From 420ae9518404c4aeda3abc8e017c8fdcc3a13d6b Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 16 Jun 2014 07:25:06 +0400
Subject: xtensa: simplify addition of new core variants

Instead of adding new Kconfig options and Makefile rules for each new
core variant provide XTENSA_VARIANT_CUSTOM variant and record variant
name in the XTENSA_VARIANT_NAME variable. Adding new core variant now
means providing directory structure under arch/xtensa/variant and
specifying correct name in kernel configuration.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/Kconfig  | 36 +++++++++++++++++++++++++++++++++++-
 arch/xtensa/Makefile |  7 ++-----
 2 files changed, 37 insertions(+), 6 deletions(-)

diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index 3a617af60d46..7feca8dfdff2 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -62,7 +62,9 @@ config TRACE_IRQFLAGS_SUPPORT
 	def_bool y
 
 config MMU
-	def_bool n
+	bool
+	default n if !XTENSA_VARIANT_CUSTOM
+	default XTENSA_VARIANT_MMU if XTENSA_VARIANT_CUSTOM
 
 config VARIANT_IRQ_SWITCH
 	def_bool n
@@ -102,8 +104,40 @@ config XTENSA_VARIANT_S6000
 	select VARIANT_IRQ_SWITCH
 	select ARCH_REQUIRE_GPIOLIB
 	select XTENSA_CALIBRATE_CCOUNT
+
+config XTENSA_VARIANT_CUSTOM
+	bool "Custom Xtensa processor configuration"
+	select MAY_HAVE_SMP
+	select HAVE_XTENSA_GPIO32
+	help
+	  Select this variant to use a custom Xtensa processor configuration.
+	  You will be prompted for a processor variant CORENAME.
 endchoice
 
+config XTENSA_VARIANT_CUSTOM_NAME
+	string "Xtensa Processor Custom Core Variant Name"
+	depends on XTENSA_VARIANT_CUSTOM
+	help
+	  Provide the name of a custom Xtensa processor variant.
+	  This CORENAME selects arch/xtensa/variant/CORENAME.
+	  Dont forget you have to select MMU if you have one.
+
+config XTENSA_VARIANT_NAME
+	string
+	default "dc232b"			if XTENSA_VARIANT_DC232B
+	default "dc233c"			if XTENSA_VARIANT_DC233C
+	default "fsf"				if XTENSA_VARIANT_FSF
+	default "s6000"				if XTENSA_VARIANT_S6000
+	default XTENSA_VARIANT_CUSTOM_NAME	if XTENSA_VARIANT_CUSTOM
+
+config XTENSA_VARIANT_MMU
+	bool "Core variant has a Full MMU (TLB, Pages, Protection, etc)"
+	depends on XTENSA_VARIANT_CUSTOM
+	default y
+	help
+	  Build a Conventional Kernel with full MMU support,
+	  ie: it supports a TLB with auto-loading, page protection.
+
 config XTENSA_UNALIGNED_USER
 	bool "Unaligned memory access in use space"
 	help
diff --git a/arch/xtensa/Makefile b/arch/xtensa/Makefile
index 81250ece3062..472533064b46 100644
--- a/arch/xtensa/Makefile
+++ b/arch/xtensa/Makefile
@@ -4,6 +4,7 @@
 # for more details.
 #
 # Copyright (C) 2001 - 2005  Tensilica Inc.
+# Copyright (C) 2014 Cadence Design Systems Inc.
 #
 # This file is included by the global makefile so that you can add your own
 # architecture-specific flags and dependencies. Remember to do have actions
@@ -13,11 +14,7 @@
 # Core configuration.
 # (Use VAR=<xtensa_config> to use another default compiler.)
 
-variant-$(CONFIG_XTENSA_VARIANT_FSF)		:= fsf
-variant-$(CONFIG_XTENSA_VARIANT_DC232B)		:= dc232b
-variant-$(CONFIG_XTENSA_VARIANT_DC233C)		:= dc233c
-variant-$(CONFIG_XTENSA_VARIANT_S6000)		:= s6000
-variant-$(CONFIG_XTENSA_VARIANT_LINUX_CUSTOM)	:= custom
+variant-y := $(patsubst "%",%,$(CONFIG_XTENSA_VARIANT_NAME))
 
 VARIANT = $(variant-y)
 export VARIANT
-- 
cgit v1.2.3


From 8a9de05954846fe854cdea47c0178bb67fc70a47 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 16 Jun 2014 08:15:43 +0400
Subject: xtensa: make MMU-related configuration options depend on MMU

MMUv3 and HIGHMEM support are available only on configurations with MMU,
don't show them otherwise.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/Kconfig | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index 7feca8dfdff2..480a7d5fb105 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -197,6 +197,7 @@ config MATH_EMULATION
 
 config INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
 	bool "Initialize Xtensa MMU inside the Linux kernel code"
+	depends on MMU
 	default y
 	help
 	  Earlier version initialized the MMU in the exception vector
@@ -226,6 +227,7 @@ config INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
 
 config HIGHMEM
 	bool "High Memory Support"
+	depends on MMU
 	help
 	  Linux can use the full amount of RAM in the system by
 	  default. However, the default MMUv2 setup only maps the
-- 
cgit v1.2.3


From 920f8a396595281037b48dff8bad6003ba6c7733 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 16 Jun 2014 08:20:17 +0400
Subject: xtensa: sort 'select' statements in Kconfig

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/Kconfig | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)

diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index 480a7d5fb105..3b486dd5d6d4 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -4,24 +4,24 @@ config ZONE_DMA
 config XTENSA
 	def_bool y
 	select ARCH_WANT_FRAME_POINTERS
-	select HAVE_IDE
-	select GENERIC_ATOMIC64
-	select GENERIC_CLOCKEVENTS
-	select VIRT_TO_BUS
-	select GENERIC_IRQ_SHOW
-	select GENERIC_SCHED_CLOCK
-	select MODULES_USE_ELF_RELA
-	select GENERIC_PCI_IOMAP
 	select ARCH_WANT_IPC_PARSE_VERSION
 	select ARCH_WANT_OPTIONAL_GPIOLIB
 	select BUILDTIME_EXTABLE_SORT
 	select CLONE_BACKWARDS
-	select IRQ_DOMAIN
-	select HAVE_OPROFILE
+	select COMMON_CLK
+	select GENERIC_ATOMIC64
+	select GENERIC_CLOCKEVENTS
+	select GENERIC_IRQ_SHOW
+	select GENERIC_PCI_IOMAP
+	select GENERIC_SCHED_CLOCK
 	select HAVE_FUNCTION_TRACER
+	select HAVE_IDE
 	select HAVE_IRQ_TIME_ACCOUNTING
+	select HAVE_OPROFILE
 	select HAVE_PERF_EVENTS
-	select COMMON_CLK
+	select IRQ_DOMAIN
+	select MODULES_USE_ELF_RELA
+	select VIRT_TO_BUS
 	help
 	  Xtensa processors are 32-bit RISC machines designed by Tensilica
 	  primarily for embedded systems.  These processors are both
-- 
cgit v1.2.3


From 4964527da82d6faacc753e8a322b1b8ef8340bd1 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 16 Jun 2014 08:25:43 +0400
Subject: xtensa: select HAVE_IDE only on platforms that may have it

HAVE_IDE is not a property of architecture but of a platform, and neither
ISS or XTFPGA support it.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/Kconfig               | 3 ++-
 arch/xtensa/configs/iss_defconfig | 2 +-
 2 files changed, 3 insertions(+), 2 deletions(-)

diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index 3b486dd5d6d4..eaa9d07907d7 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -15,7 +15,6 @@ config XTENSA
 	select GENERIC_PCI_IOMAP
 	select GENERIC_SCHED_CLOCK
 	select HAVE_FUNCTION_TRACER
-	select HAVE_IDE
 	select HAVE_IRQ_TIME_ACCOUNTING
 	select HAVE_OPROFILE
 	select HAVE_PERF_EVENTS
@@ -286,12 +285,14 @@ config XTENSA_PLATFORM_ISS
 
 config XTENSA_PLATFORM_XT2000
 	bool "XT2000"
+	select HAVE_IDE
 	help
 	  XT2000 is the name of Tensilica's feature-rich emulation platform.
 	  This hardware is capable of running a full Linux distribution.
 
 config XTENSA_PLATFORM_S6105
 	bool "S6105"
+	select HAVE_IDE
 	select SERIAL_CONSOLE
 	select NO_IOPORT_MAP
 
diff --git a/arch/xtensa/configs/iss_defconfig b/arch/xtensa/configs/iss_defconfig
index 1493c68352d1..009fe4a266dc 100644
--- a/arch/xtensa/configs/iss_defconfig
+++ b/arch/xtensa/configs/iss_defconfig
@@ -308,7 +308,7 @@ CONFIG_MISC_DEVICES=y
 # EEPROM support
 #
 # CONFIG_EEPROM_93CX6 is not set
-CONFIG_HAVE_IDE=y
+# CONFIG_HAVE_IDE is not set
 # CONFIG_IDE is not set
 
 #
-- 
cgit v1.2.3


From ad4a96b418aceed3f86882e7619bdccaf04c9ce0 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Tue, 17 Jun 2014 01:32:52 +0400
Subject: xtensa: remove orphan MATH_EMULATION symbol

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/Kconfig                  | 5 -----
 arch/xtensa/configs/common_defconfig | 1 -
 arch/xtensa/configs/iss_defconfig    | 1 -
 arch/xtensa/configs/s6105_defconfig  | 1 -
 4 files changed, 8 deletions(-)

diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index eaa9d07907d7..3d83c29c9232 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -189,11 +189,6 @@ config HOTPLUG_CPU
 
 	  Say N if you want to disable CPU hotplug.
 
