1 files changed, 348 insertions, 98 deletions

diff --git a/fs/io_uring.c b/fs/io_uring.c
index 0e9fb2cb1984..fdc18321d70c 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -4,15 +4,28 @@
  * supporting fast/efficient IO.
  *
  * A note on the read/write ordering memory barriers that are matched between
- * the application and kernel side. When the application reads the CQ ring
- * tail, it must use an appropriate smp_rmb() to order with the smp_wmb()
- * the kernel uses after writing the tail. Failure to do so could cause a
- * delay in when the application notices that completion events available.
- * This isn't a fatal condition. Likewise, the application must use an
- * appropriate smp_wmb() both before writing the SQ tail, and after writing
- * the SQ tail. The first one orders the sqe writes with the tail write, and
- * the latter is paired with the smp_rmb() the kernel will issue before
- * reading the SQ tail on submission.
+ * the application and kernel side.
+ *
+ * After the application reads the CQ ring tail, it must use an
+ * appropriate smp_rmb() to pair with the smp_wmb() the kernel uses
+ * before writing the tail (using smp_load_acquire to read the tail will
+ * do). It also needs a smp_mb() before updating CQ head (ordering the
+ * entry load(s) with the head store), pairing with an implicit barrier
+ * through a control-dependency in io_get_cqring (smp_store_release to
+ * store head will do). Failure to do so could lead to reading invalid
+ * CQ entries.
+ *
+ * Likewise, the application must use an appropriate smp_wmb() before
+ * writing the SQ tail (ordering SQ entry stores with the tail store),
+ * which pairs with smp_load_acquire in io_get_sqring (smp_store_release
+ * to store the tail will do). And it needs a barrier ordering the SQ
+ * head load before writing new SQ entries (smp_load_acquire to read
+ * head will do).
+ *
+ * When using the SQ poll thread (IORING_SETUP_SQPOLL), the application
+ * needs to check the SQ flags for IORING_SQ_NEED_WAKEUP *after*
+ * updating the SQ tail; a full memory barrier smp_mb() is needed
+ * between.
  *
  * Also see the examples in the liburing library:
  *
@@ -70,20 +83,108 @@ struct io_uring {
 	u32 tail ____cacheline_aligned_in_smp;
 };
 
+/*
+ * This data is shared with the application through the mmap at offset
+ * IORING_OFF_SQ_RING.
+ *
+ * The offsets to the member fields are published through struct
+ * io_sqring_offsets when calling io_uring_setup.
+ */
 struct io_sq_ring {
+	/*
+	 * Head and tail offsets into the ring; the offsets need to be
+	 * masked to get valid indices.
+	 *
+	 * The kernel controls head and the application controls tail.
+	 */
 	struct io_uring		r;
+	/*
+	 * Bitmask to apply to head and tail offsets (constant, equals
+	 * ring_entries - 1)
+	 */
 	u32			ring_mask;
+	/* Ring size (constant, power of 2) */
 	u32			ring_entries;
+	/*
+	 * Number of invalid entries dropped by the kernel due to
+	 * invalid index stored in array
+	 *
+	 * Written by the kernel, shouldn't be modified by the
+	 * application (i.e. get number of "new events" by comparing to
+	 * cached value).
+	 *
+	 * After a new SQ head value was read by the application this
+	 * counter includes all submissions that were dropped reaching
+	 * the new SQ head (and possibly more).
+	 */
 	u32			dropped;
+	/*
+	 * Runtime flags
+	 *
+	 * Written by the kernel, shouldn't be modified by the
+	 * application.
+	 *
+	 * The application needs a full memory barrier before checking
+	 * for IORING_SQ_NEED_WAKEUP after updating the sq tail.
+	 */
 	u32			flags;
+	/*
+	 * Ring buffer of indices into array of io_uring_sqe, which is
+	 * mmapped by the application using the IORING_OFF_SQES offset.
+	 *
+	 * This indirection could e.g. be used to assign fixed
+	 * io_uring_sqe entries to operations and only submit them to
+	 * the queue when needed.
+	 *
+	 * The kernel modifies neither the indices array nor the entries
+	 * array.
+	 */
 	u32			array[];
 };
 
