 5}},
+    {{0, 25, 4}},
+    {{16, 25, 4}},
+    {{32, 26, 5}},
+    {{0, 28, 6}},
+    {{0, 30, 6}},
+    {{48, 0, 4}},
+    {{16, 1, 4}},
+    {{32, 2, 5}},
+    {{32, 3, 5}},
+    {{32, 5, 5}},
+    {{32, 6, 5}},
+    {{32, 8, 5}},
+    {{32, 9, 5}},
+    {{32, 11, 5}},
+    {{32, 12, 5}},
+    {{0, 15, 6}},
+    {{32, 17, 5}},
+    {{32, 18, 5}},
+    {{32, 20, 5}},
+    {{32, 21, 5}},
+    {{32, 23, 5}},
+    {{32, 24, 5}},
+    {{0, 35, 6}},
+    {{0, 34, 6}},
+    {{0, 33, 6}},
+    {{0, 32, 6}},
+}; /* LL_defaultDTable */
+
+static const FSE_decode_t4 ML_defaultDTable[(1 << ML_DEFAULTNORMLOG) + 1] = {
+    {{ML_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
+    {{0, 0, 6}},		 /* 0 : base, symbol, bits */
+    {{0, 1, 4}},
+    {{32, 2, 5}},
+    {{0, 3, 5}},
+    {{0, 5, 5}},
+    {{0, 6, 5}},
+    {{0, 8, 5}},
+    {{0, 10, 6}},
+    {{0, 13, 6}},
+    {{0, 16, 6}},
+    {{0, 19, 6}},
+    {{0, 22, 6}},
+    {{0, 25, 6}},
+    {{0, 28, 6}},
+    {{0, 31, 6}},
+    {{0, 33, 6}},
+    {{0, 35, 6}},
+    {{0, 37, 6}},
+    {{0, 39, 6}},
+    {{0, 41, 6}},
+    {{0, 43, 6}},
+    {{0, 45, 6}},
+    {{16, 1, 4}},
+    {{0, 2, 4}},
+    {{32, 3, 5}},
+    {{0, 4, 5}},
+    {{32, 6, 5}},
+    {{0, 7, 5}},
+    {{0, 9, 6}},
+    {{0, 12, 6}},
+    {{0, 15, 6}},
+    {{0, 18, 6}},
+    {{0, 21, 6}},
+    {{0, 24, 6}},
+    {{0, 27, 6}},
+    {{0, 30, 6}},
+    {{0, 32, 6}},
+    {{0, 34, 6}},
+    {{0, 36, 6}},
+    {{0, 38, 6}},
+    {{0, 40, 6}},
+    {{0, 42, 6}},
+    {{0, 44, 6}},
+    {{32, 1, 4}},
+    {{48, 1, 4}},
+    {{16, 2, 4}},
+    {{32, 4, 5}},
+    {{32, 5, 5}},
+    {{32, 7, 5}},
+    {{32, 8, 5}},
+    {{0, 11, 6}},
+    {{0, 14, 6}},
+    {{0, 17, 6}},
+    {{0, 20, 6}},
+    {{0, 23, 6}},
+    {{0, 26, 6}},
+    {{0, 29, 6}},
+    {{0, 52, 6}},
+    {{0, 51, 6}},
+    {{0, 50, 6}},
+    {{0, 49, 6}},
+    {{0, 48, 6}},
+    {{0, 47, 6}},
+    {{0, 46, 6}},
+}; /* ML_defaultDTable */
+
+static const FSE_decode_t4 OF_defaultDTable[(1 << OF_DEFAULTNORMLOG) + 1] = {
+    {{OF_DEFAULTNORMLOG, 1, 1}}, /* header : tableLog, fastMode, fastMode */
+    {{0, 0, 5}},		 /* 0 : base, symbol, bits */
+    {{0, 6, 4}},
+    {{0, 9, 5}},
+    {{0, 15, 5}},
+    {{0, 21, 5}},
+    {{0, 3, 5}},
+    {{0, 7, 4}},
+    {{0, 12, 5}},
+    {{0, 18, 5}},
+    {{0, 23, 5}},
+    {{0, 5, 5}},
+    {{0, 8, 4}},
+    {{0, 14, 5}},
+    {{0, 20, 5}},
+    {{0, 2, 5}},
+    {{16, 7, 4}},
+    {{0, 11, 5}},
+    {{0, 17, 5}},
+    {{0, 22, 5}},
+    {{0, 4, 5}},
+    {{16, 8, 4}},
+    {{0, 13, 5}},
+    {{0, 19, 5}},
+    {{0, 1, 5}},
+    {{16, 6, 4}},
+    {{0, 10, 5}},
+    {{0, 16, 5}},
+    {{0, 28, 5}},
+    {{0, 27, 5}},
+    {{0, 26, 5}},
+    {{0, 25, 5}},
+    {{0, 24, 5}},
+}; /* OF_defaultDTable */
+
+/*! ZSTD_buildSeqTable() :
+	@return : nb bytes read from src,
+			  or an error code if it fails, testable with ZSTD_isError()
+*/
+static size_t ZSTD_buildSeqTable(FSE_DTable *DTableSpace, const FSE_DTable **DTablePtr, symbolEncodingType_e type, U32 max, U32 maxLog, const void *src,
+				 size_t srcSize, const FSE_decode_t4 *defaultTable, U32 flagRepeatTable, void *workspace, size_t workspaceSize)
+{
+	const void *const tmpPtr = defaultTable; /* bypass strict aliasing */
+	switch (type) {
+	case set_rle:
+		if (!srcSize)
+			return ERROR(srcSize_wrong);
+		if ((*(const BYTE *)src) > max)
+			return ERROR(corruption_detected);
+		FSE_buildDTable_rle(DTableSpace, *(const BYTE *)src);
+		*DTablePtr = DTableSpace;
+		return 1;
+	case set_basic: *DTablePtr = (const FSE_DTable *)tmpPtr; return 0;
+	case set_repeat:
+		if (!flagRepeatTable)
+			return ERROR(corruption_detected);
+		return 0;
+	default: /* impossible */
+	case set_compressed: {
+		U32 tableLog;
+		S16 *norm = (S16 *)workspace;
+		size_t const spaceUsed32 = ALIGN(sizeof(S16) * (MaxSeq + 1), sizeof(U32)) >> 2;
+
+		if ((spaceUsed32 << 2) > workspaceSize)
+			return ERROR(GENERIC);
+		workspace = (U32 *)workspace + spaceUsed32;
+		workspaceSize -= (spaceUsed32 << 2);
+		{
+			size_t const headerSize = FSE_readNCount(norm, &max, &tableLog, src, srcSize);
+			if (FSE_isError(headerSize))
+				return ERROR(corruption_detected);
+			if (tableLog > maxLog)
+				return ERROR(corruption_detected);
+			FSE_buildDTable_wksp(DTableSpace, norm, max, tableLog, workspace, workspaceSize);
+			*DTablePtr = DTableSpace;
+			return headerSize;
+		}
+	}
+	}
+}
+
+size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx *dctx, int *nbSeqPtr, const void *src, size_t srcSize)
+{
+	const BYTE *const istart = (const BYTE *const)src;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *ip = istart;
+
+	/* check */
+	if (srcSize < MIN_SEQUENCES_SIZE)
+		return ERROR(srcSize_wrong);
+
+	/* SeqHead */
+	{
+		int nbSeq = *ip++;
+		if (!nbSeq) {
+			*nbSeqPtr = 0;
+			return 1;
+		}
+		if (nbSeq > 0x7F) {
+			if (nbSeq == 0xFF) {
+				if (ip + 2 > iend)
+					return ERROR(srcSize_wrong);
+				nbSeq = ZSTD_readLE16(ip) + LONGNBSEQ, ip += 2;
+			} else {
+				if (ip >= iend)
+					return ERROR(srcSize_wrong);
+				nbSeq = ((nbSeq - 0x80) << 8) + *ip++;
+			}
+		}
+		*nbSeqPtr = nbSeq;
+	}
+
+	/* FSE table descriptors */
+	if (ip + 4 > iend)
+		return ERROR(srcSize_wrong); /* minimum possible size */
+	{
+		symbolEncodingType_e const LLtype = (symbolEncodingType_e)(*ip >> 6);
+		symbolEncodingType_e const OFtype = (symbolEncodingType_e)((*ip >> 4) & 3);
+		symbolEncodingType_e const MLtype = (symbolEncodingType_e)((*ip >> 2) & 3);
+		ip++;
+
+		/* Build DTables */
+		{
+			size_t const llhSize = ZSTD_buildSeqTable(dctx->entropy.LLTable, &dctx->LLTptr, LLtype, MaxLL, LLFSELog, ip, iend - ip,
+								  LL_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
+			if (ZSTD_isError(llhSize))
+				return ERROR(corruption_detected);
+			ip += llhSize;
+		}
+		{
+			size_t const ofhSize = ZSTD_buildSeqTable(dctx->entropy.OFTable, &dctx->OFTptr, OFtype, MaxOff, OffFSELog, ip, iend - ip,
+								  OF_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
+			if (ZSTD_isError(ofhSize))
+				return ERROR(corruption_detected);
+			ip += ofhSize;
+		}
+		{
+			size_t const mlhSize = ZSTD_buildSeqTable(dctx->entropy.MLTable, &dctx->MLTptr, MLtype, MaxML, MLFSELog, ip, iend - ip,
+								  ML_defaultDTable, dctx->fseEntropy, dctx->entropy.workspace, sizeof(dctx->entropy.workspace));
+			if (ZSTD_isError(mlhSize))
+				return ERROR(corruption_detected);
+			ip += mlhSize;
+		}
+	}
+
+	return ip - istart;
+}
+
+typedef struct {
+	size_t litLength;
+	size_t matchLength;
+	size_t offset;
+	const BYTE *match;
+} seq_t;
+
+typedef struct {
+	BIT_DStream_t DStream;
+	FSE_DState_t stateLL;
+	FSE_DState_t stateOffb;
+	FSE_DState_t stateML;
+	size_t prevOffset[ZSTD_REP_NUM];
+	const BYTE *base;
+	size_t pos;
+	uPtrDiff gotoDict;
+} seqState_t;
+
+FORCE_NOINLINE
+size_t ZSTD_execSequenceLast7(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
+			      const BYTE *const vBase, const BYTE *const dictEnd)
+{
+	BYTE *const oLitEnd = op + sequence.litLength;
+	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
+	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
+	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
+	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
+	const BYTE *match = oLitEnd - sequence.offset;
+
+	/* check */
+	if (oMatchEnd > oend)
+		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
+	if (iLitEnd > litLimit)
+		return ERROR(corruption_detected); /* over-read beyond lit buffer */
+	if (oLitEnd <= oend_w)
+		return ERROR(GENERIC); /* Precondition */
+
+	/* copy literals */
+	if (op < oend_w) {
+		ZSTD_wildcopy(op, *litPtr, oend_w - op);
+		*litPtr += oend_w - op;
+		op = oend_w;
+	}
+	while (op < oLitEnd)
+		*op++ = *(*litPtr)++;
+
+	/* copy Match */
+	if (sequence.offset > (size_t)(oLitEnd - base)) {
+		/* offset beyond prefix */
+		if (sequence.offset > (size_t)(oLitEnd - vBase))
+			return ERROR(corruption_detected);
+		match = dictEnd - (base - match);
+		if (match + sequence.matchLength <= dictEnd) {
+			memmove(oLitEnd, match, sequence.matchLength);
+			return sequenceLength;
+		}
+		/* span extDict & currPrefixSegment */
+		{
+			size_t const length1 = dictEnd - match;
+			memmove(oLitEnd, match, length1);
+			op = oLitEnd + length1;
+			sequence.matchLength -= length1;
+			match = base;
+		}
+	}
+	while (op < oMatchEnd)
+		*op++ = *match++;
+	return sequenceLength;
+}
+
+static seq_t ZSTD_decodeSequence(seqState_t *seqState)
+{
+	seq_t seq;
+
+	U32 const llCode = FSE_peekSymbol(&seqState->stateLL);
+	U32 const mlCode = FSE_peekSymbol(&seqState->stateML);
+	U32 const ofCode = FSE_peekSymbol(&seqState->stateOffb); /* <= maxOff, by table construction */
+
+	U32 const llBits = LL_bits[llCode];
+	U32 const mlBits = ML_bits[mlCode];
+	U32 const ofBits = ofCode;
+	U32 const totalBits = llBits + mlBits + ofBits;
+
+	static const U32 LL_base[MaxLL + 1] = {0,  1,  2,  3,  4,  5,  6,  7,  8,    9,     10,    11,    12,    13,     14,     15,     16,     18,
+					       20, 22, 24, 28, 32, 40, 48, 64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000, 0x8000, 0x10000};
+
+	static const U32 ML_base[MaxML + 1] = {3,  4,  5,  6,  7,  8,  9,  10,   11,    12,    13,    14,    15,     16,     17,     18,     19,     20,
+					       21, 22, 23, 24, 25, 26, 27, 28,   29,    30,    31,    32,    33,     34,     35,     37,     39,     41,
+					       43, 47, 51, 59, 67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803, 0x1003, 0x2003, 0x4003, 0x8003, 0x10003};
+
+	static const U32 OF_base[MaxOff + 1] = {0,       1,	1,	5,	0xD,      0x1D,      0x3D,      0x7D,      0xFD,     0x1FD,
+						0x3FD,   0x7FD,    0xFFD,    0x1FFD,   0x3FFD,   0x7FFD,    0xFFFD,    0x1FFFD,   0x3FFFD,  0x7FFFD,
+						0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD, 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD};
+
+	/* sequence */
+	{
+		size_t offset;
+		if (!ofCode)
+			offset = 0;
+		else {
+			offset = OF_base[ofCode] + BIT_readBitsFast(&seqState->DStream, ofBits); /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
+			if (ZSTD_32bits())
+				BIT_reloadDStream(&seqState->DStream);
+		}
+
+		if (ofCode <= 1) {
+			offset += (llCode == 0);
+			if (offset) {
+				size_t temp = (offset == 3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
+				temp += !temp; /* 0 is not valid; input is corrupted; force offset to 1 */
+				if (offset != 1)
+					seqState->prevOffset[2] = seqState->prevOffset[1];
+				seqState->prevOffset[1] = seqState->prevOffset[0];
+				seqState->prevOffset[0] = offset = temp;
+			} else {
+				offset = seqState->prevOffset[0];
+			}
+		} else {
+			seqState->prevOffset[2] = seqState->prevOffset[1];
+			seqState->prevOffset[1] = seqState->prevOffset[0];
+			seqState->prevOffset[0] = offset;
+		}
+		seq.offset = offset;
+	}
+
+	seq.matchLength = ML_base[mlCode] + ((mlCode > 31) ? BIT_readBitsFast(&seqState->DStream, mlBits) : 0); /* <=  16 bits */
+	if (ZSTD_32bits() && (mlBits + llBits > 24))
+		BIT_reloadDStream(&seqState->DStream);
+
+	seq.litLength = LL_base[llCode] + ((llCode > 15) ? BIT_readBitsFast(&seqState->DStream, llBits) : 0); /* <=  16 bits */
+	if (ZSTD_32bits() || (totalBits > 64 - 7 - (LLFSELog + MLFSELog + OffFSELog)))
+		BIT_reloadDStream(&seqState->DStream);
+
+	/* ANS state update */
+	FSE_updateState(&seqState->stateLL, &seqState->DStream); /* <=  9 bits */
+	FSE_updateState(&seqState->stateML, &seqState->DStream); /* <=  9 bits */
+	if (ZSTD_32bits())
+		BIT_reloadDStream(&seqState->DStream);		   /* <= 18 bits */
+	FSE_updateState(&seqState->stateOffb, &seqState->DStream); /* <=  8 bits */
+
+	seq.match = NULL;
+
+	return seq;
+}
+
+FORCE_INLINE
+size_t ZSTD_execSequence(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
+			 const BYTE *const vBase, const BYTE *const dictEnd)
+{
+	BYTE *const oLitEnd = op + sequence.litLength;
+	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
+	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
+	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
+	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
+	const BYTE *match = oLitEnd - sequence.offset;
+
+	/* check */
+	if (oMatchEnd > oend)
+		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
+	if (iLitEnd > litLimit)
+		return ERROR(corruption_detected); /* over-read beyond lit buffer */
+	if (oLitEnd > oend_w)
+		return ZSTD_execSequenceLast7(op, oend, sequence, litPtr, litLimit, base, vBase, dictEnd);
+
+	/* copy Literals */
+	ZSTD_copy8(op, *litPtr);
+	if (sequence.litLength > 8)
+		ZSTD_wildcopy(op + 8, (*litPtr) + 8,
+			      sequence.litLength - 8); /* note : since oLitEnd <= oend-WILDCOPY_OVERLENGTH, no risk of overwrite beyond oend */
+	op = oLitEnd;
+	*litPtr = iLitEnd; /* update for next sequence */
+
+	/* copy Match */
+	if (sequence.offset > (size_t)(oLitEnd - base)) {
+		/* offset beyond prefix */
+		if (sequence.offset > (size_t)(oLitEnd - vBase))
+			return ERROR(corruption_detected);
+		match = dictEnd + (match - base);
+		if (match + sequence.matchLength <= dictEnd) {
+			memmove(oLitEnd, match, sequence.matchLength);
+			return sequenceLength;
+		}
+		/* span extDict & currPrefixSegment */
+		{
+			size_t const length1 = dictEnd - match;
+			memmove(oLitEnd, match, length1);
+			op = oLitEnd + length1;
+			sequence.matchLength -= length1;
+			match = base;
+			if (op > oend_w || sequence.matchLength < MINMATCH) {
+				U32 i;
+				for (i = 0; i < sequence.matchLength; ++i)
+					op[i] = match[i];
+				return sequenceLength;
+			}
+		}
+	}
+	/* Requirement: op <= oend_w && sequence.matchLength >= MINMATCH */
+
+	/* match within prefix */
+	if (sequence.offset < 8) {
+		/* close range match, overlap */
+		static const U32 dec32table[] = {0, 1, 2, 1, 4, 4, 4, 4};   /* added */
+		static const int dec64table[] = {8, 8, 8, 7, 8, 9, 10, 11}; /* subtracted */
+		int const sub2 = dec64table[sequence.offset];
+		op[0] = match[0];
+		op[1] = match[1];
+		op[2] = match[2];
+		op[3] = match[3];
+		match += dec32table[sequence.offset];
+		ZSTD_copy4(op + 4, match);
+		match -= sub2;
+	} else {
+		ZSTD_copy8(op, match);
+	}
+	op += 8;
+	match += 8;
+
+	if (oMatchEnd > oend - (16 - MINMATCH)) {
+		if (op < oend_w) {
+			ZSTD_wildcopy(op, match, oend_w - op);
+			match += oend_w - op;
+			op = oend_w;
+		}
+		while (op < oMatchEnd)
+			*op++ = *match++;
+	} else {
+		ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength - 8); /* works even if matchLength < 8 */
+	}
+	return sequenceLength;
+}
+
+static size_t ZSTD_decompressSequences(ZSTD_DCtx *dctx, void *dst, size_t maxDstSize, const void *seqStart, size_t seqSize)
+{
+	const BYTE *ip = (const BYTE *)seqStart;
+	const BYTE *const iend = ip + seqSize;
+	BYTE *const ostart = (BYTE * const)dst;
+	BYTE *const oend = ostart + maxDstSize;
+	BYTE *op = ostart;
+	const BYTE *litPtr = dctx->litPtr;
+	const BYTE *const litEnd = litPtr + dctx->litSize;
+	const BYTE *const base = (const BYTE *)(dctx->base);
+	const BYTE *const vBase = (const BYTE *)(dctx->vBase);
+	const BYTE *const dictEnd = (const BYTE *)(dctx->dictEnd);
+	int nbSeq;
+
+	/* Build Decoding Tables */
+	{
+		size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, seqSize);
+		if (ZSTD_isError(seqHSize))
+			return seqHSize;
+		ip += seqHSize;
+	}
+
+	/* Regen sequences */
+	if (nbSeq) {
+		seqState_t seqState;
+		dctx->fseEntropy = 1;
+		{
+			U32 i;
+			for (i = 0; i < ZSTD_REP_NUM; i++)
+				seqState.prevOffset[i] = dctx->entropy.rep[i];
+		}
+		CHECK_E(BIT_initDStream(&seqState.DStream, ip, iend - ip), corruption_detected);
+		FSE_initDState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
+		FSE_initDState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
+		FSE_initDState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
+
+		for (; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && nbSeq;) {
+			nbSeq--;
+			{
+				seq_t const sequence = ZSTD_decodeSequence(&seqState);
+				size_t const oneSeqSize = ZSTD_execSequence(op, oend, sequence, &litPtr, litEnd, base, vBase, dictEnd);
+				if (ZSTD_isError(oneSeqSize))
+					return oneSeqSize;
+				op += oneSeqSize;
+			}
+		}
+
+		/* check if reached exact end */
+		if (nbSeq)
+			return ERROR(corruption_detected);
+		/* save reps for next block */
+		{
+			U32 i;
+			for (i = 0; i < ZSTD_REP_NUM; i++)
+				dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]);
+		}
+	}
+
+	/* last literal segment */
+	{
+		size_t const lastLLSize = litEnd - litPtr;
+		if (lastLLSize > (size_t)(oend - op))
+			return ERROR(dstSize_tooSmall);
+		memcpy(op, litPtr, lastLLSize);
+		op += lastLLSize;
+	}
+
+	return op - ostart;
+}
+
+FORCE_INLINE seq_t ZSTD_decodeSequenceLong_generic(seqState_t *seqState, int const longOffsets)
+{
+	seq_t seq;
+
+	U32 const llCode = FSE_peekSymbol(&seqState->stateLL);
+	U32 const mlCode = FSE_peekSymbol(&seqState->stateML);
+	U32 const ofCode = FSE_peekSymbol(&seqState->stateOffb); /* <= maxOff, by table construction */
+
+	U32 const llBits = LL_bits[llCode];
+	U32 const mlBits = ML_bits[mlCode];
+	U32 const ofBits = ofCode;
+	U32 const totalBits = llBits + mlBits + ofBits;
+
+	static const U32 LL_base[MaxLL + 1] = {0,  1,  2,  3,  4,  5,  6,  7,  8,    9,     10,    11,    12,    13,     14,     15,     16,     18,
+					       20, 22, 24, 28, 32, 40, 48, 64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000, 0x2000, 0x4000, 0x8000, 0x10000};
+
+	static const U32 ML_base[MaxML + 1] = {3,  4,  5,  6,  7,  8,  9,  10,   11,    12,    13,    14,    15,     16,     17,     18,     19,     20,
+					       21, 22, 23, 24, 25, 26, 27, 28,   29,    30,    31,    32,    33,     34,     35,     37,     39,     41,
+					       43, 47, 51, 59, 67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803, 0x1003, 0x2003, 0x4003, 0x8003, 0x10003};
+
+	static const U32 OF_base[MaxOff + 1] = {0,       1,	1,	5,	0xD,      0x1D,      0x3D,      0x7D,      0xFD,     0x1FD,
+						0x3FD,   0x7FD,    0xFFD,    0x1FFD,   0x3FFD,   0x7FFD,    0xFFFD,    0x1FFFD,   0x3FFFD,  0x7FFFD,
+						0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD, 0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD};
+
+	/* sequence */
+	{
+		size_t offset;
+		if (!ofCode)
+			offset = 0;
+		else {
+			if (longOffsets) {
+				int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN);
+				offset = OF_base[ofCode] + (BIT_readBitsFast(&seqState->DStream, ofBits - extraBits) << extraBits);
+				if (ZSTD_32bits() || extraBits)
+					BIT_reloadDStream(&seqState->DStream);
+				if (extraBits)
+					offset += BIT_readBitsFast(&seqState->DStream, extraBits);
+			} else {
+				offset = OF_base[ofCode] + BIT_readBitsFast(&seqState->DStream, ofBits); /* <=  (ZSTD_WINDOWLOG_MAX-1) bits */
+				if (ZSTD_32bits())
+					BIT_reloadDStream(&seqState->DStream);
+			}
+		}
+
+		if (ofCode <= 1) {
+			offset += (llCode == 0);
+			if (offset) {
+				size_t temp = (offset == 3) ? seqState->prevOffset[0] - 1 : seqState->prevOffset[offset];
+				temp += !temp; /* 0 is not valid; input is corrupted; force offset to 1 */
+				if (offset != 1)
+					seqState->prevOffset[2] = seqState->prevOffset[1];
+				seqState->prevOffset[1] = seqState->prevOffset[0];
+				seqState->prevOffset[0] = offset = temp;
+			} else {
+				offset = seqState->prevOffset[0];
+			}
+		} else {
+			seqState->prevOffset[2] = seqState->prevOffset[1];
+			seqState->prevOffset[1] = seqState->prevOffset[0];
+			seqState->prevOffset[0] = offset;
+		}
+		seq.offset = offset;
+	}
+
+	seq.matchLength = ML_base[mlCode] + ((mlCode > 31) ? BIT_readBitsFast(&seqState->DStream, mlBits) : 0); /* <=  16 bits */
+	if (ZSTD_32bits() && (mlBits + llBits > 24))
+		BIT_reloadDStream(&seqState->DStream);
+
+	seq.litLength = LL_base[llCode] + ((llCode > 15) ? BIT_readBitsFast(&seqState->DStream, llBits) : 0); /* <=  16 bits */
+	if (ZSTD_32bits() || (totalBits > 64 - 7 - (LLFSELog + MLFSELog + OffFSELog)))
+		BIT_reloadDStream(&seqState->DStream);
+
+	{
+		size_t const pos = seqState->pos + seq.litLength;
+		seq.match = seqState->base + pos - seq.offset; /* single memory segment */
+		if (seq.offset > pos)
+			seq.match += seqState->gotoDict; /* separate memory segment */
+		seqState->pos = pos + seq.matchLength;
+	}
+
+	/* ANS state update */
+	FSE_updateState(&seqState->stateLL, &seqState->DStream); /* <=  9 bits */
+	FSE_updateState(&seqState->stateML, &seqState->DStream); /* <=  9 bits */
+	if (ZSTD_32bits())
+		BIT_reloadDStream(&seqState->DStream);		   /* <= 18 bits */
+	FSE_updateState(&seqState->stateOffb, &seqState->DStream); /* <=  8 bits */
+
+	return seq;
+}
+
+static seq_t ZSTD_decodeSequenceLong(seqState_t *seqState, unsigned const windowSize)
+{
+	if (ZSTD_highbit32(windowSize) > STREAM_ACCUMULATOR_MIN) {
+		return ZSTD_decodeSequenceLong_generic(seqState, 1);
+	} else {
+		return ZSTD_decodeSequenceLong_generic(seqState, 0);
+	}
+}
+
+FORCE_INLINE
+size_t ZSTD_execSequenceLong(BYTE *op, BYTE *const oend, seq_t sequence, const BYTE **litPtr, const BYTE *const litLimit, const BYTE *const base,
+			     const BYTE *const vBase, const BYTE *const dictEnd)
+{
+	BYTE *const oLitEnd = op + sequence.litLength;
+	size_t const sequenceLength = sequence.litLength + sequence.matchLength;
+	BYTE *const oMatchEnd = op + sequenceLength; /* risk : address space overflow (32-bits) */
+	BYTE *const oend_w = oend - WILDCOPY_OVERLENGTH;
+	const BYTE *const iLitEnd = *litPtr + sequence.litLength;
+	const BYTE *match = sequence.match;
+
+	/* check */
+	if (oMatchEnd > oend)
+		return ERROR(dstSize_tooSmall); /* last match must start at a minimum distance of WILDCOPY_OVERLENGTH from oend */
+	if (iLitEnd > litLimit)
+		return ERROR(corruption_detected); /* over-read beyond lit buffer */
+	if (oLitEnd > oend_w)
+		return ZSTD_execSequenceLast7(op, oend, sequence, litPtr, litLimit, base, vBase, dictEnd);
+
+	/* copy Literals */
+	ZSTD_copy8(op, *litPtr);
+	if (sequence.litLength > 8)
+		ZSTD_wildcopy(op + 8, (*litPtr) + 8,
+			      sequence.litLength - 8); /* note : since oLitEnd <= oend-WILDCOPY_OVERLENGTH, no risk of overwrite beyond oend */
+	op = oLitEnd;
+	*litPtr = iLitEnd; /* update for next sequence */
+
+	/* copy Match */
+	if (sequence.offset > (size_t)(oLitEnd - base)) {
+		/* offset beyond prefix */
+		if (sequence.offset > (size_t)(oLitEnd - vBase))
+			return ERROR(corruption_detected);
+		if (match + sequence.matchLength <= dictEnd) {
+			memmove(oLitEnd, match, sequence.matchLength);
+			return sequenceLength;
+		}
+		/* span extDict & currPrefixSegment */
+		{
+			size_t const length1 = dictEnd - match;
+			memmove(oLitEnd, match, length1);
+			op = oLitEnd + length1;
+			sequence.matchLength -= length1;
+			match = base;
+			if (op > oend_w || sequence.matchLength < MINMATCH) {
+				U32 i;
+				for (i = 0; i < sequence.matchLength; ++i)
+					op[i] = match[i];
+				return sequenceLength;
+			}
+		}
+	}
+	/* Requirement: op <= oend_w && sequence.matchLength >= MINMATCH */
+
+	/* match within prefix */
+	if (sequence.offset < 8) {
+		/* close range match, overlap */
+		static const U32 dec32table[] = {0, 1, 2, 1, 4, 4, 4, 4};   /* added */
+		static const int dec64table[] = {8, 8, 8, 7, 8, 9, 10, 11}; /* subtracted */
+		int const sub2 = dec64table[sequence.offset];
+		op[0] = match[0];
+		op[1] = match[1];
+		op[2] = match[2];
+		op[3] = match[3];
+		match += dec32table[sequence.offset];
+		ZSTD_copy4(op + 4, match);
+		match -= sub2;
+	} else {
+		ZSTD_copy8(op, match);
+	}
+	op += 8;
+	match += 8;
+
+	if (oMatchEnd > oend - (16 - MINMATCH)) {
+		if (op < oend_w) {
+			ZSTD_wildcopy(op, match, oend_w - op);
+			match += oend_w - op;
+			op = oend_w;
+		}
+		while (op < oMatchEnd)
+			*op++ = *match++;
+	} else {
+		ZSTD_wildcopy(op, match, (ptrdiff_t)sequence.matchLength - 8); /* works even if matchLength < 8 */
+	}
+	return sequenceLength;
+}
+
+static size_t ZSTD_decompressSequencesLong(ZSTD_DCtx *dctx, void *dst, size_t maxDstSize, const void *seqStart, size_t seqSize)
+{
+	const BYTE *ip = (const BYTE *)seqStart;
+	const BYTE *const iend = ip + seqSize;
+	BYTE *const ostart = (BYTE * const)dst;
+	BYTE *const oend = ostart + maxDstSize;
+	BYTE *op = ostart;
+	const BYTE *litPtr = dctx->litPtr;
+	const BYTE *const litEnd = litPtr + dctx->litSize;
+	const BYTE *const base = (const BYTE *)(dctx->base);
+	const BYTE *const vBase = (const BYTE *)(dctx->vBase);
+	const BYTE *const dictEnd = (const BYTE *)(dctx->dictEnd);
+	unsigned const windowSize = dctx->fParams.windowSize;
+	int nbSeq;
+
+	/* Build Decoding Tables */
+	{
+		size_t const seqHSize = ZSTD_decodeSeqHeaders(dctx, &nbSeq, ip, seqSize);
+		if (ZSTD_isError(seqHSize))
+			return seqHSize;
+		ip += seqHSize;
+	}
+
+	/* Regen sequences */
+	if (nbSeq) {
+#define STORED_SEQS 4
+#define STOSEQ_MASK (STORED_SEQS - 1)
+#define ADVANCED_SEQS 4
+		seq_t *sequences = (seq_t *)dctx->entropy.workspace;
+		int const seqAdvance = MIN(nbSeq, ADVANCED_SEQS);
+		seqState_t seqState;
+		int seqNb;
+		ZSTD_STATIC_ASSERT(sizeof(dctx->entropy.workspace) >= sizeof(seq_t) * STORED_SEQS);
+		dctx->fseEntropy = 1;
+		{
+			U32 i;
+			for (i = 0; i < ZSTD_REP_NUM; i++)
+				seqState.prevOffset[i] = dctx->entropy.rep[i];
+		}
+		seqState.base = base;
+		seqState.pos = (size_t)(op - base);
+		seqState.gotoDict = (uPtrDiff)dictEnd - (uPtrDiff)base; /* cast to avoid undefined behaviour */