-config MATH_EMULATION
-	bool "Math emulation"
-	help
-	Can we use information of configuration file?
-
 config INITIALIZE_XTENSA_MMU_INSIDE_VMLINUX
 	bool "Initialize Xtensa MMU inside the Linux kernel code"
 	depends on MMU
diff --git a/arch/xtensa/configs/common_defconfig b/arch/xtensa/configs/common_defconfig
index f6000fe05119..721df1214bc3 100644
--- a/arch/xtensa/configs/common_defconfig
+++ b/arch/xtensa/configs/common_defconfig
@@ -66,7 +66,6 @@ CONFIG_XTENSA_ARCH_LINUX_BE=y
 CONFIG_MMU=y
 # CONFIG_XTENSA_UNALIGNED_USER is not set
 # CONFIG_PREEMPT is not set
-# CONFIG_MATH_EMULATION is not set
 # CONFIG_HIGHMEM is not set
 
 #
diff --git a/arch/xtensa/configs/iss_defconfig b/arch/xtensa/configs/iss_defconfig
index 009fe4a266dc..b966baf82cae 100644
--- a/arch/xtensa/configs/iss_defconfig
+++ b/arch/xtensa/configs/iss_defconfig
@@ -146,7 +146,6 @@ CONFIG_XTENSA_VARIANT_FSF=y
 # CONFIG_XTENSA_VARIANT_S6000 is not set
 # CONFIG_XTENSA_UNALIGNED_USER is not set
 # CONFIG_PREEMPT is not set
-# CONFIG_MATH_EMULATION is not set
 CONFIG_XTENSA_CALIBRATE_CCOUNT=y
 CONFIG_SERIAL_CONSOLE=y
 CONFIG_XTENSA_ISS_NETWORK=y
diff --git a/arch/xtensa/configs/s6105_defconfig b/arch/xtensa/configs/s6105_defconfig
index 12a492ab6d17..9471265b8ca6 100644
--- a/arch/xtensa/configs/s6105_defconfig
+++ b/arch/xtensa/configs/s6105_defconfig
@@ -109,7 +109,6 @@ CONFIG_VARIANT_IRQ_SWITCH=y
 CONFIG_XTENSA_VARIANT_S6000=y
 # CONFIG_XTENSA_UNALIGNED_USER is not set
 CONFIG_PREEMPT=y
-# CONFIG_MATH_EMULATION is not set
 # CONFIG_HIGHMEM is not set
 CONFIG_XTENSA_CALIBRATE_CCOUNT=y
 CONFIG_SERIAL_CONSOLE=y
-- 
cgit v1.2.3


From f61bf8e7d19e0a3456a7a9ed97c399e4353698dc Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Sun, 20 Jul 2014 03:38:53 +0400
Subject: xtensa: replace IOCTL code definitions with constants

This fixes userspace code that builds on other architectures but fails
on xtensa due to references to structures that other architectures don't
refer to. E.g. this fixes the following issue with python-2.7.8:

  python-2.7.8/Modules/termios.c:861:25: error: invalid application
     of 'sizeof' to incomplete type 'struct serial_multiport_struct'
     {"TIOCSERGETMULTI", TIOCSERGETMULTI},
  python-2.7.8/Modules/termios.c:870:25: error: invalid application
     of 'sizeof' to incomplete type 'struct serial_multiport_struct'
     {"TIOCSERSETMULTI", TIOCSERSETMULTI},
  python-2.7.8/Modules/termios.c:900:24: error: invalid application
     of 'sizeof' to incomplete type 'struct tty_struct'
     {"TIOCTTYGSTRUCT", TIOCTTYGSTRUCT},

Cc: stable@vger.kernel.org
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/include/uapi/asm/ioctls.h | 19 ++++++++++---------
 1 file changed, 10 insertions(+), 9 deletions(-)

diff --git a/arch/xtensa/include/uapi/asm/ioctls.h b/arch/xtensa/include/uapi/asm/ioctls.h
index b4cb1100c0fb..a47909f0c34b 100644
--- a/arch/xtensa/include/uapi/asm/ioctls.h
+++ b/arch/xtensa/include/uapi/asm/ioctls.h
@@ -28,17 +28,17 @@
 #define TCSETSW		0x5403
 #define TCSETSF		0x5404
 
-#define TCGETA		_IOR('t', 23, struct termio)
-#define TCSETA		_IOW('t', 24, struct termio)
-#define TCSETAW		_IOW('t', 25, struct termio)
-#define TCSETAF		_IOW('t', 28, struct termio)
+#define TCGETA		0x80127417	/* _IOR('t', 23, struct termio) */
+#define TCSETA		0x40127418	/* _IOW('t', 24, struct termio) */
+#define TCSETAW		0x40127419	/* _IOW('t', 25, struct termio) */
+#define TCSETAF		0x4012741C	/* _IOW('t', 28, struct termio) */
 
 #define TCSBRK		_IO('t', 29)
 #define TCXONC		_IO('t', 30)
 #define TCFLSH		_IO('t', 31)
 
-#define TIOCSWINSZ	_IOW('t', 103, struct winsize)
-#define TIOCGWINSZ	_IOR('t', 104, struct winsize)
+#define TIOCSWINSZ	0x40087467	/* _IOW('t', 103, struct winsize) */
+#define TIOCGWINSZ	0x80087468	/* _IOR('t', 104, struct winsize) */
 #define	TIOCSTART	_IO('t', 110)		/* start output, like ^Q */
 #define	TIOCSTOP	_IO('t', 111)		/* stop output, like ^S */
 #define TIOCOUTQ        _IOR('t', 115, int)     /* output queue size */
@@ -88,7 +88,6 @@
 #define TIOCSETD	_IOW('T', 35, int)
 #define TIOCGETD	_IOR('T', 36, int)
 #define TCSBRKP		_IOW('T', 37, int)   /* Needed for POSIX tcsendbreak()*/
-#define TIOCTTYGSTRUCT	_IOR('T', 38, struct tty_struct) /* For debugging only*/
 #define TIOCSBRK	_IO('T', 39) 	     /* BSD compatibility */
 #define TIOCCBRK	_IO('T', 40)	     /* BSD compatibility */
 #define TIOCGSID	_IOR('T', 41, pid_t) /* Return the session ID of FD*/
@@ -114,8 +113,10 @@
 #define TIOCSERGETLSR   _IOR('T', 89, unsigned int) /* Get line status reg. */
   /* ioctl (fd, TIOCSERGETLSR, &result) where result may be as below */
 # define TIOCSER_TEMT    0x01		     /* Transmitter physically empty */
-#define TIOCSERGETMULTI _IOR('T', 90, struct serial_multiport_struct) /* Get multiport config  */
-#define TIOCSERSETMULTI _IOW('T', 91, struct serial_multiport_struct) /* Set multiport config */
+#define TIOCSERGETMULTI 0x80a8545a /* Get multiport config  */
+			/* _IOR('T', 90, struct serial_multiport_struct) */
+#define TIOCSERSETMULTI 0x40a8545b /* Set multiport config */
+			/* _IOW('T', 91, struct serial_multiport_struct) */
 
 #define TIOCMIWAIT	_IO('T', 92) /* wait for a change on serial input line(s) */
 #define TIOCGICOUNT	0x545D	/* read serial port inline interrupt counts */
-- 
cgit v1.2.3


From 1ca49463c44c970b1ab1d71b0f268bfdf8427a7e Mon Sep 17 00:00:00 2001
From: Alan Douglas <adouglas@cadence.com>
Date: Wed, 23 Jul 2014 14:06:40 +0400
Subject: xtensa: fix address checks in dma_{alloc,free}_coherent

Virtual address is translated to the XCHAL_KSEG_CACHED region in the
dma_free_coherent, but is checked to be in the 0...XCHAL_KSEG_SIZE
range.

Change check for end of the range from 'addr >= X' to 'addr > X - 1' to
handle the case of X == 0.

Replace 'if (C) BUG();' construct with 'BUG_ON(C);'.

Cc: stable@vger.kernel.org
Signed-off-by: Alan Douglas <adouglas@cadence.com>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/kernel/pci-dma.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/arch/xtensa/kernel/pci-dma.c b/arch/xtensa/kernel/pci-dma.c
index 2d9cc6dbfd78..e8b76b8e4b29 100644
--- a/arch/xtensa/kernel/pci-dma.c
+++ b/arch/xtensa/kernel/pci-dma.c
@@ -49,9 +49,8 @@ dma_alloc_coherent(struct device *dev,size_t size,dma_addr_t *handle,gfp_t flag)
 
 	/* We currently don't support coherent memory outside KSEG */
 
-	if (ret < XCHAL_KSEG_CACHED_VADDR
-	    || ret >= XCHAL_KSEG_CACHED_VADDR + XCHAL_KSEG_SIZE)
-		BUG();
+	BUG_ON(ret < XCHAL_KSEG_CACHED_VADDR ||
+	       ret > XCHAL_KSEG_CACHED_VADDR + XCHAL_KSEG_SIZE - 1);
 
 
 	if (ret != 0) {
@@ -68,10 +67,11 @@ EXPORT_SYMBOL(dma_alloc_coherent);
 void dma_free_coherent(struct device *hwdev, size_t size,
 			 void *vaddr, dma_addr_t dma_handle)
 {
-	long addr=(long)vaddr+XCHAL_KSEG_CACHED_VADDR-XCHAL_KSEG_BYPASS_VADDR;
+	unsigned long addr = (unsigned long)vaddr +
+		XCHAL_KSEG_CACHED_VADDR - XCHAL_KSEG_BYPASS_VADDR;
 
-	if (addr < 0 || addr >= XCHAL_KSEG_SIZE)
-		BUG();
+	BUG_ON(addr < XCHAL_KSEG_CACHED_VADDR ||
+	       addr > XCHAL_KSEG_CACHED_VADDR + XCHAL_KSEG_SIZE - 1);
 
 	free_pages(addr, get_order(size));
 }
-- 
cgit v1.2.3


From 89f77c6f5bb4b0058f40f510809ec07255e02a7e Mon Sep 17 00:00:00 2001
From: Miklos Szeredi <mszeredi@suse.cz>
Date: Fri, 11 Apr 2014 12:25:50 +0200
Subject: xtensa: add renameat2 syscall

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Cc: Chris Zankel <chris@zankel.net>
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/include/uapi/asm/unistd.h | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/arch/xtensa/include/uapi/asm/unistd.h b/arch/xtensa/include/uapi/asm/unistd.h
index b9395529f02d..8883fc877c5c 100644
--- a/arch/xtensa/include/uapi/asm/unistd.h
+++ b/arch/xtensa/include/uapi/asm/unistd.h
@@ -739,7 +739,10 @@ __SYSCALL(334, sys_sched_setattr, 2)
 #define __NR_sched_getattr			335
 __SYSCALL(335, sys_sched_getattr, 3)
 
-#define __NR_syscall_count			336
+#define __NR_renameat2				336
+__SYSCALL(336, sys_renameat2, 5)
+
+#define __NR_syscall_count			337
 
 /*
  * sysxtensa syscall handler
-- 
cgit v1.2.3


From 52247123749cc3cbc30168b33ad8c69515c96d23 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Sun, 27 Jul 2014 07:23:41 +0400
Subject: xtensa: fix access to THREAD_RA/THREAD_SP/THREAD_DS

With SMP and a lot of debug options enabled task_struct::thread gets out
of reach of s32i/l32i instructions with base pointing at task_struct,
breaking build with the following messages:

  arch/xtensa/kernel/entry.S: Assembler messages:
  arch/xtensa/kernel/entry.S:1002: Error: operand 3 of 'l32i.n' has invalid value '1048'
  arch/xtensa/kernel/entry.S:1831: Error: operand 3 of 's32i.n' has invalid value '1040'
  arch/xtensa/kernel/entry.S:1832: Error: operand 3 of 's32i.n' has invalid value '1044'

Change base to point to task_struct::thread in such cases.
Don't use a10 in _switch_to to save/restore prev pointer as a2 is not
clobbered.