+/*
+ * This data is shared with the application through the mmap at offset
+ * IORING_OFF_CQ_RING.
+ *
+ * The offsets to the member fields are published through struct
+ * io_cqring_offsets when calling io_uring_setup.
+ */
 struct io_cq_ring {
+	/*
+	 * Head and tail offsets into the ring; the offsets need to be
+	 * masked to get valid indices.
+	 *
+	 * The application controls head and the kernel tail.
+	 */
 	struct io_uring		r;
+	/*
+	 * Bitmask to apply to head and tail offsets (constant, equals
+	 * ring_entries - 1)
+	 */
 	u32			ring_mask;
+	/* Ring size (constant, power of 2) */
 	u32			ring_entries;
+	/*
+	 * Number of completion events lost because the queue was full;
+	 * this should be avoided by the application by making sure
+	 * there are not more requests pending thatn there is space in
+	 * the completion queue.
+	 *
+	 * Written by the kernel, shouldn't be modified by the
+	 * application (i.e. get number of "new events" by comparing to
+	 * cached value).
+	 *
+	 * As completion events come in out of order this counter is not
+	 * ordered with any other data.
+	 */
 	u32			overflow;
+	/*
+	 * Ring buffer of completion events.
+	 *
+	 * The kernel writes completion events fresh every time they are
+	 * produced, so the application is allowed to modify pending
+	 * entries.
+	 */
 	struct io_uring_cqe	cqes[];
 };
 
@@ -121,6 +222,8 @@ struct io_ring_ctx {
 		unsigned		sq_mask;
 		unsigned		sq_thread_idle;
 		struct io_uring_sqe	*sq_sqes;
+
+		struct list_head	defer_list;
 	} ____cacheline_aligned_in_smp;
 
 	/* IO offload */
@@ -138,6 +241,7 @@ struct io_ring_ctx {
 		unsigned		cq_mask;
 		struct wait_queue_head	cq_wait;
 		struct fasync_struct	*cq_fasync;
+		struct eventfd_ctx	*cq_ev_fd;
 	} ____cacheline_aligned_in_smp;
 
 	/*
@@ -221,13 +325,16 @@ struct io_kiocb {
 	struct list_head	list;
 	unsigned int		flags;
 	refcount_t		refs;
-#define REQ_F_FORCE_NONBLOCK	1	/* inline submission attempt */
+#define REQ_F_NOWAIT		1	/* must not punt to workers */
 #define REQ_F_IOPOLL_COMPLETED	2	/* polled IO has completed */
 #define REQ_F_FIXED_FILE	4	/* ctx owns file */
 #define REQ_F_SEQ_PREV		8	/* sequential with previous */
 #define REQ_F_PREPPED		16	/* prep already done */
+#define REQ_F_IO_DRAIN		32	/* drain existing IO first */
+#define REQ_F_IO_DRAINED	64	/* drain done */
 	u64			user_data;
-	u64			error;
+	u32			error;	/* iopoll result from callback */
+	u32			sequence;
 
 	struct work_struct	work;
 };
@@ -255,6 +362,8 @@ struct io_submit_state {
 	unsigned int		ios_left;
 };
 
+static void io_sq_wq_submit_work(struct work_struct *work);
+
 static struct kmem_cache *req_cachep;
 
 static const struct file_operations io_uring_fops;
@@ -306,10 +415,36 @@ static struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	spin_lock_init(&ctx->completion_lock);
 	INIT_LIST_HEAD(&ctx->poll_list);
 	INIT_LIST_HEAD(&ctx->cancel_list);
+	INIT_LIST_HEAD(&ctx->defer_list);
 	return ctx;
 }
 