+		CHECK_E(BIT_initDStream(&seqState.DStream, ip, iend - ip), corruption_detected);
+		FSE_initDState(&seqState.stateLL, &seqState.DStream, dctx->LLTptr);
+		FSE_initDState(&seqState.stateOffb, &seqState.DStream, dctx->OFTptr);
+		FSE_initDState(&seqState.stateML, &seqState.DStream, dctx->MLTptr);
+
+		/* prepare in advance */
+		for (seqNb = 0; (BIT_reloadDStream(&seqState.DStream) <= BIT_DStream_completed) && seqNb < seqAdvance; seqNb++) {
+			sequences[seqNb] = ZSTD_decodeSequenceLong(&seqState, windowSize);
+		}
+		if (seqNb < seqAdvance)
+			return ERROR(corruption_detected);
+
+		/* decode and decompress */
+		for (; (BIT_reloadDStream(&(seqState.DStream)) <= BIT_DStream_completed) && seqNb < nbSeq; seqNb++) {
+			seq_t const sequence = ZSTD_decodeSequenceLong(&seqState, windowSize);
+			size_t const oneSeqSize =
+			    ZSTD_execSequenceLong(op, oend, sequences[(seqNb - ADVANCED_SEQS) & STOSEQ_MASK], &litPtr, litEnd, base, vBase, dictEnd);
+			if (ZSTD_isError(oneSeqSize))
+				return oneSeqSize;
+			ZSTD_PREFETCH(sequence.match);
+			sequences[seqNb & STOSEQ_MASK] = sequence;
+			op += oneSeqSize;
+		}
+		if (seqNb < nbSeq)
+			return ERROR(corruption_detected);
+
+		/* finish queue */
+		seqNb -= seqAdvance;
+		for (; seqNb < nbSeq; seqNb++) {
+			size_t const oneSeqSize = ZSTD_execSequenceLong(op, oend, sequences[seqNb & STOSEQ_MASK], &litPtr, litEnd, base, vBase, dictEnd);
+			if (ZSTD_isError(oneSeqSize))
+				return oneSeqSize;
+			op += oneSeqSize;
+		}
+
+		/* save reps for next block */
+		{
+			U32 i;
+			for (i = 0; i < ZSTD_REP_NUM; i++)
+				dctx->entropy.rep[i] = (U32)(seqState.prevOffset[i]);
+		}
+	}
+
+	/* last literal segment */
+	{
+		size_t const lastLLSize = litEnd - litPtr;
+		if (lastLLSize > (size_t)(oend - op))
+			return ERROR(dstSize_tooSmall);
+		memcpy(op, litPtr, lastLLSize);
+		op += lastLLSize;
+	}
+
+	return op - ostart;
+}
+
+static size_t ZSTD_decompressBlock_internal(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{ /* blockType == blockCompressed */
+	const BYTE *ip = (const BYTE *)src;
+
+	if (srcSize >= ZSTD_BLOCKSIZE_ABSOLUTEMAX)
+		return ERROR(srcSize_wrong);
+
+	/* Decode literals section */
+	{
+		size_t const litCSize = ZSTD_decodeLiteralsBlock(dctx, src, srcSize);
+		if (ZSTD_isError(litCSize))
+			return litCSize;
+		ip += litCSize;
+		srcSize -= litCSize;
+	}
+	if (sizeof(size_t) > 4) /* do not enable prefetching on 32-bits x86, as it's performance detrimental */
+				/* likely because of register pressure */
+				/* if that's the correct cause, then 32-bits ARM should be affected differently */
+				/* it would be good to test this on ARM real hardware, to see if prefetch version improves speed */
+		if (dctx->fParams.windowSize > (1 << 23))
+			return ZSTD_decompressSequencesLong(dctx, dst, dstCapacity, ip, srcSize);
+	return ZSTD_decompressSequences(dctx, dst, dstCapacity, ip, srcSize);
+}
+
+static void ZSTD_checkContinuity(ZSTD_DCtx *dctx, const void *dst)
+{
+	if (dst != dctx->previousDstEnd) { /* not contiguous */
+		dctx->dictEnd = dctx->previousDstEnd;
+		dctx->vBase = (const char *)dst - ((const char *)(dctx->previousDstEnd) - (const char *)(dctx->base));
+		dctx->base = dst;
+		dctx->previousDstEnd = dst;
+	}
+}
+
+size_t ZSTD_decompressBlock(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	size_t dSize;
+	ZSTD_checkContinuity(dctx, dst);
+	dSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize);
+	dctx->previousDstEnd = (char *)dst + dSize;
+	return dSize;
+}
+
+/** ZSTD_insertBlock() :
+	insert `src` block into `dctx` history. Useful to track uncompressed blocks. */
+size_t ZSTD_insertBlock(ZSTD_DCtx *dctx, const void *blockStart, size_t blockSize)
+{
+	ZSTD_checkContinuity(dctx, blockStart);
+	dctx->previousDstEnd = (const char *)blockStart + blockSize;
+	return blockSize;
+}
+
+size_t ZSTD_generateNxBytes(void *dst, size_t dstCapacity, BYTE byte, size_t length)
+{
+	if (length > dstCapacity)
+		return ERROR(dstSize_tooSmall);
+	memset(dst, byte, length);
+	return length;
+}
+
+/** ZSTD_findFrameCompressedSize() :
+ *  compatible with legacy mode
+ *  `src` must point to the start of a ZSTD frame, ZSTD legacy frame, or skippable frame
+ *  `srcSize` must be at least as large as the frame contained
+ *  @return : the compressed size of the frame starting at `src` */
+size_t ZSTD_findFrameCompressedSize(const void *src, size_t srcSize)
+{
+	if (srcSize >= ZSTD_skippableHeaderSize && (ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
+		return ZSTD_skippableHeaderSize + ZSTD_readLE32((const BYTE *)src + 4);
+	} else {
+		const BYTE *ip = (const BYTE *)src;
+		const BYTE *const ipstart = ip;
+		size_t remainingSize = srcSize;
+		ZSTD_frameParams fParams;
+
+		size_t const headerSize = ZSTD_frameHeaderSize(ip, remainingSize);
+		if (ZSTD_isError(headerSize))
+			return headerSize;
+
+		/* Frame Header */
+		{
+			size_t const ret = ZSTD_getFrameParams(&fParams, ip, remainingSize);
+			if (ZSTD_isError(ret))
+				return ret;
+			if (ret > 0)
+				return ERROR(srcSize_wrong);
+		}
+
+		ip += headerSize;
+		remainingSize -= headerSize;
+
+		/* Loop on each block */
+		while (1) {
+			blockProperties_t blockProperties;
+			size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
+			if (ZSTD_isError(cBlockSize))
+				return cBlockSize;
+
+			if (ZSTD_blockHeaderSize + cBlockSize > remainingSize)
+				return ERROR(srcSize_wrong);
+
+			ip += ZSTD_blockHeaderSize + cBlockSize;
+			remainingSize -= ZSTD_blockHeaderSize + cBlockSize;
+
+			if (blockProperties.lastBlock)
+				break;
+		}
+
+		if (fParams.checksumFlag) { /* Frame content checksum */
+			if (remainingSize < 4)
+				return ERROR(srcSize_wrong);
+			ip += 4;
+			remainingSize -= 4;
+		}
+
+		return ip - ipstart;
+	}
+}
+
+/*! ZSTD_decompressFrame() :
+*   @dctx must be properly initialized */
+static size_t ZSTD_decompressFrame(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void **srcPtr, size_t *srcSizePtr)
+{
+	const BYTE *ip = (const BYTE *)(*srcPtr);
+	BYTE *const ostart = (BYTE * const)dst;
+	BYTE *const oend = ostart + dstCapacity;
+	BYTE *op = ostart;
+	size_t remainingSize = *srcSizePtr;
+
+	/* check */
+	if (remainingSize < ZSTD_frameHeaderSize_min + ZSTD_blockHeaderSize)
+		return ERROR(srcSize_wrong);
+
+	/* Frame Header */
+	{
+		size_t const frameHeaderSize = ZSTD_frameHeaderSize(ip, ZSTD_frameHeaderSize_prefix);
+		if (ZSTD_isError(frameHeaderSize))
+			return frameHeaderSize;
+		if (remainingSize < frameHeaderSize + ZSTD_blockHeaderSize)
+			return ERROR(srcSize_wrong);
+		CHECK_F(ZSTD_decodeFrameHeader(dctx, ip, frameHeaderSize));
+		ip += frameHeaderSize;
+		remainingSize -= frameHeaderSize;
+	}
+
+	/* Loop on each block */
+	while (1) {
+		size_t decodedSize;
+		blockProperties_t blockProperties;
+		size_t const cBlockSize = ZSTD_getcBlockSize(ip, remainingSize, &blockProperties);
+		if (ZSTD_isError(cBlockSize))
+			return cBlockSize;
+
+		ip += ZSTD_blockHeaderSize;
+		remainingSize -= ZSTD_blockHeaderSize;
+		if (cBlockSize > remainingSize)
+			return ERROR(srcSize_wrong);
+
+		switch (blockProperties.blockType) {
+		case bt_compressed: decodedSize = ZSTD_decompressBlock_internal(dctx, op, oend - op, ip, cBlockSize); break;
+		case bt_raw: decodedSize = ZSTD_copyRawBlock(op, oend - op, ip, cBlockSize); break;
+		case bt_rle: decodedSize = ZSTD_generateNxBytes(op, oend - op, *ip, blockProperties.origSize); break;
+		case bt_reserved:
+		default: return ERROR(corruption_detected);
+		}
+
+		if (ZSTD_isError(decodedSize))
+			return decodedSize;
+		if (dctx->fParams.checksumFlag)
+			xxh64_update(&dctx->xxhState, op, decodedSize);
+		op += decodedSize;
+		ip += cBlockSize;
+		remainingSize -= cBlockSize;
+		if (blockProperties.lastBlock)
+			break;
+	}
+
+	if (dctx->fParams.checksumFlag) { /* Frame content checksum verification */
+		U32 const checkCalc = (U32)xxh64_digest(&dctx->xxhState);
+		U32 checkRead;
+		if (remainingSize < 4)
+			return ERROR(checksum_wrong);
+		checkRead = ZSTD_readLE32(ip);
+		if (checkRead != checkCalc)
+			return ERROR(checksum_wrong);
+		ip += 4;
+		remainingSize -= 4;
+	}
+
+	/* Allow caller to get size read */
+	*srcPtr = ip;
+	*srcSizePtr = remainingSize;
+	return op - ostart;
+}
+
+static const void *ZSTD_DDictDictContent(const ZSTD_DDict *ddict);
+static size_t ZSTD_DDictDictSize(const ZSTD_DDict *ddict);
+
+static size_t ZSTD_decompressMultiFrame(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize,
+					const ZSTD_DDict *ddict)
+{
+	void *const dststart = dst;
+
+	if (ddict) {
+		if (dict) {
+			/* programmer error, these two cases should be mutually exclusive */
+			return ERROR(GENERIC);
+		}
+
+		dict = ZSTD_DDictDictContent(ddict);
+		dictSize = ZSTD_DDictDictSize(ddict);
+	}
+
+	while (srcSize >= ZSTD_frameHeaderSize_prefix) {
+		U32 magicNumber;
+
+		magicNumber = ZSTD_readLE32(src);
+		if (magicNumber != ZSTD_MAGICNUMBER) {
+			if ((magicNumber & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) {
+				size_t skippableSize;
+				if (srcSize < ZSTD_skippableHeaderSize)
+					return ERROR(srcSize_wrong);
+				skippableSize = ZSTD_readLE32((const BYTE *)src + 4) + ZSTD_skippableHeaderSize;
+				if (srcSize < skippableSize) {
+					return ERROR(srcSize_wrong);
+				}
+
+				src = (const BYTE *)src + skippableSize;
+				srcSize -= skippableSize;
+				continue;
+			} else {
+				return ERROR(prefix_unknown);
+			}
+		}
+
+		if (ddict) {
+			/* we were called from ZSTD_decompress_usingDDict */
+			ZSTD_refDDict(dctx, ddict);
+		} else {
+			/* this will initialize correctly with no dict if dict == NULL, so
+			 * use this in all cases but ddict */
+			CHECK_F(ZSTD_decompressBegin_usingDict(dctx, dict, dictSize));
+		}
+		ZSTD_checkContinuity(dctx, dst);
+
+		{
+			const size_t res = ZSTD_decompressFrame(dctx, dst, dstCapacity, &src, &srcSize);
+			if (ZSTD_isError(res))
+				return res;
+			/* don't need to bounds check this, ZSTD_decompressFrame will have
+			 * already */
+			dst = (BYTE *)dst + res;
+			dstCapacity -= res;
+		}
+	}
+
+	if (srcSize)
+		return ERROR(srcSize_wrong); /* input not entirely consumed */
+
+	return (BYTE *)dst - (BYTE *)dststart;
+}
+
+size_t ZSTD_decompress_usingDict(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const void *dict, size_t dictSize)
+{
+	return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, dict, dictSize, NULL);
+}
+
+size_t ZSTD_decompressDCtx(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	return ZSTD_decompress_usingDict(dctx, dst, dstCapacity, src, srcSize, NULL, 0);
+}
+
+/*-**************************************
+*   Advanced Streaming Decompression API
+*   Bufferless and synchronous
+****************************************/
+size_t ZSTD_nextSrcSizeToDecompress(ZSTD_DCtx *dctx) { return dctx->expected; }
+
+ZSTD_nextInputType_e ZSTD_nextInputType(ZSTD_DCtx *dctx)
+{
+	switch (dctx->stage) {
+	default: /* should not happen */
+	case ZSTDds_getFrameHeaderSize:
+	case ZSTDds_decodeFrameHeader: return ZSTDnit_frameHeader;
+	case ZSTDds_decodeBlockHeader: return ZSTDnit_blockHeader;
+	case ZSTDds_decompressBlock: return ZSTDnit_block;
+	case ZSTDds_decompressLastBlock: return ZSTDnit_lastBlock;
+	case ZSTDds_checkChecksum: return ZSTDnit_checksum;
+	case ZSTDds_decodeSkippableHeader:
+	case ZSTDds_skipFrame: return ZSTDnit_skippableFrame;
+	}
+}
+
+int ZSTD_isSkipFrame(ZSTD_DCtx *dctx) { return dctx->stage == ZSTDds_skipFrame; } /* for zbuff */
+
+/** ZSTD_decompressContinue() :
+*   @return : nb of bytes generated into `dst` (necessarily <= `dstCapacity)
+*             or an error code, which can be tested using ZSTD_isError() */
+size_t ZSTD_decompressContinue(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	/* Sanity check */
+	if (srcSize != dctx->expected)
+		return ERROR(srcSize_wrong);
+	if (dstCapacity)
+		ZSTD_checkContinuity(dctx, dst);
+
+	switch (dctx->stage) {
+	case ZSTDds_getFrameHeaderSize:
+		if (srcSize != ZSTD_frameHeaderSize_prefix)
+			return ERROR(srcSize_wrong);					/* impossible */
+		if ((ZSTD_readLE32(src) & 0xFFFFFFF0U) == ZSTD_MAGIC_SKIPPABLE_START) { /* skippable frame */
+			memcpy(dctx->headerBuffer, src, ZSTD_frameHeaderSize_prefix);
+			dctx->expected = ZSTD_skippableHeaderSize - ZSTD_frameHeaderSize_prefix; /* magic number + skippable frame length */
+			dctx->stage = ZSTDds_decodeSkippableHeader;
+			return 0;
+		}
+		dctx->headerSize = ZSTD_frameHeaderSize(src, ZSTD_frameHeaderSize_prefix);
+		if (ZSTD_isError(dctx->headerSize))
+			return dctx->headerSize;
+		memcpy(dctx->headerBuffer, src, ZSTD_frameHeaderSize_prefix);
+		if (dctx->headerSize > ZSTD_frameHeaderSize_prefix) {
+			dctx->expected = dctx->headerSize - ZSTD_frameHeaderSize_prefix;
+			dctx->stage = ZSTDds_decodeFrameHeader;
+			return 0;
+		}
+		dctx->expected = 0; /* not necessary to copy more */
+
+	case ZSTDds_decodeFrameHeader:
+		memcpy(dctx->headerBuffer + ZSTD_frameHeaderSize_prefix, src, dctx->expected);
+		CHECK_F(ZSTD_decodeFrameHeader(dctx, dctx->headerBuffer, dctx->headerSize));
+		dctx->expected = ZSTD_blockHeaderSize;
+		dctx->stage = ZSTDds_decodeBlockHeader;
+		return 0;
+
+	case ZSTDds_decodeBlockHeader: {
+		blockProperties_t bp;
+		size_t const cBlockSize = ZSTD_getcBlockSize(src, ZSTD_blockHeaderSize, &bp);
+		if (ZSTD_isError(cBlockSize))
+			return cBlockSize;
+		dctx->expected = cBlockSize;
+		dctx->bType = bp.blockType;
+		dctx->rleSize = bp.origSize;
+		if (cBlockSize) {
+			dctx->stage = bp.lastBlock ? ZSTDds_decompressLastBlock : ZSTDds_decompressBlock;
+			return 0;
+		}
+		/* empty block */
+		if (bp.lastBlock) {
+			if (dctx->fParams.checksumFlag) {
+				dctx->expected = 4;
+				dctx->stage = ZSTDds_checkChecksum;
+			} else {
+				dctx->expected = 0; /* end of frame */
+				dctx->stage = ZSTDds_getFrameHeaderSize;
+			}
+		} else {
+			dctx->expected = 3; /* go directly to next header */
+			dctx->stage = ZSTDds_decodeBlockHeader;
+		}
+		return 0;
+	}
+	case ZSTDds_decompressLastBlock:
+	case ZSTDds_decompressBlock: {
+		size_t rSize;
+		switch (dctx->bType) {
+		case bt_compressed: rSize = ZSTD_decompressBlock_internal(dctx, dst, dstCapacity, src, srcSize); break;
+		case bt_raw: rSize = ZSTD_copyRawBlock(dst, dstCapacity, src, srcSize); break;
+		case bt_rle: rSize = ZSTD_setRleBlock(dst, dstCapacity, src, srcSize, dctx->rleSize); break;
+		case bt_reserved: /* should never happen */
+		default: return ERROR(corruption_detected);
+		}
+		if (ZSTD_isError(rSize))
+			return rSize;
+		if (dctx->fParams.checksumFlag)
+			xxh64_update(&dctx->xxhState, dst, rSize);
+
+		if (dctx->stage == ZSTDds_decompressLastBlock) { /* end of frame */
+			if (dctx->fParams.checksumFlag) {	/* another round for frame checksum */
+				dctx->expected = 4;
+				dctx->stage = ZSTDds_checkChecksum;
+			} else {
+				dctx->expected = 0; /* ends here */
+				dctx->stage = ZSTDds_getFrameHeaderSize;
+			}
+		} else {
+			dctx->stage = ZSTDds_decodeBlockHeader;
+			dctx->expected = ZSTD_blockHeaderSize;
+			dctx->previousDstEnd = (char *)dst + rSize;
+		}
+		return rSize;
+	}
+	case ZSTDds_checkChecksum: {
+		U32 const h32 = (U32)xxh64_digest(&dctx->xxhState);
+		U32 const check32 = ZSTD_readLE32(src); /* srcSize == 4, guaranteed by dctx->expected */
+		if (check32 != h32)
+			return ERROR(checksum_wrong);
+		dctx->expected = 0;
+		dctx->stage = ZSTDds_getFrameHeaderSize;
+		return 0;
+	}
+	case ZSTDds_decodeSkippableHeader: {
+		memcpy(dctx->headerBuffer + ZSTD_frameHeaderSize_prefix, src, dctx->expected);
+		dctx->expected = ZSTD_readLE32(dctx->headerBuffer + 4);
+		dctx->stage = ZSTDds_skipFrame;
+		return 0;
+	}
+	case ZSTDds_skipFrame: {
+		dctx->expected = 0;
+		dctx->stage = ZSTDds_getFrameHeaderSize;
+		return 0;
+	}
+	default:
+		return ERROR(GENERIC); /* impossible */
+	}
+}
+
+static size_t ZSTD_refDictContent(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
+{
+	dctx->dictEnd = dctx->previousDstEnd;
+	dctx->vBase = (const char *)dict - ((const char *)(dctx->previousDstEnd) - (const char *)(dctx->base));
+	dctx->base = dict;
+	dctx->previousDstEnd = (const char *)dict + dictSize;
+	return 0;
+}
+
+/* ZSTD_loadEntropy() :
+ * dict : must point at beginning of a valid zstd dictionary
+ * @return : size of entropy tables read */
+static size_t ZSTD_loadEntropy(ZSTD_entropyTables_t *entropy, const void *const dict, size_t const dictSize)
+{
+	const BYTE *dictPtr = (const BYTE *)dict;
+	const BYTE *const dictEnd = dictPtr + dictSize;
+
+	if (dictSize <= 8)
+		return ERROR(dictionary_corrupted);
+	dictPtr += 8; /* skip header = magic + dictID */
+
+	{
+		size_t const hSize = HUF_readDTableX4_wksp(entropy->hufTable, dictPtr, dictEnd - dictPtr, entropy->workspace, sizeof(entropy->workspace));
+		if (HUF_isError(hSize))
+			return ERROR(dictionary_corrupted);
+		dictPtr += hSize;
+	}
+
+	{
+		short offcodeNCount[MaxOff + 1];
+		U32 offcodeMaxValue = MaxOff, offcodeLog;
+		size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, dictEnd - dictPtr);
+		if (FSE_isError(offcodeHeaderSize))
+			return ERROR(dictionary_corrupted);
+		if (offcodeLog > OffFSELog)
+			return ERROR(dictionary_corrupted);
+		CHECK_E(FSE_buildDTable_wksp(entropy->OFTable, offcodeNCount, offcodeMaxValue, offcodeLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
+		dictPtr += offcodeHeaderSize;
+	}
+
+	{
+		short matchlengthNCount[MaxML + 1];
+		unsigned matchlengthMaxValue = MaxML, matchlengthLog;
+		size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd - dictPtr);
+		if (FSE_isError(matchlengthHeaderSize))
+			return ERROR(dictionary_corrupted);
+		if (matchlengthLog > MLFSELog)
+			return ERROR(dictionary_corrupted);
+		CHECK_E(FSE_buildDTable_wksp(entropy->MLTable, matchlengthNCount, matchlengthMaxValue, matchlengthLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
+		dictPtr += matchlengthHeaderSize;
+	}
+
+	{
+		short litlengthNCount[MaxLL + 1];
+		unsigned litlengthMaxValue = MaxLL, litlengthLog;
+		size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd - dictPtr);
+		if (FSE_isError(litlengthHeaderSize))
+			return ERROR(dictionary_corrupted);
+		if (litlengthLog > LLFSELog)
+			return ERROR(dictionary_corrupted);
+		CHECK_E(FSE_buildDTable_wksp(entropy->LLTable, litlengthNCount, litlengthMaxValue, litlengthLog, entropy->workspace, sizeof(entropy->workspace)), dictionary_corrupted);
+		dictPtr += litlengthHeaderSize;
+	}
+
+	if (dictPtr + 12 > dictEnd)
+		return ERROR(dictionary_corrupted);
+	{
+		int i;
+		size_t const dictContentSize = (size_t)(dictEnd - (dictPtr + 12));
+		for (i = 0; i < 3; i++) {
+			U32 const rep = ZSTD_readLE32(dictPtr);
+			dictPtr += 4;
+			if (rep == 0 || rep >= dictContentSize)
+				return ERROR(dictionary_corrupted);
+			entropy->rep[i] = rep;
+		}
+	}
+
+	return dictPtr - (const BYTE *)dict;
+}
+
+static size_t ZSTD_decompress_insertDictionary(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
+{
+	if (dictSize < 8)
+		return ZSTD_refDictContent(dctx, dict, dictSize);
+	{
+		U32 const magic = ZSTD_readLE32(dict);
+		if (magic != ZSTD_DICT_MAGIC) {
+			return ZSTD_refDictContent(dctx, dict, dictSize); /* pure content mode */
+		}
+	}
+	dctx->dictID = ZSTD_readLE32((const char *)dict + 4);
+
+	/* load entropy tables */
+	{
+		size_t const eSize = ZSTD_loadEntropy(&dctx->entropy, dict, dictSize);
+		if (ZSTD_isError(eSize))
+			return ERROR(dictionary_corrupted);
+		dict = (const char *)dict + eSize;
+		dictSize -= eSize;
+	}
+	dctx->litEntropy = dctx->fseEntropy = 1;
+
+	/* reference dictionary content */
+	return ZSTD_refDictContent(dctx, dict, dictSize);
+}
+
+size_t ZSTD_decompressBegin_usingDict(ZSTD_DCtx *dctx, const void *dict, size_t dictSize)
+{
+	CHECK_F(ZSTD_decompressBegin(dctx));
+	if (dict && dictSize)
+		CHECK_E(ZSTD_decompress_insertDictionary(dctx, dict, dictSize), dictionary_corrupted);
+	return 0;
+}
+
+/* ======   ZSTD_DDict   ====== */
+
+struct ZSTD_DDict_s {
+	void *dictBuffer;
+	const void *dictContent;
+	size_t dictSize;
+	ZSTD_entropyTables_t entropy;
+	U32 dictID;
+	U32 entropyPresent;
+	ZSTD_customMem cMem;
+}; /* typedef'd to ZSTD_DDict within "zstd.h" */
+
+size_t ZSTD_DDictWorkspaceBound(void) { return ZSTD_ALIGN(sizeof(ZSTD_stack)) + ZSTD_ALIGN(sizeof(ZSTD_DDict)); }
+
+static const void *ZSTD_DDictDictContent(const ZSTD_DDict *ddict) { return ddict->dictContent; }
+
+static size_t ZSTD_DDictDictSize(const ZSTD_DDict *ddict) { return ddict->dictSize; }
+
+static void ZSTD_refDDict(ZSTD_DCtx *dstDCtx, const ZSTD_DDict *ddict)
+{
+	ZSTD_decompressBegin(dstDCtx); /* init */
+	if (ddict) {		       /* support refDDict on NULL */
+		dstDCtx->dictID = ddict->dictID;
+		dstDCtx->base = ddict->dictContent;
+		dstDCtx->vBase = ddict->dictContent;
+		dstDCtx->dictEnd = (const BYTE *)ddict->dictContent + ddict->dictSize;
+		dstDCtx->previousDstEnd = dstDCtx->dictEnd;
+		if (ddict->entropyPresent) {
+			dstDCtx->litEntropy = 1;
+			dstDCtx->fseEntropy = 1;
+			dstDCtx->LLTptr = ddict->entropy.LLTable;
+			dstDCtx->MLTptr = ddict->entropy.MLTable;
+			dstDCtx->OFTptr = ddict->entropy.OFTable;
+			dstDCtx->HUFptr = ddict->entropy.hufTable;
+			dstDCtx->entropy.rep[0] = ddict->entropy.rep[0];
+			dstDCtx->entropy.rep[1] = ddict->entropy.rep[1];
+			dstDCtx->entropy.rep[2] = ddict->entropy.rep[2];
+		} else {
+			dstDCtx->litEntropy = 0;
+			dstDCtx->fseEntropy = 0;
+		}
+	}
+}
+
+static size_t ZSTD_loadEntropy_inDDict(ZSTD_DDict *ddict)
+{
+	ddict->dictID = 0;
+	ddict->entropyPresent = 0;
+	if (ddict->dictSize < 8)
+		return 0;
+	{
+		U32 const magic = ZSTD_readLE32(ddict->dictContent);
+		if (magic != ZSTD_DICT_MAGIC)
+			return 0; /* pure content mode */
+	}
+	ddict->dictID = ZSTD_readLE32((const char *)ddict->dictContent + 4);
+
+	/* load entropy tables */
+	CHECK_E(ZSTD_loadEntropy(&ddict->entropy, ddict->dictContent, ddict->dictSize), dictionary_corrupted);
+	ddict->entropyPresent = 1;
+	return 0;
+}
+
+static ZSTD_DDict *ZSTD_createDDict_advanced(const void *dict, size_t dictSize, unsigned byReference, ZSTD_customMem customMem)
+{
+	if (!customMem.customAlloc || !customMem.customFree)
+		return NULL;
+
+	{
+		ZSTD_DDict *const ddict = (ZSTD_DDict *)ZSTD_malloc(sizeof(ZSTD_DDict), customMem);
+		if (!ddict)
+			return NULL;
+		ddict->cMem = customMem;
+
+		if ((byReference) || (!dict) || (!dictSize)) {
+			ddict->dictBuffer = NULL;
+			ddict->dictContent = dict;
+		} else {
+			void *const internalBuffer = ZSTD_malloc(dictSize, customMem);
+			if (!internalBuffer) {
+				ZSTD_freeDDict(ddict);
+				return NULL;
+			}
+			memcpy(internalBuffer, dict, dictSize);
+			ddict->dictBuffer = internalBuffer;
+			ddict->dictContent = internalBuffer;
+		}
+		ddict->dictSize = dictSize;
+		ddict->entropy.hufTable[0] = (HUF_DTable)((HufLog)*0x1000001); /* cover both little and big endian */
+		/* parse dictionary content */
+		{
+			size_t const errorCode = ZSTD_loadEntropy_inDDict(ddict);
+			if (ZSTD_isError(errorCode)) {
+				ZSTD_freeDDict(ddict);
+				return NULL;
+			}
+		}
+
+		return ddict;
+	}
+}
+
+/*! ZSTD_initDDict() :
+*   Create a digested dictionary, to start decompression without startup delay.
+*   `dict` content is copied inside DDict.
+*   Consequently, `dict` can be released after `ZSTD_DDict` creation */
+ZSTD_DDict *ZSTD_initDDict(const void *dict, size_t dictSize, void *workspace, size_t workspaceSize)
+{
+	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
+	return ZSTD_createDDict_advanced(dict, dictSize, 1, stackMem);
+}
+
+size_t ZSTD_freeDDict(ZSTD_DDict *ddict)
+{
+	if (ddict == NULL)
+		return 0; /* support free on NULL */
+	{
+		ZSTD_customMem const cMem = ddict->cMem;
+		ZSTD_free(ddict->dictBuffer, cMem);
+		ZSTD_free(ddict, cMem);
+		return 0;
+	}
+}
+
+/*! ZSTD_getDictID_fromDict() :
+ *  Provides the dictID stored within dictionary.
+ *  if @return == 0, the dictionary is not conformant with Zstandard specification.
+ *  It can still be loaded, but as a content-only dictionary. */
+unsigned ZSTD_getDictID_fromDict(const void *dict, size_t dictSize)
+{
+	if (dictSize < 8)
+		return 0;
+	if (ZSTD_readLE32(dict) != ZSTD_DICT_MAGIC)
+		return 0;
+	return ZSTD_readLE32((const char *)dict + 4);
+}
+
+/*! ZSTD_getDictID_fromDDict() :
+ *  Provides the dictID of the dictionary loaded into `ddict`.
+ *  If @return == 0, the dictionary is not conformant to Zstandard specification, or empty.
+ *  Non-conformant dictionaries can still be loaded, but as content-only dictionaries. */
+unsigned ZSTD_getDictID_fromDDict(const ZSTD_DDict *ddict)
+{
+	if (ddict == NULL)
+		return 0;
+	return ZSTD_getDictID_fromDict(ddict->dictContent, ddict->dictSize);
+}
+
+/*! ZSTD_getDictID_fromFrame() :
+ *  Provides the dictID required to decompressed the frame stored within `src`.
+ *  If @return == 0, the dictID could not be decoded.
+ *  This could for one of the following reasons :
+ *  - The frame does not require a dictionary to be decoded (most common case).
+ *  - The frame was built with dictID intentionally removed. Whatever dictionary is necessary is a hidden information.
+ *    Note : this use case also happens when using a non-conformant dictionary.
+ *  - `srcSize` is too small, and as a result, the frame header could not be decoded (only possible if `srcSize < ZSTD_FRAMEHEADERSIZE_MAX`).
+ *  - This is not a Zstandard frame.
