diff options
author | 2020-12-16 12:44:05 -0800 | |
---|---|---|
committer | 2020-12-16 12:44:05 -0800 | |
commit | 48aba79bcf6ea05148dc82ad9c40713960b00396 (patch) | |
tree | 768ced0b8a924b8adb82771a797942cb5fd6ce1e | |
parent | Merge tag 'tif-task_work.arch-2020-12-14' of git://git.kernel.dk/linux-block (diff) | |
parent | io_uring: fix io_cqring_events()'s noflush (diff) | |
download | linux-48aba79bcf6ea05148dc82ad9c40713960b00396.tar.gz linux-48aba79bcf6ea05148dc82ad9c40713960b00396.tar.bz2 linux-48aba79bcf6ea05148dc82ad9c40713960b00396.zip |
Merge tag 'for-5.11/io_uring-2020-12-14' of git://git.kernel.dk/linux-block
Pull io_uring updates from Jens Axboe:
"Fairly light set of changes this time around, and mostly some bits
that were pushed out to 5.11 instead of 5.10, fixes/cleanups, and a
few features. In particular:
- Cleanups around iovec import (David Laight, Pavel)
- Add timeout support for io_uring_enter(2), which enables us to
clean up liburing and avoid a timeout sqe submission in the
completion path.
The big win here is that it allows setups that split SQ and CQ
handling into separate threads to avoid locking, as the CQ side
will no longer submit when timeouts are needed when waiting for
events (Hao Xu)
- Add support for socket shutdown, and renameat/unlinkat.
- SQPOLL cleanups and improvements (Xiaoguang Wang)
- Allow SQPOLL setups for CAP_SYS_NICE, and enable regular
(non-fixed) files to be used.
- Cancelation improvements (Pavel)
- Fixed file reference improvements (Pavel)
- IOPOLL related race fixes (Pavel)
- Lots of other little fixes and cleanups (mostly Pavel)"
* tag 'for-5.11/io_uring-2020-12-14' of git://git.kernel.dk/linux-block: (43 commits)
io_uring: fix io_cqring_events()'s noflush
io_uring: fix racy IOPOLL flush overflow
io_uring: fix racy IOPOLL completions
io_uring: always let io_iopoll_complete() complete polled io
io_uring: add timeout update
io_uring: restructure io_timeout_cancel()
io_uring: fix files cancellation
io_uring: use bottom half safe lock for fixed file data
io_uring: fix miscounting ios_left
io_uring: change submit file state invariant
io_uring: check kthread stopped flag when sq thread is unparked
io_uring: share fixed_file_refs b/w multiple rsrcs
io_uring: replace inflight_wait with tctx->wait
io_uring: don't take fs for recvmsg/sendmsg
io_uring: only wake up sq thread while current task is in io worker context
io_uring: don't acquire uring_lock twice
io_uring: initialize 'timeout' properly in io_sq_thread()
io_uring: refactor io_sq_thread() handling
io_uring: always batch cancel in *cancel_files()
io_uring: pass files into kill timeouts/poll
...
