for-5.12/io_uring-2021-02-17

-----BEGIN PGP SIGNATURE-----
 
 iQJEBAABCAAuFiEEwPw5LcreJtl1+l5K99NY+ylx4KYFAmAtYbYQHGF4Ym9lQGtl
 cm5lbC5kawAKCRD301j7KXHgppeWD/4xKhzBCGZWOkdycaaPhsUTOjNNIPmCBhlz
 QQj4KFSEuJNKACUg53Ak0oECJTaH5976kjKkKs7Z+hzmkEwboLBI4erkcT9MGC3M
 mPx349qBq9X3sYaFrUJF3h0sjRr+wa60nWQ01oVH8HkfI4bCNCHoqo5jDvMPWsYT
 ksFbUm8YWEZmi0K2yXFWXuJIN2bVBd72a8CrvtF3ksdEMYxbWWTOAcrhYJ4H5/U7
 BQjWIxiIVsAoJohcXWq/Swh8cgvgb5uJVpNUU8VEFob/jI3Gc3YojIToISB6soUL
 DNhDJLeyZjuXfE1Ej+ySas9bpdG4LgxzsDBl9lFl9EQkSo1c3h/lEx85aeixAZla
 QfjTOVUabzdPzvZ9H1yDQISxjVLy2PotnhVMy/rSSrnDKlowtNB9iEzd6cpzFzxU
 fxomz1d6+w8rZY9jaRIAcMNa6bEOuYmcP9V8rIzGeg3Mm3jqL7H/JgJu5s2YbjpN
 InmTNu4cwLeTO65DzqVxF8UGbZ2tHbMm5pNeVBYxuY1adRgJFlIOP5kYlNlyiY+D
 Bt41CRuK3hqpYfXh7nSK8U4BKEhMikTCS0W4aKL5EzLZ20rxjgTlaHZiOBqd9vep
 1tqNjPIvL2jWfF+5shwAZbupj3WKbuVqi4S2jXljv+Wkmk4ZVLSX3fQZv2I7JTHM
 I2qa59PB4A==
 =8MX/
 -----END PGP SIGNATURE-----

Merge tag 'for-5.12/io_uring-2021-02-17' of git://git.kernel.dk/linux-block

Pull io_uring updates from Jens Axboe:
 "Highlights from this cycles are things like request recycling and
  task_work optimizations, which net us anywhere from 10-20% of speedups
  on workloads that mostly are inline.

  This work was originally done to put io_uring under memcg, which adds
  considerable overhead. But it's a really nice win as well. Also worth
  highlighting is the LOOKUP_CACHED work in the VFS, and using it in
  io_uring. Greatly speeds up the fast path for file opens.

  Summary:

   - Put io_uring under memcg protection. We accounted just the rings
     themselves under rlimit memlock before, now we account everything.

   - Request cache recycling, persistent across invocations (Pavel, me)

   - First part of a cleanup/improvement to buffer registration (Bijan)

   - SQPOLL fixes (Hao)

   - File registration NULL pointer fixup (Dan)

   - LOOKUP_CACHED support for io_uring

   - Disable /proc/thread-self/ for io_uring, like we do for /proc/self

   - Add Pavel to the io_uring MAINTAINERS entry

   - Tons of code cleanups and optimizations (Pavel)

   - Support for skip entries in file registration (Noah)"

* tag 'for-5.12/io_uring-2021-02-17' of git://git.kernel.dk/linux-block: (103 commits)
  io_uring: tctx->task_lock should be IRQ safe
  proc: don't allow async path resolution of /proc/thread-self components
  io_uring: kill cached requests from exiting task closing the ring
  io_uring: add helper to free all request caches
  io_uring: allow task match to be passed to io_req_cache_free()
  io-wq: clear out worker ->fs and ->files
  io_uring: optimise io_init_req() flags setting
  io_uring: clean io_req_find_next() fast check
  io_uring: don't check PF_EXITING from syscall
  io_uring: don't split out consume out of SQE get
  io_uring: save ctx put/get for task_work submit
  io_uring: don't duplicate io_req_task_queue()
  io_uring: optimise SQPOLL mm/files grabbing
  io_uring: optimise out unlikely link queue
  io_uring: take compl state from submit state
  io_uring: inline io_complete_rw_common()
  io_uring: move res check out of io_rw_reissue()
  io_uring: simplify iopoll reissuing
  io_uring: clean up io_req_free_batch_finish()
  io_uring: move submit side state closer in the ring
  ...

