Merge tag 'for-5.9/io_uring-20200802' of git://git.kernel.dk/linux-block

Pull io_uring updates from Jens Axboe:
 "Lots of cleanups in here, hardening the code and/or making it easier
  to read and fixing bugs, but a core feature/change too adding support
  for real async buffered reads. With the latter in place, we just need
  buffered write async support and we're done relying on kthreads for
  the fast path. In detail:

   - Cleanup how memory accounting is done on ring setup/free (Bijan)

   - sq array offset calculation fixup (Dmitry)

   - Consistently handle blocking off O_DIRECT submission path (me)

   - Support proper async buffered reads, instead of relying on kthread
     offload for that. This uses the page waitqueue to drive retries
     from task_work, like we handle poll based retry. (me)

   - IO completion optimizations (me)

   - Fix race with accounting and ring fd install (me)

   - Support EPOLLEXCLUSIVE (Jiufei)

   - Get rid of the io_kiocb unionizing, made possible by shrinking
     other bits (Pavel)

   - Completion side cleanups (Pavel)

   - Cleanup REQ_F_ flags handling, and kill off many of them (Pavel)

   - Request environment grabbing cleanups (Pavel)

   - File and socket read/write cleanups (Pavel)

   - Improve kiocb_set_rw_flags() (Pavel)

   - Tons of fixes and cleanups (Pavel)

   - IORING_SQ_NEED_WAKEUP clear fix (Xiaoguang)"

* tag 'for-5.9/io_uring-20200802' of git://git.kernel.dk/linux-block: (127 commits)
  io_uring: flip if handling after io_setup_async_rw
  fs: optimise kiocb_set_rw_flags()
  io_uring: don't touch 'ctx' after installing file descriptor
  io_uring: get rid of atomic FAA for cq_timeouts
  io_uring: consolidate *_check_overflow accounting
  io_uring: fix stalled deferred requests
  io_uring: fix racy overflow count reporting
  io_uring: deduplicate __io_complete_rw()
  io_uring: de-unionise io_kiocb
  io-wq: update hash bits
  io_uring: fix missing io_queue_linked_timeout()
  io_uring: mark ->work uninitialised after cleanup
  io_uring: deduplicate io_grab_files() calls
  io_uring: don't do opcode prep twice
  io_uring: clear IORING_SQ_NEED_WAKEUP after executing task works
  io_uring: batch put_task_struct()
  tasks: add put_task_struct_many()
  io_uring: return locked and pinned page accounting
  io_uring: don't miscount pinned memory
  io_uring: don't open-code recv kbuf managment
  ...

14 files changed, 1611 insertions, 1148 deletions

diff --git a/block/blk-core.c b/block/blk-core.c
index 93104c7470e8..d9d632639bd1 100644
--- a/block/blk-core.c
+++ b/block/blk-core.c
@@ -960,9 +960,14 @@ static noinline_for_stack bool submit_bio_checks(struct bio *bio)
 {
 	struct request_queue *q = bio->bi_disk->queue;
 	blk_status_t status = BLK_STS_IOERR;
+	struct blk_plug *plug;
 
 	might_sleep();
 
+	plug = blk_mq_plug(q, bio);
+	if (plug && plug->nowait)
+		bio->bi_opf |= REQ_NOWAIT;
+
 	/*
 	 * For a REQ_NOWAIT based request, return -EOPNOTSUPP
 	 * if queue is not a request based queue.
@@ -1802,6 +1807,7 @@ void blk_start_plug(struct blk_plug *plug)
 	INIT_LIST_HEAD(&plug->cb_list);
 	plug->rq_count = 0;
 	plug->multiple_queues = false;
+	plug->nowait = false;
 
 	/*
 	 * Store ordering should not be needed here, since a potential
diff --git a/fs/block_dev.c b/fs/block_dev.c
index 3f94a06a0946..8ae833e00443 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -1734,7 +1734,7 @@ static int blkdev_open(struct inode * inode, struct file * filp)
 	 */
 	filp->f_flags |= O_LARGEFILE;
 
-	filp->f_mode |= FMODE_NOWAIT;
+	filp->f_mode |= FMODE_NOWAIT | FMODE_BUF_RASYNC;
 
 	if (filp->f_flags & O_NDELAY)
 		filp->f_mode |= FMODE_NDELAY;
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index 841c516079a9..bb824c7cb7c7 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -3537,7 +3537,7 @@ static loff_t btrfs_file_llseek(struct file *file, loff_t offset, int whence)
 
 static int btrfs_file_open(struct inode *inode, struct file *filp)
 {
-	filp->f_mode |= FMODE_NOWAIT;
+	filp->f_mode |= FMODE_NOWAIT | FMODE_BUF_RASYNC;
 	return generic_file_open(inode, filp);
 }
 
diff --git a/fs/io-wq.c b/fs/io-wq.c
index 47c5f3aeb460..e92c4724480c 100644
--- a/fs/io-wq.c
+++ b/fs/io-wq.c
@@ -462,6 +462,7 @@ static void io_impersonate_work(struct io_worker *worker,
 		io_wq_switch_mm(worker, work);
 	if (worker->cur_creds != work->creds)
 		io_wq_switch_creds(worker, work);
+	current->signal->rlim[RLIMIT_FSIZE].rlim_cur = work->fsize;
 }
 