Cc: stable@vger.kernel.org
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/include/asm/uaccess.h |  5 +++++
 arch/xtensa/kernel/entry.S        | 12 ++++++++----
 2 files changed, 13 insertions(+), 4 deletions(-)

diff --git a/arch/xtensa/include/asm/uaccess.h b/arch/xtensa/include/asm/uaccess.h
index fd686dc45d1a..c7211e7e182d 100644
--- a/arch/xtensa/include/asm/uaccess.h
+++ b/arch/xtensa/include/asm/uaccess.h
@@ -52,7 +52,12 @@
  */
 	.macro	get_fs	ad, sp
 	GET_CURRENT(\ad,\sp)
+#if THREAD_CURRENT_DS > 1020
+	addi	\ad, \ad, TASK_THREAD
+	l32i	\ad, \ad, THREAD_CURRENT_DS - TASK_THREAD
+#else
 	l32i	\ad, \ad, THREAD_CURRENT_DS
+#endif
 	.endm
 
 /*
diff --git a/arch/xtensa/kernel/entry.S b/arch/xtensa/kernel/entry.S
index ef7f4990722b..db96acb1362b 100644
--- a/arch/xtensa/kernel/entry.S
+++ b/arch/xtensa/kernel/entry.S
@@ -1820,7 +1820,6 @@ ENTRY(_switch_to)
 
 	entry	a1, 16
 
-	mov	a10, a2			# preserve 'prev' (a2)
 	mov	a11, a3			# and 'next' (a3)
 
 	l32i	a4, a2, TASK_THREAD_INFO
@@ -1828,8 +1827,14 @@ ENTRY(_switch_to)
 
 	save_xtregs_user a4 a6 a8 a9 a12 a13 THREAD_XTREGS_USER
 
-	s32i	a0, a10, THREAD_RA	# save return address
-	s32i	a1, a10, THREAD_SP	# save stack pointer
+#if THREAD_RA > 1020 || THREAD_SP > 1020
+	addi	a10, a2, TASK_THREAD
+	s32i	a0, a10, THREAD_RA - TASK_THREAD	# save return address
+	s32i	a1, a10, THREAD_SP - TASK_THREAD	# save stack pointer
+#else
+	s32i	a0, a2, THREAD_RA	# save return address
+	s32i	a1, a2, THREAD_SP	# save stack pointer
+#endif
 
 	/* Disable ints while we manipulate the stack pointer. */
 
@@ -1870,7 +1875,6 @@ ENTRY(_switch_to)
 	load_xtregs_user a5 a6 a8 a9 a12 a13 THREAD_XTREGS_USER
 
 	wsr	a14, ps
-	mov	a2, a10			# return 'prev'
 	rsync
 
 	retw
-- 
cgit v1.2.3


From 22def7681186f65f4f1256ae9b0b6db2a7720cb1 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Tue, 15 Jul 2014 02:27:50 +0400
Subject: xtensa: make fixmap region addressing grow with index

It's much easier to reason about alignment and coloring of regions
located in the fixmap when fixmap index is just a PFN within the fixmap
region. Change fixmap addressing so that index 0 corresponds to
FIXADDR_START instead of the FIXADDR_TOP.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/include/asm/fixmap.h | 27 ++++++++++++++++++++++++---
 arch/xtensa/mm/highmem.c         |  6 +++---
 2 files changed, 27 insertions(+), 6 deletions(-)

diff --git a/arch/xtensa/include/asm/fixmap.h b/arch/xtensa/include/asm/fixmap.h
index 9f6c33d0428a..a43cd5265556 100644
--- a/arch/xtensa/include/asm/fixmap.h
+++ b/arch/xtensa/include/asm/fixmap.h
@@ -23,8 +23,8 @@
  * Here we define all the compile-time 'special' virtual
  * addresses. The point is to have a constant address at
  * compile time, but to set the physical address only
- * in the boot process. We allocate these special  addresses
- * from the end of the consistent memory region backwards.
+ * in the boot process. We allocate these special addresses
+ * from the start of the consistent memory region upwards.
  * Also this lets us do fail-safe vmalloc(), we
  * can guarantee that these special addresses and
  * vmalloc()-ed addresses never overlap.
@@ -47,7 +47,28 @@ enum fixed_addresses {
 #define FIXADDR_SIZE	(__end_of_fixed_addresses << PAGE_SHIFT)
 #define FIXADDR_START	((FIXADDR_TOP - FIXADDR_SIZE) & PMD_MASK)
 
-#include <asm-generic/fixmap.h>
+#define __fix_to_virt(x)	(FIXADDR_START + ((x) << PAGE_SHIFT))
+#define __virt_to_fix(x)	(((x) - FIXADDR_START) >> PAGE_SHIFT)
+
+#ifndef __ASSEMBLY__
+/*
+ * 'index to address' translation. If anyone tries to use the idx
+ * directly without translation, we catch the bug with a NULL-deference
+ * kernel oops. Illegal ranges of incoming indices are caught too.
+ */
+static __always_inline unsigned long fix_to_virt(const unsigned int idx)
+{
+	BUILD_BUG_ON(idx >= __end_of_fixed_addresses);
+	return __fix_to_virt(idx);
+}
+
+static inline unsigned long virt_to_fix(const unsigned long vaddr)
+{
+	BUG_ON(vaddr >= FIXADDR_TOP || vaddr < FIXADDR_START);
+	return __virt_to_fix(vaddr);
+}
+
+#endif
 
 #define kmap_get_fixmap_pte(vaddr) \
 	pte_offset_kernel( \
diff --git a/arch/xtensa/mm/highmem.c b/arch/xtensa/mm/highmem.c
index 17a8c0d6fd17..2e95a7665bf3 100644
--- a/arch/xtensa/mm/highmem.c
+++ b/arch/xtensa/mm/highmem.c
@@ -28,9 +28,9 @@ void *kmap_atomic(struct page *page)
 	idx = type + KM_TYPE_NR * smp_processor_id();
 	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
 #ifdef CONFIG_DEBUG_HIGHMEM
-	BUG_ON(!pte_none(*(kmap_pte - idx)));
+	BUG_ON(!pte_none(*(kmap_pte + idx)));
 #endif
-	set_pte(kmap_pte - idx, mk_pte(page, PAGE_KERNEL_EXEC));
+	set_pte(kmap_pte + idx, mk_pte(page, PAGE_KERNEL_EXEC));
 
 	return (void *)vaddr;
 }
@@ -51,7 +51,7 @@ void __kunmap_atomic(void *kvaddr)
 		 * is a bad idea also, in case the page changes cacheability
 		 * attributes or becomes a protected page in a hypervisor.
 		 */
-		pte_clear(&init_mm, kvaddr, kmap_pte - idx);
+		pte_clear(&init_mm, kvaddr, kmap_pte + idx);
 		local_flush_tlb_kernel_range((unsigned long)kvaddr,
 					     (unsigned long)kvaddr + PAGE_SIZE);
 
-- 
cgit v1.2.3


From dec7305d9f752f6ad2ec30ec8a723182437c5aa5 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Tue, 15 Jul 2014 02:49:15 +0400
Subject: xtensa: allow fixmap and kmap span more than one page table

To support aliasing cache both kmap region sizes are multiplied by the
number of data cache colors. After that expansion page tables that cover
kmap regions may become larger than one page. Correctly allocate and
initialize page tables in this case.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/mm/mmu.c | 38 ++++++++++++++++++++++----------------
 1 file changed, 22 insertions(+), 16 deletions(-)

diff --git a/arch/xtensa/mm/mmu.c b/arch/xtensa/mm/mmu.c
index 3429b483d9f8..abe4513eb0dd 100644
--- a/arch/xtensa/mm/mmu.c
+++ b/arch/xtensa/mm/mmu.c
@@ -18,32 +18,38 @@
 #include <asm/io.h>
 
 #if defined(CONFIG_HIGHMEM)
-static void * __init init_pmd(unsigned long vaddr)
+static void * __init init_pmd(unsigned long vaddr, unsigned long n_pages)
 {
 	pgd_t *pgd = pgd_offset_k(vaddr);
 	pmd_t *pmd = pmd_offset(pgd, vaddr);
+	pte_t *pte;
+	unsigned long i;
 
-	if (pmd_none(*pmd)) {
-		unsigned i;
-		pte_t *pte = alloc_bootmem_low_pages(PAGE_SIZE);
+	n_pages = ALIGN(n_pages, PTRS_PER_PTE);
 
-		for (i = 0; i < 1024; i++)
-			pte_clear(NULL, 0, pte + i);
+	pr_debug("%s: vaddr: 0x%08lx, n_pages: %ld\n",
+		 __func__, vaddr, n_pages);
 
-		set_pmd(pmd, __pmd(((unsigned long)pte) & PAGE_MASK));
-		BUG_ON(pte != pte_offset_kernel(pmd, 0));
-		pr_debug("%s: vaddr: 0x%08lx, pmd: 0x%p, pte: 0x%p\n",
-			 __func__, vaddr, pmd, pte);
-		return pte;
-	} else {
-		return pte_offset_kernel(pmd, 0);
+	pte = alloc_bootmem_low_pages(n_pages * sizeof(pte_t));
+
+	for (i = 0; i < n_pages; ++i)
+		pte_clear(NULL, 0, pte + i);
+
+	for (i = 0; i < n_pages; i += PTRS_PER_PTE, ++pmd) {
+		pte_t *cur_pte = pte + i;
+
+		BUG_ON(!pmd_none(*pmd));
+		set_pmd(pmd, __pmd(((unsigned long)cur_pte) & PAGE_MASK));
+		BUG_ON(cur_pte != pte_offset_kernel(pmd, 0));
+		pr_debug("%s: pmd: 0x%p, pte: 0x%p\n",
+			 __func__, pmd, cur_pte);
 	}
+	return pte;
 }
 
 static void __init fixedrange_init(void)
 {
-	BUILD_BUG_ON(FIXADDR_SIZE > PMD_SIZE);
-	init_pmd(__fix_to_virt(__end_of_fixed_addresses - 1) & PMD_MASK);
+	init_pmd(__fix_to_virt(0), __end_of_fixed_addresses);
 }
 #endif
 
@@ -52,7 +58,7 @@ void __init paging_init(void)
 	memset(swapper_pg_dir, 0, PAGE_SIZE);
 #ifdef CONFIG_HIGHMEM
 	fixedrange_init();
-	pkmap_page_table = init_pmd(PKMAP_BASE);
+	pkmap_page_table = init_pmd(PKMAP_BASE, LAST_PKMAP);
 	kmap_init();
 #endif
 }
-- 
cgit v1.2.3


From 7128039fe2dd3d59da9e4ffa036f3aaa3ba87b9f Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 21 Jul 2014 22:01:51 +0400
Subject: xtensa: fix TLBTEMP_BASE_2 region handling in fast_second_level_miss

Current definition of TLBTEMP_BASE_2 is always 32K above the
TLBTEMP_BASE_1, whereas fast_second_level_miss handler for the TLBTEMP
region analyzes virtual address bit (PAGE_SHIFT + DCACHE_ALIAS_ORDER)
to determine TLBTEMP region where the fault happened. The size of the
TLBTEMP region is also checked incorrectly: not 64K, but twice data
cache way size (whicht may as well be less than the instruction cache
way size).