MAINTAINERS

@@ -6830,6 +6830,9 @@ F:	include/linux/fs.h
 F:	include/linux/fs_types.h
 F:	include/uapi/linux/fs.h
 F:	include/uapi/linux/openat2.h
+X:	fs/io-wq.c
+X:	fs/io-wq.h
+X:	fs/io_uring.c
 
 FINTEK F75375S HARDWARE MONITOR AND FAN CONTROLLER DRIVER
 M:	Riku Voipio <riku.voipio@iki.fi>
@@ -9263,6 +9266,7 @@ F:	include/uapi/linux/iommu.h
 
 IO_URING
 M:	Jens Axboe <axboe@kernel.dk>
+R:	Pavel Begunkov <asml.silence@gmail.com>
 L:	io-uring@vger.kernel.org
 S:	Maintained
 T:	git git://git.kernel.dk/linux-block
@@ -9270,6 +9274,7 @@ T:	git git://git.kernel.dk/liburing
 F:	fs/io-wq.c
 F:	fs/io-wq.h
 F:	fs/io_uring.c
+F:	include/linux/io_uring.h
 F:	include/uapi/linux/io_uring.h
 
 IPMI SUBSYSTEM

fs/file.c

@@ -22,6 +22,8 @@
 #include <linux/close_range.h>
 #include <net/sock.h>
 
+#include "internal.h"
+
 unsigned int sysctl_nr_open __read_mostly = 1024*1024;
 unsigned int sysctl_nr_open_min = BITS_PER_LONG;
 /* our min() is unusable in constant expressions ;-/ */
@@ -731,6 +733,32 @@ int __close_range(unsigned fd, unsigned max_fd, unsigned int flags)
 	return 0;
 }
 
+/*
+ * See close_fd_get_file() below, this variant assumes current->files->file_lock
+ * is held.
+ */
+int __close_fd_get_file(unsigned int fd, struct file **res)
+{
+	struct files_struct *files = current->files;
+	struct file *file;
+	struct fdtable *fdt;
+
+	fdt = files_fdtable(files);
+	if (fd >= fdt->max_fds)
+		goto out_err;
+	file = fdt->fd[fd];
+	if (!file)
+		goto out_err;
+	rcu_assign_pointer(fdt->fd[fd], NULL);
+	__put_unused_fd(files, fd);
+	get_file(file);
+	*res = file;
+	return 0;
+out_err:
+	*res = NULL;
+	return -ENOENT;
+}
+
 /*
  * variant of close_fd that gets a ref on the file for later fput.
  * The caller must ensure that filp_close() called on the file, and then
@@ -739,27 +767,13 @@ int __close_range(unsigned fd, unsigned max_fd, unsigned int flags)
 int close_fd_get_file(unsigned int fd, struct file **res)
 {
 	struct files_struct *files = current->files;
-	struct file *file;
-	struct fdtable *fdt;
+	int ret;
 
 	spin_lock(&files->file_lock);
-	fdt = files_fdtable(files);
-	if (fd >= fdt->max_fds)
-		goto out_unlock;
-	file = fdt->fd[fd];
-	if (!file)
-		goto out_unlock;
-	rcu_assign_pointer(fdt->fd[fd], NULL);
-	__put_unused_fd(files, fd);
+	ret = __close_fd_get_file(fd, res);
 	spin_unlock(&files->file_lock);
-	get_file(file);
-	*res = file;
-	return 0;
 