 static void io_assign_current_work(struct io_worker *worker,
@@ -489,7 +490,6 @@ static void io_worker_handle_work(struct io_worker *worker)
 
 	do {
 		struct io_wq_work *work;
-		unsigned int hash;
 get_next:
 		/*
 		 * If we got some work, mark us as busy. If we didn't, but
@@ -512,6 +512,7 @@ get_next:
 		/* handle a whole dependent link */
 		do {
 			struct io_wq_work *old_work, *next_hashed, *linked;
+			unsigned int hash = io_get_work_hash(work);
 
 			next_hashed = wq_next_work(work);
 			io_impersonate_work(worker, work);
@@ -522,10 +523,8 @@ get_next:
 			if (test_bit(IO_WQ_BIT_CANCEL, &wq->state))
 				work->flags |= IO_WQ_WORK_CANCEL;
 
-			hash = io_get_work_hash(work);
-			linked = old_work = work;
-			wq->do_work(&linked);
-			linked = (old_work == linked) ? NULL : linked;
+			old_work = work;
+			linked = wq->do_work(work);
 
 			work = next_hashed;
 			if (!work && linked && !io_wq_is_hashed(linked)) {
@@ -542,8 +541,6 @@ get_next:
 				spin_lock_irq(&wqe->lock);
 				wqe->hash_map &= ~BIT_ULL(hash);
 				wqe->flags &= ~IO_WQE_FLAG_STALLED;
-				/* dependent work is not hashed */
-				hash = -1U;
 				/* skip unnecessary unlock-lock wqe->lock */
 				if (!work)
 					goto get_next;
@@ -781,8 +778,7 @@ static void io_run_cancel(struct io_wq_work *work, struct io_wqe *wqe)
 		struct io_wq_work *old_work = work;
 
 		work->flags |= IO_WQ_WORK_CANCEL;
-		wq->do_work(&work);
-		work = (work == old_work) ? NULL : work;
+		work = wq->do_work(work);
 		wq->free_work(old_work);
 	} while (work);
 }
diff --git a/fs/io-wq.h b/fs/io-wq.h
index 071f1a997800..ddaf9614cf9b 100644
--- a/fs/io-wq.h
+++ b/fs/io-wq.h
@@ -5,10 +5,10 @@ struct io_wq;
 
 enum {
 	IO_WQ_WORK_CANCEL	= 1,
-	IO_WQ_WORK_HASHED	= 4,
-	IO_WQ_WORK_UNBOUND	= 32,
-	IO_WQ_WORK_NO_CANCEL	= 256,
-	IO_WQ_WORK_CONCURRENT	= 512,
+	IO_WQ_WORK_HASHED	= 2,
+	IO_WQ_WORK_UNBOUND	= 4,
+	IO_WQ_WORK_NO_CANCEL	= 8,
+	IO_WQ_WORK_CONCURRENT	= 16,
 
 	IO_WQ_HASH_SHIFT	= 24,	/* upper 8 bits are used for hash key */
 };
@@ -89,6 +89,7 @@ struct io_wq_work {
 	struct mm_struct *mm;
 	const struct cred *creds;
 	struct fs_struct *fs;
+	unsigned long fsize;
 	unsigned flags;
 };
 
@@ -101,7 +102,7 @@ static inline struct io_wq_work *wq_next_work(struct io_wq_work *work)
 }
 
 typedef void (free_work_fn)(struct io_wq_work *);
-typedef void (io_wq_work_fn)(struct io_wq_work **);
+typedef struct io_wq_work *(io_wq_work_fn)(struct io_wq_work *);
 
 struct io_wq_data {
 	struct user_struct *user;
diff --git a/fs/io_uring.c b/fs/io_uring.c
index 493e5047e67c..2a3af95be4ca 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -78,6 +78,7 @@
 #include <linux/fs_struct.h>
 #include <linux/splice.h>
 #include <linux/task_work.h>
+#include <linux/pagemap.h>
 
 #define CREATE_TRACE_POINTS
 #include <trace/events/io_uring.h>
@@ -226,7 +227,7 @@ struct io_ring_ctx {
 	struct {
 		unsigned int		flags;
 		unsigned int		compat: 1;
-		unsigned int		account_mem: 1;
+		unsigned int		limit_mem: 1;
 		unsigned int		cq_overflow_flushed: 1;
 		unsigned int		drain_next: 1;
 		unsigned int		eventfd_async: 1;
@@ -319,12 +320,12 @@ struct io_ring_ctx {
 		spinlock_t		completion_lock;
 