+ *  When identifying the exact failure cause, it's possible to used ZSTD_getFrameParams(), which will provide a more precise error code. */
+unsigned ZSTD_getDictID_fromFrame(const void *src, size_t srcSize)
+{
+	ZSTD_frameParams zfp = {0, 0, 0, 0};
+	size_t const hError = ZSTD_getFrameParams(&zfp, src, srcSize);
+	if (ZSTD_isError(hError))
+		return 0;
+	return zfp.dictID;
+}
+
+/*! ZSTD_decompress_usingDDict() :
+*   Decompression using a pre-digested Dictionary
+*   Use dictionary without significant overhead. */
+size_t ZSTD_decompress_usingDDict(ZSTD_DCtx *dctx, void *dst, size_t dstCapacity, const void *src, size_t srcSize, const ZSTD_DDict *ddict)
+{
+	/* pass content and size in case legacy frames are encountered */
+	return ZSTD_decompressMultiFrame(dctx, dst, dstCapacity, src, srcSize, NULL, 0, ddict);
+}
+
+/*=====================================
+*   Streaming decompression
+*====================================*/
+
+typedef enum { zdss_init, zdss_loadHeader, zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;
+
+/* *** Resource management *** */
+struct ZSTD_DStream_s {
+	ZSTD_DCtx *dctx;
+	ZSTD_DDict *ddictLocal;
+	const ZSTD_DDict *ddict;
+	ZSTD_frameParams fParams;
+	ZSTD_dStreamStage stage;
+	char *inBuff;
+	size_t inBuffSize;
+	size_t inPos;
+	size_t maxWindowSize;
+	char *outBuff;
+	size_t outBuffSize;
+	size_t outStart;
+	size_t outEnd;
+	size_t blockSize;
+	BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX]; /* tmp buffer to store frame header */
+	size_t lhSize;
+	ZSTD_customMem customMem;
+	void *legacyContext;
+	U32 previousLegacyVersion;
+	U32 legacyVersion;
+	U32 hostageByte;
+}; /* typedef'd to ZSTD_DStream within "zstd.h" */
+
+size_t ZSTD_DStreamWorkspaceBound(size_t maxWindowSize)
+{
+	size_t const blockSize = MIN(maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
+	size_t const inBuffSize = blockSize;
+	size_t const outBuffSize = maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
+	return ZSTD_DCtxWorkspaceBound() + ZSTD_ALIGN(sizeof(ZSTD_DStream)) + ZSTD_ALIGN(inBuffSize) + ZSTD_ALIGN(outBuffSize);
+}
+
+static ZSTD_DStream *ZSTD_createDStream_advanced(ZSTD_customMem customMem)
+{
+	ZSTD_DStream *zds;
+
+	if (!customMem.customAlloc || !customMem.customFree)
+		return NULL;
+
+	zds = (ZSTD_DStream *)ZSTD_malloc(sizeof(ZSTD_DStream), customMem);
+	if (zds == NULL)
+		return NULL;
+	memset(zds, 0, sizeof(ZSTD_DStream));
+	memcpy(&zds->customMem, &customMem, sizeof(ZSTD_customMem));
+	zds->dctx = ZSTD_createDCtx_advanced(customMem);
+	if (zds->dctx == NULL) {
+		ZSTD_freeDStream(zds);
+		return NULL;
+	}
+	zds->stage = zdss_init;
+	zds->maxWindowSize = ZSTD_MAXWINDOWSIZE_DEFAULT;
+	return zds;
+}
+
+ZSTD_DStream *ZSTD_initDStream(size_t maxWindowSize, void *workspace, size_t workspaceSize)
+{
+	ZSTD_customMem const stackMem = ZSTD_initStack(workspace, workspaceSize);
+	ZSTD_DStream *zds = ZSTD_createDStream_advanced(stackMem);
+	if (!zds) {
+		return NULL;
+	}
+
+	zds->maxWindowSize = maxWindowSize;
+	zds->stage = zdss_loadHeader;
+	zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
+	ZSTD_freeDDict(zds->ddictLocal);
+	zds->ddictLocal = NULL;
+	zds->ddict = zds->ddictLocal;
+	zds->legacyVersion = 0;
+	zds->hostageByte = 0;
+
+	{
+		size_t const blockSize = MIN(zds->maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
+		size_t const neededOutSize = zds->maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
+
+		zds->inBuff = (char *)ZSTD_malloc(blockSize, zds->customMem);
+		zds->inBuffSize = blockSize;
+		zds->outBuff = (char *)ZSTD_malloc(neededOutSize, zds->customMem);
+		zds->outBuffSize = neededOutSize;
+		if (zds->inBuff == NULL || zds->outBuff == NULL) {
+			ZSTD_freeDStream(zds);
+			return NULL;
+		}
+	}
+	return zds;
+}
+
+ZSTD_DStream *ZSTD_initDStream_usingDDict(size_t maxWindowSize, const ZSTD_DDict *ddict, void *workspace, size_t workspaceSize)
+{
+	ZSTD_DStream *zds = ZSTD_initDStream(maxWindowSize, workspace, workspaceSize);
+	if (zds) {
+		zds->ddict = ddict;
+	}
+	return zds;
+}
+
+size_t ZSTD_freeDStream(ZSTD_DStream *zds)
+{
+	if (zds == NULL)
+		return 0; /* support free on null */
+	{
+		ZSTD_customMem const cMem = zds->customMem;
+		ZSTD_freeDCtx(zds->dctx);
+		zds->dctx = NULL;
+		ZSTD_freeDDict(zds->ddictLocal);
+		zds->ddictLocal = NULL;
+		ZSTD_free(zds->inBuff, cMem);
+		zds->inBuff = NULL;
+		ZSTD_free(zds->outBuff, cMem);
+		zds->outBuff = NULL;
+		ZSTD_free(zds, cMem);
+		return 0;
+	}
+}
+
+/* *** Initialization *** */
+
+size_t ZSTD_DStreamInSize(void) { return ZSTD_BLOCKSIZE_ABSOLUTEMAX + ZSTD_blockHeaderSize; }
+size_t ZSTD_DStreamOutSize(void) { return ZSTD_BLOCKSIZE_ABSOLUTEMAX; }
+
+size_t ZSTD_resetDStream(ZSTD_DStream *zds)
+{
+	zds->stage = zdss_loadHeader;
+	zds->lhSize = zds->inPos = zds->outStart = zds->outEnd = 0;
+	zds->legacyVersion = 0;
+	zds->hostageByte = 0;
+	return ZSTD_frameHeaderSize_prefix;
+}
+
+/* *****   Decompression   ***** */
+
+ZSTD_STATIC size_t ZSTD_limitCopy(void *dst, size_t dstCapacity, const void *src, size_t srcSize)
+{
+	size_t const length = MIN(dstCapacity, srcSize);
+	memcpy(dst, src, length);
+	return length;
+}
+
+size_t ZSTD_decompressStream(ZSTD_DStream *zds, ZSTD_outBuffer *output, ZSTD_inBuffer *input)
+{
+	const char *const istart = (const char *)(input->src) + input->pos;
+	const char *const iend = (const char *)(input->src) + input->size;
+	const char *ip = istart;
+	char *const ostart = (char *)(output->dst) + output->pos;
+	char *const oend = (char *)(output->dst) + output->size;
+	char *op = ostart;
+	U32 someMoreWork = 1;
+
+	while (someMoreWork) {
+		switch (zds->stage) {
+		case zdss_init:
+			ZSTD_resetDStream(zds); /* transparent reset on starting decoding a new frame */
+						/* fall-through */
+
+		case zdss_loadHeader: {
+			size_t const hSize = ZSTD_getFrameParams(&zds->fParams, zds->headerBuffer, zds->lhSize);
+			if (ZSTD_isError(hSize))
+				return hSize;
+			if (hSize != 0) {				   /* need more input */
+				size_t const toLoad = hSize - zds->lhSize; /* if hSize!=0, hSize > zds->lhSize */
+				if (toLoad > (size_t)(iend - ip)) {	/* not enough input to load full header */
+					memcpy(zds->headerBuffer + zds->lhSize, ip, iend - ip);
+					zds->lhSize += iend - ip;
+					input->pos = input->size;
+					return (MAX(ZSTD_frameHeaderSize_min, hSize) - zds->lhSize) +
+					       ZSTD_blockHeaderSize; /* remaining header bytes + next block header */
+				}
+				memcpy(zds->headerBuffer + zds->lhSize, ip, toLoad);
+				zds->lhSize = hSize;
+				ip += toLoad;
+				break;
+			}
+
+			/* check for single-pass mode opportunity */
+			if (zds->fParams.frameContentSize && zds->fParams.windowSize /* skippable frame if == 0 */
+			    && (U64)(size_t)(oend - op) >= zds->fParams.frameContentSize) {
+				size_t const cSize = ZSTD_findFrameCompressedSize(istart, iend - istart);
+				if (cSize <= (size_t)(iend - istart)) {
+					size_t const decompressedSize = ZSTD_decompress_usingDDict(zds->dctx, op, oend - op, istart, cSize, zds->ddict);
+					if (ZSTD_isError(decompressedSize))
+						return decompressedSize;
+					ip = istart + cSize;
+					op += decompressedSize;
+					zds->dctx->expected = 0;
+					zds->stage = zdss_init;
+					someMoreWork = 0;
+					break;
+				}
+			}
+
+			/* Consume header */
+			ZSTD_refDDict(zds->dctx, zds->ddict);
+			{
+				size_t const h1Size = ZSTD_nextSrcSizeToDecompress(zds->dctx); /* == ZSTD_frameHeaderSize_prefix */
+				CHECK_F(ZSTD_decompressContinue(zds->dctx, NULL, 0, zds->headerBuffer, h1Size));
+				{
+					size_t const h2Size = ZSTD_nextSrcSizeToDecompress(zds->dctx);
+					CHECK_F(ZSTD_decompressContinue(zds->dctx, NULL, 0, zds->headerBuffer + h1Size, h2Size));
+				}
+			}
+
+			zds->fParams.windowSize = MAX(zds->fParams.windowSize, 1U << ZSTD_WINDOWLOG_ABSOLUTEMIN);
+			if (zds->fParams.windowSize > zds->maxWindowSize)
+				return ERROR(frameParameter_windowTooLarge);
+
+			/* Buffers are preallocated, but double check */
+			{
+				size_t const blockSize = MIN(zds->maxWindowSize, ZSTD_BLOCKSIZE_ABSOLUTEMAX);
+				size_t const neededOutSize = zds->maxWindowSize + blockSize + WILDCOPY_OVERLENGTH * 2;
+				if (zds->inBuffSize < blockSize) {
+					return ERROR(GENERIC);
+				}
+				if (zds->outBuffSize < neededOutSize) {
+					return ERROR(GENERIC);
+				}
+				zds->blockSize = blockSize;
+			}
+			zds->stage = zdss_read;
+		}
+		/* pass-through */
+
+		case zdss_read: {
+			size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds->dctx);
+			if (neededInSize == 0) { /* end of frame */
+				zds->stage = zdss_init;
+				someMoreWork = 0;
+				break;
+			}
+			if ((size_t)(iend - ip) >= neededInSize) { /* decode directly from src */
+				const int isSkipFrame = ZSTD_isSkipFrame(zds->dctx);
+				size_t const decodedSize = ZSTD_decompressContinue(zds->dctx, zds->outBuff + zds->outStart,
+										   (isSkipFrame ? 0 : zds->outBuffSize - zds->outStart), ip, neededInSize);
+				if (ZSTD_isError(decodedSize))
+					return decodedSize;
+				ip += neededInSize;
+				if (!decodedSize && !isSkipFrame)
+					break; /* this was just a header */
+				zds->outEnd = zds->outStart + decodedSize;
+				zds->stage = zdss_flush;
+				break;
+			}
+			if (ip == iend) {
+				someMoreWork = 0;
+				break;
+			} /* no more input */
+			zds->stage = zdss_load;
+			/* pass-through */
+		}
+
+		case zdss_load: {
+			size_t const neededInSize = ZSTD_nextSrcSizeToDecompress(zds->dctx);
+			size_t const toLoad = neededInSize - zds->inPos; /* should always be <= remaining space within inBuff */
+			size_t loadedSize;
+			if (toLoad > zds->inBuffSize - zds->inPos)
+				return ERROR(corruption_detected); /* should never happen */
+			loadedSize = ZSTD_limitCopy(zds->inBuff + zds->inPos, toLoad, ip, iend - ip);
+			ip += loadedSize;
+			zds->inPos += loadedSize;
+			if (loadedSize < toLoad) {
+				someMoreWork = 0;
+				break;
+			} /* not enough input, wait for more */
+
+			/* decode loaded input */
+			{
+				const int isSkipFrame = ZSTD_isSkipFrame(zds->dctx);
+				size_t const decodedSize = ZSTD_decompressContinue(zds->dctx, zds->outBuff + zds->outStart, zds->outBuffSize - zds->outStart,
+										   zds->inBuff, neededInSize);
+				if (ZSTD_isError(decodedSize))
+					return decodedSize;
+				zds->inPos = 0; /* input is consumed */
+				if (!decodedSize && !isSkipFrame) {
+					zds->stage = zdss_read;
+					break;
+				} /* this was just a header */
+				zds->outEnd = zds->outStart + decodedSize;
+				zds->stage = zdss_flush;
+				/* pass-through */
+			}
+		}
+
+		case zdss_flush: {
+			size_t const toFlushSize = zds->outEnd - zds->outStart;
+			size_t const flushedSize = ZSTD_limitCopy(op, oend - op, zds->outBuff + zds->outStart, toFlushSize);
+			op += flushedSize;
+			zds->outStart += flushedSize;
+			if (flushedSize == toFlushSize) { /* flush completed */
+				zds->stage = zdss_read;
+				if (zds->outStart + zds->blockSize > zds->outBuffSize)
+					zds->outStart = zds->outEnd = 0;
+				break;
+			}
+			/* cannot complete flush */
+			someMoreWork = 0;
+			break;
+		}
+		default:
+			return ERROR(GENERIC); /* impossible */
+		}
+	}
+
+	/* result */
+	input->pos += (size_t)(ip - istart);
+	output->pos += (size_t)(op - ostart);
+	{
+		size_t nextSrcSizeHint = ZSTD_nextSrcSizeToDecompress(zds->dctx);
+		if (!nextSrcSizeHint) {			    /* frame fully decoded */
+			if (zds->outEnd == zds->outStart) { /* output fully flushed */
+				if (zds->hostageByte) {
+					if (input->pos >= input->size) {
+						zds->stage = zdss_read;
+						return 1;
+					}	     /* can't release hostage (not present) */
+					input->pos++; /* release hostage */
+				}
+				return 0;
+			}
+			if (!zds->hostageByte) { /* output not fully flushed; keep last byte as hostage; will be released when all output is flushed */
+				input->pos--;    /* note : pos > 0, otherwise, impossible to finish reading last block */
+				zds->hostageByte = 1;
+			}
+			return 1;
+		}
+		nextSrcSizeHint += ZSTD_blockHeaderSize * (ZSTD_nextInputType(zds->dctx) == ZSTDnit_block); /* preload header of next block */
+		if (zds->inPos > nextSrcSizeHint)
+			return ERROR(GENERIC); /* should never happen */
+		nextSrcSizeHint -= zds->inPos; /* already loaded*/
+		return nextSrcSizeHint;
+	}
+}
+
+EXPORT_SYMBOL(ZSTD_DCtxWorkspaceBound);
+EXPORT_SYMBOL(ZSTD_initDCtx);
+EXPORT_SYMBOL(ZSTD_decompressDCtx);
+EXPORT_SYMBOL(ZSTD_decompress_usingDict);
+
+EXPORT_SYMBOL(ZSTD_DDictWorkspaceBound);
+EXPORT_SYMBOL(ZSTD_initDDict);
+EXPORT_SYMBOL(ZSTD_decompress_usingDDict);
+
+EXPORT_SYMBOL(ZSTD_DStreamWorkspaceBound);
+EXPORT_SYMBOL(ZSTD_initDStream);
+EXPORT_SYMBOL(ZSTD_initDStream_usingDDict);
+EXPORT_SYMBOL(ZSTD_resetDStream);
+EXPORT_SYMBOL(ZSTD_decompressStream);
+EXPORT_SYMBOL(ZSTD_DStreamInSize);
+EXPORT_SYMBOL(ZSTD_DStreamOutSize);
+
+EXPORT_SYMBOL(ZSTD_findFrameCompressedSize);
+EXPORT_SYMBOL(ZSTD_getFrameContentSize);
+EXPORT_SYMBOL(ZSTD_findDecompressedSize);
+
+EXPORT_SYMBOL(ZSTD_isFrame);
+EXPORT_SYMBOL(ZSTD_getDictID_fromDict);
+EXPORT_SYMBOL(ZSTD_getDictID_fromDDict);
+EXPORT_SYMBOL(ZSTD_getDictID_fromFrame);
+
+EXPORT_SYMBOL(ZSTD_getFrameParams);
+EXPORT_SYMBOL(ZSTD_decompressBegin);
+EXPORT_SYMBOL(ZSTD_decompressBegin_usingDict);
+EXPORT_SYMBOL(ZSTD_copyDCtx);
+EXPORT_SYMBOL(ZSTD_nextSrcSizeToDecompress);
+EXPORT_SYMBOL(ZSTD_decompressContinue);
+EXPORT_SYMBOL(ZSTD_nextInputType);
+
+EXPORT_SYMBOL(ZSTD_decompressBlock);
+EXPORT_SYMBOL(ZSTD_insertBlock);
+
+MODULE_LICENSE("Dual BSD/GPL");
+MODULE_DESCRIPTION("Zstd Decompressor");
diff --git a/lib/zstd/entropy_common.c b/lib/zstd/entropy_common.c
new file mode 100644
index 000000000000..2b0a643c32c4
--- /dev/null
+++ b/lib/zstd/entropy_common.c
@@ -0,0 +1,243 @@
+/*
+ * Common functions of New Generation Entropy library
+ * Copyright (C) 2016, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ */
+
+/* *************************************
+*  Dependencies
+***************************************/
+#include "error_private.h" /* ERR_*, ERROR */
+#include "fse.h"
+#include "huf.h"
+#include "mem.h"
+
+/*===   Version   ===*/
+unsigned FSE_versionNumber(void) { return FSE_VERSION_NUMBER; }
+
+/*===   Error Management   ===*/
+unsigned FSE_isError(size_t code) { return ERR_isError(code); }
+
+unsigned HUF_isError(size_t code) { return ERR_isError(code); }
+
+/*-**************************************************************
+*  FSE NCount encoding-decoding
+****************************************************************/
+size_t FSE_readNCount(short *normalizedCounter, unsigned *maxSVPtr, unsigned *tableLogPtr, const void *headerBuffer, size_t hbSize)
+{
+	const BYTE *const istart = (const BYTE *)headerBuffer;
+	const BYTE *const iend = istart + hbSize;
+	const BYTE *ip = istart;
+	int nbBits;
+	int remaining;
+	int threshold;
+	U32 bitStream;
+	int bitCount;
+	unsigned charnum = 0;
+	int previous0 = 0;
+
+	if (hbSize < 4)
+		return ERROR(srcSize_wrong);
+	bitStream = ZSTD_readLE32(ip);
+	nbBits = (bitStream & 0xF) + FSE_MIN_TABLELOG; /* extract tableLog */
+	if (nbBits > FSE_TABLELOG_ABSOLUTE_MAX)
+		return ERROR(tableLog_tooLarge);
+	bitStream >>= 4;
+	bitCount = 4;
+	*tableLogPtr = nbBits;
+	remaining = (1 << nbBits) + 1;
+	threshold = 1 << nbBits;
+	nbBits++;
+
+	while ((remaining > 1) & (charnum <= *maxSVPtr)) {
+		if (previous0) {
+			unsigned n0 = charnum;
+			while ((bitStream & 0xFFFF) == 0xFFFF) {
+				n0 += 24;
+				if (ip < iend - 5) {
+					ip += 2;
+					bitStream = ZSTD_readLE32(ip) >> bitCount;
+				} else {
+					bitStream >>= 16;
+					bitCount += 16;
+				}
+			}
+			while ((bitStream & 3) == 3) {
+				n0 += 3;
+				bitStream >>= 2;
+				bitCount += 2;
+			}
+			n0 += bitStream & 3;
+			bitCount += 2;
+			if (n0 > *maxSVPtr)
+				return ERROR(maxSymbolValue_tooSmall);
+			while (charnum < n0)
+				normalizedCounter[charnum++] = 0;
+			if ((ip <= iend - 7) || (ip + (bitCount >> 3) <= iend - 4)) {
+				ip += bitCount >> 3;
+				bitCount &= 7;
+				bitStream = ZSTD_readLE32(ip) >> bitCount;
+			} else {
+				bitStream >>= 2;
+			}
+		}
+		{
+			int const max = (2 * threshold - 1) - remaining;
+			int count;
+
+			if ((bitStream & (threshold - 1)) < (U32)max) {
+				count = bitStream & (threshold - 1);
+				bitCount += nbBits - 1;
+			} else {
+				count = bitStream & (2 * threshold - 1);
+				if (count >= threshold)
+					count -= max;
+				bitCount += nbBits;
+			}
+
+			count--;				 /* extra accuracy */
+			remaining -= count < 0 ? -count : count; /* -1 means +1 */
+			normalizedCounter[charnum++] = (short)count;
+			previous0 = !count;
+			while (remaining < threshold) {
+				nbBits--;
+				threshold >>= 1;
+			}
+
+			if ((ip <= iend - 7) || (ip + (bitCount >> 3) <= iend - 4)) {
+				ip += bitCount >> 3;
+				bitCount &= 7;
+			} else {
+				bitCount -= (int)(8 * (iend - 4 - ip));
+				ip = iend - 4;
+			}
+			bitStream = ZSTD_readLE32(ip) >> (bitCount & 31);
+		}
+	} /* while ((remaining>1) & (charnum<=*maxSVPtr)) */
+	if (remaining != 1)
+		return ERROR(corruption_detected);
+	if (bitCount > 32)
+		return ERROR(corruption_detected);
+	*maxSVPtr = charnum - 1;
+
+	ip += (bitCount + 7) >> 3;
+	return ip - istart;
+}
+
+/*! HUF_readStats() :
+	Read compact Huffman tree, saved by HUF_writeCTable().
+	`huffWeight` is destination buffer.
+	`rankStats` is assumed to be a table of at least HUF_TABLELOG_MAX U32.
+	@return : size read from `src` , or an error Code .
+	Note : Needed by HUF_readCTable() and HUF_readDTableX?() .
+*/
+size_t HUF_readStats_wksp(BYTE *huffWeight, size_t hwSize, U32 *rankStats, U32 *nbSymbolsPtr, U32 *tableLogPtr, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
+{
+	U32 weightTotal;
+	const BYTE *ip = (const BYTE *)src;
+	size_t iSize;
+	size_t oSize;
+
+	if (!srcSize)
+		return ERROR(srcSize_wrong);
+	iSize = ip[0];
+	/* memset(huffWeight, 0, hwSize);   */ /* is not necessary, even though some analyzer complain ... */
+
+	if (iSize >= 128) { /* special header */
+		oSize = iSize - 127;
+		iSize = ((oSize + 1) / 2);
+		if (iSize + 1 > srcSize)
+			return ERROR(srcSize_wrong);
+		if (oSize >= hwSize)
+			return ERROR(corruption_detected);
+		ip += 1;
+		{
+			U32 n;
+			for (n = 0; n < oSize; n += 2) {
+				huffWeight[n] = ip[n / 2] >> 4;
+				huffWeight[n + 1] = ip[n / 2] & 15;
+			}
+		}
+	} else {						 /* header compressed with FSE (normal case) */
+		if (iSize + 1 > srcSize)
+			return ERROR(srcSize_wrong);
+		oSize = FSE_decompress_wksp(huffWeight, hwSize - 1, ip + 1, iSize, 6, workspace, workspaceSize); /* max (hwSize-1) values decoded, as last one is implied */
+		if (FSE_isError(oSize))
+			return oSize;
+	}
+
+	/* collect weight stats */
+	memset(rankStats, 0, (HUF_TABLELOG_MAX + 1) * sizeof(U32));
+	weightTotal = 0;
+	{
+		U32 n;
+		for (n = 0; n < oSize; n++) {
+			if (huffWeight[n] >= HUF_TABLELOG_MAX)
+				return ERROR(corruption_detected);
+			rankStats[huffWeight[n]]++;
+			weightTotal += (1 << huffWeight[n]) >> 1;
+		}
+	}
+	if (weightTotal == 0)
+		return ERROR(corruption_detected);
+
+	/* get last non-null symbol weight (implied, total must be 2^n) */
+	{
+		U32 const tableLog = BIT_highbit32(weightTotal) + 1;
+		if (tableLog > HUF_TABLELOG_MAX)
+			return ERROR(corruption_detected);
+		*tableLogPtr = tableLog;
+		/* determine last weight */
+		{
+			U32 const total = 1 << tableLog;
+			U32 const rest = total - weightTotal;
+			U32 const verif = 1 << BIT_highbit32(rest);
+			U32 const lastWeight = BIT_highbit32(rest) + 1;
+			if (verif != rest)
+				return ERROR(corruption_detected); /* last value must be a clean power of 2 */
+			huffWeight[oSize] = (BYTE)lastWeight;
+			rankStats[lastWeight]++;
+		}
+	}
+
+	/* check tree construction validity */
+	if ((rankStats[1] < 2) || (rankStats[1] & 1))
+		return ERROR(corruption_detected); /* by construction : at least 2 elts of rank 1, must be even */
+
+	/* results */
+	*nbSymbolsPtr = (U32)(oSize + 1);
+	return iSize + 1;
+}
diff --git a/lib/zstd/error_private.h b/lib/zstd/error_private.h
new file mode 100644
index 000000000000..1a60b31f706c
--- /dev/null
+++ b/lib/zstd/error_private.h
@@ -0,0 +1,53 @@
+/**
+ * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd.
+ * An additional grant of patent rights can be found in the PATENTS file in the
+ * same directory.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ */
+
+/* Note : this module is expected to remain private, do not expose it */
+
+#ifndef ERROR_H_MODULE
+#define ERROR_H_MODULE
+
+/* ****************************************
+*  Dependencies
+******************************************/
+#include <linux/types.h> /* size_t */
+#include <linux/zstd.h>  /* enum list */
+
+/* ****************************************
+*  Compiler-specific
+******************************************/
+#define ERR_STATIC static __attribute__((unused))
+
+/*-****************************************
+*  Customization (error_public.h)
+******************************************/
+typedef ZSTD_ErrorCode ERR_enum;
+#define PREFIX(name) ZSTD_error_##name
+
+/*-****************************************
+*  Error codes handling
+******************************************/
+#define ERROR(name) ((size_t)-PREFIX(name))
+
+ERR_STATIC unsigned ERR_isError(size_t code) { return (code > ERROR(maxCode)); }
+
+ERR_STATIC ERR_enum ERR_getErrorCode(size_t code)
+{
+	if (!ERR_isError(code))
+		return (ERR_enum)0;
+	return (ERR_enum)(0 - code);
+}
+
+#endif /* ERROR_H_MODULE */
diff --git a/lib/zstd/fse.h b/lib/zstd/fse.h
new file mode 100644
index 000000000000..7460ab04b191
--- /dev/null
+++ b/lib/zstd/fse.h
@@ -0,0 +1,575 @@
+/*
+ * FSE : Finite State Entropy codec
+ * Public Prototypes declaration
+ * Copyright (C) 2013-2016, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ */
+#ifndef FSE_H
+#define FSE_H
+
+/*-*****************************************
+*  Dependencies
+******************************************/
+#include <linux/types.h> /* size_t, ptrdiff_t */
+
+/*-*****************************************
+*  FSE_PUBLIC_API : control library symbols visibility
+******************************************/
+#define FSE_PUBLIC_API
+
+/*------   Version   ------*/
+#define FSE_VERSION_MAJOR 0
+#define FSE_VERSION_MINOR 9
+#define FSE_VERSION_RELEASE 0
+
+#define FSE_LIB_VERSION FSE_VERSION_MAJOR.FSE_VERSION_MINOR.FSE_VERSION_RELEASE
+#define FSE_QUOTE(str) #str
+#define FSE_EXPAND_AND_QUOTE(str) FSE_QUOTE(str)
+#define FSE_VERSION_STRING FSE_EXPAND_AND_QUOTE(FSE_LIB_VERSION)
+
+#define FSE_VERSION_NUMBER (FSE_VERSION_MAJOR * 100 * 100 + FSE_VERSION_MINOR * 100 + FSE_VERSION_RELEASE)
+FSE_PUBLIC_API unsigned FSE_versionNumber(void); /**< library version number; to be used when checking dll version */
+
+/*-*****************************************
+*  Tool functions
+******************************************/
+FSE_PUBLIC_API size_t FSE_compressBound(size_t size); /* maximum compressed size */
+
+/* Error Management */
+FSE_PUBLIC_API unsigned FSE_isError(size_t code); /* tells if a return value is an error code */
+
+/*-*****************************************
+*  FSE detailed API
+******************************************/
+/*!
+FSE_compress() does the following:
+1. count symbol occurrence from source[] into table count[]
+2. normalize counters so that sum(count[]) == Power_of_2 (2^tableLog)
+3. save normalized counters to memory buffer using writeNCount()
+4. build encoding table 'CTable' from normalized counters
+5. encode the data stream using encoding table 'CTable'
+
+FSE_decompress() does the following:
+1. read normalized counters with readNCount()
+2. build decoding table 'DTable' from normalized counters
+3. decode the data stream using decoding table 'DTable'
+
+The following API allows targeting specific sub-functions for advanced tasks.
+For example, it's possible to compress several blocks using the same 'CTable',
+or to save and provide normalized distribution using external method.
+*/
+
+/* *** COMPRESSION *** */
+/*! FSE_optimalTableLog():
+	dynamically downsize 'tableLog' when conditions are met.
+	It saves CPU time, by using smaller tables, while preserving or even improving compression ratio.
+	@return : recommended tableLog (necessarily <= 'maxTableLog') */
+FSE_PUBLIC_API unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
+
+/*! FSE_normalizeCount():
+	normalize counts so that sum(count[]) == Power_of_2 (2^tableLog)
+	'normalizedCounter' is a table of short, of minimum size (maxSymbolValue+1).
+	@return : tableLog,
+			  or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_normalizeCount(short *normalizedCounter, unsigned tableLog, const unsigned *count, size_t srcSize, unsigned maxSymbolValue);
+
+/*! FSE_NCountWriteBound():
+	Provides the maximum possible size of an FSE normalized table, given 'maxSymbolValue' and 'tableLog'.
+	Typically useful for allocation purpose. */
+FSE_PUBLIC_API size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog);
+
+/*! FSE_writeNCount():
+	Compactly save 'normalizedCounter' into 'buffer'.
+	@return : size of the compressed table,
+			  or an errorCode, which can be tested using FSE_isError(). */
+FSE_PUBLIC_API size_t FSE_writeNCount(void *buffer, size_t bufferSize, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog);
+
+/*! Constructor and Destructor of FSE_CTable.
+	Note that FSE_CTable size depends on 'tableLog' and 'maxSymbolValue' */
+typedef unsigned FSE_CTable; /* don't allocate that. It's only meant to be more restrictive than void* */
+
+/*! FSE_compress_usingCTable():
+	Compress `src` using `ct` into `dst` which must be already allocated.
+	@return : size of compressed data (<= `dstCapacity`),
+			  or 0 if compressed data could not fit into `dst`,
+			  or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_compress_usingCTable(void *dst, size_t dstCapacity, const void *src, size_t srcSize, const FSE_CTable *ct);
+
+/*!
+Tutorial :
+----------
+The first step is to count all symbols. FSE_count() does this job very fast.
+Result will be saved into 'count', a table of unsigned int, which must be already allocated, and have 'maxSymbolValuePtr[0]+1' cells.
+'src' is a table of bytes of size 'srcSize'. All values within 'src' MUST be <= maxSymbolValuePtr[0]
+maxSymbolValuePtr[0] will be updated, with its real value (necessarily <= original value)
+FSE_count() will return the number of occurrence of the most frequent symbol.
+This can be used to know if there is a single symbol within 'src', and to quickly evaluate its compressibility.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
+
+The next step is to normalize the frequencies.
+FSE_normalizeCount() will ensure that sum of frequencies is == 2 ^'tableLog'.
+It also guarantees a minimum of 1 to any Symbol with frequency >= 1.
+You can use 'tableLog'==0 to mean "use default tableLog value".
+If you are unsure of which tableLog value to use, you can ask FSE_optimalTableLog(),
+which will provide the optimal valid tableLog given sourceSize, maxSymbolValue, and a user-defined maximum (0 means "default").
+
+The result of FSE_normalizeCount() will be saved into a table,
+called 'normalizedCounter', which is a table of signed short.
+'normalizedCounter' must be already allocated, and have at least 'maxSymbolValue+1' cells.
+The return value is tableLog if everything proceeded as expected.
+It is 0 if there is a single symbol within distribution.
+If there is an error (ex: invalid tableLog value), the function will return an ErrorCode (which can be tested using FSE_isError()).
+
+'normalizedCounter' can be saved in a compact manner to a memory area using FSE_writeNCount().
+'buffer' must be already allocated.
+For guaranteed success, buffer size must be at least FSE_headerBound().
+The result of the function is the number of bytes written into 'buffer'.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError(); ex : buffer size too small).
+
+'normalizedCounter' can then be used to create the compression table 'CTable'.
+The space required by 'CTable' must be already allocated, using FSE_createCTable().
+You can then use FSE_buildCTable() to fill 'CTable'.
+If there is an error, both functions will return an ErrorCode (which can be tested using FSE_isError()).
+
+'CTable' can then be used to compress 'src', with FSE_compress_usingCTable().
+Similar to FSE_count(), the convention is that 'src' is assumed to be a table of char of size 'srcSize'
+The function returns the size of compressed data (without header), necessarily <= `dstCapacity`.
+If it returns '0', compressed data could not fit into 'dst'.
+If there is an error, the function will return an ErrorCode (which can be tested using FSE_isError()).
+*/
+
+/* *** DECOMPRESSION *** */
+
+/*! FSE_readNCount():
+	Read compactly saved 'normalizedCounter' from 'rBuffer'.
+	@return : size read from 'rBuffer',
+			  or an errorCode, which can be tested using FSE_isError().
+			  maxSymbolValuePtr[0] and tableLogPtr[0] will also be updated with their respective values */
+FSE_PUBLIC_API size_t FSE_readNCount(short *normalizedCounter, unsigned *maxSymbolValuePtr, unsigned *tableLogPtr, const void *rBuffer, size_t rBuffSize);
+
+/*! Constructor and Destructor of FSE_DTable.
+	Note that its size depends on 'tableLog' */
+typedef unsigned FSE_DTable; /* don't allocate that. It's just a way to be more restrictive than void* */
+
+/*! FSE_buildDTable():
+	Builds 'dt', which must be already allocated, using FSE_createDTable().
+	return : 0, or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_buildDTable_wksp(FSE_DTable *dt, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workspace, size_t workspaceSize);
+
+/*! FSE_decompress_usingDTable():
+	Decompress compressed source `cSrc` of size `cSrcSize` using `dt`
+	into `dst` which must be already allocated.
+	@return : size of regenerated data (necessarily <= `dstCapacity`),
+			  or an errorCode, which can be tested using FSE_isError() */
+FSE_PUBLIC_API size_t FSE_decompress_usingDTable(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt);
+
+/*!
+Tutorial :
+----------
+(Note : these functions only decompress FSE-compressed blocks.
+ If block is uncompressed, use memcpy() instead
+ If block is a single repeated byte, use memset() instead )
+
+The first step is to obtain the normalized frequencies of symbols.
+This can be performed by FSE_readNCount() if it was saved using FSE_writeNCount().
+'normalizedCounter' must be already allocated, and have at least 'maxSymbolValuePtr[0]+1' cells of signed short.
+In practice, that means it's necessary to know 'maxSymbolValue' beforehand,
+or size the table to handle worst case situations (typically 256).
+FSE_readNCount() will provide 'tableLog' and 'maxSymbolValue'.
+The result of FSE_readNCount() is the number of bytes read from 'rBuffer'.
+Note that 'rBufferSize' must be at least 4 bytes, even if useful information is less than that.
+If there is an error, the function will return an error code, which can be tested using FSE_isError().
+
+The next step is to build the decompression tables 'FSE_DTable' from 'normalizedCounter'.
+This is performed by the function FSE_buildDTable().
+The space required by 'FSE_DTable' must be already allocated using FSE_createDTable().
+If there is an error, the function will return an error code, which can be tested using FSE_isError().
+
+`FSE_DTable` can then be used to decompress `cSrc`, with FSE_decompress_usingDTable().
+`cSrcSize` must be strictly correct, otherwise decompression will fail.
+FSE_decompress_usingDTable() result will tell how many bytes were regenerated (<=`dstCapacity`).
+If there is an error, the function will return an error code, which can be tested using FSE_isError(). (ex: dst buffer too small)
+*/
+
+/* *** Dependency *** */
+#include "bitstream.h"
+
+/* *****************************************
+*  Static allocation
+*******************************************/
+/* FSE buffer bounds */
+#define FSE_NCOUNTBOUND 512
+#define FSE_BLOCKBOUND(size) (size + (size >> 7))
+#define FSE_COMPRESSBOUND(size) (FSE_NCOUNTBOUND + FSE_BLOCKBOUND(size)) /* Macro version, useful for static allocation */
+
+/* It is possible to statically allocate FSE CTable/DTable as a table of FSE_CTable/FSE_DTable using below macros */
+#define FSE_CTABLE_SIZE_U32(maxTableLog, maxSymbolValue) (1 + (1 << (maxTableLog - 1)) + ((maxSymbolValue + 1) * 2))
+#define FSE_DTABLE_SIZE_U32(maxTableLog) (1 + (1 << maxTableLog))
+
+/* *****************************************
+*  FSE advanced API
+*******************************************/
+/* FSE_count_wksp() :
+ * Same as FSE_count(), but using an externally provided scratch buffer.
+ * `workSpace` size must be table of >= `1024` unsigned
+ */
+size_t FSE_count_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned *workSpace);
+
+/* FSE_countFast_wksp() :
+ * Same as FSE_countFast(), but using an externally provided scratch buffer.
+ * `workSpace` must be a table of minimum `1024` unsigned
+ */
+size_t FSE_countFast_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *src, size_t srcSize, unsigned *workSpace);
+
+/*! FSE_count_simple
+ * Same as FSE_countFast(), but does not use any additional memory (not even on stack).
+ * This function is unsafe, and will segfault if any value within `src` is `> *maxSymbolValuePtr` (presuming it's also the size of `count`).
+*/
+size_t FSE_count_simple(unsigned *count, unsigned *maxSymbolValuePtr, const void *src, size_t srcSize);
+
+unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus);
+/**< same as FSE_optimalTableLog(), which used `minus==2` */
+
+size_t FSE_buildCTable_raw(FSE_CTable *ct, unsigned nbBits);
+/**< build a fake FSE_CTable, designed for a flat distribution, where each symbol uses nbBits */
+
+size_t FSE_buildCTable_rle(FSE_CTable *ct, unsigned char symbolValue);
+/**< build a fake FSE_CTable, designed to compress always the same symbolValue */
+
+/* FSE_buildCTable_wksp() :
+ * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
+ * `wkspSize` must be >= `(1<<tableLog)`.
+ */
+size_t FSE_buildCTable_wksp(FSE_CTable *ct, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workSpace, size_t wkspSize);
+
+size_t FSE_buildDTable_raw(FSE_DTable *dt, unsigned nbBits);
+/**< build a fake FSE_DTable, designed to read a flat distribution where each symbol uses nbBits */
+
+size_t FSE_buildDTable_rle(FSE_DTable *dt, unsigned char symbolValue);
+/**< build a fake FSE_DTable, designed to always generate the same symbolValue */
+
+size_t FSE_decompress_wksp(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, unsigned maxLog, void *workspace, size_t workspaceSize);
+/**< same as FSE_decompress(), using an externally allocated `workSpace` produced with `FSE_DTABLE_SIZE_U32(maxLog)` */
+
+/* *****************************************
+*  FSE symbol compression API
+*******************************************/
+/*!
+   This API consists of small unitary functions, which highly benefit from being inlined.
+   Hence their body are included in next section.
+*/
+typedef struct {
+	ptrdiff_t value;
+	const void *stateTable;
+	const void *symbolTT;
+	unsigned stateLog;
+} FSE_CState_t;
+
+static void FSE_initCState(FSE_CState_t *CStatePtr, const FSE_CTable *ct);
+
+static void FSE_encodeSymbol(BIT_CStream_t *bitC, FSE_CState_t *CStatePtr, unsigned symbol);
+
+static void FSE_flushCState(BIT_CStream_t *bitC, const FSE_CState_t *CStatePtr);
+
+/**<
+These functions are inner components of FSE_compress_usingCTable().
+They allow the creation of custom streams, mixing multiple tables and bit sources.
+
+A key property to keep in mind is that encoding and decoding are done **in reverse direction**.
+So the first symbol you will encode is the last you will decode, like a LIFO stack.
+
+You will need a few variables to track your CStream. They are :
+
+FSE_CTable    ct;         // Provided by FSE_buildCTable()
+BIT_CStream_t bitStream;  // bitStream tracking structure
+FSE_CState_t  state;      // State tracking structure (can have several)
+
+
+The first thing to do is to init bitStream and state.
+	size_t errorCode = BIT_initCStream(&bitStream, dstBuffer, maxDstSize);
+	FSE_initCState(&state, ct);
+
+Note that BIT_initCStream() can produce an error code, so its result should be tested, using FSE_isError();
+You can then encode your input data, byte after byte.
+FSE_encodeSymbol() outputs a maximum of 'tableLog' bits at a time.
+Remember decoding will be done in reverse direction.
+	FSE_encodeByte(&bitStream, &state, symbol);
+
+At any time, you can also add any bit sequence.
+Note : maximum allowed nbBits is 25, for compatibility with 32-bits decoders
+	BIT_addBits(&bitStream, bitField, nbBits);
+
+The above methods don't commit data to memory, they just store it into local register, for speed.
+Local register size is 64-bits on 64-bits systems, 32-bits on 32-bits systems (size_t).
+Writing data to memory is a manual operation, performed by the flushBits function.
+	BIT_flushBits(&bitStream);
+
+Your last FSE encoding operation shall be to flush your last state value(s).
+	FSE_flushState(&bitStream, &state);
+
+Finally, you must close the bitStream.
+The function returns the size of CStream in bytes.
+If data couldn't fit into dstBuffer, it will return a 0 ( == not compressible)
+If there is an error, it returns an errorCode (which can be tested using FSE_isError()).
+	size_t size = BIT_closeCStream(&bitStream);
+*/
+
+/* *****************************************
+*  FSE symbol decompression API
+*******************************************/
+typedef struct {
+	size_t state;
+	const void *table; /* precise table may vary, depending on U16 */
+} FSE_DState_t;
+
+static void FSE_initDState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD, const FSE_DTable *dt);
+
+static unsigned char FSE_decodeSymbol(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD);
+
+static unsigned FSE_endOfDState(const FSE_DState_t *DStatePtr);
+
+/**<
+Let's now decompose FSE_decompress_usingDTable() into its unitary components.
+You will decode FSE-encoded symbols from the bitStream,
+and also any other bitFields you put in, **in reverse order**.
+
+You will need a few variables to track your bitStream. They are :
+
+BIT_DStream_t DStream;    // Stream context
+FSE_DState_t  DState;     // State context. Multiple ones are possible
+FSE_DTable*   DTablePtr;  // Decoding table, provided by FSE_buildDTable()
+
+The first thing to do is to init the bitStream.
+	errorCode = BIT_initDStream(&DStream, srcBuffer, srcSize);
+
+You should then retrieve your initial state(s)
+(in reverse flushing order if you have several ones) :
+	errorCode = FSE_initDState(&DState, &DStream, DTablePtr);
+
+You can then decode your data, symbol after symbol.
+For information the maximum number of bits read by FSE_decodeSymbol() is 'tableLog'.
+Keep in mind that symbols are decoded in reverse order, like a LIFO stack (last in, first out).
+	unsigned char symbol = FSE_decodeSymbol(&DState, &DStream);
+
+You can retrieve any bitfield you eventually stored into the bitStream (in reverse order)
+Note : maximum allowed nbBits is 25, for 32-bits compatibility
+	size_t bitField = BIT_readBits(&DStream, nbBits);
+
+All above operations only read from local register (which size depends on size_t).
+Refueling the register from memory is manually performed by the reload method.
+	endSignal = FSE_reloadDStream(&DStream);
+
+BIT_reloadDStream() result tells if there is still some more data to read from DStream.
+BIT_DStream_unfinished : there is still some data left into the DStream.
+BIT_DStream_endOfBuffer : Dstream reached end of buffer. Its container may no longer be completely filled.
+BIT_DStream_completed : Dstream reached its exact end, corresponding in general to decompression completed.
+BIT_DStream_tooFar : Dstream went too far. Decompression result is corrupted.
+
+When reaching end of buffer (BIT_DStream_endOfBuffer), progress slowly, notably if you decode multiple symbols per loop,
+to properly detect the exact end of stream.
+After each decoded symbol, check if DStream is fully consumed using this simple test :
+	BIT_reloadDStream(&DStream) >= BIT_DStream_completed
+
+When it's done, verify decompression is fully completed, by checking both DStream and the relevant states.
+Checking if DStream has reached its end is performed by :
+	BIT_endOfDStream(&DStream);
+Check also the states. There might be some symbols left there, if some high probability ones (>50%) are possible.
+	FSE_endOfDState(&DState);
+*/
+
+/* *****************************************
+*  FSE unsafe API
+*******************************************/
+static unsigned char FSE_decodeSymbolFast(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD);
+/* faster, but works only if nbBits is always >= 1 (otherwise, result will be corrupted) */
+
+/* *****************************************
+*  Implementation of inlined functions
+*******************************************/
+typedef struct {
+	int deltaFindState;
+	U32 deltaNbBits;
+} FSE_symbolCompressionTransform; /* total 8 bytes */
+
+ZSTD_STATIC void FSE_initCState(FSE_CState_t *statePtr, const FSE_CTable *ct)
+{
+	const void *ptr = ct;
+	const U16 *u16ptr = (const U16 *)ptr;
+	const U32 tableLog = ZSTD_read16(ptr);
+	statePtr->value = (ptrdiff_t)1 << tableLog;
+	statePtr->stateTable = u16ptr + 2;
+	statePtr->symbolTT = ((const U32 *)ct + 1 + (tableLog ? (1 << (tableLog - 1)) : 1));
+	statePtr->stateLog = tableLog;
+}
+
+/*! FSE_initCState2() :
+*   Same as FSE_initCState(), but the first symbol to include (which will be the last to be read)
+*   uses the smallest state value possible, saving the cost of this symbol */
+ZSTD_STATIC void FSE_initCState2(FSE_CState_t *statePtr, const FSE_CTable *ct, U32 symbol)
+{
+	FSE_initCState(statePtr, ct);
+	{
+		const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform *)(statePtr->symbolTT))[symbol];
+		const U16 *stateTable = (const U16 *)(statePtr->stateTable);
+		U32 nbBitsOut = (U32)((symbolTT.deltaNbBits + (1 << 15)) >> 16);
+		statePtr->value = (nbBitsOut << 16) - symbolTT.deltaNbBits;
+		statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
+	}
+}
+
+ZSTD_STATIC void FSE_encodeSymbol(BIT_CStream_t *bitC, FSE_CState_t *statePtr, U32 symbol)
+{
+	const FSE_symbolCompressionTransform symbolTT = ((const FSE_symbolCompressionTransform *)(statePtr->symbolTT))[symbol];
+	const U16 *const stateTable = (const U16 *)(statePtr->stateTable);
+	U32 nbBitsOut = (U32)((statePtr->value + symbolTT.deltaNbBits) >> 16);
+	BIT_addBits(bitC, statePtr->value, nbBitsOut);
+	statePtr->value = stateTable[(statePtr->value >> nbBitsOut) + symbolTT.deltaFindState];
+}
+
+ZSTD_STATIC void FSE_flushCState(BIT_CStream_t *bitC, const FSE_CState_t *statePtr)
+{
+	BIT_addBits(bitC, statePtr->value, statePtr->stateLog);
+	BIT_flushBits(bitC);
+}
+
+/* ======    Decompression    ====== */
+
+typedef struct {
+	U16 tableLog;
+	U16 fastMode;
+} FSE_DTableHeader; /* sizeof U32 */
+
+typedef struct {
+	unsigned short newState;
+	unsigned char symbol;
+	unsigned char nbBits;
+} FSE_decode_t; /* size == U32 */
+
+ZSTD_STATIC void FSE_initDState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD, const FSE_DTable *dt)
+{
+	const void *ptr = dt;
+	const FSE_DTableHeader *const DTableH = (const FSE_DTableHeader *)ptr;
+	DStatePtr->state = BIT_readBits(bitD, DTableH->tableLog);
+	BIT_reloadDStream(bitD);
+	DStatePtr->table = dt + 1;
+}
+
+ZSTD_STATIC BYTE FSE_peekSymbol(const FSE_DState_t *DStatePtr)
+{
+	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
+	return DInfo.symbol;
+}
+
+ZSTD_STATIC void FSE_updateState(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
+{
+	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
+	U32 const nbBits = DInfo.nbBits;
+	size_t const lowBits = BIT_readBits(bitD, nbBits);
+	DStatePtr->state = DInfo.newState + lowBits;
+}
+
+ZSTD_STATIC BYTE FSE_decodeSymbol(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
+{
+	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
+	U32 const nbBits = DInfo.nbBits;
+	BYTE const symbol = DInfo.symbol;
+	size_t const lowBits = BIT_readBits(bitD, nbBits);
+
+	DStatePtr->state = DInfo.newState + lowBits;
+	return symbol;
+}
+
+/*! FSE_decodeSymbolFast() :
+	unsafe, only works if no symbol has a probability > 50% */
+ZSTD_STATIC BYTE FSE_decodeSymbolFast(FSE_DState_t *DStatePtr, BIT_DStream_t *bitD)
+{
+	FSE_decode_t const DInfo = ((const FSE_decode_t *)(DStatePtr->table))[DStatePtr->state];
+	U32 const nbBits = DInfo.nbBits;
+	BYTE const symbol = DInfo.symbol;
+	size_t const lowBits = BIT_readBitsFast(bitD, nbBits);
+
+	DStatePtr->state = DInfo.newState + lowBits;
+	return symbol;
+}
+
+ZSTD_STATIC unsigned FSE_endOfDState(const FSE_DState_t *DStatePtr) { return DStatePtr->state == 0; }
+
+/* **************************************************************
+*  Tuning parameters
+****************************************************************/
+/*!MEMORY_USAGE :
+*  Memory usage formula : N->2^N Bytes (examples : 10 -> 1KB; 12 -> 4KB ; 16 -> 64KB; 20 -> 1MB; etc.)