Fix TLBTEMP_BASE_2 to be TLBTEMP_BASE_1 + data cache way size.
Provide TLBTEMP_SIZE that is a greater of doubled data cache way size or
the instruction cache way size, and use it to determine if the second
level TLB miss occured in the TLBTEMP region.

Practical occurence of page faults in the TLBTEMP area is extremely
rare, this code can be tested by deletion of all w[di]tlb instructions
in the tlbtemp_mapping region.

Cc: stable@vger.kernel.org
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/include/asm/pgtable.h | 7 ++++++-
 arch/xtensa/kernel/entry.S        | 2 +-
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/arch/xtensa/include/asm/pgtable.h b/arch/xtensa/include/asm/pgtable.h
index 4b0ca35a93b1..b2173e5da601 100644
--- a/arch/xtensa/include/asm/pgtable.h
+++ b/arch/xtensa/include/asm/pgtable.h
@@ -67,7 +67,12 @@
 #define VMALLOC_START		0xC0000000
 #define VMALLOC_END		0xC7FEFFFF
 #define TLBTEMP_BASE_1		0xC7FF0000
-#define TLBTEMP_BASE_2		0xC7FF8000
+#define TLBTEMP_BASE_2		(TLBTEMP_BASE_1 + DCACHE_WAY_SIZE)
+#if 2 * DCACHE_WAY_SIZE > ICACHE_WAY_SIZE
+#define TLBTEMP_SIZE		(2 * DCACHE_WAY_SIZE)
+#else
+#define TLBTEMP_SIZE		ICACHE_WAY_SIZE
+#endif
 
 /*
  * For the Xtensa architecture, the PTE layout is as follows:
diff --git a/arch/xtensa/kernel/entry.S b/arch/xtensa/kernel/entry.S
index db96acb1362b..21917e5fd53a 100644
--- a/arch/xtensa/kernel/entry.S
+++ b/arch/xtensa/kernel/entry.S
@@ -1565,7 +1565,7 @@ ENTRY(fast_second_level_miss)
 	rsr	a0, excvaddr
 	bltu	a0, a3, 2f
 
-	addi	a1, a0, -(2 << (DCACHE_ALIAS_ORDER + PAGE_SHIFT))
+	addi	a1, a0, -TLBTEMP_SIZE
 	bgeu	a1, a3, 2f
 
 	/* Check if we have to restore an ITLB mapping. */
-- 
cgit v1.2.3


From a91902db2990909ea5e6b110811b448f2e8f1571 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 21 Jul 2014 18:54:11 +0400
Subject: xtensa: implement clear_user_highpage and copy_user_highpage

Existing clear_user_page and copy_user_page cannot be used with highmem
because they calculate physical page address from its virtual address
and do it incorrectly in case of high memory page mapped with
kmap_atomic. Also kmap is not needed, as most likely userspace mapping
color would be different from the kmapped color.

Provide clear_user_highpage and copy_user_highpage functions that
determine if temporary mapping is needed for the pages. Move most of the
logic of the former clear_user_page and copy_user_page to
xtensa/mm/cache.c only leaving temporary mapping setup, invalidation and
clearing/copying in the xtensa/mm/misc.S. Rename these functions to
clear_page_alias and copy_page_alias.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/include/asm/cacheflush.h |   2 +
 arch/xtensa/include/asm/page.h       |  12 +++-
 arch/xtensa/mm/cache.c               |  63 +++++++++++++++++++
 arch/xtensa/mm/misc.S                | 116 ++++++++++++++++-------------------
 4 files changed, 127 insertions(+), 66 deletions(-)

diff --git a/arch/xtensa/include/asm/cacheflush.h b/arch/xtensa/include/asm/cacheflush.h
index 555a98a18453..e72aaca7a77f 100644
--- a/arch/xtensa/include/asm/cacheflush.h
+++ b/arch/xtensa/include/asm/cacheflush.h
@@ -37,6 +37,7 @@
  * specials for cache aliasing:
  *
  * __flush_invalidate_dcache_page_alias(vaddr,paddr)
+ * __invalidate_dcache_page_alias(vaddr,paddr)
  * __invalidate_icache_page_alias(vaddr,paddr)
  */
 
@@ -62,6 +63,7 @@ extern void __flush_invalidate_dcache_range(unsigned long, unsigned long);
 
 #if defined(CONFIG_MMU) && (DCACHE_WAY_SIZE > PAGE_SIZE)
 extern void __flush_invalidate_dcache_page_alias(unsigned long, unsigned long);
+extern void __invalidate_dcache_page_alias(unsigned long, unsigned long);
 #else
 static inline void __flush_invalidate_dcache_page_alias(unsigned long virt,
 							unsigned long phys) { }
diff --git a/arch/xtensa/include/asm/page.h b/arch/xtensa/include/asm/page.h
index 47f582333f6b..11721ccd7f23 100644
--- a/arch/xtensa/include/asm/page.h
+++ b/arch/xtensa/include/asm/page.h
@@ -134,6 +134,7 @@ static inline __attribute_const__ int get_order(unsigned long size)
 #endif
 
 struct page;
+struct vm_area_struct;
 extern void clear_page(void *page);
 extern void copy_page(void *to, void *from);
 
@@ -143,8 +144,15 @@ extern void copy_page(void *to, void *from);
  */
 
 #if DCACHE_WAY_SIZE > PAGE_SIZE
-extern void clear_user_page(void*, unsigned long, struct page*);
-extern void copy_user_page(void*, void*, unsigned long, struct page*);
+extern void clear_page_alias(void *vaddr, unsigned long paddr);
+extern void copy_page_alias(void *to, void *from,
+			    unsigned long to_paddr, unsigned long from_paddr);
+
+#define clear_user_highpage clear_user_highpage
+void clear_user_highpage(struct page *page, unsigned long vaddr);
+#define __HAVE_ARCH_COPY_USER_HIGHPAGE
+void copy_user_highpage(struct page *to, struct page *from,
+			unsigned long vaddr, struct vm_area_struct *vma);
 #else
 # define clear_user_page(page, vaddr, pg)	clear_page(page)
 # define copy_user_page(to, from, vaddr, pg)	copy_page(to, from)
diff --git a/arch/xtensa/mm/cache.c b/arch/xtensa/mm/cache.c
index 63cbb867dadd..96aea6624318 100644
--- a/arch/xtensa/mm/cache.c
+++ b/arch/xtensa/mm/cache.c
@@ -63,6 +63,69 @@
 #error "HIGHMEM is not supported on cores with aliasing cache."
 #endif
 
+#if (DCACHE_WAY_SIZE > PAGE_SIZE)
+static inline void kmap_invalidate_coherent(struct page *page,
+					    unsigned long vaddr)
+{
+	if (!DCACHE_ALIAS_EQ(page_to_phys(page), vaddr)) {
+		unsigned long kvaddr;
+
+		if (!PageHighMem(page)) {
+			kvaddr = (unsigned long)page_to_virt(page);
+
+			__invalidate_dcache_page(kvaddr);
+		} else {
+			kvaddr = TLBTEMP_BASE_1 +
+				(page_to_phys(page) & DCACHE_ALIAS_MASK);
+
+			__invalidate_dcache_page_alias(kvaddr,
+						       page_to_phys(page));
+		}
+	}
+}
+
+static inline void *coherent_kvaddr(struct page *page, unsigned long base,
+				    unsigned long vaddr, unsigned long *paddr)
+{
+	if (PageHighMem(page) || !DCACHE_ALIAS_EQ(page_to_phys(page), vaddr)) {
+		*paddr = page_to_phys(page);
+		return (void *)(base + (vaddr & DCACHE_ALIAS_MASK));
+	} else {
+		*paddr = 0;
+		return page_to_virt(page);
+	}
+}
+
+void clear_user_highpage(struct page *page, unsigned long vaddr)
+{
+	unsigned long paddr;
+	void *kvaddr = coherent_kvaddr(page, TLBTEMP_BASE_1, vaddr, &paddr);
+
+	pagefault_disable();
+	kmap_invalidate_coherent(page, vaddr);
+	set_bit(PG_arch_1, &page->flags);
+	clear_page_alias(kvaddr, paddr);
+	pagefault_enable();
+}
+
+void copy_user_highpage(struct page *dst, struct page *src,
+			unsigned long vaddr, struct vm_area_struct *vma)
+{
+	unsigned long dst_paddr, src_paddr;
+	void *dst_vaddr = coherent_kvaddr(dst, TLBTEMP_BASE_1, vaddr,
+					  &dst_paddr);
+	void *src_vaddr = coherent_kvaddr(src, TLBTEMP_BASE_2, vaddr,
+					  &src_paddr);
+
+	pagefault_disable();
+	kmap_invalidate_coherent(dst, vaddr);
+	set_bit(PG_arch_1, &dst->flags);
+	copy_page_alias(dst_vaddr, src_vaddr, dst_paddr, src_paddr);
+	pagefault_enable();
+}
+
+#endif /* DCACHE_WAY_SIZE > PAGE_SIZE */
+
 #if (DCACHE_WAY_SIZE > PAGE_SIZE) && XCHAL_DCACHE_IS_WRITEBACK
 
 /*
diff --git a/arch/xtensa/mm/misc.S b/arch/xtensa/mm/misc.S
index 1f68558dbcc2..11a01c3e9cea 100644
--- a/arch/xtensa/mm/misc.S
+++ b/arch/xtensa/mm/misc.S
@@ -110,41 +110,24 @@ ENTRY(__tlbtemp_mapping_start)
 #if (DCACHE_WAY_SIZE > PAGE_SIZE)
 
 /*
- * clear_user_page (void *addr, unsigned long vaddr, struct page *page)
- *                     a2              a3                 a4
+ * clear_page_alias(void *addr, unsigned long paddr)
+ *                     a2              a3
  */
 
-ENTRY(clear_user_page)
+ENTRY(clear_page_alias)
 
 	entry	a1, 32
 
-	/* Mark page dirty and determine alias. */
+	/* Skip setting up a temporary DTLB if not aliased low page. */
 
-	movi	a7, (1 << PG_ARCH_1)
-	l32i	a5, a4, PAGE_FLAGS
-	xor	a6, a2, a3
-	extui	a3, a3, PAGE_SHIFT, DCACHE_ALIAS_ORDER
-	extui	a6, a6, PAGE_SHIFT, DCACHE_ALIAS_ORDER
-	or	a5, a5, a7
-	slli	a3, a3, PAGE_SHIFT
-	s32i	a5, a4, PAGE_FLAGS
+	movi	a5, PAGE_OFFSET
+	movi	a6, 0
+	beqz	a3, 1f
 