 		/*
-		 * ->poll_list is protected by the ctx->uring_lock for
+		 * ->iopoll_list is protected by the ctx->uring_lock for
 		 * io_uring instances that don't use IORING_SETUP_SQPOLL.
 		 * For SQPOLL, only the single threaded io_sq_thread() will
 		 * manipulate the list, hence no extra locking is needed there.
 		 */
-		struct list_head	poll_list;
+		struct list_head	iopoll_list;
 		struct hlist_head	*cancel_hash;
 		unsigned		cancel_hash_bits;
 		bool			poll_multi_file;
@@ -395,6 +396,7 @@ struct io_timeout {
 	int				flags;
 	u32				off;
 	u32				target_seq;
+	struct list_head		list;
 };
 
 struct io_rw {
@@ -413,7 +415,7 @@ struct io_connect {
 struct io_sr_msg {
 	struct file			*file;
 	union {
-		struct user_msghdr __user *msg;
+		struct user_msghdr __user *umsg;
 		void __user		*buf;
 	};
 	int				msg_flags;
@@ -486,6 +488,12 @@ struct io_statx {
 	struct statx __user		*buffer;
 };
 
+struct io_completion {
+	struct file			*file;
+	struct list_head		list;
+	int				cflags;
+};
+
 struct io_async_connect {
 	struct sockaddr_storage		address;
 };
@@ -503,6 +511,7 @@ struct io_async_rw {
 	struct iovec			*iov;
 	ssize_t				nr_segs;
 	ssize_t				size;
+	struct wait_page_queue		wpq;
 };
 
 struct io_async_ctx {
@@ -523,23 +532,18 @@ enum {
 	REQ_F_BUFFER_SELECT_BIT	= IOSQE_BUFFER_SELECT_BIT,
 
 	REQ_F_LINK_HEAD_BIT,
-	REQ_F_LINK_NEXT_BIT,
 	REQ_F_FAIL_LINK_BIT,
 	REQ_F_INFLIGHT_BIT,
 	REQ_F_CUR_POS_BIT,
 	REQ_F_NOWAIT_BIT,
 	REQ_F_LINK_TIMEOUT_BIT,
-	REQ_F_TIMEOUT_BIT,
 	REQ_F_ISREG_BIT,
-	REQ_F_MUST_PUNT_BIT,
-	REQ_F_TIMEOUT_NOSEQ_BIT,
 	REQ_F_COMP_LOCKED_BIT,
 	REQ_F_NEED_CLEANUP_BIT,
 	REQ_F_OVERFLOW_BIT,
 	REQ_F_POLLED_BIT,
 	REQ_F_BUFFER_SELECTED_BIT,
 	REQ_F_NO_FILE_TABLE_BIT,
-	REQ_F_QUEUE_TIMEOUT_BIT,
 	REQ_F_WORK_INITIALIZED_BIT,
 	REQ_F_TASK_PINNED_BIT,
 
@@ -563,8 +567,6 @@ enum {
 
 	/* head of a link */
 	REQ_F_LINK_HEAD		= BIT(REQ_F_LINK_HEAD_BIT),
-	/* already grabbed next link */
-	REQ_F_LINK_NEXT		= BIT(REQ_F_LINK_NEXT_BIT),
 	/* fail rest of links */
 	REQ_F_FAIL_LINK		= BIT(REQ_F_FAIL_LINK_BIT),
 	/* on inflight list */
@@ -575,14 +577,8 @@ enum {
 	REQ_F_NOWAIT		= BIT(REQ_F_NOWAIT_BIT),
 	/* has linked timeout */
 	REQ_F_LINK_TIMEOUT	= BIT(REQ_F_LINK_TIMEOUT_BIT),
-	/* timeout request */
-	REQ_F_TIMEOUT		= BIT(REQ_F_TIMEOUT_BIT),
 	/* regular file */
 	REQ_F_ISREG		= BIT(REQ_F_ISREG_BIT),
-	/* must be punted even for NONBLOCK */
-	REQ_F_MUST_PUNT		= BIT(REQ_F_MUST_PUNT_BIT),
-	/* no timeout sequence */
-	REQ_F_TIMEOUT_NOSEQ	= BIT(REQ_F_TIMEOUT_NOSEQ_BIT),
 	/* completion under lock */
 	REQ_F_COMP_LOCKED	= BIT(REQ_F_COMP_LOCKED_BIT),
 	/* needs cleanup */
@@ -595,8 +591,6 @@ enum {
 	REQ_F_BUFFER_SELECTED	= BIT(REQ_F_BUFFER_SELECTED_BIT),
 	/* doesn't need file table for this request */
 	REQ_F_NO_FILE_TABLE	= BIT(REQ_F_NO_FILE_TABLE_BIT),
-	/* needs to queue linked timeout */
-	REQ_F_QUEUE_TIMEOUT	= BIT(REQ_F_QUEUE_TIMEOUT_BIT),
 	/* io_wq_work is initialized */
 	REQ_F_WORK_INITIALIZED	= BIT(REQ_F_WORK_INITIALIZED_BIT),
 	/* req->task is refcounted */
@@ -606,7 +600,6 @@ enum {
 struct async_poll {
 	struct io_poll_iocb	poll;
 	struct io_poll_iocb	*double_poll;
-	struct io_wq_work	work;
 };
 