+*  Increasing memory usage improves compression ratio
+*  Reduced memory usage can improve speed, due to cache effect
+*  Recommended max value is 14, for 16KB, which nicely fits into Intel x86 L1 cache */
+#ifndef FSE_MAX_MEMORY_USAGE
+#define FSE_MAX_MEMORY_USAGE 14
+#endif
+#ifndef FSE_DEFAULT_MEMORY_USAGE
+#define FSE_DEFAULT_MEMORY_USAGE 13
+#endif
+
+/*!FSE_MAX_SYMBOL_VALUE :
+*  Maximum symbol value authorized.
+*  Required for proper stack allocation */
+#ifndef FSE_MAX_SYMBOL_VALUE
+#define FSE_MAX_SYMBOL_VALUE 255
+#endif
+
+/* **************************************************************
+*  template functions type & suffix
+****************************************************************/
+#define FSE_FUNCTION_TYPE BYTE
+#define FSE_FUNCTION_EXTENSION
+#define FSE_DECODE_TYPE FSE_decode_t
+
+/* ***************************************************************
+*  Constants
+*****************************************************************/
+#define FSE_MAX_TABLELOG (FSE_MAX_MEMORY_USAGE - 2)
+#define FSE_MAX_TABLESIZE (1U << FSE_MAX_TABLELOG)
+#define FSE_MAXTABLESIZE_MASK (FSE_MAX_TABLESIZE - 1)
+#define FSE_DEFAULT_TABLELOG (FSE_DEFAULT_MEMORY_USAGE - 2)
+#define FSE_MIN_TABLELOG 5
+
+#define FSE_TABLELOG_ABSOLUTE_MAX 15
+#if FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX
+#error "FSE_MAX_TABLELOG > FSE_TABLELOG_ABSOLUTE_MAX is not supported"
+#endif
+
+#define FSE_TABLESTEP(tableSize) ((tableSize >> 1) + (tableSize >> 3) + 3)
+
+#endif /* FSE_H */
diff --git a/lib/zstd/fse_compress.c b/lib/zstd/fse_compress.c
new file mode 100644
index 000000000000..ef3d1741d532
--- /dev/null
+++ b/lib/zstd/fse_compress.c
@@ -0,0 +1,795 @@
+/*
+ * FSE : Finite State Entropy encoder
+ * Copyright (C) 2013-2015, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ */
+
+/* **************************************************************
+*  Compiler specifics
+****************************************************************/
+#define FORCE_INLINE static __always_inline
+
+/* **************************************************************
+*  Includes
+****************************************************************/
+#include "bitstream.h"
+#include "fse.h"
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <linux/math64.h>
+#include <linux/string.h> /* memcpy, memset */
+
+/* **************************************************************
+*  Error Management
+****************************************************************/
+#define FSE_STATIC_ASSERT(c)                                   \
+	{                                                      \
+		enum { FSE_static_assert = 1 / (int)(!!(c)) }; \
+	} /* use only *after* variable declarations */
+
+/* **************************************************************
+*  Templates
+****************************************************************/
+/*
+  designed to be included
+  for type-specific functions (template emulation in C)
+  Objective is to write these functions only once, for improved maintenance
+*/
+
+/* safety checks */
+#ifndef FSE_FUNCTION_EXTENSION
+#error "FSE_FUNCTION_EXTENSION must be defined"
+#endif
+#ifndef FSE_FUNCTION_TYPE
+#error "FSE_FUNCTION_TYPE must be defined"
+#endif
+
+/* Function names */
+#define FSE_CAT(X, Y) X##Y
+#define FSE_FUNCTION_NAME(X, Y) FSE_CAT(X, Y)
+#define FSE_TYPE_NAME(X, Y) FSE_CAT(X, Y)
+
+/* Function templates */
+
+/* FSE_buildCTable_wksp() :
+ * Same as FSE_buildCTable(), but using an externally allocated scratch buffer (`workSpace`).
+ * wkspSize should be sized to handle worst case situation, which is `1<<max_tableLog * sizeof(FSE_FUNCTION_TYPE)`
+ * workSpace must also be properly aligned with FSE_FUNCTION_TYPE requirements
+ */
+size_t FSE_buildCTable_wksp(FSE_CTable *ct, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workspace, size_t workspaceSize)
+{
+	U32 const tableSize = 1 << tableLog;
+	U32 const tableMask = tableSize - 1;
+	void *const ptr = ct;
+	U16 *const tableU16 = ((U16 *)ptr) + 2;
+	void *const FSCT = ((U32 *)ptr) + 1 /* header */ + (tableLog ? tableSize >> 1 : 1);
+	FSE_symbolCompressionTransform *const symbolTT = (FSE_symbolCompressionTransform *)(FSCT);
+	U32 const step = FSE_TABLESTEP(tableSize);
+	U32 highThreshold = tableSize - 1;
+
+	U32 *cumul;
+	FSE_FUNCTION_TYPE *tableSymbol;
+	size_t spaceUsed32 = 0;
+
+	cumul = (U32 *)workspace + spaceUsed32;
+	spaceUsed32 += FSE_MAX_SYMBOL_VALUE + 2;
+	tableSymbol = (FSE_FUNCTION_TYPE *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(sizeof(FSE_FUNCTION_TYPE) * ((size_t)1 << tableLog), sizeof(U32)) >> 2;
+
+	if ((spaceUsed32 << 2) > workspaceSize)
+		return ERROR(tableLog_tooLarge);
+	workspace = (U32 *)workspace + spaceUsed32;
+	workspaceSize -= (spaceUsed32 << 2);
+
+	/* CTable header */
+	tableU16[-2] = (U16)tableLog;
+	tableU16[-1] = (U16)maxSymbolValue;
+
+	/* For explanations on how to distribute symbol values over the table :
+	*  http://fastcompression.blogspot.fr/2014/02/fse-distributing-symbol-values.html */
+
+	/* symbol start positions */
+	{
+		U32 u;
+		cumul[0] = 0;
+		for (u = 1; u <= maxSymbolValue + 1; u++) {
+			if (normalizedCounter[u - 1] == -1) { /* Low proba symbol */
+				cumul[u] = cumul[u - 1] + 1;
+				tableSymbol[highThreshold--] = (FSE_FUNCTION_TYPE)(u - 1);
+			} else {
+				cumul[u] = cumul[u - 1] + normalizedCounter[u - 1];
+			}
+		}
+		cumul[maxSymbolValue + 1] = tableSize + 1;
+	}
+
+	/* Spread symbols */
+	{
+		U32 position = 0;
+		U32 symbol;
+		for (symbol = 0; symbol <= maxSymbolValue; symbol++) {
+			int nbOccurences;
+			for (nbOccurences = 0; nbOccurences < normalizedCounter[symbol]; nbOccurences++) {
+				tableSymbol[position] = (FSE_FUNCTION_TYPE)symbol;
+				position = (position + step) & tableMask;
+				while (position > highThreshold)
+					position = (position + step) & tableMask; /* Low proba area */
+			}
+		}
+
+		if (position != 0)
+			return ERROR(GENERIC); /* Must have gone through all positions */
+	}
+
+	/* Build table */
+	{
+		U32 u;
+		for (u = 0; u < tableSize; u++) {
+			FSE_FUNCTION_TYPE s = tableSymbol[u];	/* note : static analyzer may not understand tableSymbol is properly initialized */
+			tableU16[cumul[s]++] = (U16)(tableSize + u); /* TableU16 : sorted by symbol order; gives next state value */
+		}
+	}
+
+	/* Build Symbol Transformation Table */
+	{
+		unsigned total = 0;
+		unsigned s;
+		for (s = 0; s <= maxSymbolValue; s++) {
+			switch (normalizedCounter[s]) {
+			case 0: break;
+
+			case -1:
+			case 1:
+				symbolTT[s].deltaNbBits = (tableLog << 16) - (1 << tableLog);
+				symbolTT[s].deltaFindState = total - 1;
+				total++;
+				break;
+			default: {
+				U32 const maxBitsOut = tableLog - BIT_highbit32(normalizedCounter[s] - 1);
+				U32 const minStatePlus = normalizedCounter[s] << maxBitsOut;
+				symbolTT[s].deltaNbBits = (maxBitsOut << 16) - minStatePlus;
+				symbolTT[s].deltaFindState = total - normalizedCounter[s];
+				total += normalizedCounter[s];
+			}
+			}
+		}
+	}
+
+	return 0;
+}
+
+/*-**************************************************************
+*  FSE NCount encoding-decoding
+****************************************************************/
+size_t FSE_NCountWriteBound(unsigned maxSymbolValue, unsigned tableLog)
+{
+	size_t const maxHeaderSize = (((maxSymbolValue + 1) * tableLog) >> 3) + 3;
+	return maxSymbolValue ? maxHeaderSize : FSE_NCOUNTBOUND; /* maxSymbolValue==0 ? use default */
+}
+
+static size_t FSE_writeNCount_generic(void *header, size_t headerBufferSize, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog,
+				      unsigned writeIsSafe)
+{
+	BYTE *const ostart = (BYTE *)header;
+	BYTE *out = ostart;
+	BYTE *const oend = ostart + headerBufferSize;
+	int nbBits;
+	const int tableSize = 1 << tableLog;
+	int remaining;
+	int threshold;
+	U32 bitStream;
+	int bitCount;
+	unsigned charnum = 0;
+	int previous0 = 0;
+
+	bitStream = 0;
+	bitCount = 0;
+	/* Table Size */
+	bitStream += (tableLog - FSE_MIN_TABLELOG) << bitCount;
+	bitCount += 4;
+
+	/* Init */
+	remaining = tableSize + 1; /* +1 for extra accuracy */
+	threshold = tableSize;
+	nbBits = tableLog + 1;
+
+	while (remaining > 1) { /* stops at 1 */
+		if (previous0) {
+			unsigned start = charnum;
+			while (!normalizedCounter[charnum])
+				charnum++;
+			while (charnum >= start + 24) {
+				start += 24;
+				bitStream += 0xFFFFU << bitCount;
+				if ((!writeIsSafe) && (out > oend - 2))
+					return ERROR(dstSize_tooSmall); /* Buffer overflow */
+				out[0] = (BYTE)bitStream;
+				out[1] = (BYTE)(bitStream >> 8);
+				out += 2;
+				bitStream >>= 16;
+			}
+			while (charnum >= start + 3) {
+				start += 3;
+				bitStream += 3 << bitCount;
+				bitCount += 2;
+			}
+			bitStream += (charnum - start) << bitCount;
+			bitCount += 2;
+			if (bitCount > 16) {
+				if ((!writeIsSafe) && (out > oend - 2))
+					return ERROR(dstSize_tooSmall); /* Buffer overflow */
+				out[0] = (BYTE)bitStream;
+				out[1] = (BYTE)(bitStream >> 8);
+				out += 2;
+				bitStream >>= 16;
+				bitCount -= 16;
+			}
+		}
+		{
+			int count = normalizedCounter[charnum++];
+			int const max = (2 * threshold - 1) - remaining;
+			remaining -= count < 0 ? -count : count;
+			count++; /* +1 for extra accuracy */
+			if (count >= threshold)
+				count += max; /* [0..max[ [max..threshold[ (...) [threshold+max 2*threshold[ */
+			bitStream += count << bitCount;
+			bitCount += nbBits;
+			bitCount -= (count < max);
+			previous0 = (count == 1);
+			if (remaining < 1)
+				return ERROR(GENERIC);
+			while (remaining < threshold)
+				nbBits--, threshold >>= 1;
+		}
+		if (bitCount > 16) {
+			if ((!writeIsSafe) && (out > oend - 2))
+				return ERROR(dstSize_tooSmall); /* Buffer overflow */
+			out[0] = (BYTE)bitStream;
+			out[1] = (BYTE)(bitStream >> 8);
+			out += 2;
+			bitStream >>= 16;
+			bitCount -= 16;
+		}
+	}
+
+	/* flush remaining bitStream */
+	if ((!writeIsSafe) && (out > oend - 2))
+		return ERROR(dstSize_tooSmall); /* Buffer overflow */
+	out[0] = (BYTE)bitStream;
+	out[1] = (BYTE)(bitStream >> 8);
+	out += (bitCount + 7) / 8;
+
+	if (charnum > maxSymbolValue + 1)
+		return ERROR(GENERIC);
+
+	return (out - ostart);
+}
+
+size_t FSE_writeNCount(void *buffer, size_t bufferSize, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog)
+{
+	if (tableLog > FSE_MAX_TABLELOG)
+		return ERROR(tableLog_tooLarge); /* Unsupported */
+	if (tableLog < FSE_MIN_TABLELOG)
+		return ERROR(GENERIC); /* Unsupported */
+
+	if (bufferSize < FSE_NCountWriteBound(maxSymbolValue, tableLog))
+		return FSE_writeNCount_generic(buffer, bufferSize, normalizedCounter, maxSymbolValue, tableLog, 0);
+
+	return FSE_writeNCount_generic(buffer, bufferSize, normalizedCounter, maxSymbolValue, tableLog, 1);
+}
+
+/*-**************************************************************
+*  Counting histogram
+****************************************************************/
+/*! FSE_count_simple
+	This function counts byte values within `src`, and store the histogram into table `count`.
+	It doesn't use any additional memory.
+	But this function is unsafe : it doesn't check that all values within `src` can fit into `count`.
+	For this reason, prefer using a table `count` with 256 elements.
+	@return : count of most numerous element
+*/
+size_t FSE_count_simple(unsigned *count, unsigned *maxSymbolValuePtr, const void *src, size_t srcSize)
+{
+	const BYTE *ip = (const BYTE *)src;
+	const BYTE *const end = ip + srcSize;
+	unsigned maxSymbolValue = *maxSymbolValuePtr;
+	unsigned max = 0;
+
+	memset(count, 0, (maxSymbolValue + 1) * sizeof(*count));
+	if (srcSize == 0) {
+		*maxSymbolValuePtr = 0;
+		return 0;
+	}
+
+	while (ip < end)
+		count[*ip++]++;
+
+	while (!count[maxSymbolValue])
+		maxSymbolValue--;
+	*maxSymbolValuePtr = maxSymbolValue;
+
+	{
+		U32 s;
+		for (s = 0; s <= maxSymbolValue; s++)
+			if (count[s] > max)
+				max = count[s];
+	}
+
+	return (size_t)max;
+}
+
+/* FSE_count_parallel_wksp() :
+ * Same as FSE_count_parallel(), but using an externally provided scratch buffer.
+ * `workSpace` size must be a minimum of `1024 * sizeof(unsigned)`` */
+static size_t FSE_count_parallel_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned checkMax,
+				      unsigned *const workSpace)
+{
+	const BYTE *ip = (const BYTE *)source;
+	const BYTE *const iend = ip + sourceSize;
+	unsigned maxSymbolValue = *maxSymbolValuePtr;
+	unsigned max = 0;
+	U32 *const Counting1 = workSpace;
+	U32 *const Counting2 = Counting1 + 256;
+	U32 *const Counting3 = Counting2 + 256;
+	U32 *const Counting4 = Counting3 + 256;
+
+	memset(Counting1, 0, 4 * 256 * sizeof(unsigned));
+
+	/* safety checks */
+	if (!sourceSize) {
+		memset(count, 0, maxSymbolValue + 1);
+		*maxSymbolValuePtr = 0;
+		return 0;
+	}
+	if (!maxSymbolValue)
+		maxSymbolValue = 255; /* 0 == default */
+
+	/* by stripes of 16 bytes */
+	{
+		U32 cached = ZSTD_read32(ip);
+		ip += 4;
+		while (ip < iend - 15) {
+			U32 c = cached;
+			cached = ZSTD_read32(ip);
+			ip += 4;
+			Counting1[(BYTE)c]++;
+			Counting2[(BYTE)(c >> 8)]++;
+			Counting3[(BYTE)(c >> 16)]++;
+			Counting4[c >> 24]++;
+			c = cached;
+			cached = ZSTD_read32(ip);
+			ip += 4;
+			Counting1[(BYTE)c]++;
+			Counting2[(BYTE)(c >> 8)]++;
+			Counting3[(BYTE)(c >> 16)]++;
+			Counting4[c >> 24]++;
+			c = cached;
+			cached = ZSTD_read32(ip);
+			ip += 4;
+			Counting1[(BYTE)c]++;
+			Counting2[(BYTE)(c >> 8)]++;
+			Counting3[(BYTE)(c >> 16)]++;
+			Counting4[c >> 24]++;
+			c = cached;
+			cached = ZSTD_read32(ip);
+			ip += 4;
+			Counting1[(BYTE)c]++;
+			Counting2[(BYTE)(c >> 8)]++;
+			Counting3[(BYTE)(c >> 16)]++;
+			Counting4[c >> 24]++;
+		}
+		ip -= 4;
+	}
+
+	/* finish last symbols */
+	while (ip < iend)
+		Counting1[*ip++]++;
+
+	if (checkMax) { /* verify stats will fit into destination table */
+		U32 s;
+		for (s = 255; s > maxSymbolValue; s--) {
+			Counting1[s] += Counting2[s] + Counting3[s] + Counting4[s];
+			if (Counting1[s])
+				return ERROR(maxSymbolValue_tooSmall);
+		}
+	}
+
+	{
+		U32 s;
+		for (s = 0; s <= maxSymbolValue; s++) {
+			count[s] = Counting1[s] + Counting2[s] + Counting3[s] + Counting4[s];
+			if (count[s] > max)
+				max = count[s];
+		}
+	}
+
+	while (!count[maxSymbolValue])
+		maxSymbolValue--;
+	*maxSymbolValuePtr = maxSymbolValue;
+	return (size_t)max;
+}
+
+/* FSE_countFast_wksp() :
+ * Same as FSE_countFast(), but using an externally provided scratch buffer.
+ * `workSpace` size must be table of >= `1024` unsigned */
+size_t FSE_countFast_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned *workSpace)
+{
+	if (sourceSize < 1500)
+		return FSE_count_simple(count, maxSymbolValuePtr, source, sourceSize);
+	return FSE_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, 0, workSpace);
+}
+
+/* FSE_count_wksp() :
+ * Same as FSE_count(), but using an externally provided scratch buffer.
+ * `workSpace` size must be table of >= `1024` unsigned */
+size_t FSE_count_wksp(unsigned *count, unsigned *maxSymbolValuePtr, const void *source, size_t sourceSize, unsigned *workSpace)
+{
+	if (*maxSymbolValuePtr < 255)
+		return FSE_count_parallel_wksp(count, maxSymbolValuePtr, source, sourceSize, 1, workSpace);
+	*maxSymbolValuePtr = 255;
+	return FSE_countFast_wksp(count, maxSymbolValuePtr, source, sourceSize, workSpace);
+}
+
+/*-**************************************************************
+*  FSE Compression Code
+****************************************************************/
+/*! FSE_sizeof_CTable() :
+	FSE_CTable is a variable size structure which contains :
+	`U16 tableLog;`
+	`U16 maxSymbolValue;`
+	`U16 nextStateNumber[1 << tableLog];`                         // This size is variable
+	`FSE_symbolCompressionTransform symbolTT[maxSymbolValue+1];`  // This size is variable
+Allocation is manual (C standard does not support variable-size structures).
+*/
+size_t FSE_sizeof_CTable(unsigned maxSymbolValue, unsigned tableLog)
+{
+	if (tableLog > FSE_MAX_TABLELOG)
+		return ERROR(tableLog_tooLarge);
+	return FSE_CTABLE_SIZE_U32(tableLog, maxSymbolValue) * sizeof(U32);
+}
+
+/* provides the minimum logSize to safely represent a distribution */
+static unsigned FSE_minTableLog(size_t srcSize, unsigned maxSymbolValue)
+{
+	U32 minBitsSrc = BIT_highbit32((U32)(srcSize - 1)) + 1;
+	U32 minBitsSymbols = BIT_highbit32(maxSymbolValue) + 2;
+	U32 minBits = minBitsSrc < minBitsSymbols ? minBitsSrc : minBitsSymbols;
+	return minBits;
+}
+
+unsigned FSE_optimalTableLog_internal(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue, unsigned minus)
+{
+	U32 maxBitsSrc = BIT_highbit32((U32)(srcSize - 1)) - minus;
+	U32 tableLog = maxTableLog;
+	U32 minBits = FSE_minTableLog(srcSize, maxSymbolValue);
+	if (tableLog == 0)
+		tableLog = FSE_DEFAULT_TABLELOG;
+	if (maxBitsSrc < tableLog)
+		tableLog = maxBitsSrc; /* Accuracy can be reduced */
+	if (minBits > tableLog)
+		tableLog = minBits; /* Need a minimum to safely represent all symbol values */
+	if (tableLog < FSE_MIN_TABLELOG)
+		tableLog = FSE_MIN_TABLELOG;
+	if (tableLog > FSE_MAX_TABLELOG)
+		tableLog = FSE_MAX_TABLELOG;
+	return tableLog;
+}
+
+unsigned FSE_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
+{
+	return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 2);
+}
+
+/* Secondary normalization method.
+   To be used when primary method fails. */
+
+static size_t FSE_normalizeM2(short *norm, U32 tableLog, const unsigned *count, size_t total, U32 maxSymbolValue)
+{
+	short const NOT_YET_ASSIGNED = -2;
+	U32 s;
+	U32 distributed = 0;
+	U32 ToDistribute;
+
+	/* Init */
+	U32 const lowThreshold = (U32)(total >> tableLog);
+	U32 lowOne = (U32)((total * 3) >> (tableLog + 1));
+
+	for (s = 0; s <= maxSymbolValue; s++) {
+		if (count[s] == 0) {
+			norm[s] = 0;
+			continue;
+		}
+		if (count[s] <= lowThreshold) {
+			norm[s] = -1;
+			distributed++;
+			total -= count[s];
+			continue;
+		}
+		if (count[s] <= lowOne) {
+			norm[s] = 1;
+			distributed++;
+			total -= count[s];
+			continue;
+		}
+
+		norm[s] = NOT_YET_ASSIGNED;
+	}
+	ToDistribute = (1 << tableLog) - distributed;
+
+	if ((total / ToDistribute) > lowOne) {
+		/* risk of rounding to zero */
+		lowOne = (U32)((total * 3) / (ToDistribute * 2));
+		for (s = 0; s <= maxSymbolValue; s++) {
+			if ((norm[s] == NOT_YET_ASSIGNED) && (count[s] <= lowOne)) {
+				norm[s] = 1;
+				distributed++;
+				total -= count[s];
+				continue;
+			}
+		}
+		ToDistribute = (1 << tableLog) - distributed;
+	}
+
+	if (distributed == maxSymbolValue + 1) {
+		/* all values are pretty poor;
+		   probably incompressible data (should have already been detected);
+		   find max, then give all remaining points to max */
+		U32 maxV = 0, maxC = 0;
+		for (s = 0; s <= maxSymbolValue; s++)
+			if (count[s] > maxC)
+				maxV = s, maxC = count[s];
+		norm[maxV] += (short)ToDistribute;
+		return 0;
+	}
+
+	if (total == 0) {
+		/* all of the symbols were low enough for the lowOne or lowThreshold */
+		for (s = 0; ToDistribute > 0; s = (s + 1) % (maxSymbolValue + 1))
+			if (norm[s] > 0)
+				ToDistribute--, norm[s]++;
+		return 0;
+	}
+
+	{
+		U64 const vStepLog = 62 - tableLog;
+		U64 const mid = (1ULL << (vStepLog - 1)) - 1;
+		U64 const rStep = div_u64((((U64)1 << vStepLog) * ToDistribute) + mid, (U32)total); /* scale on remaining */
+		U64 tmpTotal = mid;
+		for (s = 0; s <= maxSymbolValue; s++) {
+			if (norm[s] == NOT_YET_ASSIGNED) {
+				U64 const end = tmpTotal + (count[s] * rStep);
+				U32 const sStart = (U32)(tmpTotal >> vStepLog);
+				U32 const sEnd = (U32)(end >> vStepLog);
+				U32 const weight = sEnd - sStart;
+				if (weight < 1)
+					return ERROR(GENERIC);
+				norm[s] = (short)weight;
+				tmpTotal = end;
+			}
+		}
+	}
+
+	return 0;
+}
+
+size_t FSE_normalizeCount(short *normalizedCounter, unsigned tableLog, const unsigned *count, size_t total, unsigned maxSymbolValue)
+{
+	/* Sanity checks */
+	if (tableLog == 0)
+		tableLog = FSE_DEFAULT_TABLELOG;
+	if (tableLog < FSE_MIN_TABLELOG)
+		return ERROR(GENERIC); /* Unsupported size */
+	if (tableLog > FSE_MAX_TABLELOG)
+		return ERROR(tableLog_tooLarge); /* Unsupported size */
+	if (tableLog < FSE_minTableLog(total, maxSymbolValue))
+		return ERROR(GENERIC); /* Too small tableLog, compression potentially impossible */
+
+	{
+		U32 const rtbTable[] = {0, 473195, 504333, 520860, 550000, 700000, 750000, 830000};
+		U64 const scale = 62 - tableLog;
+		U64 const step = div_u64((U64)1 << 62, (U32)total); /* <== here, one division ! */
+		U64 const vStep = 1ULL << (scale - 20);
+		int stillToDistribute = 1 << tableLog;
+		unsigned s;
+		unsigned largest = 0;
+		short largestP = 0;
+		U32 lowThreshold = (U32)(total >> tableLog);
+
+		for (s = 0; s <= maxSymbolValue; s++) {
+			if (count[s] == total)
+				return 0; /* rle special case */
+			if (count[s] == 0) {
+				normalizedCounter[s] = 0;
+				continue;
+			}
+			if (count[s] <= lowThreshold) {
+				normalizedCounter[s] = -1;
+				stillToDistribute--;
+			} else {
+				short proba = (short)((count[s] * step) >> scale);
+				if (proba < 8) {
+					U64 restToBeat = vStep * rtbTable[proba];
+					proba += (count[s] * step) - ((U64)proba << scale) > restToBeat;
+				}
+				if (proba > largestP)
+					largestP = proba, largest = s;
+				normalizedCounter[s] = proba;
+				stillToDistribute -= proba;
+			}
+		}
+		if (-stillToDistribute >= (normalizedCounter[largest] >> 1)) {
+			/* corner case, need another normalization method */
+			size_t const errorCode = FSE_normalizeM2(normalizedCounter, tableLog, count, total, maxSymbolValue);
+			if (FSE_isError(errorCode))
+				return errorCode;
+		} else
+			normalizedCounter[largest] += (short)stillToDistribute;
+	}
+
+	return tableLog;
+}
+
+/* fake FSE_CTable, for raw (uncompressed) input */
+size_t FSE_buildCTable_raw(FSE_CTable *ct, unsigned nbBits)
+{
+	const unsigned tableSize = 1 << nbBits;
+	const unsigned tableMask = tableSize - 1;
+	const unsigned maxSymbolValue = tableMask;
+	void *const ptr = ct;
+	U16 *const tableU16 = ((U16 *)ptr) + 2;
+	void *const FSCT = ((U32 *)ptr) + 1 /* header */ + (tableSize >> 1); /* assumption : tableLog >= 1 */
+	FSE_symbolCompressionTransform *const symbolTT = (FSE_symbolCompressionTransform *)(FSCT);
+	unsigned s;
+
+	/* Sanity checks */
+	if (nbBits < 1)
+		return ERROR(GENERIC); /* min size */
+
+	/* header */
+	tableU16[-2] = (U16)nbBits;
+	tableU16[-1] = (U16)maxSymbolValue;
+
+	/* Build table */
+	for (s = 0; s < tableSize; s++)
+		tableU16[s] = (U16)(tableSize + s);
+
+	/* Build Symbol Transformation Table */
+	{
+		const U32 deltaNbBits = (nbBits << 16) - (1 << nbBits);
+		for (s = 0; s <= maxSymbolValue; s++) {
+			symbolTT[s].deltaNbBits = deltaNbBits;
+			symbolTT[s].deltaFindState = s - 1;
+		}
+	}
+
+	return 0;
+}
+
+/* fake FSE_CTable, for rle input (always same symbol) */
+size_t FSE_buildCTable_rle(FSE_CTable *ct, BYTE symbolValue)
+{
+	void *ptr = ct;
+	U16 *tableU16 = ((U16 *)ptr) + 2;
+	void *FSCTptr = (U32 *)ptr + 2;
+	FSE_symbolCompressionTransform *symbolTT = (FSE_symbolCompressionTransform *)FSCTptr;
+
+	/* header */
+	tableU16[-2] = (U16)0;
+	tableU16[-1] = (U16)symbolValue;
+
+	/* Build table */
+	tableU16[0] = 0;
+	tableU16[1] = 0; /* just in case */
+
+	/* Build Symbol Transformation Table */
+	symbolTT[symbolValue].deltaNbBits = 0;
+	symbolTT[symbolValue].deltaFindState = 0;
+
+	return 0;
+}
+
+static size_t FSE_compress_usingCTable_generic(void *dst, size_t dstSize, const void *src, size_t srcSize, const FSE_CTable *ct, const unsigned fast)
+{
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *ip = iend;
+
+	BIT_CStream_t bitC;
+	FSE_CState_t CState1, CState2;
+
+	/* init */
+	if (srcSize <= 2)
+		return 0;
+	{
+		size_t const initError = BIT_initCStream(&bitC, dst, dstSize);
+		if (FSE_isError(initError))
+			return 0; /* not enough space available to write a bitstream */
+	}
+
+#define FSE_FLUSHBITS(s) (fast ? BIT_flushBitsFast(s) : BIT_flushBits(s))
+
+	if (srcSize & 1) {
+		FSE_initCState2(&CState1, ct, *--ip);
+		FSE_initCState2(&CState2, ct, *--ip);
+		FSE_encodeSymbol(&bitC, &CState1, *--ip);
+		FSE_FLUSHBITS(&bitC);
+	} else {
+		FSE_initCState2(&CState2, ct, *--ip);
+		FSE_initCState2(&CState1, ct, *--ip);
+	}
+
+	/* join to mod 4 */
+	srcSize -= 2;
+	if ((sizeof(bitC.bitContainer) * 8 > FSE_MAX_TABLELOG * 4 + 7) && (srcSize & 2)) { /* test bit 2 */
+		FSE_encodeSymbol(&bitC, &CState2, *--ip);
+		FSE_encodeSymbol(&bitC, &CState1, *--ip);
+		FSE_FLUSHBITS(&bitC);
+	}
+
+	/* 2 or 4 encoding per loop */
+	while (ip > istart) {
+
+		FSE_encodeSymbol(&bitC, &CState2, *--ip);
+
+		if (sizeof(bitC.bitContainer) * 8 < FSE_MAX_TABLELOG * 2 + 7) /* this test must be static */
+			FSE_FLUSHBITS(&bitC);
+
+		FSE_encodeSymbol(&bitC, &CState1, *--ip);
+
+		if (sizeof(bitC.bitContainer) * 8 > FSE_MAX_TABLELOG * 4 + 7) { /* this test must be static */
+			FSE_encodeSymbol(&bitC, &CState2, *--ip);
+			FSE_encodeSymbol(&bitC, &CState1, *--ip);
+		}
+
+		FSE_FLUSHBITS(&bitC);
+	}
+
+	FSE_flushCState(&bitC, &CState2);
+	FSE_flushCState(&bitC, &CState1);
+	return BIT_closeCStream(&bitC);
+}
+
+size_t FSE_compress_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const FSE_CTable *ct)
+{
+	unsigned const fast = (dstSize >= FSE_BLOCKBOUND(srcSize));
+
+	if (fast)
+		return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 1);
+	else
+		return FSE_compress_usingCTable_generic(dst, dstSize, src, srcSize, ct, 0);
+}
+
+size_t FSE_compressBound(size_t size) { return FSE_COMPRESSBOUND(size); }
diff --git a/lib/zstd/fse_decompress.c b/lib/zstd/fse_decompress.c
new file mode 100644
index 000000000000..a84300e5a013
--- /dev/null
+++ b/lib/zstd/fse_decompress.c
@@ -0,0 +1,332 @@
+/*
+ * FSE : Finite State Entropy decoder
+ * Copyright (C) 2013-2015, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ */
+
+/* **************************************************************
+*  Compiler specifics
+****************************************************************/
+#define FORCE_INLINE static __always_inline
+
+/* **************************************************************
+*  Includes
+****************************************************************/
+#include "bitstream.h"
+#include "fse.h"
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <linux/string.h> /* memcpy, memset */
+
+/* **************************************************************
+*  Error Management
+****************************************************************/
+#define FSE_isError ERR_isError
+#define FSE_STATIC_ASSERT(c)                                   \
+	{                                                      \
+		enum { FSE_static_assert = 1 / (int)(!!(c)) }; \
+	} /* use only *after* variable declarations */
+
+/* check and forward error code */
+#define CHECK_F(f)                  \
+	{                           \
+		size_t const e = f; \
+		if (FSE_isError(e)) \
+			return e;   \
+	}
+
+/* **************************************************************
+*  Templates
+****************************************************************/
+/*
+  designed to be included
+  for type-specific functions (template emulation in C)
+  Objective is to write these functions only once, for improved maintenance
+*/
+
+/* safety checks */
+#ifndef FSE_FUNCTION_EXTENSION
+#error "FSE_FUNCTION_EXTENSION must be defined"
+#endif
+#ifndef FSE_FUNCTION_TYPE
+#error "FSE_FUNCTION_TYPE must be defined"
+#endif
+
+/* Function names */
+#define FSE_CAT(X, Y) X##Y
+#define FSE_FUNCTION_NAME(X, Y) FSE_CAT(X, Y)
+#define FSE_TYPE_NAME(X, Y) FSE_CAT(X, Y)
+
+/* Function templates */
+
+size_t FSE_buildDTable_wksp(FSE_DTable *dt, const short *normalizedCounter, unsigned maxSymbolValue, unsigned tableLog, void *workspace, size_t workspaceSize)
+{
+	void *const tdPtr = dt + 1; /* because *dt is unsigned, 32-bits aligned on 32-bits */
+	FSE_DECODE_TYPE *const tableDecode = (FSE_DECODE_TYPE *)(tdPtr);
+	U16 *symbolNext = (U16 *)workspace;
+
+	U32 const maxSV1 = maxSymbolValue + 1;
+	U32 const tableSize = 1 << tableLog;
+	U32 highThreshold = tableSize - 1;
+
+	/* Sanity Checks */
+	if (workspaceSize < sizeof(U16) * (FSE_MAX_SYMBOL_VALUE + 1))
+		return ERROR(tableLog_tooLarge);
+	if (maxSymbolValue > FSE_MAX_SYMBOL_VALUE)
+		return ERROR(maxSymbolValue_tooLarge);
+	if (tableLog > FSE_MAX_TABLELOG)
+		return ERROR(tableLog_tooLarge);
+
+	/* Init, lay down lowprob symbols */
+	{
+		FSE_DTableHeader DTableH;
+		DTableH.tableLog = (U16)tableLog;
+		DTableH.fastMode = 1;
+		{
+			S16 const largeLimit = (S16)(1 << (tableLog - 1));
+			U32 s;
+			for (s = 0; s < maxSV1; s++) {
+				if (normalizedCounter[s] == -1) {
+					tableDecode[highThreshold--].symbol = (FSE_FUNCTION_TYPE)s;
+					symbolNext[s] = 1;
+				} else {
+					if (normalizedCounter[s] >= largeLimit)
+						DTableH.fastMode = 0;
+					symbolNext[s] = normalizedCounter[s];
+				}
+			}
+		}
+		memcpy(dt, &DTableH, sizeof(DTableH));
+	}
+
+	/* Spread symbols */
+	{
+		U32 const tableMask = tableSize - 1;
+		U32 const step = FSE_TABLESTEP(tableSize);
+		U32 s, position = 0;
+		for (s = 0; s < maxSV1; s++) {
+			int i;
+			for (i = 0; i < normalizedCounter[s]; i++) {
+				tableDecode[position].symbol = (FSE_FUNCTION_TYPE)s;
+				position = (position + step) & tableMask;
+				while (position > highThreshold)
+					position = (position + step) & tableMask; /* lowprob area */
+			}
+		}
+		if (position != 0)
+			return ERROR(GENERIC); /* position must reach all cells once, otherwise normalizedCounter is incorrect */
+	}
+
+	/* Build Decoding table */
+	{
+		U32 u;
+		for (u = 0; u < tableSize; u++) {
+			FSE_FUNCTION_TYPE const symbol = (FSE_FUNCTION_TYPE)(tableDecode[u].symbol);
+			U16 nextState = symbolNext[symbol]++;
+			tableDecode[u].nbBits = (BYTE)(tableLog - BIT_highbit32((U32)nextState));
+			tableDecode[u].newState = (U16)((nextState << tableDecode[u].nbBits) - tableSize);
+		}
+	}
+
+	return 0;
+}
+
+/*-*******************************************************
+*  Decompression (Byte symbols)
+*********************************************************/
+size_t FSE_buildDTable_rle(FSE_DTable *dt, BYTE symbolValue)
+{
+	void *ptr = dt;
+	FSE_DTableHeader *const DTableH = (FSE_DTableHeader *)ptr;
+	void *dPtr = dt + 1;
+	FSE_decode_t *const cell = (FSE_decode_t *)dPtr;
+
+	DTableH->tableLog = 0;
+	DTableH->fastMode = 0;
+
+	cell->newState = 0;
+	cell->symbol = symbolValue;
+	cell->nbBits = 0;
+
+	return 0;
+}
+
+size_t FSE_buildDTable_raw(FSE_DTable *dt, unsigned nbBits)
+{
+	void *ptr = dt;
+	FSE_DTableHeader *const DTableH = (FSE_DTableHeader *)ptr;
+	void *dPtr = dt + 1;
+	FSE_decode_t *const dinfo = (FSE_decode_t *)dPtr;
+	const unsigned tableSize = 1 << nbBits;
+	const unsigned tableMask = tableSize - 1;
+	const unsigned maxSV1 = tableMask + 1;
+	unsigned s;
+
+	/* Sanity checks */
+	if (nbBits < 1)
+		return ERROR(GENERIC); /* min size */
+
+	/* Build Decoding Table */
+	DTableH->tableLog = (U16)nbBits;
+	DTableH->fastMode = 1;
+	for (s = 0; s < maxSV1; s++) {
+		dinfo[s].newState = 0;
+		dinfo[s].symbol = (BYTE)s;
+		dinfo[s].nbBits = (BYTE)nbBits;
+	}
+
+	return 0;
+}
+
+FORCE_INLINE size_t FSE_decompress_usingDTable_generic(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt,
+						       const unsigned fast)
+{
+	BYTE *const ostart = (BYTE *)dst;
+	BYTE *op = ostart;
+	BYTE *const omax = op + maxDstSize;
+	BYTE *const olimit = omax - 3;
+
+	BIT_DStream_t bitD;
+	FSE_DState_t state1;
+	FSE_DState_t state2;
+
+	/* Init */
+	CHECK_F(BIT_initDStream(&bitD, cSrc, cSrcSize));
+
+	FSE_initDState(&state1, &bitD, dt);
+	FSE_initDState(&state2, &bitD, dt);
+
+#define FSE_GETSYMBOL(statePtr) fast ? FSE_decodeSymbolFast(statePtr, &bitD) : FSE_decodeSymbol(statePtr, &bitD)
+
+	/* 4 symbols per loop */
+	for (; (BIT_reloadDStream(&bitD) == BIT_DStream_unfinished) & (op < olimit); op += 4) {
+		op[0] = FSE_GETSYMBOL(&state1);
+
+		if (FSE_MAX_TABLELOG * 2 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
+			BIT_reloadDStream(&bitD);
+
+		op[1] = FSE_GETSYMBOL(&state2);
+
+		if (FSE_MAX_TABLELOG * 4 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
+		{
+			if (BIT_reloadDStream(&bitD) > BIT_DStream_unfinished) {
+				op += 2;
+				break;
+			}
+		}
+
+		op[2] = FSE_GETSYMBOL(&state1);
+
+		if (FSE_MAX_TABLELOG * 2 + 7 > sizeof(bitD.bitContainer) * 8) /* This test must be static */
+			BIT_reloadDStream(&bitD);
+
+		op[3] = FSE_GETSYMBOL(&state2);
+	}
+
+	/* tail */
+	/* note : BIT_reloadDStream(&bitD) >= FSE_DStream_partiallyFilled; Ends at exactly BIT_DStream_completed */
+	while (1) {
+		if (op > (omax - 2))
+			return ERROR(dstSize_tooSmall);
+		*op++ = FSE_GETSYMBOL(&state1);
+		if (BIT_reloadDStream(&bitD) == BIT_DStream_overflow) {
+			*op++ = FSE_GETSYMBOL(&state2);
+			break;
+		}
+
+		if (op > (omax - 2))
+			return ERROR(dstSize_tooSmall);
+		*op++ = FSE_GETSYMBOL(&state2);
+		if (BIT_reloadDStream(&bitD) == BIT_DStream_overflow) {
+			*op++ = FSE_GETSYMBOL(&state1);
+			break;
+		}
+	}
+
+	return op - ostart;
+}
+
+size_t FSE_decompress_usingDTable(void *dst, size_t originalSize, const void *cSrc, size_t cSrcSize, const FSE_DTable *dt)
+{
+	const void *ptr = dt;
+	const FSE_DTableHeader *DTableH = (const FSE_DTableHeader *)ptr;
+	const U32 fastMode = DTableH->fastMode;
+
+	/* select fast mode (static) */
+	if (fastMode)
+		return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 1);
+	return FSE_decompress_usingDTable_generic(dst, originalSize, cSrc, cSrcSize, dt, 0);
+}
+
+size_t FSE_decompress_wksp(void *dst, size_t dstCapacity, const void *cSrc, size_t cSrcSize, unsigned maxLog, void *workspace, size_t workspaceSize)
+{
+	const BYTE *const istart = (const BYTE *)cSrc;
+	const BYTE *ip = istart;
+	unsigned tableLog;
+	unsigned maxSymbolValue = FSE_MAX_SYMBOL_VALUE;
+	size_t NCountLength;
+
+	FSE_DTable *dt;
+	short *counting;
+	size_t spaceUsed32 = 0;
+
+	FSE_STATIC_ASSERT(sizeof(FSE_DTable) == sizeof(U32));
+
+	dt = (FSE_DTable *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += FSE_DTABLE_SIZE_U32(maxLog);
+	counting = (short *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(sizeof(short) * (FSE_MAX_SYMBOL_VALUE + 1), sizeof(U32)) >> 2;
+
+	if ((spaceUsed32 << 2) > workspaceSize)
+		return ERROR(tableLog_tooLarge);
+	workspace = (U32 *)workspace + spaceUsed32;
+	workspaceSize -= (spaceUsed32 << 2);
+
+	/* normal FSE decoding mode */
+	NCountLength = FSE_readNCount(counting, &maxSymbolValue, &tableLog, istart, cSrcSize);
+	if (FSE_isError(NCountLength))
+		return NCountLength;
+	// if (NCountLength >= cSrcSize) return ERROR(srcSize_wrong);   /* too small input size; supposed to be already checked in NCountLength, only remaining
+	// case : NCountLength==cSrcSize */
+	if (tableLog > maxLog)
+		return ERROR(tableLog_tooLarge);
+	ip += NCountLength;
+	cSrcSize -= NCountLength;
+
+	CHECK_F(FSE_buildDTable_wksp(dt, counting, maxSymbolValue, tableLog, workspace, workspaceSize));
+
+	return FSE_decompress_usingDTable(dst, dstCapacity, ip, cSrcSize, dt); /* always return, even if it is an error code */
+}
diff --git a/lib/zstd/huf.h b/lib/zstd/huf.h
new file mode 100644
index 000000000000..2143da28d952
--- /dev/null
+++ b/lib/zstd/huf.h
@@ -0,0 +1,212 @@
+/*
+ * Huffman coder, part of New Generation Entropy library
+ * header file
+ * Copyright (C) 2013-2016, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ */
+#ifndef HUF_H_298734234
+#define HUF_H_298734234
+
+/* *** Dependencies *** */
+#include <linux/types.h> /* size_t */
+
+/* ***   Tool functions *** */
+#define HUF_BLOCKSIZE_MAX (128 * 1024) /**< maximum input size for a single block compressed with HUF_compress */
+size_t HUF_compressBound(size_t size); /**< maximum compressed size (worst case) */
+
+/* Error Management */
+unsigned HUF_isError(size_t code); /**< tells if a return value is an error code */
+
+/* ***   Advanced function   *** */
+
+/** HUF_compress4X_wksp() :
+*   Same as HUF_compress2(), but uses externally allocated `workSpace`, which must be a table of >= 1024 unsigned */
+size_t HUF_compress4X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
+			   size_t wkspSize); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
+
+/* *** Dependencies *** */
+#include "mem.h" /* U32 */
+
+/* *** Constants *** */
+#define HUF_TABLELOG_MAX 12     /* max configured tableLog (for static allocation); can be modified up to HUF_ABSOLUTEMAX_TABLELOG */
+#define HUF_TABLELOG_DEFAULT 11 /* tableLog by default, when not specified */
+#define HUF_SYMBOLVALUE_MAX 255
+
+#define HUF_TABLELOG_ABSOLUTEMAX 15 /* absolute limit of HUF_MAX_TABLELOG. Beyond that value, code does not work */
+#if (HUF_TABLELOG_MAX > HUF_TABLELOG_ABSOLUTEMAX)
+#error "HUF_TABLELOG_MAX is too large !"