-	/* Skip setting up a temporary DTLB if not aliased. */
-
-	beqz	a6, 1f
-
-	/* Invalidate kernel page. */
-
-	mov	a10, a2
-	call8	__invalidate_dcache_page
-
-	/* Setup a temporary DTLB with the color of the VPN */
-
-	movi	a4, ((PAGE_KERNEL | _PAGE_HW_WRITE) - PAGE_OFFSET) & 0xffffffff
-	movi	a5, TLBTEMP_BASE_1			# virt
-	add	a6, a2, a4				# ppn
-	add	a2, a5, a3				# add 'color'
+	/* Setup a temporary DTLB for the addr. */
 
+	addi	a6, a3, (PAGE_KERNEL | _PAGE_HW_WRITE)
+	mov	a4, a2
 	wdtlb	a6, a2
 	dsync
 
@@ -165,62 +148,43 @@ ENTRY(clear_user_page)
 
 	/* We need to invalidate the temporary idtlb entry, if any. */
 
-1:	addi	a2, a2, -PAGE_SIZE
-	idtlb	a2
+1:	idtlb	a4
 	dsync
 
 	retw
 
-ENDPROC(clear_user_page)
+ENDPROC(clear_page_alias)
 
 /*
- * copy_page_user (void *to, void *from, unsigned long vaddr, struct page *page)
- *                    a2          a3	        a4		    a5
+ * copy_page_alias(void *to, void *from,
+ *			a2	  a3
+ *                 unsigned long to_paddr, unsigned long from_paddr)
+ *	        		 a4			 a5
  */
 
-ENTRY(copy_user_page)
+ENTRY(copy_page_alias)
 
 	entry	a1, 32
 
-	/* Mark page dirty and determine alias for destination. */
-
-	movi	a8, (1 << PG_ARCH_1)
-	l32i	a9, a5, PAGE_FLAGS
-	xor	a6, a2, a4
-	xor	a7, a3, a4
-	extui	a4, a4, PAGE_SHIFT, DCACHE_ALIAS_ORDER
-	extui	a6, a6, PAGE_SHIFT, DCACHE_ALIAS_ORDER
-	extui	a7, a7, PAGE_SHIFT, DCACHE_ALIAS_ORDER
-	or	a9, a9, a8
-	slli	a4, a4, PAGE_SHIFT
-	s32i	a9, a5, PAGE_FLAGS
-	movi	a5, ((PAGE_KERNEL | _PAGE_HW_WRITE) - PAGE_OFFSET) & 0xffffffff
-
-	beqz	a6, 1f
-
-	/* Invalidate dcache */
-
-	mov	a10, a2
-	call8	__invalidate_dcache_page
+	/* Skip setting up a temporary DTLB for destination if not aliased. */
 
-	/* Setup a temporary DTLB with a matching color. */
+	movi	a6, 0
+	movi	a7, 0
+	beqz	a4, 1f
 
-	movi	a8, TLBTEMP_BASE_1			# base
-	add	a6, a2, a5				# ppn
-	add	a2, a8, a4				# add 'color'
+	/* Setup a temporary DTLB for destination. */
 
+	addi	a6, a4, (PAGE_KERNEL | _PAGE_HW_WRITE)
 	wdtlb	a6, a2
 	dsync
 
-	/* Skip setting up a temporary DTLB for destination if not aliased. */
+	/* Skip setting up a temporary DTLB for source if not aliased. */
 
-1:	beqz	a7, 1f
+1:	beqz	a5, 1f
 
-	/* Setup a temporary DTLB with a matching color. */
+	/* Setup a temporary DTLB for source. */
 
-	movi	a8, TLBTEMP_BASE_2			# base
-	add	a7, a3, a5				# ppn
-	add	a3, a8, a4
+	addi	a7, a5, PAGE_KERNEL
 	addi	a8, a3, 1				# way1
 
 	wdtlb	a7, a8
@@ -271,7 +235,7 @@ ENTRY(copy_user_page)
 
 	retw
 
-ENDPROC(copy_user_page)
+ENDPROC(copy_page_alias)
 
 #endif
 
@@ -300,6 +264,30 @@ ENTRY(__flush_invalidate_dcache_page_alias)
 	retw
 
 ENDPROC(__flush_invalidate_dcache_page_alias)
+
+/*
+ * void __invalidate_dcache_page_alias (addr, phys)
+ *                                       a2    a3
+ */
+
+ENTRY(__invalidate_dcache_page_alias)
+
+	entry	sp, 16
+
+	movi	a7, 0			# required for exception handler
+	addi	a6, a3, (PAGE_KERNEL | _PAGE_HW_WRITE)
+	mov	a4, a2
+	wdtlb	a6, a2
+	dsync
+
+	___invalidate_dcache_page a2 a3
+
+	idtlb	a4
+	dsync
+
+	retw
+
+ENDPROC(__invalidate_dcache_page_alias)
 #endif
 
 ENTRY(__tlbtemp_mapping_itlb)
-- 
cgit v1.2.3


From 32544d9c10c42bac3be8b87d2fc95b0aef008795 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Tue, 15 Jul 2014 02:51:49 +0400
Subject: xtensa: support aliasing cache in k[un]map_atomic

Map high memory pages at virtual addresses with color that match color
of their physical address. Existing cache alias management mechanisms
may be used with such pages.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/include/asm/fixmap.h |  3 ++-
 arch/xtensa/include/asm/page.h   |  2 ++
 arch/xtensa/mm/highmem.c         | 17 ++++++++++-------
 3 files changed, 14 insertions(+), 8 deletions(-)

diff --git a/arch/xtensa/include/asm/fixmap.h b/arch/xtensa/include/asm/fixmap.h
index a43cd5265556..62b507deea9d 100644
--- a/arch/xtensa/include/asm/fixmap.h
+++ b/arch/xtensa/include/asm/fixmap.h
@@ -38,7 +38,8 @@ enum fixed_addresses {
 #ifdef CONFIG_HIGHMEM
 	/* reserved pte's for temporary kernel mappings */
 	FIX_KMAP_BEGIN,
-	FIX_KMAP_END = FIX_KMAP_BEGIN + (KM_TYPE_NR * NR_CPUS) - 1,
+	FIX_KMAP_END = FIX_KMAP_BEGIN +
+		(KM_TYPE_NR * NR_CPUS * DCACHE_N_COLORS) - 1,
 #endif
 	__end_of_fixed_addresses
 };
diff --git a/arch/xtensa/include/asm/page.h b/arch/xtensa/include/asm/page.h
index 11721ccd7f23..abe24c6f8b2f 100644
--- a/arch/xtensa/include/asm/page.h
+++ b/arch/xtensa/include/asm/page.h
@@ -78,7 +78,9 @@
 # define DCACHE_ALIAS_EQ(a,b)	((((a) ^ (b)) & DCACHE_ALIAS_MASK) == 0)
 #else
 # define DCACHE_ALIAS_ORDER	0
+# define DCACHE_ALIAS(a)	((void)(a), 0)
 #endif
+#define DCACHE_N_COLORS		(1 << DCACHE_ALIAS_ORDER)
 
 #if ICACHE_WAY_SIZE > PAGE_SIZE
 # define ICACHE_ALIAS_ORDER	(ICACHE_WAY_SHIFT - PAGE_SHIFT)
diff --git a/arch/xtensa/mm/highmem.c b/arch/xtensa/mm/highmem.c
index 2e95a7665bf3..466abaed5382 100644
--- a/arch/xtensa/mm/highmem.c
+++ b/arch/xtensa/mm/highmem.c
@@ -14,18 +14,23 @@
 
 static pte_t *kmap_pte;
 
+static inline enum fixed_addresses kmap_idx(int type, unsigned long color)
+{
+	return (type + KM_TYPE_NR * smp_processor_id()) * DCACHE_N_COLORS +
+		color;
+}
+
 void *kmap_atomic(struct page *page)
 {
 	enum fixed_addresses idx;
 	unsigned long vaddr;
-	int type;
 
 	pagefault_disable();
 	if (!PageHighMem(page))
 		return page_address(page);
 
-	type = kmap_atomic_idx_push();
-	idx = type + KM_TYPE_NR * smp_processor_id();
+	idx = kmap_idx(kmap_atomic_idx_push(),
+		       DCACHE_ALIAS(page_to_phys(page)));
 	vaddr = __fix_to_virt(FIX_KMAP_BEGIN + idx);
 #ifdef CONFIG_DEBUG_HIGHMEM
 	BUG_ON(!pte_none(*(kmap_pte + idx)));
@@ -38,12 +43,10 @@ EXPORT_SYMBOL(kmap_atomic);
 
 void __kunmap_atomic(void *kvaddr)
 {
-	int idx, type;
-
 	if (kvaddr >= (void *)FIXADDR_START &&
 	    kvaddr < (void *)FIXADDR_TOP) {
-		type = kmap_atomic_idx();
-		idx = type + KM_TYPE_NR * smp_processor_id();
+		int idx = kmap_idx(kmap_atomic_idx(),
+				   DCACHE_ALIAS((unsigned long)kvaddr));
 
 		/*
 		 * Force other mappings to Oops if they'll try to access this
-- 
cgit v1.2.3


From 8504b503dfa86f698a38f9ee1fc2876ab012b776 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Thu, 17 Jul 2014 05:04:49 +0400
Subject: xtensa: support aliasing cache in kmap

Define ARCH_PKMAP_COLORING and provide corresponding macro definitions
on cores with aliasing data cache.

Instead of single last_pkmap_nr maintain an array last_pkmap_nr_arr of
pkmap counters for each page color. Make sure that kmap maps physical
page at virtual address with color matching its physical address.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/include/asm/highmem.h | 40 +++++++++++++++++++++++++++++++++++++--
 arch/xtensa/mm/highmem.c          | 18 ++++++++++++++++++
 2 files changed, 56 insertions(+), 2 deletions(-)

diff --git a/arch/xtensa/include/asm/highmem.h b/arch/xtensa/include/asm/highmem.h
index 2653ef5d55f1..2c7901edffaf 100644
--- a/arch/xtensa/include/asm/highmem.h
+++ b/arch/xtensa/include/asm/highmem.h
@@ -12,19 +12,55 @@
 #ifndef _XTENSA_HIGHMEM_H
 #define _XTENSA_HIGHMEM_H
 
+#include <linux/wait.h>
 #include <asm/cacheflush.h>
 #include <asm/fixmap.h>
 #include <asm/kmap_types.h>
 #include <asm/pgtable.h>
 
-#define PKMAP_BASE		(FIXADDR_START - PMD_SIZE)
-#define LAST_PKMAP		PTRS_PER_PTE
+#define PKMAP_BASE		((FIXADDR_START - \
+				  (LAST_PKMAP + 1) * PAGE_SIZE) & PMD_MASK)
+#define LAST_PKMAP		(PTRS_PER_PTE * DCACHE_N_COLORS)
 #define LAST_PKMAP_MASK		(LAST_PKMAP - 1)
 #define PKMAP_NR(virt)		(((virt) - PKMAP_BASE) >> PAGE_SHIFT)
 #define PKMAP_ADDR(nr)		(PKMAP_BASE + ((nr) << PAGE_SHIFT))
 