 /*
@@ -635,51 +628,54 @@ struct io_kiocb {
 		struct io_splice	splice;
 		struct io_provide_buf	pbuf;
 		struct io_statx		statx;
+		/* use only after cleaning per-op data, see io_clean_op() */
+		struct io_completion	compl;
 	};
 
 	struct io_async_ctx		*io;
-	int				cflags;
 	u8				opcode;
 	/* polled IO has completed */
 	u8				iopoll_completed;
 
 	u16				buf_index;
+	u32				result;
 
-	struct io_ring_ctx	*ctx;
-	struct list_head	list;
-	unsigned int		flags;
-	refcount_t		refs;
-	struct task_struct	*task;
-	unsigned long		fsize;
-	u64			user_data;
-	u32			result;
-	u32			sequence;
-
-	struct list_head	link_list;
+	struct io_ring_ctx		*ctx;
+	unsigned int			flags;
+	refcount_t			refs;
+	struct task_struct		*task;
+	u64				user_data;
 
-	struct list_head	inflight_entry;
+	struct list_head		link_list;
 
-	struct percpu_ref	*fixed_file_refs;
+	/*
+	 * 1. used with ctx->iopoll_list with reads/writes
+	 * 2. to track reqs with ->files (see io_op_def::file_table)
+	 */
+	struct list_head		inflight_entry;
+
+	struct percpu_ref		*fixed_file_refs;
+	struct callback_head		task_work;
+	/* for polled requests, i.e. IORING_OP_POLL_ADD and async armed poll */
+	struct hlist_node		hash_node;
+	struct async_poll		*apoll;
+	struct io_wq_work		work;
+};
 
-	union {
-		/*
-		 * Only commands that never go async can use the below fields,
-		 * obviously. Right now only IORING_OP_POLL_ADD uses them, and
-		 * async armed poll handlers for regular commands. The latter
-		 * restore the work, if needed.
-		 */
-		struct {
-			struct callback_head	task_work;
-			struct hlist_node	hash_node;
-			struct async_poll	*apoll;
-		};
-		struct io_wq_work	work;
-	};
+struct io_defer_entry {
+	struct list_head	list;
+	struct io_kiocb		*req;
+	u32			seq;
 };
 
-#define IO_PLUG_THRESHOLD		2
 #define IO_IOPOLL_BATCH			8
 
+struct io_comp_state {
+	unsigned int		nr;
+	struct list_head	list;
+	struct io_ring_ctx	*ctx;
+};
+
 struct io_submit_state {
 	struct blk_plug		plug;
 
@@ -690,12 +686,16 @@ struct io_submit_state {
 	unsigned int		free_reqs;
 
 	/*
+	 * Batch completion logic
+	 */
+	struct io_comp_state	comp;
+
+	/*
 	 * File reference cache
 	 */
 	struct file		*file;
 	unsigned int		fd;
 	unsigned int		has_refs;
-	unsigned int		used_refs;
 	unsigned int		ios_left;
 };
 
@@ -723,6 +723,7 @@ struct io_op_def {
 	unsigned		pollout : 1;
 	/* op supports buffer selection */
 	unsigned		buffer_select : 1;
+	unsigned		needs_fsize : 1;
 };
 
 static const struct io_op_def io_op_defs[] = {
@@ -742,6 +743,7 @@ static const struct io_op_def io_op_defs[] = {
 		.hash_reg_file		= 1,
 		.unbound_nonreg_file	= 1,
 		.pollout		= 1,
+		.needs_fsize		= 1,
 	},
 	[IORING_OP_FSYNC] = {
 		.needs_file		= 1,
@@ -756,6 +758,7 @@ static const struct io_op_def io_op_defs[] = {
 		.hash_reg_file		= 1,
 		.unbound_nonreg_file	= 1,
 		.pollout		= 1,
+		.needs_fsize		= 1,
 	},
 	[IORING_OP_POLL_ADD] = {
 		.needs_file		= 1,
@@ -808,6 +811,7 @@ static const struct io_op_def io_op_defs[] = {
 	},
 	[IORING_OP_FALLOCATE] = {
 		.needs_file		= 1,
+		.needs_fsize		= 1,
 	},
 	[IORING_OP_OPENAT] = {
 		.file_table		= 1,
@@ -839,6 +843,7 @@ static const struct io_op_def io_op_defs[] = {
 		.needs_file		= 1,
 		.unbound_nonreg_file	= 1,
 		.pollout		= 1,
+		.needs_fsize		= 1,
 	},
 	[IORING_OP_FADVISE] = {
 		.needs_file		= 1,
@@ -881,22 +886,37 @@ static const struct io_op_def io_op_defs[] = {
 	},
 };
 