-out_unlock:
-	spin_unlock(&files->file_lock);
-	*res = NULL;
-	return -ENOENT;
+	return ret;
 }
 
 void do_close_on_exec(struct files_struct *files)

fs/internal.h

@@ -133,6 +133,7 @@ extern struct file *do_file_open_root(struct dentry *, struct vfsmount *,
 		const char *, const struct open_flags *);
 extern struct open_how build_open_how(int flags, umode_t mode);
 extern int build_open_flags(const struct open_how *how, struct open_flags *op);
+extern int __close_fd_get_file(unsigned int fd, struct file **res);
 
 long do_sys_ftruncate(unsigned int fd, loff_t length, int small);
 int chmod_common(const struct path *path, umode_t mode);

fs/io-wq.c

@@ -64,9 +64,7 @@ struct io_worker {
 #endif
 	const struct cred *cur_creds;
 	const struct cred *saved_creds;
-	struct files_struct *restore_files;
 	struct nsproxy *restore_nsproxy;
-	struct fs_struct *restore_fs;
 };
 
 #if BITS_PER_LONG == 64
@@ -156,19 +154,19 @@ static bool __io_worker_unuse(struct io_wqe *wqe, struct io_worker *worker)
 		worker->cur_creds = worker->saved_creds = NULL;
 	}
 
-	if (current->files != worker->restore_files) {
+	if (current->files) {
 		__acquire(&wqe->lock);
 		raw_spin_unlock_irq(&wqe->lock);
 		dropped_lock = true;
 
 		task_lock(current);
-		current->files = worker->restore_files;
+		current->files = NULL;
 		current->nsproxy = worker->restore_nsproxy;
 		task_unlock(current);
 	}
 
-	if (current->fs != worker->restore_fs)
-		current->fs = worker->restore_fs;
+	if (current->fs)
+		current->fs = NULL;
 
 	/*
 	 * If we have an active mm, we need to drop the wq lock before unusing
@@ -329,11 +327,11 @@ static void io_worker_start(struct io_wqe *wqe, struct io_worker *worker)
 	allow_kernel_signal(SIGINT);
 
 	current->flags |= PF_IO_WORKER;
+	current->fs = NULL;
+	current->files = NULL;
 
 	worker->flags |= (IO_WORKER_F_UP | IO_WORKER_F_RUNNING);
-	worker->restore_files = current->files;
 	worker->restore_nsproxy = current->nsproxy;
-	worker->restore_fs = current->fs;
 	io_wqe_inc_running(wqe, worker);
 }
 
@@ -555,23 +553,21 @@ static void io_worker_handle_work(struct io_worker *worker)
 
 		/* handle a whole dependent link */
 		do {
-			struct io_wq_work *old_work, *next_hashed, *linked;
+			struct io_wq_work *next_hashed, *linked;
 			unsigned int hash = io_get_work_hash(work);
 
 			next_hashed = wq_next_work(work);
 			io_impersonate_work(worker, work);
+			wq->do_work(work);
+			io_assign_current_work(worker, NULL);
 
-			old_work = work;
-			linked = wq->do_work(work);
-
+			linked = wq->free_work(work);
 			work = next_hashed;
 			if (!work && linked && !io_wq_is_hashed(linked)) {
 				work = linked;
 				linked = NULL;
 			}
 			io_assign_current_work(worker, work);
-			wq->free_work(old_work);
-
 			if (linked)
 				io_wqe_enqueue(wqe, linked);
 
@@ -850,11 +846,9 @@ static void io_run_cancel(struct io_wq_work *work, struct io_wqe *wqe)
 	struct io_wq *wq = wqe->wq;
 
 	do {
-		struct io_wq_work *old_work = work;
-
 		work->flags |= IO_WQ_WORK_CANCEL;
-		work = wq->do_work(work);
-		wq->free_work(old_work);
+		wq->do_work(work);
+		work = wq->free_work(work);
 	} while (work);
 }
 
@@ -944,7 +938,6 @@ static bool io_wq_worker_cancel(struct io_worker *worker, void *data)
 	 */
 	spin_lock_irqsave(&worker->lock, flags);
 	if (worker->cur_work &&
-	    !(worker->cur_work->flags & IO_WQ_WORK_NO_CANCEL) &&
 	    match->fn(worker->cur_work, match->data)) {
 		send_sig(SIGINT, worker->task, 1);
 		match->nr_running++;

fs/io-wq.h

@@ -9,7 +9,6 @@ enum {
 	IO_WQ_WORK_CANCEL	= 1,
 	IO_WQ_WORK_HASHED	= 2,
 	IO_WQ_WORK_UNBOUND	= 4,
-	IO_WQ_WORK_NO_CANCEL	= 8,
 	IO_WQ_WORK_CONCURRENT	= 16,
 