-static void io_commit_cqring(struct io_ring_ctx *ctx)
+static inline bool io_sequence_defer(struct io_ring_ctx *ctx,
+				     struct io_kiocb *req)
+{
+	if ((req->flags & (REQ_F_IO_DRAIN|REQ_F_IO_DRAINED)) != REQ_F_IO_DRAIN)
+		return false;
+
+	return req->sequence > ctx->cached_cq_tail + ctx->sq_ring->dropped;
+}
+
+static struct io_kiocb *io_get_deferred_req(struct io_ring_ctx *ctx)
+{
+	struct io_kiocb *req;
+
+	if (list_empty(&ctx->defer_list))
+		return NULL;
+
+	req = list_first_entry(&ctx->defer_list, struct io_kiocb, list);
+	if (!io_sequence_defer(ctx, req)) {
+		list_del_init(&req->list);
+		return req;
+	}
+
+	return NULL;
+}
+
+static void __io_commit_cqring(struct io_ring_ctx *ctx)
 {
 	struct io_cq_ring *ring = ctx->cq_ring;
 
@@ -317,12 +452,6 @@ static void io_commit_cqring(struct io_ring_ctx *ctx)
 		/* order cqe stores with ring update */
 		smp_store_release(&ring->r.tail, ctx->cached_cq_tail);
 
-		/*
-		 * Write sider barrier of tail update, app has read side. See
-		 * comment at the top of this file.
-		 */
-		smp_wmb();
-
 		if (wq_has_sleeper(&ctx->cq_wait)) {
 			wake_up_interruptible(&ctx->cq_wait);
 			kill_fasync(&ctx->cq_fasync, SIGIO, POLL_IN);
@@ -330,14 +459,29 @@ static void io_commit_cqring(struct io_ring_ctx *ctx)
 	}
 }
 
+static void io_commit_cqring(struct io_ring_ctx *ctx)
+{
+	struct io_kiocb *req;
+
+	__io_commit_cqring(ctx);
+
+	while ((req = io_get_deferred_req(ctx)) != NULL) {
+		req->flags |= REQ_F_IO_DRAINED;
+		queue_work(ctx->sqo_wq, &req->work);
+	}
+}
+
 static struct io_uring_cqe *io_get_cqring(struct io_ring_ctx *ctx)
 {
 	struct io_cq_ring *ring = ctx->cq_ring;
 	unsigned tail;
 
 	tail = ctx->cached_cq_tail;
-	/* See comment at the top of the file */
-	smp_rmb();
+	/*
+	 * writes to the cq entry need to come after reading head; the
+	 * control dependency is enough as we're using WRITE_ONCE to
+	 * fill the cq entry
+	 */
 	if (tail - READ_ONCE(ring->r.head) == ring->ring_entries)
 		return NULL;
 
@@ -373,6 +517,8 @@ static void io_cqring_ev_posted(struct io_ring_ctx *ctx)
 		wake_up(&ctx->wait);
 	if (waitqueue_active(&ctx->sqo_wait))
 		wake_up(&ctx->sqo_wait);
+	if (ctx->cq_ev_fd)
+		eventfd_signal(ctx->cq_ev_fd, 1);
 }
 
 static void io_cqring_add_event(struct io_ring_ctx *ctx, u64 user_data,
@@ -774,10 +920,14 @@ static int io_prep_rw(struct io_kiocb *req, const struct sqe_submit *s,
 	ret = kiocb_set_rw_flags(kiocb, READ_ONCE(sqe->rw_flags));
 	if (unlikely(ret))
 		return ret;
-	if (force_nonblock) {
+
+	/* don't allow async punt if RWF_NOWAIT was requested */
+	if (kiocb->ki_flags & IOCB_NOWAIT)
+		req->flags |= REQ_F_NOWAIT;
+
+	if (force_nonblock)
 		kiocb->ki_flags |= IOCB_NOWAIT;
-		req->flags |= REQ_F_FORCE_NONBLOCK;
-	}
+
 	if (ctx->flags & IORING_SETUP_IOPOLL) {
 		if (!(kiocb->ki_flags & IOCB_DIRECT) ||
 		    !kiocb->ki_filp->f_op->iopoll)
@@ -1120,6 +1270,54 @@ static int io_fsync(struct io_kiocb *req, const struct io_uring_sqe *sqe,
 	return 0;
 }
 