-static void io_wq_submit_work(struct io_wq_work **workptr);
+enum io_mem_account {
+	ACCT_LOCKED,
+	ACCT_PINNED,
+};
+
+static void __io_complete_rw(struct io_kiocb *req, long res, long res2,
+			     struct io_comp_state *cs);
 static void io_cqring_fill_event(struct io_kiocb *req, long res);
 static void io_put_req(struct io_kiocb *req);
+static void io_double_put_req(struct io_kiocb *req);
 static void __io_double_put_req(struct io_kiocb *req);
 static struct io_kiocb *io_prep_linked_timeout(struct io_kiocb *req);
 static void io_queue_linked_timeout(struct io_kiocb *req);
 static int __io_sqe_files_update(struct io_ring_ctx *ctx,
 				 struct io_uring_files_update *ip,
 				 unsigned nr_args);
-static int io_grab_files(struct io_kiocb *req);
-static void io_complete_rw_common(struct kiocb *kiocb, long res);
-static void io_cleanup_req(struct io_kiocb *req);
+static int io_prep_work_files(struct io_kiocb *req);
+static void __io_clean_op(struct io_kiocb *req);
 static int io_file_get(struct io_submit_state *state, struct io_kiocb *req,
 		       int fd, struct file **out_file, bool fixed);
 static void __io_queue_sqe(struct io_kiocb *req,
-			   const struct io_uring_sqe *sqe);
+			   const struct io_uring_sqe *sqe,
+			   struct io_comp_state *cs);
+static void io_file_put_work(struct work_struct *work);
+
+static ssize_t io_import_iovec(int rw, struct io_kiocb *req,
+			       struct iovec **iovec, struct iov_iter *iter,
+			       bool needs_lock);
+static int io_setup_async_rw(struct io_kiocb *req, ssize_t io_size,
+			     struct iovec *iovec, struct iovec *fast_iov,
+			     struct iov_iter *iter);
 
 static struct kmem_cache *req_cachep;
 
@@ -923,6 +943,12 @@ static void io_get_req_task(struct io_kiocb *req)
 	req->flags |= REQ_F_TASK_PINNED;
 }
 
+static inline void io_clean_op(struct io_kiocb *req)
+{
+	if (req->flags & (REQ_F_NEED_CLEANUP | REQ_F_BUFFER_SELECTED))
+		__io_clean_op(req);
+}
+
 /* not idempotent -- it doesn't clear REQ_F_TASK_PINNED */
 static void __io_put_req_task(struct io_kiocb *req)
 {
@@ -930,7 +956,41 @@ static void __io_put_req_task(struct io_kiocb *req)
 		put_task_struct(req->task);
 }
 
-static void io_file_put_work(struct work_struct *work);
+static void io_sq_thread_drop_mm(void)
+{
+	struct mm_struct *mm = current->mm;
+
+	if (mm) {
+		kthread_unuse_mm(mm);
+		mmput(mm);
+	}
+}
+
+static int __io_sq_thread_acquire_mm(struct io_ring_ctx *ctx)
+{
+	if (!current->mm) {
+		if (unlikely(!(ctx->flags & IORING_SETUP_SQPOLL) ||
+			     !mmget_not_zero(ctx->sqo_mm)))
+			return -EFAULT;
+		kthread_use_mm(ctx->sqo_mm);
+	}
+
+	return 0;
+}
+
+static int io_sq_thread_acquire_mm(struct io_ring_ctx *ctx,
+				   struct io_kiocb *req)
+{
+	if (!io_op_defs[req->opcode].needs_mm)
+		return 0;
+	return __io_sq_thread_acquire_mm(ctx);
+}
+
+static inline void req_set_fail_links(struct io_kiocb *req)
+{
+	if ((req->flags & (REQ_F_LINK | REQ_F_HARDLINK)) == REQ_F_LINK)
+		req->flags |= REQ_F_FAIL_LINK;
+}
 
 /*
  * Note: must call io_req_init_async() for the first time you
@@ -957,6 +1017,11 @@ static void io_ring_ctx_ref_free(struct percpu_ref *ref)
 	complete(&ctx->ref_comp);
 }
 
+static inline bool io_is_timeout_noseq(struct io_kiocb *req)
+{
+	return !req->timeout.off;
+}
+
 static struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 {
 	struct io_ring_ctx *ctx;
@@ -1000,7 +1065,7 @@ static struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p)
 	mutex_init(&ctx->uring_lock);
 	init_waitqueue_head(&ctx->wait);
 	spin_lock_init(&ctx->completion_lock);
-	INIT_LIST_HEAD(&ctx->poll_list);
+	INIT_LIST_HEAD(&ctx->iopoll_list);
 	INIT_LIST_HEAD(&ctx->defer_list);
 	INIT_LIST_HEAD(&ctx->timeout_list);
 	init_waitqueue_head(&ctx->inflight_wait);
@@ -1017,18 +1082,14 @@ err:
 	return NULL;
 }
 
-static inline bool __req_need_defer(struct io_kiocb *req)
+static bool req_need_defer(struct io_kiocb *req, u32 seq)
 {
-	struct io_ring_ctx *ctx = req->ctx;
+	if (unlikely(req->flags & REQ_F_IO_DRAIN)) {
+		struct io_ring_ctx *ctx = req->ctx;
 