 #define kmap_prot		PAGE_KERNEL
 
+#if DCACHE_WAY_SIZE > PAGE_SIZE
+#define get_pkmap_color get_pkmap_color
+static inline int get_pkmap_color(struct page *page)
+{
+	return DCACHE_ALIAS(page_to_phys(page));
+}
+
+extern unsigned int last_pkmap_nr_arr[];
+
+static inline unsigned int get_next_pkmap_nr(unsigned int color)
+{
+	last_pkmap_nr_arr[color] =
+		(last_pkmap_nr_arr[color] + DCACHE_N_COLORS) & LAST_PKMAP_MASK;
+	return last_pkmap_nr_arr[color] + color;
+}
+
+static inline int no_more_pkmaps(unsigned int pkmap_nr, unsigned int color)
+{
+	return pkmap_nr < DCACHE_N_COLORS;
+}
+
+static inline int get_pkmap_entries_count(unsigned int color)
+{
+	return LAST_PKMAP / DCACHE_N_COLORS;
+}
+
+extern wait_queue_head_t pkmap_map_wait_arr[];
+
+static inline wait_queue_head_t *get_pkmap_wait_queue_head(unsigned int color)
+{
+	return pkmap_map_wait_arr + color;
+}
+#endif
+
 extern pte_t *pkmap_page_table;
 
 void *kmap_high(struct page *page);
diff --git a/arch/xtensa/mm/highmem.c b/arch/xtensa/mm/highmem.c
index 466abaed5382..8cfb71ec0937 100644
--- a/arch/xtensa/mm/highmem.c
+++ b/arch/xtensa/mm/highmem.c
@@ -14,6 +14,23 @@
 
 static pte_t *kmap_pte;
 
+#if DCACHE_WAY_SIZE > PAGE_SIZE
+unsigned int last_pkmap_nr_arr[DCACHE_N_COLORS];
+wait_queue_head_t pkmap_map_wait_arr[DCACHE_N_COLORS];
+
+static void __init kmap_waitqueues_init(void)
+{
+	unsigned int i;
+
+	for (i = 0; i < ARRAY_SIZE(pkmap_map_wait_arr); ++i)
+		init_waitqueue_head(pkmap_map_wait_arr + i);
+}
+#else
+static inline void kmap_waitqueues_init(void)
+{
+}
+#endif
+
 static inline enum fixed_addresses kmap_idx(int type, unsigned long color)
 {
 	return (type + KM_TYPE_NR * smp_processor_id()) * DCACHE_N_COLORS +
@@ -72,4 +89,5 @@ void __init kmap_init(void)
 	/* cache the first kmap pte */
 	kmap_vstart = __fix_to_virt(FIX_KMAP_BEGIN);
 	kmap_pte = kmap_get_fixmap_pte(kmap_vstart);
+	kmap_waitqueues_init();
 }
-- 
cgit v1.2.3


From 270eec76de2557c9df01d74bc4c948d0924fc007 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 21 Jul 2014 04:24:40 +0400
Subject: xtensa: support highmem in aliasing cache flushing code

Use __flush_invalidate_dcache_page_alias with alias set to color of the
page physical address instead of __flush_invalidate_dcache_page: this
works for high memory pages and mapping/unmapping to the TLBTEMP area is
virtually free.

Allow building configurations with aliasing cache and highmem enabled.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/mm/cache.c | 16 ++++++----------
 1 file changed, 6 insertions(+), 10 deletions(-)

diff --git a/arch/xtensa/mm/cache.c b/arch/xtensa/mm/cache.c
index 96aea6624318..d75aa1476da7 100644
--- a/arch/xtensa/mm/cache.c
+++ b/arch/xtensa/mm/cache.c
@@ -59,10 +59,6 @@
  *
  */
 
-#if (DCACHE_WAY_SIZE > PAGE_SIZE) && defined(CONFIG_HIGHMEM)
-#error "HIGHMEM is not supported on cores with aliasing cache."
-#endif
-
 #if (DCACHE_WAY_SIZE > PAGE_SIZE)
 static inline void kmap_invalidate_coherent(struct page *page,
 					    unsigned long vaddr)
@@ -166,7 +162,8 @@ void flush_dcache_page(struct page *page)
 		if (!alias && !mapping)
 			return;
 
-		__flush_invalidate_dcache_page((long)page_address(page));
+		virt = TLBTEMP_BASE_1 + (phys & DCACHE_ALIAS_MASK);
+		__flush_invalidate_dcache_page_alias(virt, phys);
 
 		virt = TLBTEMP_BASE_1 + (temp & DCACHE_ALIAS_MASK);
 
@@ -231,13 +228,12 @@ update_mmu_cache(struct vm_area_struct * vma, unsigned long addr, pte_t *ptep)
 #if (DCACHE_WAY_SIZE > PAGE_SIZE) && XCHAL_DCACHE_IS_WRITEBACK
 
 	if (!PageReserved(page) && test_bit(PG_arch_1, &page->flags)) {
-
-		unsigned long paddr = (unsigned long) page_address(page);
 		unsigned long phys = page_to_phys(page);
-		unsigned long tmp = TLBTEMP_BASE_1 + (addr & DCACHE_ALIAS_MASK);
-
-		__flush_invalidate_dcache_page(paddr);
+		unsigned long tmp;
 
+		tmp = TLBTEMP_BASE_1 + (phys & DCACHE_ALIAS_MASK);
+		__flush_invalidate_dcache_page_alias(tmp, phys);
+		tmp = TLBTEMP_BASE_1 + (addr & DCACHE_ALIAS_MASK);
 		__flush_invalidate_dcache_page_alias(tmp, phys);
 		__invalidate_icache_page_alias(tmp, phys);
 
-- 
cgit v1.2.3


From b82837c772c3a1d8778295ab094cf46ecddc8057 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 24 Mar 2014 05:16:02 +0400
Subject: xtensa: configure kc705 for highmem

Enable all memory available on KC705 (1G - 128M) by default. Update memory
node in DTS and also limit usable memory in bootargs in case memmap is
passed from the bootloader.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/boot/dts/kc705.dts | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/arch/xtensa/boot/dts/kc705.dts b/arch/xtensa/boot/dts/kc705.dts
index 742a347be67a..c4d17a34ab86 100644
--- a/arch/xtensa/boot/dts/kc705.dts
+++ b/arch/xtensa/boot/dts/kc705.dts
@@ -4,8 +4,11 @@
 
 / {
 	compatible = "cdns,xtensa-kc705";
+	chosen {
+		bootargs = "earlycon=uart8250,mmio32,0xfd050020,115200n8 console=ttyS0,115200n8 ip=dhcp root=/dev/nfs rw debug memmap=0x38000000";
+	};
 	memory@0 {
 		device_type = "memory";
-		reg = <0x00000000 0x08000000>;
+		reg = <0x00000000 0x38000000>;
 	};
 };
-- 
cgit v1.2.3


From a450dc69dc57e2bd9de5a970f5015502e6950c73 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Fri, 1 Aug 2014 19:07:10 +0400
Subject: xtensa: fix kernel/user jump out of fast_unaligned

Use correct register (a0, just read from the PS) to check user mode bit.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/kernel/align.S | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/xtensa/kernel/align.S b/arch/xtensa/kernel/align.S
index d4cef6039a5c..25a65938dda8 100644
--- a/arch/xtensa/kernel/align.S
+++ b/arch/xtensa/kernel/align.S
@@ -441,7 +441,7 @@ ENTRY(fast_unaligned)
 	mov	a1, a2
 
 	rsr	a0, ps
-	bbsi.l  a2, PS_UM_BIT, 1f     # jump if user mode
+	bbsi.l  a0, PS_UM_BIT, 1f     # jump if user mode
 
 	movi	a0, _kernel_exception
 	jx	a0
-- 
cgit v1.2.3


From c3ef1f4d379cbc79daf80ffb8d43c611da090b82 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Wed, 22 Jan 2014 09:16:37 +0400
Subject: xtensa: add double exception fixup handler for fast_unaligned

fast_unaligned_fixup restores user registers and runs normal exception
handler in the current stack frame. Unaligned load/store is retried
after that.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/kernel/align.S | 43 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 43 insertions(+)

diff --git a/arch/xtensa/kernel/align.S b/arch/xtensa/kernel/align.S
index 25a65938dda8..904f32f05c09 100644
--- a/arch/xtensa/kernel/align.S
+++ b/arch/xtensa/kernel/align.S
@@ -8,6 +8,7 @@
  * this archive for more details.
  *
  * Copyright (C) 2001 - 2005 Tensilica, Inc.
+ * Copyright (C) 2014 Cadence Design Systems Inc.
  *
  * Rewritten by Chris Zankel <chris@zankel.net>
  *
@@ -174,6 +175,10 @@ ENTRY(fast_unaligned)
 	s32i	a0, a2, PT_AREG2
 	s32i	a3, a2, PT_AREG3
 
+	rsr	a3, excsave1
+	movi	a4, fast_unaligned_fixup
+	s32i	a4, a3, EXC_TABLE_FIXUP
+
 	/* Keep value of SAR in a0 */
 
 	rsr	a0, sar
@@ -430,6 +435,10 @@ ENTRY(fast_unaligned)
 .Linvalid_instruction_store:
 .Linvalid_instruction:
 
+	movi	a4, 0
+	rsr	a3, excsave1
+	s32i	a4, a3, EXC_TABLE_FIXUP
+
 	/* Restore a4...a8 and SAR, set SP, and jump to default exception. */
 
 	l32i	a8, a2, PT_AREG8
@@ -451,4 +460,38 @@ ENTRY(fast_unaligned)
 
 ENDPROC(fast_unaligned)
 
+ENTRY(fast_unaligned_fixup)
+
+	l32i	a2, a3, EXC_TABLE_DOUBLE_SAVE
+	wsr	a3, excsave1
+
+	l32i	a8, a2, PT_AREG8
+	l32i	a7, a2, PT_AREG7
+	l32i	a6, a2, PT_AREG6
+	l32i	a5, a2, PT_AREG5
+	l32i	a4, a2, PT_AREG4
+	l32i	a0, a2, PT_AREG2
+	xsr	a0, depc			# restore depc and a0
+	wsr	a0, sar
+
+	rsr	a0, exccause
+	s32i	a0, a2, PT_DEPC			# mark as a regular exception
+
+	rsr	a0, ps
+	bbsi.l  a0, PS_UM_BIT, 1f		# jump if user mode
+
+	rsr	a0, exccause
+	addx4	a0, a0, a3              	# find entry in table
+	l32i	a0, a0, EXC_TABLE_FAST_KERNEL   # load handler
+	l32i	a3, a2, PT_AREG3
+	jx	a0
+1:
+	rsr	a0, exccause
+	addx4	a0, a0, a3              	# find entry in table
+	l32i	a0, a0, EXC_TABLE_FAST_USER     # load handler
+	l32i	a3, a2, PT_AREG3
+	jx	a0
+
+ENDPROC(fast_unaligned_fixup)
+
 #endif /* XCHAL_UNALIGNED_LOAD_EXCEPTION || XCHAL_UNALIGNED_STORE_EXCEPTION */
-- 
cgit v1.2.3