+#endif
+
+/* ****************************************
+*  Static allocation
+******************************************/
+/* HUF buffer bounds */
+#define HUF_CTABLEBOUND 129
+#define HUF_BLOCKBOUND(size) (size + (size >> 8) + 8)			 /* only true if incompressible pre-filtered with fast heuristic */
+#define HUF_COMPRESSBOUND(size) (HUF_CTABLEBOUND + HUF_BLOCKBOUND(size)) /* Macro version, useful for static allocation */
+
+/* static allocation of HUF's Compression Table */
+#define HUF_CREATE_STATIC_CTABLE(name, maxSymbolValue) \
+	U32 name##hb[maxSymbolValue + 1];              \
+	void *name##hv = &(name##hb);                  \
+	HUF_CElt *name = (HUF_CElt *)(name##hv) /* no final ; */
+
+/* static allocation of HUF's DTable */
+typedef U32 HUF_DTable;
+#define HUF_DTABLE_SIZE(maxTableLog) (1 + (1 << (maxTableLog)))
+#define HUF_CREATE_STATIC_DTABLEX2(DTable, maxTableLog) HUF_DTable DTable[HUF_DTABLE_SIZE((maxTableLog)-1)] = {((U32)((maxTableLog)-1) * 0x01000001)}
+#define HUF_CREATE_STATIC_DTABLEX4(DTable, maxTableLog) HUF_DTable DTable[HUF_DTABLE_SIZE(maxTableLog)] = {((U32)(maxTableLog)*0x01000001)}
+
+/* The workspace must have alignment at least 4 and be at least this large */
+#define HUF_COMPRESS_WORKSPACE_SIZE (6 << 10)
+#define HUF_COMPRESS_WORKSPACE_SIZE_U32 (HUF_COMPRESS_WORKSPACE_SIZE / sizeof(U32))
+
+/* The workspace must have alignment at least 4 and be at least this large */
+#define HUF_DECOMPRESS_WORKSPACE_SIZE (3 << 10)
+#define HUF_DECOMPRESS_WORKSPACE_SIZE_U32 (HUF_DECOMPRESS_WORKSPACE_SIZE / sizeof(U32))
+
+/* ****************************************
+*  Advanced decompression functions
+******************************************/
+size_t HUF_decompress4X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize); /**< decodes RLE and uncompressed */
+size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
+				size_t workspaceSize);							       /**< considers RLE and uncompressed as errors */
+size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
+				   size_t workspaceSize); /**< single-symbol decoder */
+size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
+				   size_t workspaceSize); /**< double-symbols decoder */
+
+/* ****************************************
+*  HUF detailed API
+******************************************/
+/*!
+HUF_compress() does the following:
+1. count symbol occurrence from source[] into table count[] using FSE_count()
+2. (optional) refine tableLog using HUF_optimalTableLog()
+3. build Huffman table from count using HUF_buildCTable()
+4. save Huffman table to memory buffer using HUF_writeCTable_wksp()
+5. encode the data stream using HUF_compress4X_usingCTable()
+
+The following API allows targeting specific sub-functions for advanced tasks.
+For example, it's possible to compress several blocks using the same 'CTable',
+or to save and regenerate 'CTable' using external methods.
+*/
+/* FSE_count() : find it within "fse.h" */
+unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue);
+typedef struct HUF_CElt_s HUF_CElt; /* incomplete type */
+size_t HUF_writeCTable_wksp(void *dst, size_t maxDstSize, const HUF_CElt *CTable, unsigned maxSymbolValue, unsigned huffLog, void *workspace, size_t workspaceSize);
+size_t HUF_compress4X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable);
+
+typedef enum {
+	HUF_repeat_none,  /**< Cannot use the previous table */
+	HUF_repeat_check, /**< Can use the previous table but it must be checked. Note : The previous table must have been constructed by HUF_compress{1,
+			     4}X_repeat */
+	HUF_repeat_valid  /**< Can use the previous table and it is asumed to be valid */
+} HUF_repeat;
+/** HUF_compress4X_repeat() :
+*   Same as HUF_compress4X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
+*   If it uses hufTable it does not modify hufTable or repeat.
+*   If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
+*   If preferRepeat then the old table will always be used if valid. */
+size_t HUF_compress4X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
+			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat,
+			     int preferRepeat); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
+
+/** HUF_buildCTable_wksp() :
+ *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
+ *  `workSpace` must be aligned on 4-bytes boundaries, and be at least as large as a table of 1024 unsigned.
+ */
+size_t HUF_buildCTable_wksp(HUF_CElt *tree, const U32 *count, U32 maxSymbolValue, U32 maxNbBits, void *workSpace, size_t wkspSize);
+
+/*! HUF_readStats() :
+	Read compact Huffman tree, saved by HUF_writeCTable().
+	`huffWeight` is destination buffer.
+	@return : size read from `src` , or an error Code .
+	Note : Needed by HUF_readCTable() and HUF_readDTableXn() . */
+size_t HUF_readStats_wksp(BYTE *huffWeight, size_t hwSize, U32 *rankStats, U32 *nbSymbolsPtr, U32 *tableLogPtr, const void *src, size_t srcSize,
+			  void *workspace, size_t workspaceSize);
+
+/** HUF_readCTable() :
+*   Loading a CTable saved with HUF_writeCTable() */
+size_t HUF_readCTable_wksp(HUF_CElt *CTable, unsigned maxSymbolValue, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
+
+/*
+HUF_decompress() does the following:
+1. select the decompression algorithm (X2, X4) based on pre-computed heuristics
+2. build Huffman table from save, using HUF_readDTableXn()
+3. decode 1 or 4 segments in parallel using HUF_decompressSXn_usingDTable
+*/
+
+/** HUF_selectDecoder() :
+*   Tells which decoder is likely to decode faster,
+*   based on a set of pre-determined metrics.
+*   @return : 0==HUF_decompress4X2, 1==HUF_decompress4X4 .
+*   Assumption : 0 < cSrcSize < dstSize <= 128 KB */
+U32 HUF_selectDecoder(size_t dstSize, size_t cSrcSize);
+
+size_t HUF_readDTableX2_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
+size_t HUF_readDTableX4_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize);
+
+size_t HUF_decompress4X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
+size_t HUF_decompress4X2_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
+size_t HUF_decompress4X4_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
+
+/* single stream variants */
+
+size_t HUF_compress1X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
+			   size_t wkspSize); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
+size_t HUF_compress1X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable);
+/** HUF_compress1X_repeat() :
+*   Same as HUF_compress1X_wksp(), but considers using hufTable if *repeat != HUF_repeat_none.
+*   If it uses hufTable it does not modify hufTable or repeat.
+*   If it doesn't, it sets *repeat = HUF_repeat_none, and it sets hufTable to the table used.
+*   If preferRepeat then the old table will always be used if valid. */
+size_t HUF_compress1X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned tableLog, void *workSpace,
+			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat,
+			     int preferRepeat); /**< `workSpace` must be a table of at least HUF_COMPRESS_WORKSPACE_SIZE_U32 unsigned */
+
+size_t HUF_decompress1X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize);
+size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
+				   size_t workspaceSize); /**< single-symbol decoder */
+size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace,
+				   size_t workspaceSize); /**< double-symbols decoder */
+
+size_t HUF_decompress1X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize,
+				    const HUF_DTable *DTable); /**< automatic selection of sing or double symbol decoder, based on DTable */
+size_t HUF_decompress1X2_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
+size_t HUF_decompress1X4_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable);
+
+#endif /* HUF_H_298734234 */
diff --git a/lib/zstd/huf_compress.c b/lib/zstd/huf_compress.c
new file mode 100644
index 000000000000..40055a7016e6
--- /dev/null
+++ b/lib/zstd/huf_compress.c
@@ -0,0 +1,770 @@
+/*
+ * Huffman encoder, part of New Generation Entropy library
+ * Copyright (C) 2013-2016, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ */
+
+/* **************************************************************
+*  Includes
+****************************************************************/
+#include "bitstream.h"
+#include "fse.h" /* header compression */
+#include "huf.h"
+#include <linux/kernel.h>
+#include <linux/string.h> /* memcpy, memset */
+
+/* **************************************************************
+*  Error Management
+****************************************************************/
+#define HUF_STATIC_ASSERT(c)                                   \
+	{                                                      \
+		enum { HUF_static_assert = 1 / (int)(!!(c)) }; \
+	} /* use only *after* variable declarations */
+#define CHECK_V_F(e, f)     \
+	size_t const e = f; \
+	if (ERR_isError(e)) \
+	return f
+#define CHECK_F(f)                        \
+	{                                 \
+		CHECK_V_F(_var_err__, f); \
+	}
+
+/* **************************************************************
+*  Utils
+****************************************************************/
+unsigned HUF_optimalTableLog(unsigned maxTableLog, size_t srcSize, unsigned maxSymbolValue)
+{
+	return FSE_optimalTableLog_internal(maxTableLog, srcSize, maxSymbolValue, 1);
+}
+
+/* *******************************************************
+*  HUF : Huffman block compression
+*********************************************************/
+/* HUF_compressWeights() :
+ * Same as FSE_compress(), but dedicated to huff0's weights compression.
+ * The use case needs much less stack memory.
+ * Note : all elements within weightTable are supposed to be <= HUF_TABLELOG_MAX.
+ */
+#define MAX_FSE_TABLELOG_FOR_HUFF_HEADER 6
+size_t HUF_compressWeights_wksp(void *dst, size_t dstSize, const void *weightTable, size_t wtSize, void *workspace, size_t workspaceSize)
+{
+	BYTE *const ostart = (BYTE *)dst;
+	BYTE *op = ostart;
+	BYTE *const oend = ostart + dstSize;
+
+	U32 maxSymbolValue = HUF_TABLELOG_MAX;
+	U32 tableLog = MAX_FSE_TABLELOG_FOR_HUFF_HEADER;
+
+	FSE_CTable *CTable;
+	U32 *count;
+	S16 *norm;
+	size_t spaceUsed32 = 0;
+
+	HUF_STATIC_ASSERT(sizeof(FSE_CTable) == sizeof(U32));
+
+	CTable = (FSE_CTable *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += FSE_CTABLE_SIZE_U32(MAX_FSE_TABLELOG_FOR_HUFF_HEADER, HUF_TABLELOG_MAX);
+	count = (U32 *)workspace + spaceUsed32;
+	spaceUsed32 += HUF_TABLELOG_MAX + 1;
+	norm = (S16 *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(sizeof(S16) * (HUF_TABLELOG_MAX + 1), sizeof(U32)) >> 2;
+
+	if ((spaceUsed32 << 2) > workspaceSize)
+		return ERROR(tableLog_tooLarge);
+	workspace = (U32 *)workspace + spaceUsed32;
+	workspaceSize -= (spaceUsed32 << 2);
+
+	/* init conditions */
+	if (wtSize <= 1)
+		return 0; /* Not compressible */
+
+	/* Scan input and build symbol stats */
+	{
+		CHECK_V_F(maxCount, FSE_count_simple(count, &maxSymbolValue, weightTable, wtSize));
+		if (maxCount == wtSize)
+			return 1; /* only a single symbol in src : rle */
+		if (maxCount == 1)
+			return 0; /* each symbol present maximum once => not compressible */
+	}
+
+	tableLog = FSE_optimalTableLog(tableLog, wtSize, maxSymbolValue);
+	CHECK_F(FSE_normalizeCount(norm, tableLog, count, wtSize, maxSymbolValue));
+
+	/* Write table description header */
+	{
+		CHECK_V_F(hSize, FSE_writeNCount(op, oend - op, norm, maxSymbolValue, tableLog));
+		op += hSize;
+	}
+
+	/* Compress */
+	CHECK_F(FSE_buildCTable_wksp(CTable, norm, maxSymbolValue, tableLog, workspace, workspaceSize));
+	{
+		CHECK_V_F(cSize, FSE_compress_usingCTable(op, oend - op, weightTable, wtSize, CTable));
+		if (cSize == 0)
+			return 0; /* not enough space for compressed data */
+		op += cSize;
+	}
+
+	return op - ostart;
+}
+
+struct HUF_CElt_s {
+	U16 val;
+	BYTE nbBits;
+}; /* typedef'd to HUF_CElt within "huf.h" */
+
+/*! HUF_writeCTable_wksp() :
+	`CTable` : Huffman tree to save, using huf representation.
+	@return : size of saved CTable */
+size_t HUF_writeCTable_wksp(void *dst, size_t maxDstSize, const HUF_CElt *CTable, U32 maxSymbolValue, U32 huffLog, void *workspace, size_t workspaceSize)
+{
+	BYTE *op = (BYTE *)dst;
+	U32 n;
+
+	BYTE *bitsToWeight;
+	BYTE *huffWeight;
+	size_t spaceUsed32 = 0;
+
+	bitsToWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(HUF_TABLELOG_MAX + 1, sizeof(U32)) >> 2;
+	huffWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX, sizeof(U32)) >> 2;
+
+	if ((spaceUsed32 << 2) > workspaceSize)
+		return ERROR(tableLog_tooLarge);
+	workspace = (U32 *)workspace + spaceUsed32;
+	workspaceSize -= (spaceUsed32 << 2);
+
+	/* check conditions */
+	if (maxSymbolValue > HUF_SYMBOLVALUE_MAX)
+		return ERROR(maxSymbolValue_tooLarge);
+
+	/* convert to weight */
+	bitsToWeight[0] = 0;
+	for (n = 1; n < huffLog + 1; n++)
+		bitsToWeight[n] = (BYTE)(huffLog + 1 - n);
+	for (n = 0; n < maxSymbolValue; n++)
+		huffWeight[n] = bitsToWeight[CTable[n].nbBits];
+
+	/* attempt weights compression by FSE */
+	{
+		CHECK_V_F(hSize, HUF_compressWeights_wksp(op + 1, maxDstSize - 1, huffWeight, maxSymbolValue, workspace, workspaceSize));
+		if ((hSize > 1) & (hSize < maxSymbolValue / 2)) { /* FSE compressed */
+			op[0] = (BYTE)hSize;
+			return hSize + 1;
+		}
+	}
+
+	/* write raw values as 4-bits (max : 15) */
+	if (maxSymbolValue > (256 - 128))
+		return ERROR(GENERIC); /* should not happen : likely means source cannot be compressed */
+	if (((maxSymbolValue + 1) / 2) + 1 > maxDstSize)
+		return ERROR(dstSize_tooSmall); /* not enough space within dst buffer */
+	op[0] = (BYTE)(128 /*special case*/ + (maxSymbolValue - 1));
+	huffWeight[maxSymbolValue] = 0; /* to be sure it doesn't cause msan issue in final combination */
+	for (n = 0; n < maxSymbolValue; n += 2)
+		op[(n / 2) + 1] = (BYTE)((huffWeight[n] << 4) + huffWeight[n + 1]);
+	return ((maxSymbolValue + 1) / 2) + 1;
+}
+
+size_t HUF_readCTable_wksp(HUF_CElt *CTable, U32 maxSymbolValue, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
+{
+	U32 *rankVal;
+	BYTE *huffWeight;
+	U32 tableLog = 0;
+	U32 nbSymbols = 0;
+	size_t readSize;
+	size_t spaceUsed32 = 0;
+
+	rankVal = (U32 *)workspace + spaceUsed32;
+	spaceUsed32 += HUF_TABLELOG_ABSOLUTEMAX + 1;
+	huffWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
+
+	if ((spaceUsed32 << 2) > workspaceSize)
+		return ERROR(tableLog_tooLarge);
+	workspace = (U32 *)workspace + spaceUsed32;
+	workspaceSize -= (spaceUsed32 << 2);
+
+	/* get symbol weights */
+	readSize = HUF_readStats_wksp(huffWeight, HUF_SYMBOLVALUE_MAX + 1, rankVal, &nbSymbols, &tableLog, src, srcSize, workspace, workspaceSize);
+	if (ERR_isError(readSize))
+		return readSize;
+
+	/* check result */
+	if (tableLog > HUF_TABLELOG_MAX)
+		return ERROR(tableLog_tooLarge);
+	if (nbSymbols > maxSymbolValue + 1)
+		return ERROR(maxSymbolValue_tooSmall);
+
+	/* Prepare base value per rank */
+	{
+		U32 n, nextRankStart = 0;
+		for (n = 1; n <= tableLog; n++) {
+			U32 curr = nextRankStart;
+			nextRankStart += (rankVal[n] << (n - 1));
+			rankVal[n] = curr;
+		}
+	}
+
+	/* fill nbBits */
+	{
+		U32 n;
+		for (n = 0; n < nbSymbols; n++) {
+			const U32 w = huffWeight[n];
+			CTable[n].nbBits = (BYTE)(tableLog + 1 - w);
+		}
+	}
+
+	/* fill val */
+	{
+		U16 nbPerRank[HUF_TABLELOG_MAX + 2] = {0}; /* support w=0=>n=tableLog+1 */
+		U16 valPerRank[HUF_TABLELOG_MAX + 2] = {0};
+		{
+			U32 n;
+			for (n = 0; n < nbSymbols; n++)
+				nbPerRank[CTable[n].nbBits]++;
+		}
+		/* determine stating value per rank */
+		valPerRank[tableLog + 1] = 0; /* for w==0 */
+		{
+			U16 min = 0;
+			U32 n;
+			for (n = tableLog; n > 0; n--) { /* start at n=tablelog <-> w=1 */
+				valPerRank[n] = min;     /* get starting value within each rank */
+				min += nbPerRank[n];
+				min >>= 1;
+			}
+		}
+		/* assign value within rank, symbol order */
+		{
+			U32 n;
+			for (n = 0; n <= maxSymbolValue; n++)
+				CTable[n].val = valPerRank[CTable[n].nbBits]++;
+		}
+	}
+
+	return readSize;
+}
+
+typedef struct nodeElt_s {
+	U32 count;
+	U16 parent;
+	BYTE byte;
+	BYTE nbBits;
+} nodeElt;
+
+static U32 HUF_setMaxHeight(nodeElt *huffNode, U32 lastNonNull, U32 maxNbBits)
+{
+	const U32 largestBits = huffNode[lastNonNull].nbBits;
+	if (largestBits <= maxNbBits)
+		return largestBits; /* early exit : no elt > maxNbBits */
+
+	/* there are several too large elements (at least >= 2) */
+	{
+		int totalCost = 0;
+		const U32 baseCost = 1 << (largestBits - maxNbBits);
+		U32 n = lastNonNull;
+
+		while (huffNode[n].nbBits > maxNbBits) {
+			totalCost += baseCost - (1 << (largestBits - huffNode[n].nbBits));
+			huffNode[n].nbBits = (BYTE)maxNbBits;
+			n--;
+		} /* n stops at huffNode[n].nbBits <= maxNbBits */
+		while (huffNode[n].nbBits == maxNbBits)
+			n--; /* n end at index of smallest symbol using < maxNbBits */
+
+		/* renorm totalCost */
+		totalCost >>= (largestBits - maxNbBits); /* note : totalCost is necessarily a multiple of baseCost */
+
+		/* repay normalized cost */
+		{
+			U32 const noSymbol = 0xF0F0F0F0;
+			U32 rankLast[HUF_TABLELOG_MAX + 2];
+			int pos;
+
+			/* Get pos of last (smallest) symbol per rank */
+			memset(rankLast, 0xF0, sizeof(rankLast));
+			{
+				U32 currNbBits = maxNbBits;
+				for (pos = n; pos >= 0; pos--) {
+					if (huffNode[pos].nbBits >= currNbBits)
+						continue;
+					currNbBits = huffNode[pos].nbBits; /* < maxNbBits */
+					rankLast[maxNbBits - currNbBits] = pos;
+				}
+			}
+
+			while (totalCost > 0) {
+				U32 nBitsToDecrease = BIT_highbit32(totalCost) + 1;
+				for (; nBitsToDecrease > 1; nBitsToDecrease--) {
+					U32 highPos = rankLast[nBitsToDecrease];
+					U32 lowPos = rankLast[nBitsToDecrease - 1];
+					if (highPos == noSymbol)
+						continue;
+					if (lowPos == noSymbol)
+						break;
+					{
+						U32 const highTotal = huffNode[highPos].count;
+						U32 const lowTotal = 2 * huffNode[lowPos].count;
+						if (highTotal <= lowTotal)
+							break;
+					}
+				}
+				/* only triggered when no more rank 1 symbol left => find closest one (note : there is necessarily at least one !) */
+				/* HUF_MAX_TABLELOG test just to please gcc 5+; but it should not be necessary */
+				while ((nBitsToDecrease <= HUF_TABLELOG_MAX) && (rankLast[nBitsToDecrease] == noSymbol))
+					nBitsToDecrease++;
+				totalCost -= 1 << (nBitsToDecrease - 1);
+				if (rankLast[nBitsToDecrease - 1] == noSymbol)
+					rankLast[nBitsToDecrease - 1] = rankLast[nBitsToDecrease]; /* this rank is no longer empty */
+				huffNode[rankLast[nBitsToDecrease]].nbBits++;
+				if (rankLast[nBitsToDecrease] == 0) /* special case, reached largest symbol */
+					rankLast[nBitsToDecrease] = noSymbol;
+				else {
+					rankLast[nBitsToDecrease]--;
+					if (huffNode[rankLast[nBitsToDecrease]].nbBits != maxNbBits - nBitsToDecrease)
+						rankLast[nBitsToDecrease] = noSymbol; /* this rank is now empty */
+				}
+			} /* while (totalCost > 0) */
+
+			while (totalCost < 0) {		       /* Sometimes, cost correction overshoot */
+				if (rankLast[1] == noSymbol) { /* special case : no rank 1 symbol (using maxNbBits-1); let's create one from largest rank 0
+								  (using maxNbBits) */
+					while (huffNode[n].nbBits == maxNbBits)
+						n--;
+					huffNode[n + 1].nbBits--;
+					rankLast[1] = n + 1;
+					totalCost++;
+					continue;
+				}
+				huffNode[rankLast[1] + 1].nbBits--;
+				rankLast[1]++;
+				totalCost++;
+			}
+		}
+	} /* there are several too large elements (at least >= 2) */
+
+	return maxNbBits;
+}
+
+typedef struct {
+	U32 base;
+	U32 curr;
+} rankPos;
+
+static void HUF_sort(nodeElt *huffNode, const U32 *count, U32 maxSymbolValue)
+{
+	rankPos rank[32];
+	U32 n;
+
+	memset(rank, 0, sizeof(rank));
+	for (n = 0; n <= maxSymbolValue; n++) {
+		U32 r = BIT_highbit32(count[n] + 1);
+		rank[r].base++;
+	}
+	for (n = 30; n > 0; n--)
+		rank[n - 1].base += rank[n].base;
+	for (n = 0; n < 32; n++)
+		rank[n].curr = rank[n].base;
+	for (n = 0; n <= maxSymbolValue; n++) {
+		U32 const c = count[n];
+		U32 const r = BIT_highbit32(c + 1) + 1;
+		U32 pos = rank[r].curr++;
+		while ((pos > rank[r].base) && (c > huffNode[pos - 1].count))
+			huffNode[pos] = huffNode[pos - 1], pos--;
+		huffNode[pos].count = c;
+		huffNode[pos].byte = (BYTE)n;
+	}
+}
+
+/** HUF_buildCTable_wksp() :
+ *  Same as HUF_buildCTable(), but using externally allocated scratch buffer.
+ *  `workSpace` must be aligned on 4-bytes boundaries, and be at least as large as a table of 1024 unsigned.