-	return req->sequence != ctx->cached_cq_tail
+		return seq != ctx->cached_cq_tail
 				+ atomic_read(&ctx->cached_cq_overflow);
-}
-
-static inline bool req_need_defer(struct io_kiocb *req)
-{
-	if (unlikely(req->flags & REQ_F_IO_DRAIN))
-		return __req_need_defer(req);
+	}
 
 	return false;
 }
@@ -1046,28 +1107,7 @@ static void __io_commit_cqring(struct io_ring_ctx *ctx)
 	}
 }
 
-static inline void io_req_work_grab_env(struct io_kiocb *req,
-					const struct io_op_def *def)
-{
-	if (!req->work.mm && def->needs_mm) {
-		mmgrab(current->mm);
-		req->work.mm = current->mm;
-	}
-	if (!req->work.creds)
-		req->work.creds = get_current_cred();
-	if (!req->work.fs && def->needs_fs) {
-		spin_lock(&current->fs->lock);
-		if (!current->fs->in_exec) {
-			req->work.fs = current->fs;
-			req->work.fs->users++;
-		} else {
-			req->work.flags |= IO_WQ_WORK_CANCEL;
-		}
-		spin_unlock(&current->fs->lock);
-	}
-}
-
-static inline void io_req_work_drop_env(struct io_kiocb *req)
+static void io_req_clean_work(struct io_kiocb *req)
 {
 	if (!(req->flags & REQ_F_WORK_INITIALIZED))
 		return;
@@ -1089,11 +1129,12 @@ static inline void io_req_work_drop_env(struct io_kiocb *req)
 		spin_unlock(&req->work.fs->lock);
 		if (fs)
 			free_fs_struct(fs);
+		req->work.fs = NULL;
 	}
+	req->flags &= ~REQ_F_WORK_INITIALIZED;
 }
 
-static inline void io_prep_async_work(struct io_kiocb *req,
-				      struct io_kiocb **link)
+static void io_prep_async_work(struct io_kiocb *req)
 {
 	const struct io_op_def *def = &io_op_defs[req->opcode];
 
@@ -1106,18 +1147,42 @@ static inline void io_prep_async_work(struct io_kiocb *req,
 		if (def->unbound_nonreg_file)
 			req->work.flags |= IO_WQ_WORK_UNBOUND;
 	}
+	if (!req->work.mm && def->needs_mm) {
+		mmgrab(current->mm);
+		req->work.mm = current->mm;
+	}
+	if (!req->work.creds)
+		req->work.creds = get_current_cred();
+	if (!req->work.fs && def->needs_fs) {
+		spin_lock(&current->fs->lock);
+		if (!current->fs->in_exec) {
+			req->work.fs = current->fs;
+			req->work.fs->users++;
+		} else {
+			req->work.flags |= IO_WQ_WORK_CANCEL;
+		}
+		spin_unlock(&current->fs->lock);
+	}
+	if (def->needs_fsize)
+		req->work.fsize = rlimit(RLIMIT_FSIZE);
+	else
+		req->work.fsize = RLIM_INFINITY;
+}
 
-	io_req_work_grab_env(req, def);
+static void io_prep_async_link(struct io_kiocb *req)
+{
+	struct io_kiocb *cur;
 
-	*link = io_prep_linked_timeout(req);
+	io_prep_async_work(req);
+	if (req->flags & REQ_F_LINK_HEAD)
+		list_for_each_entry(cur, &req->link_list, link_list)
+			io_prep_async_work(cur);
 }
 
-static inline void io_queue_async_work(struct io_kiocb *req)
+static void __io_queue_async_work(struct io_kiocb *req)
 {
 	struct io_ring_ctx *ctx = req->ctx;
-	struct io_kiocb *link;
-
-	io_prep_async_work(req, &link);
+	struct io_kiocb *link = io_prep_linked_timeout(req);
 
 	trace_io_uring_queue_async_work(ctx, io_wq_is_hashed(&req->work), req,
 					&req->work, req->flags);
@@ -1127,14 +1192,22 @@ static inline void io_queue_async_work(struct io_kiocb *req)
 		io_queue_linked_timeout(link);
 }
 
+static void io_queue_async_work(struct io_kiocb *req)
+{
+	/* init ->work of the whole link before punting */
+	io_prep_async_link(req);
+	__io_queue_async_work(req);
+}
+
 static void io_kill_timeout(struct io_kiocb *req)
 {
 	int ret;
 
 	ret = hrtimer_try_to_cancel(&req->io->timeout.timer);
 	if (ret != -1) {
-		atomic_inc(&req->ctx->cq_timeouts);
-		list_del_init(&req->list);
+		atomic_set(&req->ctx->cq_timeouts,
+			atomic_read(&req->ctx->cq_timeouts) + 1);
+		list_del_init(&req->timeout.list);
 		req->flags |= REQ_F_COMP_LOCKED;
 		io_cqring_fill_event(req, 0);
 		io_put_req(req);
@@ -1146,7 +1219,7 @@ static void io_kill_timeouts(struct io_ring_ctx *ctx)
 	struct io_kiocb *req, *tmp;
 
 	spin_lock_irq(&ctx->completion_lock);
-	list_for_each_entry_safe(req, tmp, &ctx->timeout_list, list)
+	list_for_each_entry_safe(req, tmp, &ctx->timeout_list, timeout.list)
 		io_kill_timeout(req);
 	spin_unlock_irq(&ctx->completion_lock);
 }
@@ -1154,13 +1227,15 @@ static void io_kill_timeouts(struct io_ring_ctx *ctx)
 static void __io_queue_deferred(struct io_ring_ctx *ctx)
 {
 	do {
-		struct io_kiocb *req = list_first_entry(&ctx->defer_list,
-							struct io_kiocb, list);
+		struct io_defer_entry *de = list_first_entry(&ctx->defer_list,
+						struct io_defer_entry, list);
 