+static int io_prep_sfr(struct io_kiocb *req, const struct io_uring_sqe *sqe)
+{
+	struct io_ring_ctx *ctx = req->ctx;
+	int ret = 0;
+
+	if (!req->file)
+		return -EBADF;
+	/* Prep already done (EAGAIN retry) */
+	if (req->flags & REQ_F_PREPPED)
+		return 0;
+
+	if (unlikely(ctx->flags & IORING_SETUP_IOPOLL))
+		return -EINVAL;
+	if (unlikely(sqe->addr || sqe->ioprio || sqe->buf_index))
+		return -EINVAL;
+
+	req->flags |= REQ_F_PREPPED;
+	return ret;
+}
+
+static int io_sync_file_range(struct io_kiocb *req,
+			      const struct io_uring_sqe *sqe,
+			      bool force_nonblock)
+{
+	loff_t sqe_off;
+	loff_t sqe_len;
+	unsigned flags;
+	int ret;
+
+	ret = io_prep_sfr(req, sqe);
+	if (ret)
+		return ret;
+
+	/* sync_file_range always requires a blocking context */
+	if (force_nonblock)
+		return -EAGAIN;
+
+	sqe_off = READ_ONCE(sqe->off);
+	sqe_len = READ_ONCE(sqe->len);
+	flags = READ_ONCE(sqe->sync_range_flags);
+
+	ret = sync_file_range(req->rw.ki_filp, sqe_off, sqe_len, flags);
+
+	io_cqring_add_event(req->ctx, sqe->user_data, ret, 0);
+	io_put_req(req);
+	return 0;
+}
+
 static void io_poll_remove_one(struct io_kiocb *req)
 {
 	struct io_poll_iocb *poll = &req->poll;
@@ -1322,7 +1520,6 @@ static int io_poll_add(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 		spin_unlock(&poll->head->lock);
 	}
 	if (mask) { /* no async, we'd stolen it */
-		req->error = mangle_poll(mask);
 		ipt.error = 0;
 		io_poll_complete(ctx, req, mask);
 	}
@@ -1335,6 +1532,34 @@ static int io_poll_add(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 	return ipt.error;
 }
 
+static int io_req_defer(struct io_ring_ctx *ctx, struct io_kiocb *req,
+			const struct io_uring_sqe *sqe)
+{
+	struct io_uring_sqe *sqe_copy;
+
+	if (!io_sequence_defer(ctx, req) && list_empty(&ctx->defer_list))
+		return 0;
+
+	sqe_copy = kmalloc(sizeof(*sqe_copy), GFP_KERNEL);
+	if (!sqe_copy)
+		return -EAGAIN;
+
+	spin_lock_irq(&ctx->completion_lock);
+	if (!io_sequence_defer(ctx, req) && list_empty(&ctx->defer_list)) {
+		spin_unlock_irq(&ctx->completion_lock);
+		kfree(sqe_copy);
+		return 0;
+	}
+
+	memcpy(sqe_copy, sqe, sizeof(*sqe_copy));
+	req->submit.sqe = sqe_copy;
+
+	INIT_WORK(&req->work, io_sq_wq_submit_work);
+	list_add_tail(&req->list, &ctx->defer_list);
+	spin_unlock_irq(&ctx->completion_lock);
+	return -EIOCBQUEUED;
+}
+
 static int __io_submit_sqe(struct io_ring_ctx *ctx, struct io_kiocb *req,
 			   const struct sqe_submit *s, bool force_nonblock)
 {
@@ -1374,6 +1599,9 @@ static int __io_submit_sqe(struct io_ring_ctx *ctx, struct io_kiocb *req,
 	case IORING_OP_POLL_REMOVE:
 		ret = io_poll_remove(req, s->sqe);
 		break;
+	case IORING_OP_SYNC_FILE_RANGE:
+		ret = io_sync_file_range(req, s->sqe, force_nonblock);
+		break;
 	default:
 		ret = -EINVAL;
 		break;
@@ -1436,8 +1664,7 @@ restart:
 		struct sqe_submit *s = &req->submit;
 		const struct io_uring_sqe *sqe = s->sqe;
 