+ */
+#define STARTNODE (HUF_SYMBOLVALUE_MAX + 1)
+typedef nodeElt huffNodeTable[2 * HUF_SYMBOLVALUE_MAX + 1 + 1];
+size_t HUF_buildCTable_wksp(HUF_CElt *tree, const U32 *count, U32 maxSymbolValue, U32 maxNbBits, void *workSpace, size_t wkspSize)
+{
+	nodeElt *const huffNode0 = (nodeElt *)workSpace;
+	nodeElt *const huffNode = huffNode0 + 1;
+	U32 n, nonNullRank;
+	int lowS, lowN;
+	U16 nodeNb = STARTNODE;
+	U32 nodeRoot;
+
+	/* safety checks */
+	if (wkspSize < sizeof(huffNodeTable))
+		return ERROR(GENERIC); /* workSpace is not large enough */
+	if (maxNbBits == 0)
+		maxNbBits = HUF_TABLELOG_DEFAULT;
+	if (maxSymbolValue > HUF_SYMBOLVALUE_MAX)
+		return ERROR(GENERIC);
+	memset(huffNode0, 0, sizeof(huffNodeTable));
+
+	/* sort, decreasing order */
+	HUF_sort(huffNode, count, maxSymbolValue);
+
+	/* init for parents */
+	nonNullRank = maxSymbolValue;
+	while (huffNode[nonNullRank].count == 0)
+		nonNullRank--;
+	lowS = nonNullRank;
+	nodeRoot = nodeNb + lowS - 1;
+	lowN = nodeNb;
+	huffNode[nodeNb].count = huffNode[lowS].count + huffNode[lowS - 1].count;
+	huffNode[lowS].parent = huffNode[lowS - 1].parent = nodeNb;
+	nodeNb++;
+	lowS -= 2;
+	for (n = nodeNb; n <= nodeRoot; n++)
+		huffNode[n].count = (U32)(1U << 30);
+	huffNode0[0].count = (U32)(1U << 31); /* fake entry, strong barrier */
+
+	/* create parents */
+	while (nodeNb <= nodeRoot) {
+		U32 n1 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++;
+		U32 n2 = (huffNode[lowS].count < huffNode[lowN].count) ? lowS-- : lowN++;
+		huffNode[nodeNb].count = huffNode[n1].count + huffNode[n2].count;
+		huffNode[n1].parent = huffNode[n2].parent = nodeNb;
+		nodeNb++;
+	}
+
+	/* distribute weights (unlimited tree height) */
+	huffNode[nodeRoot].nbBits = 0;
+	for (n = nodeRoot - 1; n >= STARTNODE; n--)
+		huffNode[n].nbBits = huffNode[huffNode[n].parent].nbBits + 1;
+	for (n = 0; n <= nonNullRank; n++)
+		huffNode[n].nbBits = huffNode[huffNode[n].parent].nbBits + 1;
+
+	/* enforce maxTableLog */
+	maxNbBits = HUF_setMaxHeight(huffNode, nonNullRank, maxNbBits);
+
+	/* fill result into tree (val, nbBits) */
+	{
+		U16 nbPerRank[HUF_TABLELOG_MAX + 1] = {0};
+		U16 valPerRank[HUF_TABLELOG_MAX + 1] = {0};
+		if (maxNbBits > HUF_TABLELOG_MAX)
+			return ERROR(GENERIC); /* check fit into table */
+		for (n = 0; n <= nonNullRank; n++)
+			nbPerRank[huffNode[n].nbBits]++;
+		/* determine stating value per rank */
+		{
+			U16 min = 0;
+			for (n = maxNbBits; n > 0; n--) {
+				valPerRank[n] = min; /* get starting value within each rank */
+				min += nbPerRank[n];
+				min >>= 1;
+			}
+		}
+		for (n = 0; n <= maxSymbolValue; n++)
+			tree[huffNode[n].byte].nbBits = huffNode[n].nbBits; /* push nbBits per symbol, symbol order */
+		for (n = 0; n <= maxSymbolValue; n++)
+			tree[n].val = valPerRank[tree[n].nbBits]++; /* assign value within rank, symbol order */
+	}
+
+	return maxNbBits;
+}
+
+static size_t HUF_estimateCompressedSize(HUF_CElt *CTable, const unsigned *count, unsigned maxSymbolValue)
+{
+	size_t nbBits = 0;
+	int s;
+	for (s = 0; s <= (int)maxSymbolValue; ++s) {
+		nbBits += CTable[s].nbBits * count[s];
+	}
+	return nbBits >> 3;
+}
+
+static int HUF_validateCTable(const HUF_CElt *CTable, const unsigned *count, unsigned maxSymbolValue)
+{
+	int bad = 0;
+	int s;
+	for (s = 0; s <= (int)maxSymbolValue; ++s) {
+		bad |= (count[s] != 0) & (CTable[s].nbBits == 0);
+	}
+	return !bad;
+}
+
+static void HUF_encodeSymbol(BIT_CStream_t *bitCPtr, U32 symbol, const HUF_CElt *CTable)
+{
+	BIT_addBitsFast(bitCPtr, CTable[symbol].val, CTable[symbol].nbBits);
+}
+
+size_t HUF_compressBound(size_t size) { return HUF_COMPRESSBOUND(size); }
+
+#define HUF_FLUSHBITS(s)  BIT_flushBits(s)
+
+#define HUF_FLUSHBITS_1(stream)                                            \
+	if (sizeof((stream)->bitContainer) * 8 < HUF_TABLELOG_MAX * 2 + 7) \
+	HUF_FLUSHBITS(stream)
+
+#define HUF_FLUSHBITS_2(stream)                                            \
+	if (sizeof((stream)->bitContainer) * 8 < HUF_TABLELOG_MAX * 4 + 7) \
+	HUF_FLUSHBITS(stream)
+
+size_t HUF_compress1X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable)
+{
+	const BYTE *ip = (const BYTE *)src;
+	BYTE *const ostart = (BYTE *)dst;
+	BYTE *const oend = ostart + dstSize;
+	BYTE *op = ostart;
+	size_t n;
+	BIT_CStream_t bitC;
+
+	/* init */
+	if (dstSize < 8)
+		return 0; /* not enough space to compress */
+	{
+		size_t const initErr = BIT_initCStream(&bitC, op, oend - op);
+		if (HUF_isError(initErr))
+			return 0;
+	}
+
+	n = srcSize & ~3; /* join to mod 4 */
+	switch (srcSize & 3) {
+	case 3: HUF_encodeSymbol(&bitC, ip[n + 2], CTable); HUF_FLUSHBITS_2(&bitC);
+	case 2: HUF_encodeSymbol(&bitC, ip[n + 1], CTable); HUF_FLUSHBITS_1(&bitC);
+	case 1: HUF_encodeSymbol(&bitC, ip[n + 0], CTable); HUF_FLUSHBITS(&bitC);
+	case 0:
+	default:;
+	}
+
+	for (; n > 0; n -= 4) { /* note : n&3==0 at this stage */
+		HUF_encodeSymbol(&bitC, ip[n - 1], CTable);
+		HUF_FLUSHBITS_1(&bitC);
+		HUF_encodeSymbol(&bitC, ip[n - 2], CTable);
+		HUF_FLUSHBITS_2(&bitC);
+		HUF_encodeSymbol(&bitC, ip[n - 3], CTable);
+		HUF_FLUSHBITS_1(&bitC);
+		HUF_encodeSymbol(&bitC, ip[n - 4], CTable);
+		HUF_FLUSHBITS(&bitC);
+	}
+
+	return BIT_closeCStream(&bitC);
+}
+
+size_t HUF_compress4X_usingCTable(void *dst, size_t dstSize, const void *src, size_t srcSize, const HUF_CElt *CTable)
+{
+	size_t const segmentSize = (srcSize + 3) / 4; /* first 3 segments */
+	const BYTE *ip = (const BYTE *)src;
+	const BYTE *const iend = ip + srcSize;
+	BYTE *const ostart = (BYTE *)dst;
+	BYTE *const oend = ostart + dstSize;
+	BYTE *op = ostart;
+
+	if (dstSize < 6 + 1 + 1 + 1 + 8)
+		return 0; /* minimum space to compress successfully */
+	if (srcSize < 12)
+		return 0; /* no saving possible : too small input */
+	op += 6;	  /* jumpTable */
+
+	{
+		CHECK_V_F(cSize, HUF_compress1X_usingCTable(op, oend - op, ip, segmentSize, CTable));
+		if (cSize == 0)
+			return 0;
+		ZSTD_writeLE16(ostart, (U16)cSize);
+		op += cSize;
+	}
+
+	ip += segmentSize;
+	{
+		CHECK_V_F(cSize, HUF_compress1X_usingCTable(op, oend - op, ip, segmentSize, CTable));
+		if (cSize == 0)
+			return 0;
+		ZSTD_writeLE16(ostart + 2, (U16)cSize);
+		op += cSize;
+	}
+
+	ip += segmentSize;
+	{
+		CHECK_V_F(cSize, HUF_compress1X_usingCTable(op, oend - op, ip, segmentSize, CTable));
+		if (cSize == 0)
+			return 0;
+		ZSTD_writeLE16(ostart + 4, (U16)cSize);
+		op += cSize;
+	}
+
+	ip += segmentSize;
+	{
+		CHECK_V_F(cSize, HUF_compress1X_usingCTable(op, oend - op, ip, iend - ip, CTable));
+		if (cSize == 0)
+			return 0;
+		op += cSize;
+	}
+
+	return op - ostart;
+}
+
+static size_t HUF_compressCTable_internal(BYTE *const ostart, BYTE *op, BYTE *const oend, const void *src, size_t srcSize, unsigned singleStream,
+					  const HUF_CElt *CTable)
+{
+	size_t const cSize =
+	    singleStream ? HUF_compress1X_usingCTable(op, oend - op, src, srcSize, CTable) : HUF_compress4X_usingCTable(op, oend - op, src, srcSize, CTable);
+	if (HUF_isError(cSize)) {
+		return cSize;
+	}
+	if (cSize == 0) {
+		return 0;
+	} /* uncompressible */
+	op += cSize;
+	/* check compressibility */
+	if ((size_t)(op - ostart) >= srcSize - 1) {
+		return 0;
+	}
+	return op - ostart;
+}
+
+/* `workSpace` must a table of at least 1024 unsigned */
+static size_t HUF_compress_internal(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog,
+				    unsigned singleStream, void *workSpace, size_t wkspSize, HUF_CElt *oldHufTable, HUF_repeat *repeat, int preferRepeat)
+{
+	BYTE *const ostart = (BYTE *)dst;
+	BYTE *const oend = ostart + dstSize;
+	BYTE *op = ostart;
+
+	U32 *count;
+	size_t const countSize = sizeof(U32) * (HUF_SYMBOLVALUE_MAX + 1);
+	HUF_CElt *CTable;
+	size_t const CTableSize = sizeof(HUF_CElt) * (HUF_SYMBOLVALUE_MAX + 1);
+
+	/* checks & inits */
+	if (wkspSize < sizeof(huffNodeTable) + countSize + CTableSize)
+		return ERROR(GENERIC);
+	if (!srcSize)
+		return 0; /* Uncompressed (note : 1 means rle, so first byte must be correct) */
+	if (!dstSize)
+		return 0; /* cannot fit within dst budget */
+	if (srcSize > HUF_BLOCKSIZE_MAX)
+		return ERROR(srcSize_wrong); /* curr block size limit */
+	if (huffLog > HUF_TABLELOG_MAX)
+		return ERROR(tableLog_tooLarge);
+	if (!maxSymbolValue)
+		maxSymbolValue = HUF_SYMBOLVALUE_MAX;
+	if (!huffLog)
+		huffLog = HUF_TABLELOG_DEFAULT;
+
+	count = (U32 *)workSpace;
+	workSpace = (BYTE *)workSpace + countSize;
+	wkspSize -= countSize;
+	CTable = (HUF_CElt *)workSpace;
+	workSpace = (BYTE *)workSpace + CTableSize;
+	wkspSize -= CTableSize;
+
+	/* Heuristic : If we don't need to check the validity of the old table use the old table for small inputs */
+	if (preferRepeat && repeat && *repeat == HUF_repeat_valid) {
+		return HUF_compressCTable_internal(ostart, op, oend, src, srcSize, singleStream, oldHufTable);
+	}
+
+	/* Scan input and build symbol stats */
+	{
+		CHECK_V_F(largest, FSE_count_wksp(count, &maxSymbolValue, (const BYTE *)src, srcSize, (U32 *)workSpace));
+		if (largest == srcSize) {
+			*ostart = ((const BYTE *)src)[0];
+			return 1;
+		} /* single symbol, rle */
+		if (largest <= (srcSize >> 7) + 1)
+			return 0; /* Fast heuristic : not compressible enough */
+	}
+
+	/* Check validity of previous table */
+	if (repeat && *repeat == HUF_repeat_check && !HUF_validateCTable(oldHufTable, count, maxSymbolValue)) {
+		*repeat = HUF_repeat_none;
+	}
+	/* Heuristic : use existing table for small inputs */
+	if (preferRepeat && repeat && *repeat != HUF_repeat_none) {
+		return HUF_compressCTable_internal(ostart, op, oend, src, srcSize, singleStream, oldHufTable);
+	}
+
+	/* Build Huffman Tree */
+	huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
+	{
+		CHECK_V_F(maxBits, HUF_buildCTable_wksp(CTable, count, maxSymbolValue, huffLog, workSpace, wkspSize));
+		huffLog = (U32)maxBits;
+		/* Zero the unused symbols so we can check it for validity */
+		memset(CTable + maxSymbolValue + 1, 0, CTableSize - (maxSymbolValue + 1) * sizeof(HUF_CElt));
+	}
+
+	/* Write table description header */
+	{
+		CHECK_V_F(hSize, HUF_writeCTable_wksp(op, dstSize, CTable, maxSymbolValue, huffLog, workSpace, wkspSize));
+		/* Check if using the previous table will be beneficial */
+		if (repeat && *repeat != HUF_repeat_none) {
+			size_t const oldSize = HUF_estimateCompressedSize(oldHufTable, count, maxSymbolValue);
+			size_t const newSize = HUF_estimateCompressedSize(CTable, count, maxSymbolValue);
+			if (oldSize <= hSize + newSize || hSize + 12 >= srcSize) {
+				return HUF_compressCTable_internal(ostart, op, oend, src, srcSize, singleStream, oldHufTable);
+			}
+		}
+		/* Use the new table */
+		if (hSize + 12ul >= srcSize) {
+			return 0;
+		}
+		op += hSize;
+		if (repeat) {
+			*repeat = HUF_repeat_none;
+		}
+		if (oldHufTable) {
+			memcpy(oldHufTable, CTable, CTableSize);
+		} /* Save the new table */
+	}
+	return HUF_compressCTable_internal(ostart, op, oend, src, srcSize, singleStream, CTable);
+}
+
+size_t HUF_compress1X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog, void *workSpace,
+			   size_t wkspSize)
+{
+	return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 1 /* single stream */, workSpace, wkspSize, NULL, NULL, 0);
+}
+
+size_t HUF_compress1X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog, void *workSpace,
+			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat, int preferRepeat)
+{
+	return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 1 /* single stream */, workSpace, wkspSize, hufTable, repeat,
+				     preferRepeat);
+}
+
+size_t HUF_compress4X_wksp(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog, void *workSpace,
+			   size_t wkspSize)
+{
+	return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 0 /* 4 streams */, workSpace, wkspSize, NULL, NULL, 0);
+}
+
+size_t HUF_compress4X_repeat(void *dst, size_t dstSize, const void *src, size_t srcSize, unsigned maxSymbolValue, unsigned huffLog, void *workSpace,
+			     size_t wkspSize, HUF_CElt *hufTable, HUF_repeat *repeat, int preferRepeat)
+{
+	return HUF_compress_internal(dst, dstSize, src, srcSize, maxSymbolValue, huffLog, 0 /* 4 streams */, workSpace, wkspSize, hufTable, repeat,
+				     preferRepeat);
+}
diff --git a/lib/zstd/huf_decompress.c b/lib/zstd/huf_decompress.c
new file mode 100644
index 000000000000..6526482047dc
--- /dev/null
+++ b/lib/zstd/huf_decompress.c
@@ -0,0 +1,960 @@
+/*
+ * Huffman decoder, part of New Generation Entropy library
+ * Copyright (C) 2013-2016, Yann Collet.
+ *
+ * BSD 2-Clause License (http://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *   * Redistributions of source code must retain the above copyright
+ * notice, this list of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following disclaimer
+ * in the documentation and/or other materials provided with the
+ * distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ *
+ * You can contact the author at :
+ * - Source repository : https://github.com/Cyan4973/FiniteStateEntropy
+ */
+
+/* **************************************************************
+*  Compiler specifics
+****************************************************************/
+#define FORCE_INLINE static __always_inline
+
+/* **************************************************************
+*  Dependencies
+****************************************************************/
+#include "bitstream.h" /* BIT_* */
+#include "fse.h"       /* header compression */
+#include "huf.h"
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <linux/string.h> /* memcpy, memset */
+
+/* **************************************************************
+*  Error Management
+****************************************************************/
+#define HUF_STATIC_ASSERT(c)                                   \
+	{                                                      \
+		enum { HUF_static_assert = 1 / (int)(!!(c)) }; \
+	} /* use only *after* variable declarations */
+
+/*-***************************/
+/*  generic DTableDesc       */
+/*-***************************/
+
+typedef struct {
+	BYTE maxTableLog;
+	BYTE tableType;
+	BYTE tableLog;
+	BYTE reserved;
+} DTableDesc;
+
+static DTableDesc HUF_getDTableDesc(const HUF_DTable *table)
+{
+	DTableDesc dtd;
+	memcpy(&dtd, table, sizeof(dtd));
+	return dtd;
+}
+
+/*-***************************/
+/*  single-symbol decoding   */
+/*-***************************/
+
+typedef struct {
+	BYTE byte;
+	BYTE nbBits;
+} HUF_DEltX2; /* single-symbol decoding */
+
+size_t HUF_readDTableX2_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
+{
+	U32 tableLog = 0;
+	U32 nbSymbols = 0;
+	size_t iSize;
+	void *const dtPtr = DTable + 1;
+	HUF_DEltX2 *const dt = (HUF_DEltX2 *)dtPtr;
+
+	U32 *rankVal;
+	BYTE *huffWeight;
+	size_t spaceUsed32 = 0;
+
+	rankVal = (U32 *)workspace + spaceUsed32;
+	spaceUsed32 += HUF_TABLELOG_ABSOLUTEMAX + 1;
+	huffWeight = (BYTE *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
+
+	if ((spaceUsed32 << 2) > workspaceSize)
+		return ERROR(tableLog_tooLarge);
+	workspace = (U32 *)workspace + spaceUsed32;
+	workspaceSize -= (spaceUsed32 << 2);
+
+	HUF_STATIC_ASSERT(sizeof(DTableDesc) == sizeof(HUF_DTable));
+	/* memset(huffWeight, 0, sizeof(huffWeight)); */ /* is not necessary, even though some analyzer complain ... */
+
+	iSize = HUF_readStats_wksp(huffWeight, HUF_SYMBOLVALUE_MAX + 1, rankVal, &nbSymbols, &tableLog, src, srcSize, workspace, workspaceSize);
+	if (HUF_isError(iSize))
+		return iSize;
+
+	/* Table header */
+	{
+		DTableDesc dtd = HUF_getDTableDesc(DTable);
+		if (tableLog > (U32)(dtd.maxTableLog + 1))
+			return ERROR(tableLog_tooLarge); /* DTable too small, Huffman tree cannot fit in */
+		dtd.tableType = 0;
+		dtd.tableLog = (BYTE)tableLog;
+		memcpy(DTable, &dtd, sizeof(dtd));
+	}
+
+	/* Calculate starting value for each rank */
+	{
+		U32 n, nextRankStart = 0;
+		for (n = 1; n < tableLog + 1; n++) {
+			U32 const curr = nextRankStart;
+			nextRankStart += (rankVal[n] << (n - 1));
+			rankVal[n] = curr;
+		}
+	}
+
+	/* fill DTable */
+	{
+		U32 n;
+		for (n = 0; n < nbSymbols; n++) {
+			U32 const w = huffWeight[n];
+			U32 const length = (1 << w) >> 1;
+			U32 u;
+			HUF_DEltX2 D;
+			D.byte = (BYTE)n;
+			D.nbBits = (BYTE)(tableLog + 1 - w);
+			for (u = rankVal[w]; u < rankVal[w] + length; u++)
+				dt[u] = D;
+			rankVal[w] += length;
+		}
+	}
+
+	return iSize;
+}
+
+static BYTE HUF_decodeSymbolX2(BIT_DStream_t *Dstream, const HUF_DEltX2 *dt, const U32 dtLog)
+{
+	size_t const val = BIT_lookBitsFast(Dstream, dtLog); /* note : dtLog >= 1 */
+	BYTE const c = dt[val].byte;
+	BIT_skipBits(Dstream, dt[val].nbBits);
+	return c;
+}
+
+#define HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr) *ptr++ = HUF_decodeSymbolX2(DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX2_1(ptr, DStreamPtr)         \
+	if (ZSTD_64bits() || (HUF_TABLELOG_MAX <= 12)) \
+	HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
+
+#define HUF_DECODE_SYMBOLX2_2(ptr, DStreamPtr) \
+	if (ZSTD_64bits())                     \
+	HUF_DECODE_SYMBOLX2_0(ptr, DStreamPtr)
+
+FORCE_INLINE size_t HUF_decodeStreamX2(BYTE *p, BIT_DStream_t *const bitDPtr, BYTE *const pEnd, const HUF_DEltX2 *const dt, const U32 dtLog)
+{
+	BYTE *const pStart = p;
+
+	/* up to 4 symbols at a time */
+	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p <= pEnd - 4)) {
+		HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
+		HUF_DECODE_SYMBOLX2_1(p, bitDPtr);
+		HUF_DECODE_SYMBOLX2_2(p, bitDPtr);
+		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
+	}
+
+	/* closer to the end */
+	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) && (p < pEnd))
+		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
+
+	/* no more data to retrieve from bitstream, hence no need to reload */
+	while (p < pEnd)
+		HUF_DECODE_SYMBOLX2_0(p, bitDPtr);
+
+	return pEnd - pStart;
+}
+
+static size_t HUF_decompress1X2_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	BYTE *op = (BYTE *)dst;
+	BYTE *const oend = op + dstSize;
+	const void *dtPtr = DTable + 1;
+	const HUF_DEltX2 *const dt = (const HUF_DEltX2 *)dtPtr;
+	BIT_DStream_t bitD;
+	DTableDesc const dtd = HUF_getDTableDesc(DTable);
+	U32 const dtLog = dtd.tableLog;
+
+	{
+		size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
+		if (HUF_isError(errorCode))
+			return errorCode;
+	}
+
+	HUF_decodeStreamX2(op, &bitD, oend, dt, dtLog);
+
+	/* check */
+	if (!BIT_endOfDStream(&bitD))
+		return ERROR(corruption_detected);
+
+	return dstSize;
+}
+
+size_t HUF_decompress1X2_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	DTableDesc dtd = HUF_getDTableDesc(DTable);
+	if (dtd.tableType != 0)
+		return ERROR(GENERIC);
+	return HUF_decompress1X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
+}
+
+size_t HUF_decompress1X2_DCtx_wksp(HUF_DTable *DCtx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
+{
+	const BYTE *ip = (const BYTE *)cSrc;
+
+	size_t const hSize = HUF_readDTableX2_wksp(DCtx, cSrc, cSrcSize, workspace, workspaceSize);
+	if (HUF_isError(hSize))
+		return hSize;
+	if (hSize >= cSrcSize)
+		return ERROR(srcSize_wrong);
+	ip += hSize;
+	cSrcSize -= hSize;
+
+	return HUF_decompress1X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx);
+}
+
+static size_t HUF_decompress4X2_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	/* Check */
+	if (cSrcSize < 10)
+		return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
+
+	{
+		const BYTE *const istart = (const BYTE *)cSrc;
+		BYTE *const ostart = (BYTE *)dst;
+		BYTE *const oend = ostart + dstSize;
+		const void *const dtPtr = DTable + 1;
+		const HUF_DEltX2 *const dt = (const HUF_DEltX2 *)dtPtr;
+
+		/* Init */
+		BIT_DStream_t bitD1;
+		BIT_DStream_t bitD2;
+		BIT_DStream_t bitD3;
+		BIT_DStream_t bitD4;
+		size_t const length1 = ZSTD_readLE16(istart);
+		size_t const length2 = ZSTD_readLE16(istart + 2);
+		size_t const length3 = ZSTD_readLE16(istart + 4);
+		size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
+		const BYTE *const istart1 = istart + 6; /* jumpTable */
+		const BYTE *const istart2 = istart1 + length1;
+		const BYTE *const istart3 = istart2 + length2;
+		const BYTE *const istart4 = istart3 + length3;
+		const size_t segmentSize = (dstSize + 3) / 4;
+		BYTE *const opStart2 = ostart + segmentSize;
+		BYTE *const opStart3 = opStart2 + segmentSize;
+		BYTE *const opStart4 = opStart3 + segmentSize;
+		BYTE *op1 = ostart;
+		BYTE *op2 = opStart2;
+		BYTE *op3 = opStart3;
+		BYTE *op4 = opStart4;
+		U32 endSignal;
+		DTableDesc const dtd = HUF_getDTableDesc(DTable);
+		U32 const dtLog = dtd.tableLog;
+
+		if (length4 > cSrcSize)
+			return ERROR(corruption_detected); /* overflow */
+		{
+			size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
+			if (HUF_isError(errorCode))
+				return errorCode;
+		}
+		{
+			size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
+			if (HUF_isError(errorCode))
+				return errorCode;
+		}
+		{
+			size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
+			if (HUF_isError(errorCode))
+				return errorCode;
+		}
+		{
+			size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
+			if (HUF_isError(errorCode))
+				return errorCode;
+		}
+
+		/* 16-32 symbols per loop (4-8 symbols per stream) */
+		endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
+		for (; (endSignal == BIT_DStream_unfinished) && (op4 < (oend - 7));) {
+			HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+			HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+			HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+			HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+			HUF_DECODE_SYMBOLX2_1(op1, &bitD1);
+			HUF_DECODE_SYMBOLX2_1(op2, &bitD2);
+			HUF_DECODE_SYMBOLX2_1(op3, &bitD3);
+			HUF_DECODE_SYMBOLX2_1(op4, &bitD4);
+			HUF_DECODE_SYMBOLX2_2(op1, &bitD1);
+			HUF_DECODE_SYMBOLX2_2(op2, &bitD2);
+			HUF_DECODE_SYMBOLX2_2(op3, &bitD3);
+			HUF_DECODE_SYMBOLX2_2(op4, &bitD4);
+			HUF_DECODE_SYMBOLX2_0(op1, &bitD1);
+			HUF_DECODE_SYMBOLX2_0(op2, &bitD2);
+			HUF_DECODE_SYMBOLX2_0(op3, &bitD3);
+			HUF_DECODE_SYMBOLX2_0(op4, &bitD4);
+			endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
+		}
+
+		/* check corruption */
+		if (op1 > opStart2)
+			return ERROR(corruption_detected);
+		if (op2 > opStart3)
+			return ERROR(corruption_detected);
+		if (op3 > opStart4)
+			return ERROR(corruption_detected);
+		/* note : op4 supposed already verified within main loop */
+
+		/* finish bitStreams one by one */
+		HUF_decodeStreamX2(op1, &bitD1, opStart2, dt, dtLog);
+		HUF_decodeStreamX2(op2, &bitD2, opStart3, dt, dtLog);
+		HUF_decodeStreamX2(op3, &bitD3, opStart4, dt, dtLog);
+		HUF_decodeStreamX2(op4, &bitD4, oend, dt, dtLog);
+
+		/* check */
+		endSignal = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
+		if (!endSignal)
+			return ERROR(corruption_detected);
+
+		/* decoded size */
+		return dstSize;
+	}
+}
+
+size_t HUF_decompress4X2_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	DTableDesc dtd = HUF_getDTableDesc(DTable);
+	if (dtd.tableType != 0)
+		return ERROR(GENERIC);
+	return HUF_decompress4X2_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
+}
+
+size_t HUF_decompress4X2_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
+{
+	const BYTE *ip = (const BYTE *)cSrc;
+
+	size_t const hSize = HUF_readDTableX2_wksp(dctx, cSrc, cSrcSize, workspace, workspaceSize);
+	if (HUF_isError(hSize))
+		return hSize;
+	if (hSize >= cSrcSize)
+		return ERROR(srcSize_wrong);
+	ip += hSize;
+	cSrcSize -= hSize;
+
+	return HUF_decompress4X2_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx);
+}
+
+/* *************************/
+/* double-symbols decoding */
+/* *************************/
+typedef struct {
+	U16 sequence;
+	BYTE nbBits;
+	BYTE length;
+} HUF_DEltX4; /* double-symbols decoding */
+
+typedef struct {
+	BYTE symbol;
+	BYTE weight;
+} sortedSymbol_t;
+
+/* HUF_fillDTableX4Level2() :
+ * `rankValOrigin` must be a table of at least (HUF_TABLELOG_MAX + 1) U32 */
+static void HUF_fillDTableX4Level2(HUF_DEltX4 *DTable, U32 sizeLog, const U32 consumed, const U32 *rankValOrigin, const int minWeight,
+				   const sortedSymbol_t *sortedSymbols, const U32 sortedListSize, U32 nbBitsBaseline, U16 baseSeq)
+{
+	HUF_DEltX4 DElt;
+	U32 rankVal[HUF_TABLELOG_MAX + 1];
+
+	/* get pre-calculated rankVal */
+	memcpy(rankVal, rankValOrigin, sizeof(rankVal));
+
+	/* fill skipped values */
+	if (minWeight > 1) {
+		U32 i, skipSize = rankVal[minWeight];
+		ZSTD_writeLE16(&(DElt.sequence), baseSeq);
+		DElt.nbBits = (BYTE)(consumed);
+		DElt.length = 1;
+		for (i = 0; i < skipSize; i++)
+			DTable[i] = DElt;
+	}
+
+	/* fill DTable */
+	{
+		U32 s;
+		for (s = 0; s < sortedListSize; s++) { /* note : sortedSymbols already skipped */
+			const U32 symbol = sortedSymbols[s].symbol;
+			const U32 weight = sortedSymbols[s].weight;
+			const U32 nbBits = nbBitsBaseline - weight;
+			const U32 length = 1 << (sizeLog - nbBits);
+			const U32 start = rankVal[weight];
+			U32 i = start;
+			const U32 end = start + length;
+
+			ZSTD_writeLE16(&(DElt.sequence), (U16)(baseSeq + (symbol << 8)));
+			DElt.nbBits = (BYTE)(nbBits + consumed);
+			DElt.length = 2;
+			do {
+				DTable[i++] = DElt;
+			} while (i < end); /* since length >= 1 */
+
+			rankVal[weight] += length;
+		}
+	}
+}
+
+typedef U32 rankVal_t[HUF_TABLELOG_MAX][HUF_TABLELOG_MAX + 1];
+typedef U32 rankValCol_t[HUF_TABLELOG_MAX + 1];
+
+static void HUF_fillDTableX4(HUF_DEltX4 *DTable, const U32 targetLog, const sortedSymbol_t *sortedList, const U32 sortedListSize, const U32 *rankStart,
+			     rankVal_t rankValOrigin, const U32 maxWeight, const U32 nbBitsBaseline)
+{
+	U32 rankVal[HUF_TABLELOG_MAX + 1];
+	const int scaleLog = nbBitsBaseline - targetLog; /* note : targetLog >= srcLog, hence scaleLog <= 1 */
+	const U32 minBits = nbBitsBaseline - maxWeight;
+	U32 s;
+
+	memcpy(rankVal, rankValOrigin, sizeof(rankVal));
+
+	/* fill DTable */
+	for (s = 0; s < sortedListSize; s++) {
+		const U16 symbol = sortedList[s].symbol;
+		const U32 weight = sortedList[s].weight;
+		const U32 nbBits = nbBitsBaseline - weight;
+		const U32 start = rankVal[weight];
+		const U32 length = 1 << (targetLog - nbBits);
+
+		if (targetLog - nbBits >= minBits) { /* enough room for a second symbol */
+			U32 sortedRank;
+			int minWeight = nbBits + scaleLog;
+			if (minWeight < 1)
+				minWeight = 1;
+			sortedRank = rankStart[minWeight];
+			HUF_fillDTableX4Level2(DTable + start, targetLog - nbBits, nbBits, rankValOrigin[nbBits], minWeight, sortedList + sortedRank,
+					       sortedListSize - sortedRank, nbBitsBaseline, symbol);
+		} else {
+			HUF_DEltX4 DElt;
+			ZSTD_writeLE16(&(DElt.sequence), symbol);
+			DElt.nbBits = (BYTE)(nbBits);
+			DElt.length = 1;
+			{
+				U32 const end = start + length;
+				U32 u;
+				for (u = start; u < end; u++)
+					DTable[u] = DElt;
+			}
+		}
+		rankVal[weight] += length;
+	}
+}
+
+size_t HUF_readDTableX4_wksp(HUF_DTable *DTable, const void *src, size_t srcSize, void *workspace, size_t workspaceSize)
+{
+	U32 tableLog, maxW, sizeOfSort, nbSymbols;
+	DTableDesc dtd = HUF_getDTableDesc(DTable);
+	U32 const maxTableLog = dtd.maxTableLog;
+	size_t iSize;
+	void *dtPtr = DTable + 1; /* force compiler to avoid strict-aliasing */
+	HUF_DEltX4 *const dt = (HUF_DEltX4 *)dtPtr;
+	U32 *rankStart;
+
+	rankValCol_t *rankVal;
+	U32 *rankStats;
+	U32 *rankStart0;
+	sortedSymbol_t *sortedSymbol;
+	BYTE *weightList;
+	size_t spaceUsed32 = 0;
+
+	HUF_STATIC_ASSERT((sizeof(rankValCol_t) & 3) == 0);
+
+	rankVal = (rankValCol_t *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += (sizeof(rankValCol_t) * HUF_TABLELOG_MAX) >> 2;
+	rankStats = (U32 *)workspace + spaceUsed32;
+	spaceUsed32 += HUF_TABLELOG_MAX + 1;
+	rankStart0 = (U32 *)workspace + spaceUsed32;
+	spaceUsed32 += HUF_TABLELOG_MAX + 2;
+	sortedSymbol = (sortedSymbol_t *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(sizeof(sortedSymbol_t) * (HUF_SYMBOLVALUE_MAX + 1), sizeof(U32)) >> 2;
+	weightList = (BYTE *)((U32 *)workspace + spaceUsed32);
+	spaceUsed32 += ALIGN(HUF_SYMBOLVALUE_MAX + 1, sizeof(U32)) >> 2;
+
+	if ((spaceUsed32 << 2) > workspaceSize)
+		return ERROR(tableLog_tooLarge);
+	workspace = (U32 *)workspace + spaceUsed32;
+	workspaceSize -= (spaceUsed32 << 2);
+
+	rankStart = rankStart0 + 1;
+	memset(rankStats, 0, sizeof(U32) * (2 * HUF_TABLELOG_MAX + 2 + 1));
+
+	HUF_STATIC_ASSERT(sizeof(HUF_DEltX4) == sizeof(HUF_DTable)); /* if compiler fails here, assertion is wrong */
+	if (maxTableLog > HUF_TABLELOG_MAX)
+		return ERROR(tableLog_tooLarge);
+	/* memset(weightList, 0, sizeof(weightList)); */ /* is not necessary, even though some analyzer complain ... */
+
+	iSize = HUF_readStats_wksp(weightList, HUF_SYMBOLVALUE_MAX + 1, rankStats, &nbSymbols, &tableLog, src, srcSize, workspace, workspaceSize);
+	if (HUF_isError(iSize))
+		return iSize;
+
+	/* check result */
+	if (tableLog > maxTableLog)
+		return ERROR(tableLog_tooLarge); /* DTable can't fit code depth */
+
+	/* find maxWeight */
+	for (maxW = tableLog; rankStats[maxW] == 0; maxW--) {
+	} /* necessarily finds a solution before 0 */
+
+	/* Get start index of each weight */
+	{
+		U32 w, nextRankStart = 0;
+		for (w = 1; w < maxW + 1; w++) {
+			U32 curr = nextRankStart;
+			nextRankStart += rankStats[w];
+			rankStart[w] = curr;
+		}
+		rankStart[0] = nextRankStart; /* put all 0w symbols at the end of sorted list*/
+		sizeOfSort = nextRankStart;
+	}
+
+	/* sort symbols by weight */
+	{
+		U32 s;
+		for (s = 0; s < nbSymbols; s++) {
+			U32 const w = weightList[s];
+			U32 const r = rankStart[w]++;
+			sortedSymbol[r].symbol = (BYTE)s;
+			sortedSymbol[r].weight = (BYTE)w;
+		}
+		rankStart[0] = 0; /* forget 0w symbols; this is beginning of weight(1) */
+	}
+
+	/* Build rankVal */
+	{
+		U32 *const rankVal0 = rankVal[0];
+		{
+			int const rescale = (maxTableLog - tableLog) - 1; /* tableLog <= maxTableLog */
+			U32 nextRankVal = 0;
+			U32 w;
+			for (w = 1; w < maxW + 1; w++) {
+				U32 curr = nextRankVal;
+				nextRankVal += rankStats[w] << (w + rescale);
+				rankVal0[w] = curr;
+			}
+		}
+		{
+			U32 const minBits = tableLog + 1 - maxW;
+			U32 consumed;
+			for (consumed = minBits; consumed < maxTableLog - minBits + 1; consumed++) {
+				U32 *const rankValPtr = rankVal[consumed];
+				U32 w;
+				for (w = 1; w < maxW + 1; w++) {
+					rankValPtr[w] = rankVal0[w] >> consumed;
+				}
+			}
+		}
+	}
+
+	HUF_fillDTableX4(dt, maxTableLog, sortedSymbol, sizeOfSort, rankStart0, rankVal, maxW, tableLog + 1);
+
+	dtd.tableLog = (BYTE)maxTableLog;
+	dtd.tableType = 1;
+	memcpy(DTable, &dtd, sizeof(dtd));
+	return iSize;
+}
+
+static U32 HUF_decodeSymbolX4(void *op, BIT_DStream_t *DStream, const HUF_DEltX4 *dt, const U32 dtLog)
+{
+	size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
+	memcpy(op, dt + val, 2);
+	BIT_skipBits(DStream, dt[val].nbBits);
+	return dt[val].length;
+}
+
+static U32 HUF_decodeLastSymbolX4(void *op, BIT_DStream_t *DStream, const HUF_DEltX4 *dt, const U32 dtLog)
+{
+	size_t const val = BIT_lookBitsFast(DStream, dtLog); /* note : dtLog >= 1 */
+	memcpy(op, dt + val, 1);
+	if (dt[val].length == 1)
+		BIT_skipBits(DStream, dt[val].nbBits);
+	else {
+		if (DStream->bitsConsumed < (sizeof(DStream->bitContainer) * 8)) {
+			BIT_skipBits(DStream, dt[val].nbBits);
+			if (DStream->bitsConsumed > (sizeof(DStream->bitContainer) * 8))
+				/* ugly hack; works only because it's the last symbol. Note : can't easily extract nbBits from just this symbol */
+				DStream->bitsConsumed = (sizeof(DStream->bitContainer) * 8);
+		}
+	}
+	return 1;
+}
+
+#define HUF_DECODE_SYMBOLX4_0(ptr, DStreamPtr) ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX4_1(ptr, DStreamPtr)         \
+	if (ZSTD_64bits() || (HUF_TABLELOG_MAX <= 12)) \
+	ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
+
+#define HUF_DECODE_SYMBOLX4_2(ptr, DStreamPtr) \
+	if (ZSTD_64bits())                     \
+	ptr += HUF_decodeSymbolX4(ptr, DStreamPtr, dt, dtLog)
+
+FORCE_INLINE size_t HUF_decodeStreamX4(BYTE *p, BIT_DStream_t *bitDPtr, BYTE *const pEnd, const HUF_DEltX4 *const dt, const U32 dtLog)
+{
+	BYTE *const pStart = p;
+
+	/* up to 8 symbols at a time */
+	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p < pEnd - (sizeof(bitDPtr->bitContainer) - 1))) {
+		HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
+		HUF_DECODE_SYMBOLX4_1(p, bitDPtr);
+		HUF_DECODE_SYMBOLX4_2(p, bitDPtr);
+		HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
+	}
+
+	/* closer to end : up to 2 symbols at a time */
+	while ((BIT_reloadDStream(bitDPtr) == BIT_DStream_unfinished) & (p <= pEnd - 2))
+		HUF_DECODE_SYMBOLX4_0(p, bitDPtr);
+
+	while (p <= pEnd - 2)
+		HUF_DECODE_SYMBOLX4_0(p, bitDPtr); /* no need to reload : reached the end of DStream */
+
+	if (p < pEnd)
+		p += HUF_decodeLastSymbolX4(p, bitDPtr, dt, dtLog);
+
+	return p - pStart;
+}
+
+static size_t HUF_decompress1X4_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	BIT_DStream_t bitD;
+
+	/* Init */
+	{
+		size_t const errorCode = BIT_initDStream(&bitD, cSrc, cSrcSize);
+		if (HUF_isError(errorCode))
+			return errorCode;
+	}
+
+	/* decode */
+	{
+		BYTE *const ostart = (BYTE *)dst;
+		BYTE *const oend = ostart + dstSize;
+		const void *const dtPtr = DTable + 1; /* force compiler to not use strict-aliasing */
+		const HUF_DEltX4 *const dt = (const HUF_DEltX4 *)dtPtr;
+		DTableDesc const dtd = HUF_getDTableDesc(DTable);
+		HUF_decodeStreamX4(ostart, &bitD, oend, dt, dtd.tableLog);
+	}
+
+	/* check */
+	if (!BIT_endOfDStream(&bitD))
+		return ERROR(corruption_detected);
+
+	/* decoded size */
+	return dstSize;
+}
+
+size_t HUF_decompress1X4_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	DTableDesc dtd = HUF_getDTableDesc(DTable);
+	if (dtd.tableType != 1)
+		return ERROR(GENERIC);
+	return HUF_decompress1X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
+}
+
+size_t HUF_decompress1X4_DCtx_wksp(HUF_DTable *DCtx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
+{
+	const BYTE *ip = (const BYTE *)cSrc;
+
+	size_t const hSize = HUF_readDTableX4_wksp(DCtx, cSrc, cSrcSize, workspace, workspaceSize);
+	if (HUF_isError(hSize))
+		return hSize;
+	if (hSize >= cSrcSize)
+		return ERROR(srcSize_wrong);
+	ip += hSize;
+	cSrcSize -= hSize;
+
+	return HUF_decompress1X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, DCtx);
+}
+
+static size_t HUF_decompress4X4_usingDTable_internal(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	if (cSrcSize < 10)
+		return ERROR(corruption_detected); /* strict minimum : jump table + 1 byte per stream */
+
+	{
+		const BYTE *const istart = (const BYTE *)cSrc;
+		BYTE *const ostart = (BYTE *)dst;
+		BYTE *const oend = ostart + dstSize;
+		const void *const dtPtr = DTable + 1;
+		const HUF_DEltX4 *const dt = (const HUF_DEltX4 *)dtPtr;
+
+		/* Init */
+		BIT_DStream_t bitD1;
+		BIT_DStream_t bitD2;
+		BIT_DStream_t bitD3;
+		BIT_DStream_t bitD4;
+		size_t const length1 = ZSTD_readLE16(istart);
+		size_t const length2 = ZSTD_readLE16(istart + 2);
+		size_t const length3 = ZSTD_readLE16(istart + 4);
+		size_t const length4 = cSrcSize - (length1 + length2 + length3 + 6);
+		const BYTE *const istart1 = istart + 6; /* jumpTable */
+		const BYTE *const istart2 = istart1 + length1;
+		const BYTE *const istart3 = istart2 + length2;
+		const BYTE *const istart4 = istart3 + length3;
+		size_t const segmentSize = (dstSize + 3) / 4;
+		BYTE *const opStart2 = ostart + segmentSize;
+		BYTE *const opStart3 = opStart2 + segmentSize;
+		BYTE *const opStart4 = opStart3 + segmentSize;
+		BYTE *op1 = ostart;
+		BYTE *op2 = opStart2;
+		BYTE *op3 = opStart3;
+		BYTE *op4 = opStart4;
+		U32 endSignal;
+		DTableDesc const dtd = HUF_getDTableDesc(DTable);
+		U32 const dtLog = dtd.tableLog;
+
+		if (length4 > cSrcSize)
+			return ERROR(corruption_detected); /* overflow */
+		{
+			size_t const errorCode = BIT_initDStream(&bitD1, istart1, length1);
+			if (HUF_isError(errorCode))
+				return errorCode;
+		}
+		{
+			size_t const errorCode = BIT_initDStream(&bitD2, istart2, length2);
+			if (HUF_isError(errorCode))
+				return errorCode;
+		}
+		{
+			size_t const errorCode = BIT_initDStream(&bitD3, istart3, length3);
+			if (HUF_isError(errorCode))
+				return errorCode;
+		}
+		{
+			size_t const errorCode = BIT_initDStream(&bitD4, istart4, length4);
+			if (HUF_isError(errorCode))
+				return errorCode;
+		}
+
+		/* 16-32 symbols per loop (4-8 symbols per stream) */
+		endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
+		for (; (endSignal == BIT_DStream_unfinished) & (op4 < (oend - (sizeof(bitD4.bitContainer) - 1)));) {
+			HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
+			HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
+			HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
+			HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
+			HUF_DECODE_SYMBOLX4_1(op1, &bitD1);
+			HUF_DECODE_SYMBOLX4_1(op2, &bitD2);
+			HUF_DECODE_SYMBOLX4_1(op3, &bitD3);
+			HUF_DECODE_SYMBOLX4_1(op4, &bitD4);
+			HUF_DECODE_SYMBOLX4_2(op1, &bitD1);
+			HUF_DECODE_SYMBOLX4_2(op2, &bitD2);
+			HUF_DECODE_SYMBOLX4_2(op3, &bitD3);
+			HUF_DECODE_SYMBOLX4_2(op4, &bitD4);
+			HUF_DECODE_SYMBOLX4_0(op1, &bitD1);
+			HUF_DECODE_SYMBOLX4_0(op2, &bitD2);
+			HUF_DECODE_SYMBOLX4_0(op3, &bitD3);
+			HUF_DECODE_SYMBOLX4_0(op4, &bitD4);
+
+			endSignal = BIT_reloadDStream(&bitD1) | BIT_reloadDStream(&bitD2) | BIT_reloadDStream(&bitD3) | BIT_reloadDStream(&bitD4);
+		}
+
+		/* check corruption */
+		if (op1 > opStart2)
+			return ERROR(corruption_detected);
+		if (op2 > opStart3)
+			return ERROR(corruption_detected);
+		if (op3 > opStart4)
+			return ERROR(corruption_detected);
+		/* note : op4 already verified within main loop */
+
+		/* finish bitStreams one by one */
+		HUF_decodeStreamX4(op1, &bitD1, opStart2, dt, dtLog);
+		HUF_decodeStreamX4(op2, &bitD2, opStart3, dt, dtLog);
+		HUF_decodeStreamX4(op3, &bitD3, opStart4, dt, dtLog);
+		HUF_decodeStreamX4(op4, &bitD4, oend, dt, dtLog);
+
+		/* check */
+		{
+			U32 const endCheck = BIT_endOfDStream(&bitD1) & BIT_endOfDStream(&bitD2) & BIT_endOfDStream(&bitD3) & BIT_endOfDStream(&bitD4);
+			if (!endCheck)
+				return ERROR(corruption_detected);
+		}
+
+		/* decoded size */
+		return dstSize;
+	}
+}
+
+size_t HUF_decompress4X4_usingDTable(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	DTableDesc dtd = HUF_getDTableDesc(DTable);
+	if (dtd.tableType != 1)
+		return ERROR(GENERIC);
+	return HUF_decompress4X4_usingDTable_internal(dst, dstSize, cSrc, cSrcSize, DTable);
+}
+
+size_t HUF_decompress4X4_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
+{
+	const BYTE *ip = (const BYTE *)cSrc;
+
+	size_t hSize = HUF_readDTableX4_wksp(dctx, cSrc, cSrcSize, workspace, workspaceSize);
+	if (HUF_isError(hSize))
+		return hSize;
+	if (hSize >= cSrcSize)
+		return ERROR(srcSize_wrong);
+	ip += hSize;
+	cSrcSize -= hSize;
+
+	return HUF_decompress4X4_usingDTable_internal(dst, dstSize, ip, cSrcSize, dctx);
+}
+
+/* ********************************/
+/* Generic decompression selector */
+/* ********************************/
+
+size_t HUF_decompress1X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	DTableDesc const dtd = HUF_getDTableDesc(DTable);
+	return dtd.tableType ? HUF_decompress1X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable)
+			     : HUF_decompress1X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable);
+}
+
+size_t HUF_decompress4X_usingDTable(void *dst, size_t maxDstSize, const void *cSrc, size_t cSrcSize, const HUF_DTable *DTable)
+{
+	DTableDesc const dtd = HUF_getDTableDesc(DTable);
+	return dtd.tableType ? HUF_decompress4X4_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable)
+			     : HUF_decompress4X2_usingDTable_internal(dst, maxDstSize, cSrc, cSrcSize, DTable);
+}
+
+typedef struct {
+	U32 tableTime;
+	U32 decode256Time;
+} algo_time_t;
+static const algo_time_t algoTime[16 /* Quantization */][3 /* single, double, quad */] = {
+    /* single, double, quad */
+    {{0, 0}, {1, 1}, {2, 2}},		     /* Q==0 : impossible */
+    {{0, 0}, {1, 1}, {2, 2}},		     /* Q==1 : impossible */
+    {{38, 130}, {1313, 74}, {2151, 38}},     /* Q == 2 : 12-18% */
+    {{448, 128}, {1353, 74}, {2238, 41}},    /* Q == 3 : 18-25% */
+    {{556, 128}, {1353, 74}, {2238, 47}},    /* Q == 4 : 25-32% */
+    {{714, 128}, {1418, 74}, {2436, 53}},    /* Q == 5 : 32-38% */
+    {{883, 128}, {1437, 74}, {2464, 61}},    /* Q == 6 : 38-44% */
+    {{897, 128}, {1515, 75}, {2622, 68}},    /* Q == 7 : 44-50% */
+    {{926, 128}, {1613, 75}, {2730, 75}},    /* Q == 8 : 50-56% */
+    {{947, 128}, {1729, 77}, {3359, 77}},    /* Q == 9 : 56-62% */
+    {{1107, 128}, {2083, 81}, {4006, 84}},   /* Q ==10 : 62-69% */
+    {{1177, 128}, {2379, 87}, {4785, 88}},   /* Q ==11 : 69-75% */
+    {{1242, 128}, {2415, 93}, {5155, 84}},   /* Q ==12 : 75-81% */
+    {{1349, 128}, {2644, 106}, {5260, 106}}, /* Q ==13 : 81-87% */
+    {{1455, 128}, {2422, 124}, {4174, 124}}, /* Q ==14 : 87-93% */
+    {{722, 128}, {1891, 145}, {1936, 146}},  /* Q ==15 : 93-99% */
+};
+
+/** HUF_selectDecoder() :
+*   Tells which decoder is likely to decode faster,
+*   based on a set of pre-determined metrics.