-		if (req_need_defer(req))
+		if (req_need_defer(de->req, de->seq))
 			break;
-		list_del_init(&req->list);
-		io_queue_async_work(req);
+		list_del_init(&de->list);
+		/* punt-init is done before queueing for defer */
+		__io_queue_async_work(de->req);
+		kfree(de);
 	} while (!list_empty(&ctx->defer_list));
 }
 
@@ -1168,15 +1243,15 @@ static void io_flush_timeouts(struct io_ring_ctx *ctx)
 {
 	while (!list_empty(&ctx->timeout_list)) {
 		struct io_kiocb *req = list_first_entry(&ctx->timeout_list,
-							struct io_kiocb, list);
+						struct io_kiocb, timeout.list);
 
-		if (req->flags & REQ_F_TIMEOUT_NOSEQ)
+		if (io_is_timeout_noseq(req))
 			break;
 		if (req->timeout.target_seq != ctx->cached_cq_tail
 					- atomic_read(&ctx->cq_timeouts))
 			break;
 
-		list_del_init(&req->list);
+		list_del_init(&req->timeout.list);
 		io_kill_timeout(req);
 	}
 }
@@ -1229,6 +1304,15 @@ static void io_cqring_ev_posted(struct io_ring_ctx *ctx)
 		eventfd_signal(ctx->cq_ev_fd, 1);
 }
 
+static void io_cqring_mark_overflow(struct io_ring_ctx *ctx)
+{
+	if (list_empty(&ctx->cq_overflow_list)) {
+		clear_bit(0, &ctx->sq_check_overflow);
+		clear_bit(0, &ctx->cq_check_overflow);
+		ctx->rings->sq_flags &= ~IORING_SQ_CQ_OVERFLOW;
+	}
+}
+
 /* Returns true if there are no backlogged entries after the flush */
 static bool io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force)
 {
@@ -1259,13 +1343,13 @@ static bool io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force)
 			break;
 
 		req = list_first_entry(&ctx->cq_overflow_list, struct io_kiocb,
-						list);
-		list_move(&req->list, &list);
+						compl.list);
+		list_move(&req->compl.list, &list);
 		req->flags &= ~REQ_F_OVERFLOW;
 		if (cqe) {
 			WRITE_ONCE(cqe->user_data, req->user_data);
 			WRITE_ONCE(cqe->res, req->result);
-			WRITE_ONCE(cqe->flags, req->cflags);
+			WRITE_ONCE(cqe->flags, req->compl.cflags);
 		} else {
 			WRITE_ONCE(ctx->rings->cq_overflow,
 				atomic_inc_return(&ctx->cached_cq_overflow));
@@ -1273,17 +1357,14 @@ static bool io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force)
 	}
 
 	io_commit_cqring(ctx);
-	if (cqe) {
-		clear_bit(0, &ctx->sq_check_overflow);
-		clear_bit(0, &ctx->cq_check_overflow);
-		ctx->rings->sq_flags &= ~IORING_SQ_CQ_OVERFLOW;
-	}
+	io_cqring_mark_overflow(ctx);
+
 	spin_unlock_irqrestore(&ctx->completion_lock, flags);
 	io_cqring_ev_posted(ctx);
 
 	while (!list_empty(&list)) {
-		req = list_first_entry(&list, struct io_kiocb, list);
-		list_del(&req->list);
+		req = list_first_entry(&list, struct io_kiocb, compl.list);
+		list_del(&req->compl.list);
 		io_put_req(req);
 	}
 
@@ -1316,11 +1397,12 @@ static void __io_cqring_fill_event(struct io_kiocb *req, long res, long cflags)
 			set_bit(0, &ctx->cq_check_overflow);
 			ctx->rings->sq_flags |= IORING_SQ_CQ_OVERFLOW;
 		}
+		io_clean_op(req);
 		req->flags |= REQ_F_OVERFLOW;
-		refcount_inc(&req->refs);
 		req->result = res;
-		req->cflags = cflags;
-		list_add_tail(&req->list, &ctx->cq_overflow_list);
+		req->compl.cflags = cflags;
+		refcount_inc(&req->refs);
+		list_add_tail(&req->compl.list, &ctx->cq_overflow_list);
 	}
 }
 
@@ -1329,7 +1411,7 @@ static void io_cqring_fill_event(struct io_kiocb *req, long res)
 	__io_cqring_fill_event(req, res, 0);
 }
 