-		/* Ensure we clear previously set forced non-block flag */
-		req->flags &= ~REQ_F_FORCE_NONBLOCK;
+		/* Ensure we clear previously set non-block flag */
 		req->rw.ki_flags &= ~IOCB_NOWAIT;
 
 		ret = 0;
@@ -1467,10 +1694,11 @@ restart:
 					break;
 				cond_resched();
 			} while (1);
-
-			/* drop submission reference */
-			io_put_req(req);
 		}
+
+		/* drop submission reference */
+		io_put_req(req);
+
 		if (ret) {
 			io_cqring_add_event(ctx, sqe->user_data, ret, 0);
 			io_put_req(req);
@@ -1582,6 +1810,11 @@ static int io_req_set_file(struct io_ring_ctx *ctx, const struct sqe_submit *s,
 	flags = READ_ONCE(s->sqe->flags);
 	fd = READ_ONCE(s->sqe->fd);
 
+	if (flags & IOSQE_IO_DRAIN) {
+		req->flags |= REQ_F_IO_DRAIN;
+		req->sequence = ctx->cached_sq_head - 1;
+	}
+
 	if (!io_op_needs_file(s->sqe)) {
 		req->file = NULL;
 		return 0;
@@ -1611,7 +1844,7 @@ static int io_submit_sqe(struct io_ring_ctx *ctx, struct sqe_submit *s,
 	int ret;
 
 	/* enforce forwards compatibility on users */
-	if (unlikely(s->sqe->flags & ~IOSQE_FIXED_FILE))
+	if (unlikely(s->sqe->flags & ~(IOSQE_FIXED_FILE | IOSQE_IO_DRAIN)))
 		return -EINVAL;
 
 	req = io_get_req(ctx, state);
@@ -1622,8 +1855,15 @@ static int io_submit_sqe(struct io_ring_ctx *ctx, struct sqe_submit *s,
 	if (unlikely(ret))
 		goto out;
 
+	ret = io_req_defer(ctx, req, s->sqe);
+	if (ret) {
+		if (ret == -EIOCBQUEUED)
+			ret = 0;
+		return ret;
+	}
+
 	ret = __io_submit_sqe(ctx, req, s, true);
-	if (ret == -EAGAIN) {
+	if (ret == -EAGAIN && !(req->flags & REQ_F_NOWAIT)) {
 		struct io_uring_sqe *sqe_copy;
 
 		sqe_copy = kmalloc(sizeof(*sqe_copy), GFP_KERNEL);
@@ -1697,24 +1937,10 @@ static void io_commit_sqring(struct io_ring_ctx *ctx)
 		 * write new data to them.
 		 */
 		smp_store_release(&ring->r.head, ctx->cached_sq_head);
-
-		/*
-		 * write side barrier of head update, app has read side. See
-		 * comment at the top of this file
-		 */
-		smp_wmb();
 	}
 }
 