+*   @return : 0==HUF_decompress4X2, 1==HUF_decompress4X4 .
+*   Assumption : 0 < cSrcSize < dstSize <= 128 KB */
+U32 HUF_selectDecoder(size_t dstSize, size_t cSrcSize)
+{
+	/* decoder timing evaluation */
+	U32 const Q = (U32)(cSrcSize * 16 / dstSize); /* Q < 16 since dstSize > cSrcSize */
+	U32 const D256 = (U32)(dstSize >> 8);
+	U32 const DTime0 = algoTime[Q][0].tableTime + (algoTime[Q][0].decode256Time * D256);
+	U32 DTime1 = algoTime[Q][1].tableTime + (algoTime[Q][1].decode256Time * D256);
+	DTime1 += DTime1 >> 3; /* advantage to algorithm using less memory, for cache eviction */
+
+	return DTime1 < DTime0;
+}
+
+typedef size_t (*decompressionAlgo)(void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize);
+
+size_t HUF_decompress4X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
+{
+	/* validation checks */
+	if (dstSize == 0)
+		return ERROR(dstSize_tooSmall);
+	if (cSrcSize > dstSize)
+		return ERROR(corruption_detected); /* invalid */
+	if (cSrcSize == dstSize) {
+		memcpy(dst, cSrc, dstSize);
+		return dstSize;
+	} /* not compressed */
+	if (cSrcSize == 1) {
+		memset(dst, *(const BYTE *)cSrc, dstSize);
+		return dstSize;
+	} /* RLE */
+
+	{
+		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+		return algoNb ? HUF_decompress4X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
+			      : HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
+	}
+}
+
+size_t HUF_decompress4X_hufOnly_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
+{
+	/* validation checks */
+	if (dstSize == 0)
+		return ERROR(dstSize_tooSmall);
+	if ((cSrcSize >= dstSize) || (cSrcSize <= 1))
+		return ERROR(corruption_detected); /* invalid */
+
+	{
+		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+		return algoNb ? HUF_decompress4X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
+			      : HUF_decompress4X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
+	}
+}
+
+size_t HUF_decompress1X_DCtx_wksp(HUF_DTable *dctx, void *dst, size_t dstSize, const void *cSrc, size_t cSrcSize, void *workspace, size_t workspaceSize)
+{
+	/* validation checks */
+	if (dstSize == 0)
+		return ERROR(dstSize_tooSmall);
+	if (cSrcSize > dstSize)
+		return ERROR(corruption_detected); /* invalid */
+	if (cSrcSize == dstSize) {
+		memcpy(dst, cSrc, dstSize);
+		return dstSize;
+	} /* not compressed */
+	if (cSrcSize == 1) {
+		memset(dst, *(const BYTE *)cSrc, dstSize);
+		return dstSize;
+	} /* RLE */
+
+	{
+		U32 const algoNb = HUF_selectDecoder(dstSize, cSrcSize);
+		return algoNb ? HUF_decompress1X4_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize)
+			      : HUF_decompress1X2_DCtx_wksp(dctx, dst, dstSize, cSrc, cSrcSize, workspace, workspaceSize);
+	}
+}
diff --git a/lib/zstd/mem.h b/lib/zstd/mem.h
new file mode 100644
index 000000000000..3a0f34c8706c
--- /dev/null
+++ b/lib/zstd/mem.h
@@ -0,0 +1,151 @@
+/**
+ * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd.
+ * An additional grant of patent rights can be found in the PATENTS file in the
+ * same directory.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ */
+
+#ifndef MEM_H_MODULE
+#define MEM_H_MODULE
+
+/*-****************************************
+*  Dependencies
+******************************************/
+#include <asm/unaligned.h>
+#include <linux/string.h> /* memcpy */
+#include <linux/types.h>  /* size_t, ptrdiff_t */
+
+/*-****************************************
+*  Compiler specifics
+******************************************/
+#define ZSTD_STATIC static __inline __attribute__((unused))
+
+/*-**************************************************************
+*  Basic Types
+*****************************************************************/
+typedef uint8_t BYTE;
+typedef uint16_t U16;
+typedef int16_t S16;
+typedef uint32_t U32;
+typedef int32_t S32;
+typedef uint64_t U64;
+typedef int64_t S64;
+typedef ptrdiff_t iPtrDiff;
+typedef uintptr_t uPtrDiff;
+
+/*-**************************************************************
+*  Memory I/O
+*****************************************************************/
+ZSTD_STATIC unsigned ZSTD_32bits(void) { return sizeof(size_t) == 4; }
+ZSTD_STATIC unsigned ZSTD_64bits(void) { return sizeof(size_t) == 8; }
+
+#if defined(__LITTLE_ENDIAN)
+#define ZSTD_LITTLE_ENDIAN 1
+#else
+#define ZSTD_LITTLE_ENDIAN 0
+#endif
+
+ZSTD_STATIC unsigned ZSTD_isLittleEndian(void) { return ZSTD_LITTLE_ENDIAN; }
+
+ZSTD_STATIC U16 ZSTD_read16(const void *memPtr) { return get_unaligned((const U16 *)memPtr); }
+
+ZSTD_STATIC U32 ZSTD_read32(const void *memPtr) { return get_unaligned((const U32 *)memPtr); }
+
+ZSTD_STATIC U64 ZSTD_read64(const void *memPtr) { return get_unaligned((const U64 *)memPtr); }
+
+ZSTD_STATIC size_t ZSTD_readST(const void *memPtr) { return get_unaligned((const size_t *)memPtr); }
+
+ZSTD_STATIC void ZSTD_write16(void *memPtr, U16 value) { put_unaligned(value, (U16 *)memPtr); }
+
+ZSTD_STATIC void ZSTD_write32(void *memPtr, U32 value) { put_unaligned(value, (U32 *)memPtr); }
+
+ZSTD_STATIC void ZSTD_write64(void *memPtr, U64 value) { put_unaligned(value, (U64 *)memPtr); }
+
+/*=== Little endian r/w ===*/
+
+ZSTD_STATIC U16 ZSTD_readLE16(const void *memPtr) { return get_unaligned_le16(memPtr); }
+
+ZSTD_STATIC void ZSTD_writeLE16(void *memPtr, U16 val) { put_unaligned_le16(val, memPtr); }
+
+ZSTD_STATIC U32 ZSTD_readLE24(const void *memPtr) { return ZSTD_readLE16(memPtr) + (((const BYTE *)memPtr)[2] << 16); }
+
+ZSTD_STATIC void ZSTD_writeLE24(void *memPtr, U32 val)
+{
+	ZSTD_writeLE16(memPtr, (U16)val);
+	((BYTE *)memPtr)[2] = (BYTE)(val >> 16);
+}
+
+ZSTD_STATIC U32 ZSTD_readLE32(const void *memPtr) { return get_unaligned_le32(memPtr); }
+
+ZSTD_STATIC void ZSTD_writeLE32(void *memPtr, U32 val32) { put_unaligned_le32(val32, memPtr); }
+
+ZSTD_STATIC U64 ZSTD_readLE64(const void *memPtr) { return get_unaligned_le64(memPtr); }
+
+ZSTD_STATIC void ZSTD_writeLE64(void *memPtr, U64 val64) { put_unaligned_le64(val64, memPtr); }
+
+ZSTD_STATIC size_t ZSTD_readLEST(const void *memPtr)
+{
+	if (ZSTD_32bits())
+		return (size_t)ZSTD_readLE32(memPtr);
+	else
+		return (size_t)ZSTD_readLE64(memPtr);
+}
+
+ZSTD_STATIC void ZSTD_writeLEST(void *memPtr, size_t val)
+{
+	if (ZSTD_32bits())
+		ZSTD_writeLE32(memPtr, (U32)val);
+	else
+		ZSTD_writeLE64(memPtr, (U64)val);
+}
+
+/*=== Big endian r/w ===*/
+
+ZSTD_STATIC U32 ZSTD_readBE32(const void *memPtr) { return get_unaligned_be32(memPtr); }
+
+ZSTD_STATIC void ZSTD_writeBE32(void *memPtr, U32 val32) { put_unaligned_be32(val32, memPtr); }
+
+ZSTD_STATIC U64 ZSTD_readBE64(const void *memPtr) { return get_unaligned_be64(memPtr); }
+
+ZSTD_STATIC void ZSTD_writeBE64(void *memPtr, U64 val64) { put_unaligned_be64(val64, memPtr); }
+
+ZSTD_STATIC size_t ZSTD_readBEST(const void *memPtr)
+{
+	if (ZSTD_32bits())
+		return (size_t)ZSTD_readBE32(memPtr);
+	else
+		return (size_t)ZSTD_readBE64(memPtr);
+}
+
+ZSTD_STATIC void ZSTD_writeBEST(void *memPtr, size_t val)
+{
+	if (ZSTD_32bits())
+		ZSTD_writeBE32(memPtr, (U32)val);
+	else
+		ZSTD_writeBE64(memPtr, (U64)val);
+}
+
+/* function safe only for comparisons */
+ZSTD_STATIC U32 ZSTD_readMINMATCH(const void *memPtr, U32 length)
+{
+	switch (length) {
+	default:
+	case 4: return ZSTD_read32(memPtr);
+	case 3:
+		if (ZSTD_isLittleEndian())
+			return ZSTD_read32(memPtr) << 8;
+		else
+			return ZSTD_read32(memPtr) >> 8;
+	}
+}
+
+#endif /* MEM_H_MODULE */
diff --git a/lib/zstd/zstd_common.c b/lib/zstd/zstd_common.c
new file mode 100644
index 000000000000..a282624ee155
--- /dev/null
+++ b/lib/zstd/zstd_common.c
@@ -0,0 +1,75 @@
+/**
+ * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd.
+ * An additional grant of patent rights can be found in the PATENTS file in the
+ * same directory.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ */
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#include "error_private.h"
+#include "zstd_internal.h" /* declaration of ZSTD_isError, ZSTD_getErrorName, ZSTD_getErrorCode, ZSTD_getErrorString, ZSTD_versionNumber */
+#include <linux/kernel.h>
+
+/*=**************************************************************
+*  Custom allocator
+****************************************************************/
+
+#define stack_push(stack, size)                                 \
+	({                                                      \
+		void *const ptr = ZSTD_PTR_ALIGN((stack)->ptr); \
+		(stack)->ptr = (char *)ptr + (size);            \
+		(stack)->ptr <= (stack)->end ? ptr : NULL;      \
+	})
+
+ZSTD_customMem ZSTD_initStack(void *workspace, size_t workspaceSize)
+{
+	ZSTD_customMem stackMem = {ZSTD_stackAlloc, ZSTD_stackFree, workspace};
+	ZSTD_stack *stack = (ZSTD_stack *)workspace;
+	/* Verify preconditions */
+	if (!workspace || workspaceSize < sizeof(ZSTD_stack) || workspace != ZSTD_PTR_ALIGN(workspace)) {
+		ZSTD_customMem error = {NULL, NULL, NULL};
+		return error;
+	}
+	/* Initialize the stack */
+	stack->ptr = workspace;
+	stack->end = (char *)workspace + workspaceSize;
+	stack_push(stack, sizeof(ZSTD_stack));
+	return stackMem;
+}
+
+void *ZSTD_stackAllocAll(void *opaque, size_t *size)
+{
+	ZSTD_stack *stack = (ZSTD_stack *)opaque;
+	*size = (BYTE const *)stack->end - (BYTE *)ZSTD_PTR_ALIGN(stack->ptr);
+	return stack_push(stack, *size);
+}
+
+void *ZSTD_stackAlloc(void *opaque, size_t size)
+{
+	ZSTD_stack *stack = (ZSTD_stack *)opaque;
+	return stack_push(stack, size);
+}
+void ZSTD_stackFree(void *opaque, void *address)
+{
+	(void)opaque;
+	(void)address;
+}
+
+void *ZSTD_malloc(size_t size, ZSTD_customMem customMem) { return customMem.customAlloc(customMem.opaque, size); }
+
+void ZSTD_free(void *ptr, ZSTD_customMem customMem)
+{
+	if (ptr != NULL)
+		customMem.customFree(customMem.opaque, ptr);
+}
diff --git a/lib/zstd/zstd_internal.h b/lib/zstd/zstd_internal.h
new file mode 100644
index 000000000000..1a79fab9e13a
--- /dev/null
+++ b/lib/zstd/zstd_internal.h
@@ -0,0 +1,263 @@
+/**
+ * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd.
+ * An additional grant of patent rights can be found in the PATENTS file in the
+ * same directory.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ */
+
+#ifndef ZSTD_CCOMMON_H_MODULE
+#define ZSTD_CCOMMON_H_MODULE
+
+/*-*******************************************************
+*  Compiler specifics
+*********************************************************/
+#define FORCE_INLINE static __always_inline
+#define FORCE_NOINLINE static noinline
+
+/*-*************************************
+*  Dependencies
+***************************************/
+#include "error_private.h"
+#include "mem.h"
+#include <linux/compiler.h>
+#include <linux/kernel.h>
+#include <linux/xxhash.h>
+#include <linux/zstd.h>
+
+/*-*************************************
+*  shared macros
+***************************************/
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+#define CHECK_F(f)                       \
+	{                                \
+		size_t const errcod = f; \
+		if (ERR_isError(errcod)) \
+			return errcod;   \
+	} /* check and Forward error code */
+#define CHECK_E(f, e)                    \
+	{                                \
+		size_t const errcod = f; \
+		if (ERR_isError(errcod)) \
+			return ERROR(e); \
+	} /* check and send Error code */
+#define ZSTD_STATIC_ASSERT(c)                                   \
+	{                                                       \
+		enum { ZSTD_static_assert = 1 / (int)(!!(c)) }; \
+	}
+
+/*-*************************************
+*  Common constants
+***************************************/
+#define ZSTD_OPT_NUM (1 << 12)
+#define ZSTD_DICT_MAGIC 0xEC30A437 /* v0.7+ */
+
+#define ZSTD_REP_NUM 3		      /* number of repcodes */
+#define ZSTD_REP_CHECK (ZSTD_REP_NUM) /* number of repcodes to check by the optimal parser */
+#define ZSTD_REP_MOVE (ZSTD_REP_NUM - 1)
+#define ZSTD_REP_MOVE_OPT (ZSTD_REP_NUM)
+static const U32 repStartValue[ZSTD_REP_NUM] = {1, 4, 8};
+
+#define KB *(1 << 10)
+#define MB *(1 << 20)
+#define GB *(1U << 30)
+
+#define BIT7 128
+#define BIT6 64
+#define BIT5 32
+#define BIT4 16
+#define BIT1 2
+#define BIT0 1
+
+#define ZSTD_WINDOWLOG_ABSOLUTEMIN 10
+static const size_t ZSTD_fcs_fieldSize[4] = {0, 2, 4, 8};
+static const size_t ZSTD_did_fieldSize[4] = {0, 1, 2, 4};
+
+#define ZSTD_BLOCKHEADERSIZE 3 /* C standard doesn't allow `static const` variable to be init using another `static const` variable */
+static const size_t ZSTD_blockHeaderSize = ZSTD_BLOCKHEADERSIZE;
+typedef enum { bt_raw, bt_rle, bt_compressed, bt_reserved } blockType_e;
+
+#define MIN_SEQUENCES_SIZE 1									  /* nbSeq==0 */
+#define MIN_CBLOCK_SIZE (1 /*litCSize*/ + 1 /* RLE or RAW */ + MIN_SEQUENCES_SIZE /* nbSeq==0 */) /* for a non-null block */
+
+#define HufLog 12
+typedef enum { set_basic, set_rle, set_compressed, set_repeat } symbolEncodingType_e;
+
+#define LONGNBSEQ 0x7F00
+
+#define MINMATCH 3
+#define EQUAL_READ32 4
+
+#define Litbits 8
+#define MaxLit ((1 << Litbits) - 1)
+#define MaxML 52
+#define MaxLL 35
+#define MaxOff 28
+#define MaxSeq MAX(MaxLL, MaxML) /* Assumption : MaxOff < MaxLL,MaxML */
+#define MLFSELog 9
+#define LLFSELog 9
+#define OffFSELog 8
+
+static const U32 LL_bits[MaxLL + 1] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
+static const S16 LL_defaultNorm[MaxLL + 1] = {4, 3, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 2, 1, 1, 1, 1, 1, -1, -1, -1, -1};
+#define LL_DEFAULTNORMLOG 6 /* for static allocation */
+static const U32 LL_defaultNormLog = LL_DEFAULTNORMLOG;
+
+static const U32 ML_bits[MaxML + 1] = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,  0,  0,  0,  0,  0,  0, 0,
+				       0, 0, 0, 0, 0, 1, 1, 1, 1, 2, 2, 3, 3, 4, 4, 5, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16};
+static const S16 ML_defaultNorm[MaxML + 1] = {1, 4, 3, 2, 2, 2, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,  1,  1,  1,  1,  1,  1, 1,
+					      1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, -1, -1, -1, -1, -1, -1, -1};
+#define ML_DEFAULTNORMLOG 6 /* for static allocation */
+static const U32 ML_defaultNormLog = ML_DEFAULTNORMLOG;
+
+static const S16 OF_defaultNorm[MaxOff + 1] = {1, 1, 1, 1, 1, 1, 2, 2, 2, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, -1, -1, -1, -1, -1};
+#define OF_DEFAULTNORMLOG 5 /* for static allocation */
+static const U32 OF_defaultNormLog = OF_DEFAULTNORMLOG;
+
+/*-*******************************************
+*  Shared functions to include for inlining
+*********************************************/
+ZSTD_STATIC void ZSTD_copy8(void *dst, const void *src) {
+	memcpy(dst, src, 8);
+}
+/*! ZSTD_wildcopy() :
+*   custom version of memcpy(), can copy up to 7 bytes too many (8 bytes if length==0) */
+#define WILDCOPY_OVERLENGTH 8
+ZSTD_STATIC void ZSTD_wildcopy(void *dst, const void *src, ptrdiff_t length)
+{
+	const BYTE* ip = (const BYTE*)src;
+	BYTE* op = (BYTE*)dst;
+	BYTE* const oend = op + length;
+	/* Work around https://gcc.gnu.org/bugzilla/show_bug.cgi?id=81388.
+	 * Avoid the bad case where the loop only runs once by handling the
+	 * special case separately. This doesn't trigger the bug because it
+	 * doesn't involve pointer/integer overflow.
+	 */
+	if (length <= 8)
+		return ZSTD_copy8(dst, src);
+	do {
+		ZSTD_copy8(op, ip);
+		op += 8;
+		ip += 8;
+	} while (op < oend);
+}
+
+/*-*******************************************
+*  Private interfaces
+*********************************************/
+typedef struct ZSTD_stats_s ZSTD_stats_t;
+
+typedef struct {
+	U32 off;
+	U32 len;
+} ZSTD_match_t;
+
+typedef struct {
+	U32 price;
+	U32 off;
+	U32 mlen;
+	U32 litlen;
+	U32 rep[ZSTD_REP_NUM];
+} ZSTD_optimal_t;
+
+typedef struct seqDef_s {
+	U32 offset;
+	U16 litLength;
+	U16 matchLength;
+} seqDef;
+
+typedef struct {
+	seqDef *sequencesStart;
+	seqDef *sequences;
+	BYTE *litStart;
+	BYTE *lit;
+	BYTE *llCode;
+	BYTE *mlCode;
+	BYTE *ofCode;
+	U32 longLengthID; /* 0 == no longLength; 1 == Lit.longLength; 2 == Match.longLength; */
+	U32 longLengthPos;
+	/* opt */
+	ZSTD_optimal_t *priceTable;
+	ZSTD_match_t *matchTable;
+	U32 *matchLengthFreq;
+	U32 *litLengthFreq;
+	U32 *litFreq;
+	U32 *offCodeFreq;
+	U32 matchLengthSum;
+	U32 matchSum;
+	U32 litLengthSum;
+	U32 litSum;
+	U32 offCodeSum;
+	U32 log2matchLengthSum;
+	U32 log2matchSum;
+	U32 log2litLengthSum;
+	U32 log2litSum;
+	U32 log2offCodeSum;
+	U32 factor;
+	U32 staticPrices;
+	U32 cachedPrice;
+	U32 cachedLitLength;
+	const BYTE *cachedLiterals;
+} seqStore_t;
+
+const seqStore_t *ZSTD_getSeqStore(const ZSTD_CCtx *ctx);
+void ZSTD_seqToCodes(const seqStore_t *seqStorePtr);
+int ZSTD_isSkipFrame(ZSTD_DCtx *dctx);
+
+/*= Custom memory allocation functions */
+typedef void *(*ZSTD_allocFunction)(void *opaque, size_t size);
+typedef void (*ZSTD_freeFunction)(void *opaque, void *address);
+typedef struct {
+	ZSTD_allocFunction customAlloc;
+	ZSTD_freeFunction customFree;
+	void *opaque;
+} ZSTD_customMem;
+
+void *ZSTD_malloc(size_t size, ZSTD_customMem customMem);
+void ZSTD_free(void *ptr, ZSTD_customMem customMem);
+
+/*====== stack allocation  ======*/
+
+typedef struct {
+	void *ptr;
+	const void *end;
+} ZSTD_stack;
+
+#define ZSTD_ALIGN(x) ALIGN(x, sizeof(size_t))
+#define ZSTD_PTR_ALIGN(p) PTR_ALIGN(p, sizeof(size_t))
+
+ZSTD_customMem ZSTD_initStack(void *workspace, size_t workspaceSize);
+
+void *ZSTD_stackAllocAll(void *opaque, size_t *size);
+void *ZSTD_stackAlloc(void *opaque, size_t size);
+void ZSTD_stackFree(void *opaque, void *address);
+
+/*======  common function  ======*/
+
+ZSTD_STATIC U32 ZSTD_highbit32(U32 val) { return 31 - __builtin_clz(val); }
+
+/* hidden functions */
+
+/* ZSTD_invalidateRepCodes() :
+ * ensures next compression will not use repcodes from previous block.
+ * Note : only works with regular variant;
+ *        do not use with extDict variant ! */
+void ZSTD_invalidateRepCodes(ZSTD_CCtx *cctx);
+
+size_t ZSTD_freeCCtx(ZSTD_CCtx *cctx);
+size_t ZSTD_freeDCtx(ZSTD_DCtx *dctx);
+size_t ZSTD_freeCDict(ZSTD_CDict *cdict);
+size_t ZSTD_freeDDict(ZSTD_DDict *cdict);
+size_t ZSTD_freeCStream(ZSTD_CStream *zcs);
+size_t ZSTD_freeDStream(ZSTD_DStream *zds);
+
+#endif /* ZSTD_CCOMMON_H_MODULE */
diff --git a/lib/zstd/zstd_opt.h b/lib/zstd/zstd_opt.h
new file mode 100644
index 000000000000..55e1b4cba808
--- /dev/null
+++ b/lib/zstd/zstd_opt.h
@@ -0,0 +1,1014 @@
+/**
+ * Copyright (c) 2016-present, Przemyslaw Skibinski, Yann Collet, Facebook, Inc.
+ * All rights reserved.
+ *
+ * This source code is licensed under the BSD-style license found in the
+ * LICENSE file in the root directory of https://github.com/facebook/zstd.
+ * An additional grant of patent rights can be found in the PATENTS file in the
+ * same directory.
+ *
+ * This program is free software; you can redistribute it and/or modify it under
+ * the terms of the GNU General Public License version 2 as published by the
+ * Free Software Foundation. This program is dual-licensed; you may select
+ * either version 2 of the GNU General Public License ("GPL") or BSD license
+ * ("BSD").