-static void __io_cqring_add_event(struct io_kiocb *req, long res, long cflags)
+static void io_cqring_add_event(struct io_kiocb *req, long res, long cflags)
 {
 	struct io_ring_ctx *ctx = req->ctx;
 	unsigned long flags;
@@ -1342,9 +1424,52 @@ static void __io_cqring_add_event(struct io_kiocb *req, long res, long cflags)
 	io_cqring_ev_posted(ctx);
 }
 
-static void io_cqring_add_event(struct io_kiocb *req, long res)
+static void io_submit_flush_completions(struct io_comp_state *cs)
+{
+	struct io_ring_ctx *ctx = cs->ctx;
+
+	spin_lock_irq(&ctx->completion_lock);
+	while (!list_empty(&cs->list)) {
+		struct io_kiocb *req;
+
+		req = list_first_entry(&cs->list, struct io_kiocb, compl.list);
+		list_del(&req->compl.list);
+		__io_cqring_fill_event(req, req->result, req->compl.cflags);
+		if (!(req->flags & REQ_F_LINK_HEAD)) {
+			req->flags |= REQ_F_COMP_LOCKED;
+			io_put_req(req);
+		} else {
+			spin_unlock_irq(&ctx->completion_lock);
+			io_put_req(req);
+			spin_lock_irq(&ctx->completion_lock);
+		}
+	}
+	io_commit_cqring(ctx);
+	spin_unlock_irq(&ctx->completion_lock);
+
+	io_cqring_ev_posted(ctx);
+	cs->nr = 0;
+}
+
+static void __io_req_complete(struct io_kiocb *req, long res, unsigned cflags,
+			      struct io_comp_state *cs)
 {
-	__io_cqring_add_event(req, res, 0);
+	if (!cs) {
+		io_cqring_add_event(req, res, cflags);
+		io_put_req(req);
+	} else {
+		io_clean_op(req);
+		req->result = res;
+		req->compl.cflags = cflags;
+		list_add_tail(&req->compl.list, &cs->list);
+		if (++cs->nr >= 32)
+			io_submit_flush_completions(cs);
+	}
+}
+
+static void io_req_complete(struct io_kiocb *req, long res)
+{
+	__io_req_complete(req, res, 0, NULL);
 }
 
 static inline bool io_is_fallback_req(struct io_kiocb *req)
@@ -1370,11 +1495,7 @@ static struct io_kiocb *io_alloc_req(struct io_ring_ctx *ctx,
 	gfp_t gfp = GFP_KERNEL | __GFP_NOWARN;
 	struct io_kiocb *req;
 
-	if (!state) {
-		req = kmem_cache_alloc(req_cachep, gfp);
-		if (unlikely(!req))
-			goto fallback;
-	} else if (!state->free_reqs) {
+	if (!state->free_reqs) {
 		size_t sz;
 		int ret;
 
@@ -1412,21 +1533,15 @@ static inline void io_put_file(struct io_kiocb *req, struct file *file,
 		fput(file);
 }
 
-static void __io_req_aux_free(struct io_kiocb *req)
+static void io_dismantle_req(struct io_kiocb *req)
 {
-	if (req->flags & REQ_F_NEED_CLEANUP)
-		io_cleanup_req(req);
+	io_clean_op(req);
 
-	kfree(req->io);
+	if (req->io)
+		kfree(req->io);
 	if (req->file)
 		io_put_file(req, req->file, (req->flags & REQ_F_FIXED_FILE));
-	__io_put_req_task(req);
-	io_req_work_drop_env(req);
-}
-
-static void __io_free_req(struct io_kiocb *req)
-{
-	__io_req_aux_free(req);
+	io_req_clean_work(req);
 
 	if (req->flags & REQ_F_INFLIGHT) {
 		struct io_ring_ctx *ctx = req->ctx;
@@ -1438,57 +1553,20 @@ static void __io_free_req(struct io_kiocb *req)
 			wake_up(&ctx->inflight_wait);
 		spin_unlock_irqrestore(&ctx->inflight_lock, flags);
 	}
-
-	percpu_ref_put(&req->ctx->refs);
-	if (likely(!io_is_fallback_req(req)))
-		kmem_cache_free(req_cachep, req);
-	else
-		clear_bit_unlock(0, (unsigned long *) &req->ctx->fallback_req);
 }
 
-struct req_batch {
-	void *reqs[IO_IOPOLL_BATCH];
-	int to_free;
-	int need_iter;
-};
-
-static void io_free_req_many(struct io_ring_ctx *ctx, struct req_batch *rb)
+static void __io_free_req(struct io_kiocb *req)
 {
-	if (!rb->to_free)
-		return;
-	if (rb->need_iter) {
-		int i, inflight = 0;
-		unsigned long flags;
-
-		for (i = 0; i < rb->to_free; i++) {
-			struct io_kiocb *req = rb->reqs[i];
-
-			if (req->flags & REQ_F_INFLIGHT)
-				inflight++;
-			__io_req_aux_free(req);
-		}
-		if (!inflight)
-			goto do_free;
-
-		spin_lock_irqsave(&ctx->inflight_lock, flags);
-		for (i = 0; i < rb->to_free; i