 /*
- * Undo last io_get_sqring()
- */
-static void io_drop_sqring(struct io_ring_ctx *ctx)
-{
-	ctx->cached_sq_head--;
-}
-
-/*
  * Fetch an sqe, if one is available. Note that s->sqe will point to memory
  * that is mapped by userspace. This means that care needs to be taken to
  * ensure that reads are stable, as we cannot rely on userspace always
@@ -1736,8 +1962,6 @@ static bool io_get_sqring(struct io_ring_ctx *ctx, struct sqe_submit *s)
 	 *    though the application is the one updating it.
 	 */
 	head = ctx->cached_sq_head;
-	/* See comment at the top of this file */
-	smp_rmb();
 	/* make sure SQ entry isn't read before tail */
 	if (head == smp_load_acquire(&ring->r.tail))
 		return false;
@@ -1753,8 +1977,6 @@ static bool io_get_sqring(struct io_ring_ctx *ctx, struct sqe_submit *s)
 	/* drop invalid entries */
 	ctx->cached_sq_head++;
 	ring->dropped++;
-	/* See comment at the top of this file */
-	smp_wmb();
 	return false;
 }
 
@@ -1878,13 +2100,11 @@ static int io_sq_thread(void *data)
 				finish_wait(&ctx->sqo_wait, &wait);
 
 				ctx->sq_ring->flags &= ~IORING_SQ_NEED_WAKEUP;
-				smp_wmb();
 				continue;
 			}
 			finish_wait(&ctx->sqo_wait, &wait);
 
 			ctx->sq_ring->flags &= ~IORING_SQ_NEED_WAKEUP;
-			smp_wmb();
 		}
 
 		i = 0;
@@ -1929,7 +2149,7 @@ static int io_sq_thread(void *data)
 static int io_ring_submit(struct io_ring_ctx *ctx, unsigned int to_submit)
 {
 	struct io_submit_state state, *statep = NULL;
-	int i, ret = 0, submit = 0;
+	int i, submit = 0;
 
 	if (to_submit > IO_PLUG_THRESHOLD) {
 		io_submit_state_start(&state, ctx, to_submit);
@@ -1938,6 +2158,7 @@ static int io_ring_submit(struct io_ring_ctx *ctx, unsigned int to_submit)
 
 	for (i = 0; i < to_submit; i++) {
 		struct sqe_submit s;
+		int ret;
 
 		if (!io_get_sqring(ctx, &s))
 			break;
@@ -1945,21 +2166,18 @@ static int io_ring_submit(struct io_ring_ctx *ctx, unsigned int to_submit)
 		s.has_user = true;
 		s.needs_lock = false;
 		s.needs_fixed_file = false;
+		submit++;
 
 		ret = io_submit_sqe(ctx, &s, statep);
-		if (ret) {
-			io_drop_sqring(ctx);
-			break;
-		}
-
-		submit++;
+		if (ret)
+			io_cqring_add_event(ctx, s.sqe->user_data, ret, 0);
 	}
 	io_commit_sqring(ctx);
 
 	if (statep)
 		io_submit_state_end(statep);
 
-	return submit ? submit : ret;
+	return submit;
 }
 
 static unsigned io_cqring_events(struct io_cq_ring *ring)
@@ -2145,7 +2363,6 @@ static int io_sqe_files_scm(struct io_ring_ctx *ctx)
 	left = ctx->nr_user_files;
 	while (left) {
 		unsigned this_files = min_t(unsigned, left, SCM_MAX_FD);
-		int ret;
 
 		ret = __io_sqe_files_scm(ctx, this_files, total);
 		if (ret)
@@ -2240,10 +2457,6 @@ static int io_sq_offload_start(struct io_ring_ctx *ctx,
 	mmgrab(current->mm);
 	ctx->sqo_mm = current->mm;
 
-	ret = -EINVAL;
-	if (!cpu_possible(p->sq_thread_cpu))
-		goto err;
-
 	if (ctx->flags & IORING_SETUP_SQPOLL) {
 		ret = -EPERM;
 		if (!capable(CAP_SYS_ADMIN))
@@ -2254,11 +2467,11 @@ static int io_sq_offload_start(struct io_ring_ctx *ctx,
 			ctx->sq_thread_idle = HZ;
 
 		if (p->flags & IORING_SETUP_SQ_AFF) {
-			int cpu;
+			int cpu = array_index_nospec(p->sq_thread_cpu,
+							nr_cpu_ids);
 