-rw-r--r-- | fs/internal.h | 2 | ||||
-rw-r--r-- | fs/io-wq.c | 10 | ||||
-rw-r--r-- | fs/io-wq.h | 1 | ||||
-rw-r--r-- | fs/io_uring.c | 1327 | ||||
-rw-r--r-- | fs/namei.c | 40 | ||||
-rw-r--r-- | include/linux/socket.h | 1 | ||||
-rw-r--r-- | include/linux/syscalls.h | 2 | ||||
-rw-r--r-- | include/uapi/linux/io_uring.h | 16 | ||||
-rw-r--r-- | net/socket.c | 15 |
9 files changed, 824 insertions, 590 deletions
diff --git a/fs/internal.h b/fs/internal.h index a7cd0f64faa4..6fd14ea213c3 100644 --- a/fs/internal.h +++ b/fs/internal.h @@ -78,6 +78,8 @@ extern int vfs_path_lookup(struct dentry *, struct vfsmount *, long do_rmdir(int dfd, struct filename *name); long do_unlinkat(int dfd, struct filename *name); int may_linkat(struct path *link); +int do_renameat2(int olddfd, struct filename *oldname, int newdfd, + struct filename *newname, unsigned int flags); /* * namespace.c diff --git a/fs/io-wq.c b/fs/io-wq.c index b53c055bea6a..f72d53848dcb 100644 --- a/fs/io-wq.c +++ b/fs/io-wq.c @@ -1078,16 +1078,6 @@ enum io_wq_cancel io_wq_cancel_cb(struct io_wq *wq, work_cancel_fn *cancel, return IO_WQ_CANCEL_NOTFOUND; } -static bool io_wq_io_cb_cancel_data(struct io_wq_work *work, void *data) -{ - return work == data; -} - -enum io_wq_cancel io_wq_cancel_work(struct io_wq *wq, struct io_wq_work *cwork) -{ - return io_wq_cancel_cb(wq, io_wq_io_cb_cancel_data, (void *)cwork, false); -} - struct io_wq *io_wq_create(unsigned bounded, struct io_wq_data *data) { int ret = -ENOMEM, node; diff --git a/fs/io-wq.h b/fs/io-wq.h index cba36f03c355..069496c6d4f9 100644 --- a/fs/io-wq.h +++ b/fs/io-wq.h @@ -129,7 +129,6 @@ static inline bool io_wq_is_hashed(struct io_wq_work *work) } void io_wq_cancel_all(struct io_wq *wq); -enum io_wq_cancel io_wq_cancel_work(struct io_wq *wq, struct io_wq_work *cwork); typedef bool (work_cancel_fn)(struct io_wq_work *, void *); diff --git a/fs/io_uring.c b/fs/io_uring.c index 0e8902be6b96..22e31050f33e 100644 --- a/fs/io_uring.c +++ b/fs/io_uring.c @@ -245,6 +245,8 @@ struct io_sq_data { struct task_struct *thread; struct wait_queue_head wait; + + unsigned sq_thread_idle; }; struct io_ring_ctx { @@ -285,7 +287,6 @@ struct io_ring_ctx { struct list_head timeout_list; struct list_head cq_overflow_list; - wait_queue_head_t inflight_wait; struct io_uring_sqe *sq_sqes; } ____cacheline_aligned_in_smp; @@ -310,7 +311,6 @@ struct io_ring_ctx { struct io_sq_data *sq_data; /* if using sq thread polling */ struct wait_queue_head sqo_sq_wait; - struct wait_queue_entry sqo_wait_entry; struct list_head sqd_list; /* @@ -395,16 +395,18 @@ struct io_ring_ctx { */ struct io_poll_iocb { struct file *file; - union { - struct wait_queue_head *head; - u64 addr; - }; + struct wait_queue_head *head; __poll_t events; bool done; bool canceled; struct wait_queue_entry wait; }; +struct io_poll_remove { + struct file *file; + u64 addr; +}; + struct io_close { struct file *file; struct file *put_file; @@ -444,11 +446,17 @@ struct io_timeout { u32 off; u32 target_seq; struct list_head list; + /* head of the link, used by linked timeouts only */ + struct io_kiocb *head; }; struct io_timeout_rem { struct file *file; u64 addr; + + /* timeout update */ + struct timespec64 ts; + u32 flags; }; struct io_rw { @@ -541,6 +549,27 @@ struct io_statx { struct statx __user *buffer; }; +struct io_shutdown { + struct file *file; + int how; +}; + +struct io_rename { + struct file *file; + int old_dfd; + int new_dfd; + struct filename *oldpath; + struct filename *newpath; + int flags; +}; + +struct io_unlink { + struct file *file; + int dfd; + int flags; + struct filename *filename; +}; + struct io_completion { struct file *file; struct list_head list; @@ -575,7 +604,6 @@ enum { REQ_F_FORCE_ASYNC_BIT = IOSQE_ASYNC_BIT, REQ_F_BUFFER_SELECT_BIT = IOSQE_BUFFER_SELECT_BIT, - REQ_F_LINK_HEAD_BIT, REQ_F_FAIL_LINK_BIT, REQ_F_INFLIGHT_BIT, REQ_F_CUR_POS_BIT, @@ -607,8 +635,6 @@ enum { /* IOSQE_BUFFER_SELECT */ REQ_F_BUFFER_SELECT = BIT(REQ_F_BUFFER_SELECT_BIT), - /* head of a link */ - REQ_F_LINK_HEAD = BIT(REQ_F_LINK_HEAD_BIT), /* fail rest of links */ REQ_F_FAIL_LINK = BIT(REQ_F_FAIL_LINK_BIT), /* on inflight list */ @@ -651,6 +677,7 @@ struct io_kiocb { struct file *file; struct io_rw rw; struct io_poll_iocb poll; + struct io_poll_remove poll_remove; struct io_accept accept; struct io_sync sync; struct io_cancel cancel; @@ -667,6 +694,9 @@ struct io_kiocb { struct io_splice splice; struct io_provide_buf pbuf; struct io_statx statx; + struct io_shutdown shutdown; + struct io_rename rename; + struct io_unlink unlink; /* use only after cleaning per-op data, see io_clean_op() */ struct io_completion compl; }; @@ -686,15 +716,14 @@ struct io_kiocb { struct task_struct *task; u64 user_data; - struct list_head link_list; + struct io_kiocb *link; + 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; @@ -725,6 +754,8 @@ struct io_submit_state { void *reqs[IO_IOPOLL_BATCH]; unsigned int free_reqs; + bool plug_started; + /* * Batch completion logic */ @@ -735,7 +766,7 @@ struct io_submit_state { */ struct file *file; unsigned int fd; - unsigned int has_refs; + unsigned int file_refs; unsigned int ios_left; }; @@ -757,6 +788,8 @@ struct io_op_def { unsigned buffer_select : 1; /* must always have async data allocated */ unsigned needs_async_data : 1; + /* should block plug */ + unsigned plug : 1; /* size of async data needed, if any */ unsigned short async_size; unsigned work_flags; @@ -770,6 +803,7 @@ static const struct io_op_def io_op_defs[] = { .pollin = 1, .buffer_select = 1, .needs_async_data = 1, + .plug = 1, .async_size = sizeof(struct io_async_rw), .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_BLKCG, }, @@ -779,6 +813,7 @@ static const struct io_op_def io_op_defs[] = { .unbound_nonreg_file = 1, .pollout = 1, .needs_async_data = 1, + .plug = 1, .async_size = sizeof(struct io_async_rw), .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_BLKCG | IO_WQ_WORK_FSIZE, @@ -791,6 +826,7 @@ static const struct io_op_def io_op_defs[] = { .needs_file = 1, .unbound_nonreg_file = 1, .pollin = 1, + .plug = 1, .async_size = sizeof(struct io_async_rw), .work_flags = IO_WQ_WORK_BLKCG | IO_WQ_WORK_MM, }, @@ -799,6 +835,7 @@ static const struct io_op_def io_op_defs[] = { .hash_reg_file = 1, .unbound_nonreg_file = 1, .pollout = 1, + .plug = 1, .async_size = sizeof(struct io_async_rw), .work_flags = IO_WQ_WORK_BLKCG | IO_WQ_WORK_FSIZE | IO_WQ_WORK_MM, @@ -818,8 +855,7 @@ static const struct io_op_def io_op_defs[] = { .pollout = 1, .needs_async_data = 1, .async_size = sizeof(struct io_async_msghdr), - .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_BLKCG | - IO_WQ_WORK_FS, + .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_BLKCG, }, [IORING_OP_RECVMSG] = { .needs_file = 1, @@ -828,15 +864,17 @@ static const struct io_op_def io_op_defs[] = { .buffer_select = 1, .needs_async_data = 1, .async_size = sizeof(struct io_async_msghdr), - .