From e9500dd852ca6ede346500010545975bf10244dc Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Sun, 3 Aug 2014 00:42:38 +0400
Subject: xtensa: make fast_unaligned store restartable

fast_unaligned may encounter DTLB miss or SEGFAULT during the store
emulation. Don't update epc1 and lcount until after the store emulation
is complete, so that the faulting store instruction could be replayed.
Remove duplicate code handling zero overhead loops and calculate new
epc1 and lcount in one place.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/kernel/align.S | 51 +++++++++++++++++-----------------------------
 1 file changed, 19 insertions(+), 32 deletions(-)

diff --git a/arch/xtensa/kernel/align.S b/arch/xtensa/kernel/align.S
index 904f32f05c09..2c7c13d8e91a 100644
--- a/arch/xtensa/kernel/align.S
+++ b/arch/xtensa/kernel/align.S
@@ -277,18 +277,6 @@ ENTRY(fast_unaligned)
 	/* Set target register. */
 
 1:
-
-#if XCHAL_HAVE_LOOPS
-	rsr	a5, lend		# check if we reached LEND
-	bne	a7, a5, 1f
-	rsr	a5, lcount		# and LCOUNT != 0
-	beqz	a5, 1f
-	addi	a5, a5, -1		# decrement LCOUNT and set
-	rsr	a7, lbeg		# set PC to LBEGIN
-	wsr	a5, lcount
-#endif
-
-1:	wsr	a7, epc1		# skip load instruction
 	extui	a4, a4, INSN_T, 4	# extract target register
 	movi	a5, .Lload_table
 	addx8	a4, a4, a5
@@ -358,17 +346,6 @@ ENTRY(fast_unaligned)
 	/* Get memory address */
 
 1:
-#if XCHAL_HAVE_LOOPS
-	rsr	a4, lend		# check if we reached LEND
-	bne	a7, a4, 1f
-	rsr	a4, lcount		# and LCOUNT != 0
-	beqz	a4, 1f
-	addi	a4, a4, -1		# decrement LCOUNT and set
-	rsr	a7, lbeg		# set PC to LBEGIN
-	wsr	a4, lcount
-#endif
-
-1:	wsr	a7, epc1		# skip store instruction
 	movi	a4, ~3
 	and	a4, a4, a8		# align memory address
 
@@ -380,25 +357,25 @@ ENTRY(fast_unaligned)
 #endif
 
 	__ssa8r a8
-	__src_b	a7, a5, a6		# lo-mask  F..F0..0 (BE) 0..0F..F (LE)
+	__src_b	a8, a5, a6		# lo-mask  F..F0..0 (BE) 0..0F..F (LE)
 	__src_b	a6, a6, a5		# hi-mask  0..0F..F (BE) F..F0..0 (LE)
 #ifdef UNALIGNED_USER_EXCEPTION
 	l32e	a5, a4, -8
 #else
 	l32i	a5, a4, 0		# load lower address word
 #endif
-	and	a5, a5, a7		# mask
-	__sh	a7, a3 			# shift value
-	or	a5, a5, a7		# or with original value
+	and	a5, a5, a8		# mask
+	__sh	a8, a3 			# shift value
+	or	a5, a5, a8		# or with original value
 #ifdef UNALIGNED_USER_EXCEPTION
 	s32e	a5, a4, -8
-	l32e	a7, a4, -4
+	l32e	a8, a4, -4
 #else
 	s32i	a5, a4, 0		# store
-	l32i	a7, a4, 4		# same for upper address word
+	l32i	a8, a4, 4		# same for upper address word
 #endif
 	__sl	a5, a3
-	and	a6, a7, a6
+	and	a6, a8, a6
 	or	a6, a6, a5
 #ifdef UNALIGNED_USER_EXCEPTION
 	s32e	a6, a4, -4
@@ -406,9 +383,19 @@ ENTRY(fast_unaligned)
 	s32i	a6, a4, 4
 #endif
 
-	/* Done. restore stack and return */
-
 .Lexit:
+#if XCHAL_HAVE_LOOPS
+	rsr	a4, lend		# check if we reached LEND
+	bne	a7, a4, 1f
+	rsr	a4, lcount		# and LCOUNT != 0
+	beqz	a4, 1f
+	addi	a4, a4, -1		# decrement LCOUNT and set
+	rsr	a7, lbeg		# set PC to LBEGIN
+	wsr	a4, lcount
+#endif
+
+1:	wsr	a7, epc1		# skip emulated instruction
+
 	movi	a4, 0
 	rsr	a3, excsave1
 	s32i	a4, a3, EXC_TABLE_FIXUP
-- 
cgit v1.2.3


From 21570465a30f13197991eb2637d6ffc6c6880eef Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 4 Aug 2014 15:24:58 +0400
Subject: xtensa: move invalid unaligned instruction handler closer to its
 users

With this change a threaded jump from .Linvalid_instruction_load to
.Linvalid_instruction can be removed and more code may be added to
common load/store exit path.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/kernel/align.S | 62 ++++++++++++++++++++++------------------------
 1 file changed, 29 insertions(+), 33 deletions(-)

diff --git a/arch/xtensa/kernel/align.S b/arch/xtensa/kernel/align.S
index 2c7c13d8e91a..87d80d8c33e5 100644
--- a/arch/xtensa/kernel/align.S
+++ b/arch/xtensa/kernel/align.S
@@ -230,10 +230,6 @@ ENTRY(fast_unaligned)
 	addx8	a5, a6, a5
 	jx	a5			# jump into table
 
-	/* Invalid instruction, CRITICAL! */
-.Linvalid_instruction_load:
-	j	.Linvalid_instruction
-
 	/* Load: Load memory address. */
 
 .Lload: movi	a3, ~3
@@ -319,6 +315,35 @@ ENTRY(fast_unaligned)
 	mov	a3, a14		;	_j 1f;	.align 8
 	mov	a3, a15		;	_j 1f;	.align 8
 
+	/* We cannot handle this exception. */
+
+	.extern _kernel_exception
+.Linvalid_instruction_load:
+.Linvalid_instruction_store:
+
+	movi	a4, 0
+	rsr	a3, excsave1
+	s32i	a4, a3, EXC_TABLE_FIXUP
+
+	/* Restore a4...a8 and SAR, set SP, and jump to default exception. */
+
+	l32i	a8, a2, PT_AREG8
+	l32i	a7, a2, PT_AREG7
+	l32i	a6, a2, PT_AREG6
+	l32i	a5, a2, PT_AREG5
+	l32i	a4, a2, PT_AREG4
+	wsr	a0, sar
+	mov	a1, a2
+
+	rsr	a0, ps
+	bbsi.l  a0, PS_UM_BIT, 2f     # jump if user mode
+
+	movi	a0, _kernel_exception
+	jx	a0
+
+2:	movi	a0, _user_exception
+	jx	a0
+
 1: 	# a7: instruction pointer, a4: instruction, a3: value
 
 	movi	a6, 0			# mask: ffffffff:00000000
@@ -416,35 +441,6 @@ ENTRY(fast_unaligned)
 	l32i	a2, a2, PT_AREG2
 	rfe
 
-	/* We cannot handle this exception. */
-
-	.extern _kernel_exception
-.Linvalid_instruction_store:
-.Linvalid_instruction:
-
-	movi	a4, 0
-	rsr	a3, excsave1
-	s32i	a4, a3, EXC_TABLE_FIXUP
-
-	/* Restore a4...a8 and SAR, set SP, and jump to default exception. */
-
-	l32i	a8, a2, PT_AREG8
-	l32i	a7, a2, PT_AREG7
-	l32i	a6, a2, PT_AREG6
-	l32i	a5, a2, PT_AREG5
-	l32i	a4, a2, PT_AREG4
-	wsr	a0, sar
-	mov	a1, a2
-
-	rsr	a0, ps
-	bbsi.l  a0, PS_UM_BIT, 1f     # jump if user mode
-
-	movi	a0, _kernel_exception
-	jx	a0
-
-1:	movi	a0, _user_exception
-	jx	a0
-
 ENDPROC(fast_unaligned)
 
 ENTRY(fast_unaligned_fixup)
-- 
cgit v1.2.3


From a83b02e9bd0c28d27b6c6e5b184585f7a1b8bf86 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Mon, 4 Aug 2014 05:55:53 +0400
Subject: xtensa: allow single-stepping through unaligned load/store

Update icount when icountlevel is non-zero but not greater than EXCM level
when load/store instruction is successfully emulated. This allows
single-stepping over such instruction in userspace debugger.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/kernel/align.S | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/arch/xtensa/kernel/align.S b/arch/xtensa/kernel/align.S
index 87d80d8c33e5..890004af03a9 100644
--- a/arch/xtensa/kernel/align.S
+++ b/arch/xtensa/kernel/align.S
@@ -421,6 +421,14 @@ ENTRY(fast_unaligned)
 
 1:	wsr	a7, epc1		# skip emulated instruction
 
+	/* Update icount if we're single-stepping in userspace. */
+	rsr	a4, icountlevel
+	beqz	a4, 1f
+	bgeui	a4, LOCKLEVEL + 1, 1f
+	rsr	a4, icount
+	addi	a4, a4, 1
+	wsr	a4, icount
+1:
 	movi	a4, 0
 	rsr	a3, excsave1
 	s32i	a4, a3, EXC_TABLE_FIXUP
-- 
cgit v1.2.3


From d1b6ba82a50cecf94be540a3a153aa89d97511a0 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Thu, 31 Jul 2014 22:40:57 +0400
Subject: xtensa: fix a6 and a7 handling in fast_syscall_xtensa

Remove restoring a6 on some return paths and instead modify and restore
it in a single place, using symbolic name.
Correctly restore a7 from PT_AREG7 in case of illegal a6 value.