-			cpu = array_index_nospec(p->sq_thread_cpu, NR_CPUS);
 			ret = -EINVAL;
-			if (!cpu_possible(p->sq_thread_cpu))
+			if (!cpu_online(cpu))
 				goto err;
 
 			ctx->sqo_thread = kthread_create_on_cpu(io_sq_thread,
@@ -2321,8 +2534,12 @@ static int io_account_mem(struct user_struct *user, unsigned long nr_pages)
 
 static void io_mem_free(void *ptr)
 {
-	struct page *page = virt_to_head_page(ptr);
+	struct page *page;
 
+	if (!ptr)
+		return;
+
+	page = virt_to_head_page(ptr);
 	if (put_page_testzero(page))
 		free_compound_page(page);
 }
@@ -2363,7 +2580,7 @@ static int io_sqe_buffer_unregister(struct io_ring_ctx *ctx)
 
 		if (ctx->account_mem)
 			io_unaccount_mem(ctx->user, imu->nr_bvecs);
-		kfree(imu->bvec);
+		kvfree(imu->bvec);
 		imu->nr_bvecs = 0;
 	}
 
@@ -2455,9 +2672,9 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
 		if (!pages || nr_pages > got_pages) {
 			kfree(vmas);
 			kfree(pages);
-			pages = kmalloc_array(nr_pages, sizeof(struct page *),
+			pages = kvmalloc_array(nr_pages, sizeof(struct page *),
 						GFP_KERNEL);
-			vmas = kmalloc_array(nr_pages,
+			vmas = kvmalloc_array(nr_pages,
 					sizeof(struct vm_area_struct *),
 					GFP_KERNEL);
 			if (!pages || !vmas) {
@@ -2469,7 +2686,7 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
 			got_pages = nr_pages;
 		}
 
-		imu->bvec = kmalloc_array(nr_pages, sizeof(struct bio_vec),
+		imu->bvec = kvmalloc_array(nr_pages, sizeof(struct bio_vec),
 						GFP_KERNEL);
 		ret = -ENOMEM;
 		if (!imu->bvec) {
@@ -2480,8 +2697,9 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
 
 		ret = 0;
 		down_read(&current->mm->mmap_sem);
-		pret = get_user_pages_longterm(ubuf, nr_pages, FOLL_WRITE,
-						pages, vmas);
+		pret = get_user_pages(ubuf, nr_pages,
+				      FOLL_WRITE | FOLL_LONGTERM,
+				      pages, vmas);
 		if (pret == nr_pages) {
 			/* don't support file backed memory */
 			for (j = 0; j < nr_pages; j++) {
@@ -2508,6 +2726,7 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
 			}
 			if (ctx->account_mem)
 				io_unaccount_mem(ctx->user, nr_pages);
+			kvfree(imu->bvec);
 			goto err;
 		}
 
@@ -2530,16 +2749,48 @@ static int io_sqe_buffer_register(struct io_ring_ctx *ctx, void __user *arg,
 
 		ctx->nr_user_bufs++;
 	}
-	kfree(pages);
-	kfree(vmas);
+	kvfree(pages);
+	kvfree(vmas);
 	return 0;
 err:
-	kfree(pages);
-	kfree(vmas);
+	kvfree(pages);
+	kvfree(vmas);
 	io_sqe_buffer_unregister(ctx);
 	return ret;
 }
 