 	IO_WQ_WORK_FILES	= 32,
@@ -28,15 +27,6 @@ enum io_wq_cancel {
 	IO_WQ_CANCEL_NOTFOUND,	/* work not found */
 };
 
-struct io_wq_work_node {
-	struct io_wq_work_node *next;
-};
-
-struct io_wq_work_list {
-	struct io_wq_work_node *first;
-	struct io_wq_work_node *last;
-};
-
 static inline void wq_list_add_after(struct io_wq_work_node *node,
 				     struct io_wq_work_node *pos,
 				     struct io_wq_work_list *list)
@@ -107,8 +97,8 @@ static inline struct io_wq_work *wq_next_work(struct io_wq_work *work)
 	return container_of(work->list.next, struct io_wq_work, list);
 }
 
-typedef void (free_work_fn)(struct io_wq_work *);
-typedef struct io_wq_work *(io_wq_work_fn)(struct io_wq_work *);
+typedef struct io_wq_work *(free_work_fn)(struct io_wq_work *);
+typedef void (io_wq_work_fn)(struct io_wq_work *);
 
 struct io_wq_data {
 	struct user_struct *user;

fs/proc/self.c

@@ -20,7 +20,7 @@ static const char *proc_self_get_link(struct dentry *dentry,
 	 * Not currently supported. Once we can inherit all of struct pid,
 	 * we can allow this.
 	 */
-	if (current->flags & PF_KTHREAD)
+	if (current->flags & PF_IO_WORKER)
 		return ERR_PTR(-EOPNOTSUPP);
 
 	if (!tgid)

fs/proc/thread_self.c

@@ -17,6 +17,13 @@ static const char *proc_thread_self_get_link(struct dentry *dentry,
 	pid_t pid = task_pid_nr_ns(current, ns);
 	char *name;
 
+	/*
+	 * Not currently supported. Once we can inherit all of struct pid,
+	 * we can allow this.
+	 */
+	if (current->flags & PF_IO_WORKER)
+		return ERR_PTR(-EOPNOTSUPP);
+
 	if (!pid)
 		return ERR_PTR(-ENOENT);
 	name = kmalloc(10 + 6 + 10 + 1, dentry ? GFP_KERNEL : GFP_ATOMIC);

include/linux/io_uring.h

@@ -22,6 +22,15 @@ struct io_identity {
 	refcount_t			count;
 };
 
+struct io_wq_work_node {
+	struct io_wq_work_node *next;
+};
+
+struct io_wq_work_list {
+	struct io_wq_work_node *first;
+	struct io_wq_work_node *last;
+};
+
 struct io_uring_task {
 	/* submission side */
 	struct xarray		xa;
@@ -32,6 +41,11 @@ struct io_uring_task {
 	struct io_identity	*identity;
 	atomic_t		in_idle;
 	bool			sqpoll;
+
+	spinlock_t		task_lock;
+	struct io_wq_work_list	task_list;
+	unsigned long		task_state;
+	struct callback_head	task_work;
 };
 
 #if defined(CONFIG_IO_URING)

include/uapi/linux/io_uring.h

@@ -285,12 +285,22 @@ enum {
 	IORING_REGISTER_LAST
 };
 
+/* deprecated, see struct io_uring_rsrc_update */
 struct io_uring_files_update {
 	__u32 offset;
 	__u32 resv;
 	__aligned_u64 /* __s32 * */ fds;
 };
 
+struct io_uring_rsrc_update {
+	__u32 offset;
+	__u32 resv;
+	__aligned_u64 data;
+};
+
+/* Skip updating fd indexes set to this value in the fd table */
+#define IORING_REGISTER_FILES_SKIP	(-2)
+
 #define IO_URING_OP_SUPPORTED	(1U << 0)
 
 struct io_uring_probe_op {