+ */
+
+/* Note : this file is intended to be included within zstd_compress.c */
+
+#ifndef ZSTD_OPT_H_91842398743
+#define ZSTD_OPT_H_91842398743
+
+#define ZSTD_LITFREQ_ADD 2
+#define ZSTD_FREQ_DIV 4
+#define ZSTD_MAX_PRICE (1 << 30)
+
+/*-*************************************
+*  Price functions for optimal parser
+***************************************/
+FORCE_INLINE void ZSTD_setLog2Prices(seqStore_t *ssPtr)
+{
+	ssPtr->log2matchLengthSum = ZSTD_highbit32(ssPtr->matchLengthSum + 1);
+	ssPtr->log2litLengthSum = ZSTD_highbit32(ssPtr->litLengthSum + 1);
+	ssPtr->log2litSum = ZSTD_highbit32(ssPtr->litSum + 1);
+	ssPtr->log2offCodeSum = ZSTD_highbit32(ssPtr->offCodeSum + 1);
+	ssPtr->factor = 1 + ((ssPtr->litSum >> 5) / ssPtr->litLengthSum) + ((ssPtr->litSum << 1) / (ssPtr->litSum + ssPtr->matchSum));
+}
+
+ZSTD_STATIC void ZSTD_rescaleFreqs(seqStore_t *ssPtr, const BYTE *src, size_t srcSize)
+{
+	unsigned u;
+
+	ssPtr->cachedLiterals = NULL;
+	ssPtr->cachedPrice = ssPtr->cachedLitLength = 0;
+	ssPtr->staticPrices = 0;
+
+	if (ssPtr->litLengthSum == 0) {
+		if (srcSize <= 1024)
+			ssPtr->staticPrices = 1;
+
+		for (u = 0; u <= MaxLit; u++)
+			ssPtr->litFreq[u] = 0;
+		for (u = 0; u < srcSize; u++)
+			ssPtr->litFreq[src[u]]++;
+
+		ssPtr->litSum = 0;
+		ssPtr->litLengthSum = MaxLL + 1;
+		ssPtr->matchLengthSum = MaxML + 1;
+		ssPtr->offCodeSum = (MaxOff + 1);
+		ssPtr->matchSum = (ZSTD_LITFREQ_ADD << Litbits);
+
+		for (u = 0; u <= MaxLit; u++) {
+			ssPtr->litFreq[u] = 1 + (ssPtr->litFreq[u] >> ZSTD_FREQ_DIV);
+			ssPtr->litSum += ssPtr->litFreq[u];
+		}
+		for (u = 0; u <= MaxLL; u++)
+			ssPtr->litLengthFreq[u] = 1;
+		for (u = 0; u <= MaxML; u++)
+			ssPtr->matchLengthFreq[u] = 1;
+		for (u = 0; u <= MaxOff; u++)
+			ssPtr->offCodeFreq[u] = 1;
+	} else {
+		ssPtr->matchLengthSum = 0;
+		ssPtr->litLengthSum = 0;
+		ssPtr->offCodeSum = 0;
+		ssPtr->matchSum = 0;
+		ssPtr->litSum = 0;
+
+		for (u = 0; u <= MaxLit; u++) {
+			ssPtr->litFreq[u] = 1 + (ssPtr->litFreq[u] >> (ZSTD_FREQ_DIV + 1));
+			ssPtr->litSum += ssPtr->litFreq[u];
+		}
+		for (u = 0; u <= MaxLL; u++) {
+			ssPtr->litLengthFreq[u] = 1 + (ssPtr->litLengthFreq[u] >> (ZSTD_FREQ_DIV + 1));
+			ssPtr->litLengthSum += ssPtr->litLengthFreq[u];
+		}
+		for (u = 0; u <= MaxML; u++) {
+			ssPtr->matchLengthFreq[u] = 1 + (ssPtr->matchLengthFreq[u] >> ZSTD_FREQ_DIV);
+			ssPtr->matchLengthSum += ssPtr->matchLengthFreq[u];
+			ssPtr->matchSum += ssPtr->matchLengthFreq[u] * (u + 3);
+		}
+		ssPtr->matchSum *= ZSTD_LITFREQ_ADD;
+		for (u = 0; u <= MaxOff; u++) {
+			ssPtr->offCodeFreq[u] = 1 + (ssPtr->offCodeFreq[u] >> ZSTD_FREQ_DIV);
+			ssPtr->offCodeSum += ssPtr->offCodeFreq[u];
+		}
+	}
+
+	ZSTD_setLog2Prices(ssPtr);
+}
+
+FORCE_INLINE U32 ZSTD_getLiteralPrice(seqStore_t *ssPtr, U32 litLength, const BYTE *literals)
+{
+	U32 price, u;
+
+	if (ssPtr->staticPrices)
+		return ZSTD_highbit32((U32)litLength + 1) + (litLength * 6);
+
+	if (litLength == 0)
+		return ssPtr->log2litLengthSum - ZSTD_highbit32(ssPtr->litLengthFreq[0] + 1);
+
+	/* literals */
+	if (ssPtr->cachedLiterals == literals) {
+		U32 const additional = litLength - ssPtr->cachedLitLength;
+		const BYTE *literals2 = ssPtr->cachedLiterals + ssPtr->cachedLitLength;
+		price = ssPtr->cachedPrice + additional * ssPtr->log2litSum;
+		for (u = 0; u < additional; u++)
+			price -= ZSTD_highbit32(ssPtr->litFreq[literals2[u]] + 1);
+		ssPtr->cachedPrice = price;
+		ssPtr->cachedLitLength = litLength;
+	} else {
+		price = litLength * ssPtr->log2litSum;
+		for (u = 0; u < litLength; u++)
+			price -= ZSTD_highbit32(ssPtr->litFreq[literals[u]] + 1);
+
+		if (litLength >= 12) {
+			ssPtr->cachedLiterals = literals;
+			ssPtr->cachedPrice = price;
+			ssPtr->cachedLitLength = litLength;
+		}
+	}
+
+	/* literal Length */
+	{
+		const BYTE LL_deltaCode = 19;
+		const BYTE llCode = (litLength > 63) ? (BYTE)ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength];
+		price += LL_bits[llCode] + ssPtr->log2litLengthSum - ZSTD_highbit32(ssPtr->litLengthFreq[llCode] + 1);
+	}
+
+	return price;
+}
+
+FORCE_INLINE U32 ZSTD_getPrice(seqStore_t *seqStorePtr, U32 litLength, const BYTE *literals, U32 offset, U32 matchLength, const int ultra)
+{
+	/* offset */
+	U32 price;
+	BYTE const offCode = (BYTE)ZSTD_highbit32(offset + 1);
+
+	if (seqStorePtr->staticPrices)
+		return ZSTD_getLiteralPrice(seqStorePtr, litLength, literals) + ZSTD_highbit32((U32)matchLength + 1) + 16 + offCode;
+
+	price = offCode + seqStorePtr->log2offCodeSum - ZSTD_highbit32(seqStorePtr->offCodeFreq[offCode] + 1);
+	if (!ultra && offCode >= 20)
+		price += (offCode - 19) * 2;
+
+	/* match Length */
+	{
+		const BYTE ML_deltaCode = 36;
+		const BYTE mlCode = (matchLength > 127) ? (BYTE)ZSTD_highbit32(matchLength) + ML_deltaCode : ML_Code[matchLength];
+		price += ML_bits[mlCode] + seqStorePtr->log2matchLengthSum - ZSTD_highbit32(seqStorePtr->matchLengthFreq[mlCode] + 1);
+	}
+
+	return price + ZSTD_getLiteralPrice(seqStorePtr, litLength, literals) + seqStorePtr->factor;
+}
+
+ZSTD_STATIC void ZSTD_updatePrice(seqStore_t *seqStorePtr, U32 litLength, const BYTE *literals, U32 offset, U32 matchLength)
+{
+	U32 u;
+
+	/* literals */
+	seqStorePtr->litSum += litLength * ZSTD_LITFREQ_ADD;
+	for (u = 0; u < litLength; u++)
+		seqStorePtr->litFreq[literals[u]] += ZSTD_LITFREQ_ADD;
+
+	/* literal Length */
+	{
+		const BYTE LL_deltaCode = 19;
+		const BYTE llCode = (litLength > 63) ? (BYTE)ZSTD_highbit32(litLength) + LL_deltaCode : LL_Code[litLength];
+		seqStorePtr->litLengthFreq[llCode]++;
+		seqStorePtr->litLengthSum++;
+	}
+
+	/* match offset */
+	{
+		BYTE const offCode = (BYTE)ZSTD_highbit32(offset + 1);
+		seqStorePtr->offCodeSum++;
+		seqStorePtr->offCodeFreq[offCode]++;
+	}
+
+	/* match Length */
+	{
+		const BYTE ML_deltaCode = 36;
+		const BYTE mlCode = (matchLength > 127) ? (BYTE)ZSTD_highbit32(matchLength) + ML_deltaCode : ML_Code[matchLength];
+		seqStorePtr->matchLengthFreq[mlCode]++;
+		seqStorePtr->matchLengthSum++;
+	}
+
+	ZSTD_setLog2Prices(seqStorePtr);
+}
+
+#define SET_PRICE(pos, mlen_, offset_, litlen_, price_)           \
+	{                                                         \
+		while (last_pos < pos) {                          \
+			opt[last_pos + 1].price = ZSTD_MAX_PRICE; \
+			last_pos++;                               \
+		}                                                 \
+		opt[pos].mlen = mlen_;                            \
+		opt[pos].off = offset_;                           \
+		opt[pos].litlen = litlen_;                        \
+		opt[pos].price = price_;                          \
+	}
+
+/* Update hashTable3 up to ip (excluded)
+   Assumption : always within prefix (i.e. not within extDict) */
+FORCE_INLINE
+U32 ZSTD_insertAndFindFirstIndexHash3(ZSTD_CCtx *zc, const BYTE *ip)
+{
+	U32 *const hashTable3 = zc->hashTable3;
+	U32 const hashLog3 = zc->hashLog3;
+	const BYTE *const base = zc->base;
+	U32 idx = zc->nextToUpdate3;
+	const U32 target = zc->nextToUpdate3 = (U32)(ip - base);
+	const size_t hash3 = ZSTD_hash3Ptr(ip, hashLog3);
+
+	while (idx < target) {
+		hashTable3[ZSTD_hash3Ptr(base + idx, hashLog3)] = idx;
+		idx++;
+	}
+
+	return hashTable3[hash3];
+}
+
+/*-*************************************
+*  Binary Tree search
+***************************************/
+static U32 ZSTD_insertBtAndGetAllMatches(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, U32 nbCompares, const U32 mls, U32 extDict,
+					 ZSTD_match_t *matches, const U32 minMatchLen)
+{
+	const BYTE *const base = zc->base;
+	const U32 curr = (U32)(ip - base);
+	const U32 hashLog = zc->params.cParams.hashLog;
+	const size_t h = ZSTD_hashPtr(ip, hashLog, mls);
+	U32 *const hashTable = zc->hashTable;
+	U32 matchIndex = hashTable[h];
+	U32 *const bt = zc->chainTable;
+	const U32 btLog = zc->params.cParams.chainLog - 1;
+	const U32 btMask = (1U << btLog) - 1;
+	size_t commonLengthSmaller = 0, commonLengthLarger = 0;
+	const BYTE *const dictBase = zc->dictBase;
+	const U32 dictLimit = zc->dictLimit;
+	const BYTE *const dictEnd = dictBase + dictLimit;
+	const BYTE *const prefixStart = base + dictLimit;
+	const U32 btLow = btMask >= curr ? 0 : curr - btMask;
+	const U32 windowLow = zc->lowLimit;
+	U32 *smallerPtr = bt + 2 * (curr & btMask);
+	U32 *largerPtr = bt + 2 * (curr & btMask) + 1;
+	U32 matchEndIdx = curr + 8;
+	U32 dummy32; /* to be nullified at the end */
+	U32 mnum = 0;
+
+	const U32 minMatch = (mls == 3) ? 3 : 4;
+	size_t bestLength = minMatchLen - 1;
+
+	if (minMatch == 3) { /* HC3 match finder */
+		U32 const matchIndex3 = ZSTD_insertAndFindFirstIndexHash3(zc, ip);
+		if (matchIndex3 > windowLow && (curr - matchIndex3 < (1 << 18))) {
+			const BYTE *match;
+			size_t currMl = 0;
+			if ((!extDict) || matchIndex3 >= dictLimit) {
+				match = base + matchIndex3;
+				if (match[bestLength] == ip[bestLength])
+					currMl = ZSTD_count(ip, match, iLimit);
+			} else {
+				match = dictBase + matchIndex3;
+				if (ZSTD_readMINMATCH(match, MINMATCH) ==
+				    ZSTD_readMINMATCH(ip, MINMATCH)) /* assumption : matchIndex3 <= dictLimit-4 (by table construction) */
+					currMl = ZSTD_count_2segments(ip + MINMATCH, match + MINMATCH, iLimit, dictEnd, prefixStart) + MINMATCH;
+			}
+
+			/* save best solution */
+			if (currMl > bestLength) {
+				bestLength = currMl;
+				matches[mnum].off = ZSTD_REP_MOVE_OPT + curr - matchIndex3;
+				matches[mnum].len = (U32)currMl;
+				mnum++;
+				if (currMl > ZSTD_OPT_NUM)
+					goto update;
+				if (ip + currMl == iLimit)
+					goto update; /* best possible, and avoid read overflow*/
+			}
+		}
+	}
+
+	hashTable[h] = curr; /* Update Hash Table */
+
+	while (nbCompares-- && (matchIndex > windowLow)) {
+		U32 *nextPtr = bt + 2 * (matchIndex & btMask);
+		size_t matchLength = MIN(commonLengthSmaller, commonLengthLarger); /* guaranteed minimum nb of common bytes */
+		const BYTE *match;
+
+		if ((!extDict) || (matchIndex + matchLength >= dictLimit)) {
+			match = base + matchIndex;
+			if (match[matchLength] == ip[matchLength]) {
+				matchLength += ZSTD_count(ip + matchLength + 1, match + matchLength + 1, iLimit) + 1;
+			}
+		} else {
+			match = dictBase + matchIndex;
+			matchLength += ZSTD_count_2segments(ip + matchLength, match + matchLength, iLimit, dictEnd, prefixStart);
+			if (matchIndex + matchLength >= dictLimit)
+				match = base + matchIndex; /* to prepare for next usage of match[matchLength] */
+		}
+
+		if (matchLength > bestLength) {
+			if (matchLength > matchEndIdx - matchIndex)
+				matchEndIdx = matchIndex + (U32)matchLength;
+			bestLength = matchLength;
+			matches[mnum].off = ZSTD_REP_MOVE_OPT + curr - matchIndex;
+			matches[mnum].len = (U32)matchLength;
+			mnum++;
+			if (matchLength > ZSTD_OPT_NUM)
+				break;
+			if (ip + matchLength == iLimit) /* equal : no way to know if inf or sup */
+				break;			/* drop, to guarantee consistency (miss a little bit of compression) */
+		}
+
+		if (match[matchLength] < ip[matchLength]) {
+			/* match is smaller than curr */
+			*smallerPtr = matchIndex;	  /* update smaller idx */
+			commonLengthSmaller = matchLength; /* all smaller will now have at least this guaranteed common length */
+			if (matchIndex <= btLow) {
+				smallerPtr = &dummy32;
+				break;
+			}			  /* beyond tree size, stop the search */
+			smallerPtr = nextPtr + 1; /* new "smaller" => larger of match */
+			matchIndex = nextPtr[1];  /* new matchIndex larger than previous (closer to curr) */
+		} else {
+			/* match is larger than curr */
+			*largerPtr = matchIndex;
+			commonLengthLarger = matchLength;
+			if (matchIndex <= btLow) {
+				largerPtr = &dummy32;
+				break;
+			} /* beyond tree size, stop the search */
+			largerPtr = nextPtr;
+			matchIndex = nextPtr[0];
+		}
+	}
+
+	*smallerPtr = *largerPtr = 0;
+
+update:
+	zc->nextToUpdate = (matchEndIdx > curr + 8) ? matchEndIdx - 8 : curr + 1;
+	return mnum;
+}
+
+/** Tree updater, providing best match */
+static U32 ZSTD_BtGetAllMatches(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, const U32 maxNbAttempts, const U32 mls, ZSTD_match_t *matches,
+				const U32 minMatchLen)
+{
+	if (ip < zc->base + zc->nextToUpdate)
+		return 0; /* skipped area */
+	ZSTD_updateTree(zc, ip, iLimit, maxNbAttempts, mls);
+	return ZSTD_insertBtAndGetAllMatches(zc, ip, iLimit, maxNbAttempts, mls, 0, matches, minMatchLen);
+}
+
+static U32 ZSTD_BtGetAllMatches_selectMLS(ZSTD_CCtx *zc, /* Index table will be updated */
+					  const BYTE *ip, const BYTE *const iHighLimit, const U32 maxNbAttempts, const U32 matchLengthSearch,
+					  ZSTD_match_t *matches, const U32 minMatchLen)
+{
+	switch (matchLengthSearch) {
+	case 3: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 3, matches, minMatchLen);
+	default:
+	case 4: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 4, matches, minMatchLen);
+	case 5: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 5, matches, minMatchLen);
+	case 7:
+	case 6: return ZSTD_BtGetAllMatches(zc, ip, iHighLimit, maxNbAttempts, 6, matches, minMatchLen);
+	}
+}
+
+/** Tree updater, providing best match */
+static U32 ZSTD_BtGetAllMatches_extDict(ZSTD_CCtx *zc, const BYTE *const ip, const BYTE *const iLimit, const U32 maxNbAttempts, const U32 mls,
+					ZSTD_match_t *matches, const U32 minMatchLen)
+{
+	if (ip < zc->base + zc->nextToUpdate)
+		return 0; /* skipped area */
+	ZSTD_updateTree_extDict(zc, ip, iLimit, maxNbAttempts, mls);
+	return ZSTD_insertBtAndGetAllMatches(zc, ip, iLimit, maxNbAttempts, mls, 1, matches, minMatchLen);
+}
+
+static U32 ZSTD_BtGetAllMatches_selectMLS_extDict(ZSTD_CCtx *zc, /* Index table will be updated */
+						  const BYTE *ip, const BYTE *const iHighLimit, const U32 maxNbAttempts, const U32 matchLengthSearch,
+						  ZSTD_match_t *matches, const U32 minMatchLen)
+{
+	switch (matchLengthSearch) {
+	case 3: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 3, matches, minMatchLen);
+	default:
+	case 4: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 4, matches, minMatchLen);
+	case 5: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 5, matches, minMatchLen);
+	case 7:
+	case 6: return ZSTD_BtGetAllMatches_extDict(zc, ip, iHighLimit, maxNbAttempts, 6, matches, minMatchLen);
+	}
+}
+
+/*-*******************************
+*  Optimal parser
+*********************************/
+FORCE_INLINE
+void ZSTD_compressBlock_opt_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const int ultra)
+{
+	seqStore_t *seqStorePtr = &(ctx->seqStore);
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *ip = istart;
+	const BYTE *anchor = istart;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *const ilimit = iend - 8;
+	const BYTE *const base = ctx->base;
+	const BYTE *const prefixStart = base + ctx->dictLimit;
+
+	const U32 maxSearches = 1U << ctx->params.cParams.searchLog;
+	const U32 sufficient_len = ctx->params.cParams.targetLength;
+	const U32 mls = ctx->params.cParams.searchLength;
+	const U32 minMatch = (ctx->params.cParams.searchLength == 3) ? 3 : 4;
+
+	ZSTD_optimal_t *opt = seqStorePtr->priceTable;
+	ZSTD_match_t *matches = seqStorePtr->matchTable;
+	const BYTE *inr;
+	U32 offset, rep[ZSTD_REP_NUM];
+
+	/* init */
+	ctx->nextToUpdate3 = ctx->nextToUpdate;
+	ZSTD_rescaleFreqs(seqStorePtr, (const BYTE *)src, srcSize);
+	ip += (ip == prefixStart);
+	{
+		U32 i;
+		for (i = 0; i < ZSTD_REP_NUM; i++)
+			rep[i] = ctx->rep[i];
+	}
+
+	/* Match Loop */
+	while (ip < ilimit) {
+		U32 cur, match_num, last_pos, litlen, price;
+		U32 u, mlen, best_mlen, best_off, litLength;
+		memset(opt, 0, sizeof(ZSTD_optimal_t));
+		last_pos = 0;
+		litlen = (U32)(ip - anchor);
+
+		/* check repCode */
+		{
+			U32 i, last_i = ZSTD_REP_CHECK + (ip == anchor);
+			for (i = (ip == anchor); i < last_i; i++) {
+				const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : rep[i];
+				if ((repCur > 0) && (repCur < (S32)(ip - prefixStart)) &&
+				    (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(ip - repCur, minMatch))) {
+					mlen = (U32)ZSTD_count(ip + minMatch, ip + minMatch - repCur, iend) + minMatch;
+					if (mlen > sufficient_len || mlen >= ZSTD_OPT_NUM) {
+						best_mlen = mlen;
+						best_off = i;
+						cur = 0;
+						last_pos = 1;
+						goto _storeSequence;
+					}
+					best_off = i - (ip == anchor);
+					do {
+						price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
+						if (mlen > last_pos || price < opt[mlen].price)
+							SET_PRICE(mlen, mlen, i, litlen, price); /* note : macro modifies last_pos */
+						mlen--;
+					} while (mlen >= minMatch);
+				}
+			}
+		}
+
+		match_num = ZSTD_BtGetAllMatches_selectMLS(ctx, ip, iend, maxSearches, mls, matches, minMatch);
+
+		if (!last_pos && !match_num) {
+			ip++;
+			continue;
+		}
+
+		if (match_num && (matches[match_num - 1].len > sufficient_len || matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
+			best_mlen = matches[match_num - 1].len;
+			best_off = matches[match_num - 1].off;
+			cur = 0;
+			last_pos = 1;
+			goto _storeSequence;
+		}
+
+		/* set prices using matches at position = 0 */
+		best_mlen = (last_pos) ? last_pos : minMatch;
+		for (u = 0; u < match_num; u++) {
+			mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
+			best_mlen = matches[u].len;
+			while (mlen <= best_mlen) {
+				price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
+				if (mlen > last_pos || price < opt[mlen].price)
+					SET_PRICE(mlen, mlen, matches[u].off, litlen, price); /* note : macro modifies last_pos */
+				mlen++;
+			}
+		}
+
+		if (last_pos < minMatch) {
+			ip++;
+			continue;
+		}
+
+		/* initialize opt[0] */
+		{
+			U32 i;
+			for (i = 0; i < ZSTD_REP_NUM; i++)
+				opt[0].rep[i] = rep[i];
+		}
+		opt[0].mlen = 1;
+		opt[0].litlen = litlen;
+
+		/* check further positions */
+		for (cur = 1; cur <= last_pos; cur++) {
+			inr = ip + cur;
+
+			if (opt[cur - 1].mlen == 1) {
+				litlen = opt[cur - 1].litlen + 1;
+				if (cur > litlen) {
+					price = opt[cur - litlen].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - litlen);
+				} else
+					price = ZSTD_getLiteralPrice(seqStorePtr, litlen, anchor);
+			} else {
+				litlen = 1;
+				price = opt[cur - 1].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - 1);
+			}
+
+			if (cur > last_pos || price <= opt[cur].price)
+				SET_PRICE(cur, 1, 0, litlen, price);
+
+			if (cur == last_pos)
+				break;
+
+			if (inr > ilimit) /* last match must start at a minimum distance of 8 from oend */
+				continue;
+
+			mlen = opt[cur].mlen;
+			if (opt[cur].off > ZSTD_REP_MOVE_OPT) {
+				opt[cur].rep[2] = opt[cur - mlen].rep[1];
+				opt[cur].rep[1] = opt[cur - mlen].rep[0];
+				opt[cur].rep[0] = opt[cur].off - ZSTD_REP_MOVE_OPT;
+			} else {
+				opt[cur].rep[2] = (opt[cur].off > 1) ? opt[cur - mlen].rep[1] : opt[cur - mlen].rep[2];
+				opt[cur].rep[1] = (opt[cur].off > 0) ? opt[cur - mlen].rep[0] : opt[cur - mlen].rep[1];
+				opt[cur].rep[0] =
+				    ((opt[cur].off == ZSTD_REP_MOVE_OPT) && (mlen != 1)) ? (opt[cur - mlen].rep[0] - 1) : (opt[cur - mlen].rep[opt[cur].off]);
+			}
+
+			best_mlen = minMatch;
+			{
+				U32 i, last_i = ZSTD_REP_CHECK + (mlen != 1);
+				for (i = (opt[cur].mlen != 1); i < last_i; i++) { /* check rep */
+					const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (opt[cur].rep[0] - 1) : opt[cur].rep[i];
+					if ((repCur > 0) && (repCur < (S32)(inr - prefixStart)) &&
+					    (ZSTD_readMINMATCH(inr, minMatch) == ZSTD_readMINMATCH(inr - repCur, minMatch))) {
+						mlen = (U32)ZSTD_count(inr + minMatch, inr + minMatch - repCur, iend) + minMatch;
+
+						if (mlen > sufficient_len || cur + mlen >= ZSTD_OPT_NUM) {
+							best_mlen = mlen;
+							best_off = i;
+							last_pos = cur + 1;
+							goto _storeSequence;
+						}
+
+						best_off = i - (opt[cur].mlen != 1);
+						if (mlen > best_mlen)
+							best_mlen = mlen;
+
+						do {
+							if (opt[cur].mlen == 1) {
+								litlen = opt[cur].litlen;
+								if (cur > litlen) {
+									price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, inr - litlen,
+															best_off, mlen - MINMATCH, ultra);
+								} else
+									price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
+							} else {
+								litlen = 0;
+								price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, best_off, mlen - MINMATCH, ultra);
+							}
+
+							if (cur + mlen > last_pos || price <= opt[cur + mlen].price)
+								SET_PRICE(cur + mlen, mlen, i, litlen, price);
+							mlen--;
+						} while (mlen >= minMatch);
+					}
+				}
+			}
+
+			match_num = ZSTD_BtGetAllMatches_selectMLS(ctx, inr, iend, maxSearches, mls, matches, best_mlen);
+
+			if (match_num > 0 && (matches[match_num - 1].len > sufficient_len || cur + matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
+				best_mlen = matches[match_num - 1].len;
+				best_off = matches[match_num - 1].off;
+				last_pos = cur + 1;
+				goto _storeSequence;
+			}
+
+			/* set prices using matches at position = cur */
+			for (u = 0; u < match_num; u++) {
+				mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
+				best_mlen = matches[u].len;
+
+				while (mlen <= best_mlen) {
+					if (opt[cur].mlen == 1) {
+						litlen = opt[cur].litlen;
+						if (cur > litlen)
+							price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, ip + cur - litlen,
+													matches[u].off - 1, mlen - MINMATCH, ultra);
+						else
+							price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
+					} else {
+						litlen = 0;
+						price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, matches[u].off - 1, mlen - MINMATCH, ultra);
+					}
+
+					if (cur + mlen > last_pos || (price < opt[cur + mlen].price))
+						SET_PRICE(cur + mlen, mlen, matches[u].off, litlen, price);
+
+					mlen++;
+				}
+			}
+		}
+
+		best_mlen = opt[last_pos].mlen;
+		best_off = opt[last_pos].off;
+		cur = last_pos - best_mlen;
+
+	/* store sequence */
+_storeSequence: /* cur, last_pos, best_mlen, best_off have to be set */
+		opt[0].mlen = 1;
+
+		while (1) {
+			mlen = opt[cur].mlen;
+			offset = opt[cur].off;
+			opt[cur].mlen = best_mlen;
+			opt[cur].off = best_off;
+			best_mlen = mlen;
+			best_off = offset;
+			if (mlen > cur)
+				break;
+			cur -= mlen;
+		}
+
+		for (u = 0; u <= last_pos;) {
+			u += opt[u].mlen;
+		}
+
+		for (cur = 0; cur < last_pos;) {
+			mlen = opt[cur].mlen;
+			if (mlen == 1) {
+				ip++;
+				cur++;
+				continue;
+			}
+			offset = opt[cur].off;
+			cur += mlen;
+			litLength = (U32)(ip - anchor);
+
+			if (offset > ZSTD_REP_MOVE_OPT) {
+				rep[2] = rep[1];
+				rep[1] = rep[0];
+				rep[0] = offset - ZSTD_REP_MOVE_OPT;
+				offset--;
+			} else {
+				if (offset != 0) {
+					best_off = (offset == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : (rep[offset]);
+					if (offset != 1)
+						rep[2] = rep[1];
+					rep[1] = rep[0];
+					rep[0] = best_off;
+				}
+				if (litLength == 0)
+					offset--;
+			}
+
+			ZSTD_updatePrice(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
+			ZSTD_storeSeq(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
+			anchor = ip = ip + mlen;
+		}
+	} /* for (cur=0; cur < last_pos; ) */
+
+	/* Save reps for next block */
+	{
+		int i;
+		for (i = 0; i < ZSTD_REP_NUM; i++)
+			ctx->repToConfirm[i] = rep[i];
+	}
+
+	/* Last Literals */
+	{
+		size_t const lastLLSize = iend - anchor;
+		memcpy(seqStorePtr->lit, anchor, lastLLSize);
+		seqStorePtr->lit += lastLLSize;
+	}
+}
+
+FORCE_INLINE
+void ZSTD_compressBlock_opt_extDict_generic(ZSTD_CCtx *ctx, const void *src, size_t srcSize, const int ultra)
+{
+	seqStore_t *seqStorePtr = &(ctx->seqStore);
+	const BYTE *const istart = (const BYTE *)src;
+	const BYTE *ip = istart;
+	const BYTE *anchor = istart;
+	const BYTE *const iend = istart + srcSize;
+	const BYTE *const ilimit = iend - 8;
+	const BYTE *const base = ctx->base;
+	const U32 lowestIndex = ctx->lowLimit;
+	const U32 dictLimit = ctx->dictLimit;
+	const BYTE *const prefixStart = base + dictLimit;
+	const BYTE *const dictBase = ctx->dictBase;
+	const BYTE *const dictEnd = dictBase + dictLimit;
+
+	const U32 maxSearches = 1U << ctx->params.cParams.searchLog;
+	const U32 sufficient_len = ctx->params.cParams.targetLength;
+	const U32 mls = ctx->params.cParams.searchLength;
+	const U32 minMatch = (ctx->params.cParams.searchLength == 3) ? 3 : 4;
+
+	ZSTD_optimal_t *opt = seqStorePtr->priceTable;
+	ZSTD_match_t *matches = seqStorePtr->matchTable;
+	const BYTE *inr;
+
+	/* init */
+	U32 offset, rep[ZSTD_REP_NUM];
+	{
+		U32 i;
+		for (i = 0; i < ZSTD_REP_NUM; i++)
+			rep[i] = ctx->rep[i];
+	}
+
+	ctx->nextToUpdate3 = ctx->nextToUpdate;
+	ZSTD_rescaleFreqs(seqStorePtr, (const BYTE *)src, srcSize);
+	ip += (ip == prefixStart);
+
+	/* Match Loop */
+	while (ip < ilimit) {
+		U32 cur, match_num, last_pos, litlen, price;
+		U32 u, mlen, best_mlen, best_off, litLength;
+		U32 curr = (U32)(ip - base);
+		memset(opt, 0, sizeof(ZSTD_optimal_t));
+		last_pos = 0;
+		opt[0].litlen = (U32)(ip - anchor);
+
+		/* check repCode */
+		{
+			U32 i, last_i = ZSTD_REP_CHECK + (ip == anchor);
+			for (i = (ip == anchor); i < last_i; i++) {
+				const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : rep[i];
+				const U32 repIndex = (U32)(curr - repCur);
+				const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
+				const BYTE *const repMatch = repBase + repIndex;
+				if ((repCur > 0 && repCur <= (S32)curr) &&
+				    (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
+				    && (ZSTD_readMINMATCH(ip, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch))) {
+					/* repcode detected we should take it */
+					const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
+					mlen = (U32)ZSTD_count_2segments(ip + minMatch, repMatch + minMatch, iend, repEnd, prefixStart) + minMatch;
+
+					if (mlen > sufficient_len || mlen >= ZSTD_OPT_NUM) {
+						best_mlen = mlen;
+						best_off = i;
+						cur = 0;
+						last_pos = 1;
+						goto _storeSequence;
+					}
+
+					best_off = i - (ip == anchor);
+					litlen = opt[0].litlen;
+					do {
+						price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
+						if (mlen > last_pos || price < opt[mlen].price)
+							SET_PRICE(mlen, mlen, i, litlen, price); /* note : macro modifies last_pos */
+						mlen--;
+					} while (mlen >= minMatch);
+				}
+			}
+		}
+
+		match_num = ZSTD_BtGetAllMatches_selectMLS_extDict(ctx, ip, iend, maxSearches, mls, matches, minMatch); /* first search (depth 0) */
+
+		if (!last_pos && !match_num) {
+			ip++;
+			continue;
+		}
+
+		{
+			U32 i;
+			for (i = 0; i < ZSTD_REP_NUM; i++)
+				opt[0].rep[i] = rep[i];
+		}
+		opt[0].mlen = 1;
+
+		if (match_num && (matches[match_num - 1].len > sufficient_len || matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
+			best_mlen = matches[match_num - 1].len;
+			best_off = matches[match_num - 1].off;
+			cur = 0;
+			last_pos = 1;
+			goto _storeSequence;
+		}
+
+		best_mlen = (last_pos) ? last_pos : minMatch;
+
+		/* set prices using matches at position = 0 */
+		for (u = 0; u < match_num; u++) {
+			mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
+			best_mlen = matches[u].len;
+			litlen = opt[0].litlen;
+			while (mlen <= best_mlen) {
+				price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
+				if (mlen > last_pos || price < opt[mlen].price)
+					SET_PRICE(mlen, mlen, matches[u].off, litlen, price);
+				mlen++;
+			}
+		}
+
+		if (last_pos < minMatch) {
+			ip++;
+			continue;
+		}
+
+		/* check further positions */
+		for (cur = 1; cur <= last_pos; cur++) {
+			inr = ip + cur;
+
+			if (opt[cur - 1].mlen == 1) {
+				litlen = opt[cur - 1].litlen + 1;
+				if (cur > litlen) {
+					price = opt[cur - litlen].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - litlen);
+				} else
+					price = ZSTD_getLiteralPrice(seqStorePtr, litlen, anchor);
+			} else {
+				litlen = 1;
+				price = opt[cur - 1].price + ZSTD_getLiteralPrice(seqStorePtr, litlen, inr - 1);
+			}
+
+			if (cur > last_pos || price <= opt[cur].price)
+				SET_PRICE(cur, 1, 0, litlen, price);
+
+			if (cur == last_pos)
+				break;
+
+			if (inr > ilimit) /* last match must start at a minimum distance of 8 from oend */
+				continue;
+
+			mlen = opt[cur].mlen;
+			if (opt[cur].off > ZSTD_REP_MOVE_OPT) {
+				opt[cur].rep[2] = opt[cur - mlen].rep[1];
+				opt[cur].rep[1] = opt[cur - mlen].rep[0];
+				opt[cur].rep[0] = opt[cur].off - ZSTD_REP_MOVE_OPT;
+			} else {
+				opt[cur].rep[2] = (opt[cur].off > 1) ? opt[cur - mlen].rep[1] : opt[cur - mlen].rep[2];
+				opt[cur].rep[1] = (opt[cur].off > 0) ? opt[cur - mlen].rep[0] : opt[cur - mlen].rep[1];
+				opt[cur].rep[0] =
+				    ((opt[cur].off == ZSTD_REP_MOVE_OPT) && (mlen != 1)) ? (opt[cur - mlen].rep[0] - 1) : (opt[cur - mlen].rep[opt[cur].off]);
+			}
+
+			best_mlen = minMatch;
+			{
+				U32 i, last_i = ZSTD_REP_CHECK + (mlen != 1);
+				for (i = (mlen != 1); i < last_i; i++) {
+					const S32 repCur = (i == ZSTD_REP_MOVE_OPT) ? (opt[cur].rep[0] - 1) : opt[cur].rep[i];
+					const U32 repIndex = (U32)(curr + cur - repCur);
+					const BYTE *const repBase = repIndex < dictLimit ? dictBase : base;
+					const BYTE *const repMatch = repBase + repIndex;
+					if ((repCur > 0 && repCur <= (S32)(curr + cur)) &&
+					    (((U32)((dictLimit - 1) - repIndex) >= 3) & (repIndex > lowestIndex)) /* intentional overflow */
+					    && (ZSTD_readMINMATCH(inr, minMatch) == ZSTD_readMINMATCH(repMatch, minMatch))) {
+						/* repcode detected */
+						const BYTE *const repEnd = repIndex < dictLimit ? dictEnd : iend;
+						mlen = (U32)ZSTD_count_2segments(inr + minMatch, repMatch + minMatch, iend, repEnd, prefixStart) + minMatch;
+
+						if (mlen > sufficient_len || cur + mlen >= ZSTD_OPT_NUM) {
+							best_mlen = mlen;
+							best_off = i;
+							last_pos = cur + 1;
+							goto _storeSequence;
+						}
+
+						best_off = i - (opt[cur].mlen != 1);
+						if (mlen > best_mlen)
+							best_mlen = mlen;
+
+						do {
+							if (opt[cur].mlen == 1) {
+								litlen = opt[cur].litlen;
+								if (cur > litlen) {
+									price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, inr - litlen,
+															best_off, mlen - MINMATCH, ultra);
+								} else
+									price = ZSTD_getPrice(seqStorePtr, litlen, anchor, best_off, mlen - MINMATCH, ultra);
+							} else {
+								litlen = 0;
+								price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, best_off, mlen - MINMATCH, ultra);
+							}
+
+							if (cur + mlen > last_pos || price <= opt[cur + mlen].price)
+								SET_PRICE(cur + mlen, mlen, i, litlen, price);
+							mlen--;
+						} while (mlen >= minMatch);
+					}
+				}
+			}
+
+			match_num = ZSTD_BtGetAllMatches_selectMLS_extDict(ctx, inr, iend, maxSearches, mls, matches, minMatch);
+
+			if (match_num > 0 && (matches[match_num - 1].len > sufficient_len || cur + matches[match_num - 1].len >= ZSTD_OPT_NUM)) {
+				best_mlen = matches[match_num - 1].len;
+				best_off = matches[match_num - 1].off;
+				last_pos = cur + 1;
+				goto _storeSequence;
+			}
+
+			/* set prices using matches at position = cur */
+			for (u = 0; u < match_num; u++) {
+				mlen = (u > 0) ? matches[u - 1].len + 1 : best_mlen;
+				best_mlen = matches[u].len;
+
+				while (mlen <= best_mlen) {
+					if (opt[cur].mlen == 1) {
+						litlen = opt[cur].litlen;
+						if (cur > litlen)
+							price = opt[cur - litlen].price + ZSTD_getPrice(seqStorePtr, litlen, ip + cur - litlen,
+													matches[u].off - 1, mlen - MINMATCH, ultra);
+						else
+							price = ZSTD_getPrice(seqStorePtr, litlen, anchor, matches[u].off - 1, mlen - MINMATCH, ultra);
+					} else {
+						litlen = 0;
+						price = opt[cur].price + ZSTD_getPrice(seqStorePtr, 0, NULL, matches[u].off - 1, mlen - MINMATCH, ultra);
+					}
+
+					if (cur + mlen > last_pos || (price < opt[cur + mlen].price))
+						SET_PRICE(cur + mlen, mlen, matches[u].off, litlen, price);
+
+					mlen++;
+				}
+			}
+		} /* for (cur = 1; cur <= last_pos; cur++) */
+
+		best_mlen = opt[last_pos].mlen;
+		best_off = opt[last_pos].off;
+		cur = last_pos - best_mlen;
+
+	/* store sequence */
+_storeSequence: /* cur, last_pos, best_mlen, best_off have to be set */
+		opt[0].mlen = 1;
+
+		while (1) {
+			mlen = opt[cur].mlen;
+			offset = opt[cur].off;
+			opt[cur].mlen = best_mlen;
+			opt[cur].off = best_off;
+			best_mlen = mlen;
+			best_off = offset;
+			if (mlen > cur)
+				break;
+			cur -= mlen;
+		}
+
+		for (u = 0; u <= last_pos;) {
+			u += opt[u].mlen;
+		}
+
+		for (cur = 0; cur < last_pos;) {
+			mlen = opt[cur].mlen;
+			if (mlen == 1) {
+				ip++;
+				cur++;
+				continue;
+			}
+			offset = opt[cur].off;
+			cur += mlen;
+			litLength = (U32)(ip - anchor);
+
+			if (offset > ZSTD_REP_MOVE_OPT) {
+				rep[2] = rep[1];
+				rep[1] = rep[0];
+				rep[0] = offset - ZSTD_REP_MOVE_OPT;
+				offset--;
+			} else {
+				if (offset != 0) {
+					best_off = (offset == ZSTD_REP_MOVE_OPT) ? (rep[0] - 1) : (rep[offset]);
+					if (offset != 1)
+						rep[2] = rep[1];
+					rep[1] = rep[0];
+					rep[0] = best_off;
+				}
+
+				if (litLength == 0)
+					offset--;
+			}
+
+			ZSTD_updatePrice(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
+			ZSTD_storeSeq(seqStorePtr, litLength, anchor, offset, mlen - MINMATCH);
+			anchor = ip = ip + mlen;
+		}
+	} /* for (cur=0; cur < last_pos; ) */
+
+	/* Save reps for next block */
+	{
+		int i;
+		for (i = 0; i < ZSTD_REP_NUM; i++)
+			ctx->repToConfirm[i] = rep[i];
+	}
+
+	/* Last Literals */
+	{
+		size_t lastLLSize = iend - anchor;
+		memcpy(seqStorePtr->lit, anchor, lastLLSize);
+		seqStorePtr->lit += lastLLSize;
+	}
+}
+
+#endif /* ZSTD_OPT_H_91842398743 */