Cc: stable@vger.kernel.org
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/kernel/entry.S | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/arch/xtensa/kernel/entry.S b/arch/xtensa/kernel/entry.S
index 21917e5fd53a..a06b7efaae82 100644
--- a/arch/xtensa/kernel/entry.S
+++ b/arch/xtensa/kernel/entry.S
@@ -1001,9 +1001,8 @@ ENTRY(fast_syscall_xtensa)
 	movi	a7, 4			# sizeof(unsigned int)
 	access_ok a3, a7, a0, a2, .Leac	# a0: scratch reg, a2: sp
 
-	addi	a6, a6, -1		# assuming SYS_XTENSA_ATOMIC_SET = 1
-	_bgeui	a6, SYS_XTENSA_COUNT - 1, .Lill
-	_bnei	a6, SYS_XTENSA_ATOMIC_CMP_SWP - 1, .Lnswp
+	_bgeui	a6, SYS_XTENSA_COUNT, .Lill
+	_bnei	a6, SYS_XTENSA_ATOMIC_CMP_SWP, .Lnswp
 
 	/* Fall through for ATOMIC_CMP_SWP. */
 
@@ -1015,27 +1014,26 @@ TRY	s32i	a5, a3, 0		# different, modify value
 	l32i	a7, a2, PT_AREG7	# restore a7
 	l32i	a0, a2, PT_AREG0	# restore a0
 	movi	a2, 1			# and return 1
-	addi	a6, a6, 1		# restore a6 (really necessary?)
 	rfe
 
 1:	l32i	a7, a2, PT_AREG7	# restore a7
 	l32i	a0, a2, PT_AREG0	# restore a0
 	movi	a2, 0			# return 0 (note that we cannot set
-	addi	a6, a6, 1		# restore a6 (really necessary?)
 	rfe
 
 .Lnswp:	/* Atomic set, add, and exg_add. */
 
 TRY	l32i	a7, a3, 0		# orig
+	addi	a6, a6, -SYS_XTENSA_ATOMIC_SET
 	add	a0, a4, a7		# + arg
 	moveqz	a0, a4, a6		# set
+	addi	a6, a6, SYS_XTENSA_ATOMIC_SET
 TRY	s32i	a0, a3, 0		# write new value
 
 	mov	a0, a2
 	mov	a2, a7
 	l32i	a7, a0, PT_AREG7	# restore a7
 	l32i	a0, a0, PT_AREG0	# restore a0
-	addi	a6, a6, 1		# restore a6 (really necessary?)
 	rfe
 
 CATCH
@@ -1044,7 +1042,7 @@ CATCH
 	movi	a2, -EFAULT
 	rfe
 
-.Lill:	l32i	a7, a2, PT_AREG0	# restore a7
+.Lill:	l32i	a7, a2, PT_AREG7	# restore a7
 	l32i	a0, a2, PT_AREG0	# restore a0
 	movi	a2, -EINVAL
 	rfe
-- 
cgit v1.2.3


From 3cfc096e4c4fbc234634cf8a30d40348a25fc9ba Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Thu, 7 Aug 2014 01:03:01 +0400
Subject: xtensa: don't allow overflow/underflow on unaligned stack

Double exceptions that happen during register window overflow/underflow
are handled in the topmost stack frame, as if it was the only exception
that occured. However unaligned access exception handler is special
because it needs to analyze instruction that caused the exception, but
the userspace instruction that triggered window exception is completely
irrelevant. Unaligned data access is rather normal in the generic
userspace code, but stack pointer manipulation must always be done by
architecture-aware code and thus unaligned stack means a serious problem
anyway.
Use the default unaligned access handler that raises SIGBUS in case
of unaligned access in window overflow/underflow handler.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/kernel/traps.c       | 5 +----
 arch/xtensa/kernel/vectors.S     | 8 +++++++-
 arch/xtensa/kernel/vmlinux.lds.S | 4 ++--
 3 files changed, 10 insertions(+), 7 deletions(-)

diff --git a/arch/xtensa/kernel/traps.c b/arch/xtensa/kernel/traps.c
index eebbfd8c26fc..9d2f45f010ef 100644
--- a/arch/xtensa/kernel/traps.c
+++ b/arch/xtensa/kernel/traps.c
@@ -101,9 +101,8 @@ static dispatch_init_table_t __initdata dispatch_init_table[] = {
 #if XCHAL_UNALIGNED_LOAD_EXCEPTION || XCHAL_UNALIGNED_STORE_EXCEPTION
 #ifdef CONFIG_XTENSA_UNALIGNED_USER
 { EXCCAUSE_UNALIGNED,		USER,	   fast_unaligned },
-#else
-{ EXCCAUSE_UNALIGNED,		0,	   do_unaligned_user },
 #endif
+{ EXCCAUSE_UNALIGNED,		0,	   do_unaligned_user },
 { EXCCAUSE_UNALIGNED,		KRNL,	   fast_unaligned },
 #endif
 #ifdef CONFIG_MMU
@@ -264,7 +263,6 @@ do_illegal_instruction(struct pt_regs *regs)
  */
 
 #if XCHAL_UNALIGNED_LOAD_EXCEPTION || XCHAL_UNALIGNED_STORE_EXCEPTION
-#ifndef CONFIG_XTENSA_UNALIGNED_USER
 void
 do_unaligned_user (struct pt_regs *regs)
 {
@@ -286,7 +284,6 @@ do_unaligned_user (struct pt_regs *regs)
 
 }
 #endif
-#endif
 
 void
 do_debug(struct pt_regs *regs)
diff --git a/arch/xtensa/kernel/vectors.S b/arch/xtensa/kernel/vectors.S
index 8453e6e39895..1b397a902292 100644
--- a/arch/xtensa/kernel/vectors.S
+++ b/arch/xtensa/kernel/vectors.S
@@ -454,8 +454,14 @@ _DoubleExceptionVector_WindowOverflow:
 	s32i	a0, a2, PT_DEPC
 
 _DoubleExceptionVector_handle_exception:
+	addi	a0, a0, -EXCCAUSE_UNALIGNED
+	beqz	a0, 2f
 	addx4	a0, a0, a3
-	l32i	a0, a0, EXC_TABLE_FAST_USER
+	l32i	a0, a0, EXC_TABLE_FAST_USER + 4 * EXCCAUSE_UNALIGNED
+	xsr	a3, excsave1
+	jx	a0
+2:
+	movi	a0, user_exception
 	xsr	a3, excsave1
 	jx	a0
 
diff --git a/arch/xtensa/kernel/vmlinux.lds.S b/arch/xtensa/kernel/vmlinux.lds.S
index d16db6df86f8..fc1bc2ba8d5d 100644
--- a/arch/xtensa/kernel/vmlinux.lds.S
+++ b/arch/xtensa/kernel/vmlinux.lds.S
@@ -269,13 +269,13 @@ SECTIONS
 		  .UserExceptionVector.literal)
   SECTION_VECTOR (_DoubleExceptionVector_literal,
 		  .DoubleExceptionVector.literal,
-		  DOUBLEEXC_VECTOR_VADDR - 40,
+		  DOUBLEEXC_VECTOR_VADDR - 48,
 		  SIZEOF(.UserExceptionVector.text),
 		  .UserExceptionVector.text)
   SECTION_VECTOR (_DoubleExceptionVector_text,
 		  .DoubleExceptionVector.text,
 		  DOUBLEEXC_VECTOR_VADDR,
-		  40,
+		  48,
 		  .DoubleExceptionVector.literal)
 
   . = (LOADADDR( .DoubleExceptionVector.text ) + SIZEOF( .DoubleExceptionVector.text ) + 3) & ~ 3;
-- 
cgit v1.2.3


From 9184289c979e78ce466993b53fc951633441e571 Mon Sep 17 00:00:00 2001
From: Max Filippov <jcmvbkbc@gmail.com>
Date: Thu, 7 Aug 2014 03:32:30 +0400
Subject: xtensa: deprecate fast_xtensa and fast_spill_registers syscalls

These syscalls are not used by userspace tools for some time now, and
they have issues when called with invalid arguments. It's not worth
changing signal delivery mechanism as we don't expect any new users for
these syscalls. Let's keep them for backwards compatibility under #ifdef,
disabled by default.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
---
 arch/xtensa/Kconfig        | 26 ++++++++++++++++++++++++++
 arch/xtensa/kernel/entry.S | 28 ++++++++++++++++++++++++++++
 2 files changed, 54 insertions(+)

diff --git a/arch/xtensa/Kconfig b/arch/xtensa/Kconfig
index 3d83c29c9232..49c6c3d94449 100644
--- a/arch/xtensa/Kconfig
+++ b/arch/xtensa/Kconfig
@@ -238,6 +238,32 @@ config HIGHMEM
 
 	  If unsure, say Y.
 
+config FAST_SYSCALL_XTENSA
+	bool "Enable fast atomic syscalls"
+	default n
+	help
+	  fast_syscall_xtensa is a syscall that can make atomic operations
+	  on UP kernel when processor has no s32c1i support.
+
+	  This syscall is deprecated. It may have issues when called with
+	  invalid arguments. It is provided only for backwards compatibility.
+	  Only enable it if your userspace software requires it.
+
+	  If unsure, say N.
+
+config FAST_SYSCALL_SPILL_REGISTERS
+	bool "Enable spill registers syscall"
+	default n
+	help
+	  fast_syscall_spill_registers is a syscall that spills all active
+	  register windows of a calling userspace task onto its stack.
+
+	  This syscall is deprecated. It may have issues when called with
+	  invalid arguments. It is provided only for backwards compatibility.
+	  Only enable it if your userspace software requires it.
+
+	  If unsure, say N.
+
 endmenu
 
 config XTENSA_CALIBRATE_CCOUNT
diff --git a/arch/xtensa/kernel/entry.S b/arch/xtensa/kernel/entry.S
index a06b7efaae82..82bbfa5a05b3 100644
--- a/arch/xtensa/kernel/entry.S
+++ b/arch/xtensa/kernel/entry.S
@@ -986,6 +986,8 @@ ENDPROC(fast_syscall_unrecoverable)
  *		j done
  */
 
+#ifdef CONFIG_FAST_SYSCALL_XTENSA
+
 #define TRY								\
 	.section __ex_table, "a";					\
 	.word	66f, 67f;						\
@@ -1049,6 +1051,18 @@ CATCH
 
 ENDPROC(fast_syscall_xtensa)
 
+#else /* CONFIG_FAST_SYSCALL_XTENSA */
+
+ENTRY(fast_syscall_xtensa)
+
+	l32i    a0, a2, PT_AREG0        # restore a0
+	movi	a2, -ENOSYS
+	rfe
+
+ENDPROC(fast_syscall_xtensa)
+
+#endif /* CONFIG_FAST_SYSCALL_XTENSA */
+
 
 /* fast_syscall_spill_registers.
  *
@@ -1064,6 +1078,8 @@ ENDPROC(fast_syscall_xtensa)
  * Note: We assume the stack pointer is EXC_TABLE_KSTK in the fixup handler.
  */
 
+#ifdef CONFIG_FAST_SYSCALL_SPILL_REGISTERS
+
 ENTRY(fast_syscall_spill_registers)
 
 	/* Register a FIXUP handler (pass current wb as a parameter) */
@@ -1398,6 +1414,18 @@ ENTRY(fast_syscall_spill_registers_fixup_return)
 
 ENDPROC(fast_syscall_spill_registers_fixup_return)
 
+#else /* CONFIG_FAST_SYSCALL_SPILL_REGISTERS */
+
+ENTRY(fast_syscall_spill_registers)
+
+	l32i    a0, a2, PT_AREG0        # restore a0
+	movi	a2, -ENOSYS
+	rfe
+
+ENDPROC(fast_syscall_spill_registers)
+
+#endif /* CONFIG_FAST_SYSCALL_SPILL_REGISTERS */
+
 #ifdef CONFIG_MMU
 /*
  * We should never get here. Bail out!
-- 
cgit v1.2.3