+static int io_eventfd_register(struct io_ring_ctx *ctx, void __user *arg)
+{
+	__s32 __user *fds = arg;
+	int fd;
+
+	if (ctx->cq_ev_fd)
+		return -EBUSY;
+
+	if (copy_from_user(&fd, fds, sizeof(*fds)))
+		return -EFAULT;
+
+	ctx->cq_ev_fd = eventfd_ctx_fdget(fd);
+	if (IS_ERR(ctx->cq_ev_fd)) {
+		int ret = PTR_ERR(ctx->cq_ev_fd);
+		ctx->cq_ev_fd = NULL;
+		return ret;
+	}
+
+	return 0;
+}
+
+static int io_eventfd_unregister(struct io_ring_ctx *ctx)
+{
+	if (ctx->cq_ev_fd) {
+		eventfd_ctx_put(ctx->cq_ev_fd);
+		ctx->cq_ev_fd = NULL;
+		return 0;
+	}
+
+	return -ENXIO;
+}
+
 static void io_ring_ctx_free(struct io_ring_ctx *ctx)
 {
 	io_finish_async(ctx);
@@ -2549,6 +2800,7 @@ static void io_ring_ctx_free(struct io_ring_ctx *ctx)
 	io_iopoll_reap_events(ctx);
 	io_sqe_buffer_unregister(ctx);
 	io_sqe_files_unregister(ctx);
+	io_eventfd_unregister(ctx);
 
 #if defined(CONFIG_UNIX)
 	if (ctx->ring_sock)
@@ -2573,7 +2825,10 @@ static __poll_t io_uring_poll(struct file *file, poll_table *wait)
 	__poll_t mask = 0;
 
 	poll_wait(file, &ctx->cq_wait, wait);
-	/* See comment at the top of this file */
+	/*
+	 * synchronizes with barrier from wq_has_sleeper call in
+	 * io_commit_cqring
+	 */
 	smp_rmb();
 	if (READ_ONCE(ctx->sq_ring->r.tail) - ctx->cached_sq_head !=
 	    ctx->sq_ring->ring_entries)
@@ -2687,24 +2942,12 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit,
 		mutex_lock(&ctx->uring_lock);
 		submitted = io_ring_submit(ctx, to_submit);
 		mutex_unlock(&ctx->uring_lock);
-
-		if (submitted < 0)
-			goto out_ctx;
 	}
 	if (flags & IORING_ENTER_GETEVENTS) {
 		unsigned nr_events = 0;
 
 		min_complete = min(min_complete, ctx->cq_entries);
 
-		/*
-		 * The application could have included the 'to_submit' count
-		 * in how many events it wanted to wait for. If we failed to
-		 * submit the desired count, we may need to adjust the number
-		 * of events to poll/wait for.
-		 */
-		if (submitted < to_submit)
-			min_complete = min_t(unsigned, submitted, min_complete);
-
 		if (ctx->flags & IORING_SETUP_IOPOLL) {
 			mutex_lock(&ctx->uring_lock);
 			ret = io_iopoll_check(ctx, &nr_events, min_complete);
@@ -2750,17 +2993,12 @@ static int io_allocate_scq_urings(struct io_ring_ctx *ctx,
 		return -EOVERFLOW;
 
 	ctx->sq_sqes = io_mem_alloc(size);
-	if (!ctx->sq_sqes) {
-		io_mem_free(ctx->sq_ring);
+	if (!ctx->sq_sqes)
 		return -ENOMEM;
-	}
 
 	cq_ring = io_mem_alloc(struct_size(cq_ring, cqes, p->cq_entries));
-	if (!cq_ring) {
-		io_mem_free(ctx->sq_ring);
-		io_mem_free(ctx->sq_sqes);
+	if (!cq_ring)
 		return -ENOMEM;
-	}
 
 	ctx->cq_ring = cq_ring;
 	cq_ring->ring_mask = p->cq_entries - 1;
@@ -2976,6 +3214,18 @@ static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
 			break;
 		ret = io_sqe_files_unregister(ctx);
 		break;
+	case IORING_REGISTER_EVENTFD:
+		ret = -EINVAL;
+		if (nr_args != 1)
+			break;
+		ret = io_eventfd_register(ctx, arg);
+		break;
+	case IORING_UNREGISTER_EVENTFD:
+		ret = -EINVAL;
+		if (arg || nr_args)
+			break;
+		ret = io_eventfd_unregister(ctx);
+		break;
 	default:
 		ret = -EINVAL;
 		break;