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_BLKCG | - IO_WQ_WORK_FS, + .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_BLKCG, }, [IORING_OP_TIMEOUT] = { .needs_async_data = 1, .async_size = sizeof(struct io_timeout_data), .work_flags = IO_WQ_WORK_MM, }, - [IORING_OP_TIMEOUT_REMOVE] = {}, + [IORING_OP_TIMEOUT_REMOVE] = { + /* used by timeout updates' prep() */ + .work_flags = IO_WQ_WORK_MM, + }, [IORING_OP_ACCEPT] = { .needs_file = 1, .unbound_nonreg_file = 1, @@ -863,7 +901,7 @@ static const struct io_op_def io_op_defs[] = { }, [IORING_OP_OPENAT] = { .work_flags = IO_WQ_WORK_FILES | IO_WQ_WORK_BLKCG | - IO_WQ_WORK_FS, + IO_WQ_WORK_FS | IO_WQ_WORK_MM, }, [IORING_OP_CLOSE] = { .needs_file = 1, @@ -882,6 +920,7 @@ static const struct io_op_def io_op_defs[] = { .unbound_nonreg_file = 1, .pollin = 1, .buffer_select = 1, + .plug = 1, .async_size = sizeof(struct io_async_rw), .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_BLKCG, }, @@ -889,6 +928,7 @@ static const struct io_op_def io_op_defs[] = { .needs_file = 1, .unbound_nonreg_file = 1, .pollout = 1, + .plug = 1, .async_size = sizeof(struct io_async_rw), .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_BLKCG | IO_WQ_WORK_FSIZE, @@ -915,7 +955,7 @@ static const struct io_op_def io_op_defs[] = { }, [IORING_OP_OPENAT2] = { .work_flags = IO_WQ_WORK_FILES | IO_WQ_WORK_FS | - IO_WQ_WORK_BLKCG, + IO_WQ_WORK_BLKCG | IO_WQ_WORK_MM, }, [IORING_OP_EPOLL_CTL] = { .unbound_nonreg_file = 1, @@ -934,6 +974,17 @@ static const struct io_op_def io_op_defs[] = { .hash_reg_file = 1, .unbound_nonreg_file = 1, }, + [IORING_OP_SHUTDOWN] = { + .needs_file = 1, + }, + [IORING_OP_RENAMEAT] = { + .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_FILES | + IO_WQ_WORK_FS | IO_WQ_WORK_BLKCG, + }, + [IORING_OP_UNLINKAT] = { + .work_flags = IO_WQ_WORK_MM | IO_WQ_WORK_FILES | + IO_WQ_WORK_FS | IO_WQ_WORK_BLKCG, + }, }; enum io_mem_account { @@ -983,6 +1034,9 @@ struct sock *io_uring_get_socket(struct file *file) } EXPORT_SYMBOL(io_uring_get_socket); +#define io_for_each_link(pos, head) \ + for (pos = (head); pos; pos = pos->link) + static inline void io_clean_op(struct io_kiocb *req) { if (req->flags & (REQ_F_NEED_CLEANUP | REQ_F_BUFFER_SELECTED | @@ -990,8 +1044,39 @@ static inline void io_clean_op(struct io_kiocb *req) __io_clean_op(req); } -static void io_sq_thread_drop_mm(void) +static inline void io_set_resource_node(struct io_kiocb *req) +{ + struct io_ring_ctx *ctx = req->ctx; + + if (!req->fixed_file_refs) { + req->fixed_file_refs = &ctx->file_data->node->refs; + percpu_ref_get(req->fixed_file_refs); + } +} + +static bool io_match_task(struct io_kiocb *head, + struct task_struct *task, + struct files_struct *files) { + struct io_kiocb *req; + + if (task && head->task != task) + return false; + if (!files) + return true; + + io_for_each_link(req, head) { + if ((req->flags & REQ_F_WORK_INITIALIZED) && + (req->work.flags & IO_WQ_WORK_FILES) && + req->work.identity->files == files) + return true; + } + return false; +} + +static void io_sq_thread_drop_mm_files(void) +{ + struct files_struct *files = current->files; struct mm_struct *mm = current->mm; if (mm) { @@ -999,6 +1084,41 @@ static void io_sq_thread_drop_mm(void) mmput(mm); current->mm = NULL; } + if (files) { + struct nsproxy *nsproxy = current->nsproxy; + + task_lock(current); + current->files = NULL; + current->nsproxy = NULL; + task_unlock(current); + put_files_struct(files); + put_nsproxy(nsproxy); + } +} + +static int __io_sq_thread_acquire_files(struct io_ring_ctx *ctx) +{ + if (!current->files) { + struct files_struct *files; + struct nsproxy *nsproxy; + + task_lock(ctx->sqo_task); + files = ctx->sqo_task->files; + if (!files) { + task_unlock(ctx->sqo_task); + return -EOWNERDEAD; + } + atomic_inc(&files->count); + get_nsproxy(ctx->sqo_task->nsproxy); + nsproxy = ctx->sqo_task->nsproxy; + task_unlock(ctx->sqo_task); + + task_lock(current); + current->files = files; + current->nsproxy = nsproxy; + task_unlock(current); + } + return 0; } static int __io_sq_thread_acquire_mm(struct io_ring_ctx *ctx) @@ -1026,12 +1146,25 @@ static int __io_sq_thread_acquire_mm(struct io_ring_ctx *ctx) return -EFAULT; } -static int io_sq_thread_acquire_mm(struct io_ring_ctx *ctx, - struct io_kiocb *req) +static int io_sq_thread_acquire_mm_files(struct io_ring_ctx *ctx, + struct io_kiocb *req) { - if (!(io_op_defs[req->opcode].work_flags & IO_WQ_WORK_MM)) - return 0; - return __io_sq_thread_acquire_mm(ctx); + const struct io_op_def *def = &io_op_defs[req->opcode]; + int ret; + + if (def->work_flags & IO_WQ_WORK_MM) { + ret = __io_sq_thread_acquire_mm(ctx); + if (unlikely(ret)) + return ret; + } + + if (def->needs_file || (def->work_flags & IO_WQ_WORK_FILES)) { + ret = __io_sq_thread_acquire_files(ctx); + if (unlikely(ret)) + return ret; + } + + return 0; } static void io_sq_thread_associate_blkcg(struct io_ring_ctx *ctx, @@ -1174,7 +1307,6 @@ static struct io_ring_ctx *io_ring_ctx_alloc(struct io_uring_params *p) INIT_LIST_HEAD(&ctx->iopoll_list); INIT_LIST_HEAD(&ctx->defer_list); INIT_LIST_HEAD(&ctx->timeout_list); - init_waitqueue_head(&ctx->inflight_wait); spin_lock_init(&ctx->inflight_lock); INIT_LIST_HEAD(&ctx->inflight_list); INIT_DELAYED_WORK(&ctx->file_put_work, io_file_put_work); @@ -1416,10 +1548,8 @@ static void io_prep_async_link(struct io_kiocb *req) { struct io_kiocb *cur; - 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); + io_for_each_link(cur, req) + io_prep_async_work(cur); } static struct io_kiocb *__io_queue_async_work(struct io_kiocb *req) @@ -1460,30 +1590,18 @@ static void io_kill_timeout(struct io_kiocb *req) } } -static bool io_task_match(struct io_kiocb *req, struct task_struct *tsk) -{ - struct io_ring_ctx *ctx = req->ctx; - - if (!tsk || req->task == tsk) - return true; - if (ctx->flags & IORING_SETUP_SQPOLL) { - if (ctx->sq_data && req->task == ctx->sq_data->thread) - return true; - } - return false; -} - /* * Returns true if we found and killed one or more timeouts */ -static bool io_kill_timeouts(struct io_ring_ctx *ctx, struct task_struct *tsk) +static bool io_kill_timeouts(struct io_ring_ctx *ctx, struct task_struct *tsk, + struct files_struct *files) { struct io_kiocb *req, *tmp; int canceled = 0; spin_lock_irq(&ctx->completion_lock); list_for_each_entry_safe(req, tmp, &ctx->timeout_list, timeout.list) { - if (io_task_match(req, tsk)) { + if (io_match_task(req, tsk, files)) { io_kill_timeout(req); canceled++; } @@ -1594,32 +1712,6 @@ static void io_cqring_mark_overflow(struct io_ring_ctx *ctx) } } -static inline bool __io_match_files(struct io_kiocb *req, - struct files_struct *files) -{ - return ((req->flags & REQ_F_WORK_INITIALIZED) && - (req->work.flags & IO_WQ_WORK_FILES)) && - req->work.identity->files == files; -} - -static bool io_match_files(struct io_kiocb *req, - struct files_struct *files) -{ - struct io_kiocb *link; - - if (!files) - return true; - if (__io_match_files(req, files)) - return true; - if (req->flags & REQ_F_LINK_HEAD) { - list_for_each_entry(link, &req->link_list, link_list) { - if (__io_match_files(link, files)) - return true; - } - } - return false; -} - /* Returns true if there are no backlogged entries after the flush */ static bool io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force, struct task_struct *tsk, @@ -1647,9 +1739,7 @@ static bool io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force, cqe = NULL; list_for_each_entry_safe(req, tmp, &ctx->cq_overflow_list, compl.list) { - if (tsk && req->task != tsk) - continue; - if (!io_match_files(req, files)) + if (!io_match_task(req, tsk, files)) continue; cqe = io_get_cqring(ctx); @@ -1845,9 +1935,7 @@ fallback: static inline void io_put_file(struct io_kiocb *req, struct file *file, bool fixed) { - if (fixed) - percpu_ref_put(req->fixed_file_refs); - else + if (!fixed) fput(file); } @@ -1859,7 +1947,8 @@ static void io_dismantle_req(struct io_kiocb *req) kfree(req->async_data); if (req->file) io_put_file(req, req->file, (req->flags & REQ_F_FIXED_FILE)); - + if (req->fixed_file_refs) + percpu_ref_put(req->fixed_file_refs); io_req_clean_work(req); } @@ -1882,6 +1971,14 @@ static void __io_free_req(struct io_kiocb *req) percpu_ref_put(&ctx->refs); } +static inline void io_remove_next_linked(struct io_kiocb *req) +{ + struct io_kiocb *nxt = req->link; + + req->link = nxt->link; + nxt->link = NULL; +} + static void io_kill_linked_timeout(struct io_kiocb *req) { struct io_ring_ctx *ctx = req->ctx; @@ -1890,8 +1987,8 @@ static void io_kill_linked_timeout(struct io_kiocb *req) unsigned long flags; spin_lock_irqsave(&ctx->completion_lock, flags); - link = list_first_entry_or_null(&req->link_list, struct io_kiocb, - link_list); + link = req->link; + /* * Can happen if a linked timeout fired and link had been like * req -> link t-out -> link t-out [-> ...] @@ -1900,7 +1997,8 @@ static void io_kill_linked_timeout(struct io_kiocb *req) struct io_timeout_data *io = link->async_data; int ret; - list_del_init(&link->link_list); + io_remove_next_linked(req); + link->timeout.head = NULL; ret = hrtimer_try_to_cancel(&io->timer); if (ret != -1) { io_cqring_fill_event(link, -ECANCELED); @@ -1917,41 +2015,22 @@ static void io_kill_linked_timeout(struct io_kiocb *req) } } -static struct io_kiocb *io_req_link_next(struct io_kiocb *req) -{ - struct io_kiocb *nxt; - /* - * The list should never be empty when we are called here. But could - * potentially happen if the chain is messed up, check to be on the - * safe side. - */ - if (unlikely(list_empty(&req->link_list))) - return NULL; - - nxt = list_first_entry(&req->link_list, struct io_kiocb, link_list); - list_del_init(&req->link_list); - if (!list_empty(&nxt->link_list)) - nxt->flags |= REQ_F_LINK_HEAD; - return nxt; -} - -/* - * Called if REQ_F_LINK_HEAD is set, and we fail the head request - */ static void io_fail_links(struct io_kiocb *req) { + struct io_kiocb *link, *nxt; struct io_ring_ctx *ctx = req->ctx; unsigned long flags; spin_lock_irqsave(&ctx->completion_lock, flags); - while (!list_empty(&req->link_list)) { - struct io_kiocb *link = list_first_entry(&req->link_list, - struct io_kiocb, link_list); + link = req->link; + req->link = NULL; - list_del_init(&link->link_list); - trace_io_uring_fail_link(req, link); + while (link) { + nxt = link->link; + link->link = NULL; + trace_io_uring_fail_link(req, link); io_cqring_fill_event(link, -ECANCELED); /* @@ -1963,8 +2042,8 @@ static void io_fail_links(struct io_kiocb *req) io_put_req_deferred(link, 2); else io_double_put_req(link); + link = nxt; } - io_commit_cqring(ctx); spin_unlock_irqrestore(&ctx->completion_lock, flags); @@ -1973,7 +2052,6 @@ static void io_fail_links(struct io_kiocb *req) static struct io_kiocb *__io_req_find_next(struct io_kiocb *req) { - req->flags &= ~REQ_F_LINK_HEAD; if (req->flags & REQ_F_LINK_TIMEOUT) io_kill_linked_timeout(req); @@ -1983,15 +2061,19 @@ static struct io_kiocb *__io_req_find_next(struct io_kiocb *req) * dependencies to the next request. In case of failure, fail the rest * of the chain. */ - if (likely(!(req->flags & REQ_F_FAIL_LINK))) - return io_req_link_next(req); + if (likely(!(req->flags & REQ_F_FAIL_LINK))) { + struct io_kiocb *nxt = req->link; + + req->link = NULL; + return nxt; + } io_fail_links(req); return NULL; } -static struct io_kiocb *io_req_find_next(struct io_kiocb *req) +static inline struct io_kiocb *io_req_find_next(struct io_kiocb *req) { - if (likely(!(req->flags & REQ_F_LINK_HEAD))) + if (likely(!(req->link) && !(req->flags & REQ_F_LINK_TIMEOUT))) return NULL; return __io_req_find_next(req); } @@ -2050,7 +2132,8 @@ static void __io_req_task_submit(struct io_kiocb *req) { struct io_ring_ctx *ctx = req->ctx; - if (!__io_sq_thread_acquire_mm(ctx)) { + if (!__io_sq_thread_acquire_mm(ctx) && + !__io_sq_thread_acquire_files(ctx)) { mutex_lock(&ctx->uring_lock); __io_queue_sqe(req, NULL); mutex_unlock(&ctx->uring_lock); @@ -2086,7 +2169,7 @@ static void io_req_task_queue(struct io_kiocb *req) } } -static void io_queue_next(struct io_kiocb *req) +static inline void io_queue_next(struct io_kiocb *req) { struct io_kiocb *nxt = io_req_find_next(req); @@ -2143,8 +2226,7 @@ static void io_req_free_batch(struct req_batch *rb, struct io_kiocb *req) io_free_req(req); return; } - if (req->flags & REQ_F_LINK_HEAD) - io_queue_next(req); + io_queue_next(req); if (req->task != rb->task) { if (rb->task) { @@ -2246,7 +2328,7 @@ static unsigned io_cqring_events(struct io_ring_ctx *ctx, bool noflush) * we wake up the task, and the next invocation will flush the * entries. We cannot safely to it from here. */ - if (noflush && !list_empty(&ctx->cq_overflow_list)) + if (noflush) return -1U; io_cqring_overflow_flush(ctx, false, NULL, NULL); @@ -2593,7 +2675,7 @@ static bool io_rw_reissue(struct io_kiocb *req, long res) if ((res != -EAGAIN && res != -EOPNOTSUPP) || io_wq_current_is_worker()) return false; - ret = io_sq_thread_acquire_mm(req->ctx, req); + ret = io_sq_thread_acquire_mm_files(req->ctx, req); if (io_resubmit_prep(req, ret)) { refcount_inc(&req->refs); @@ -2641,7 +2723,7 @@ static void io_complete_rw_iopoll(struct kiocb *kiocb, long res, long res2) * find it from a io_iopoll_getevents() thread before the issuer is done * accessing the kiocb cookie. */ -static void io_iopoll_req_issued(struct io_kiocb *req) +static void io_iopoll_req_issued(struct io_kiocb *req, bool in_async) { struct io_ring_ctx *ctx = req->ctx; @@ -2670,21 +2752,25 @@ static void io_iopoll_req_issued(struct io_kiocb *req) else list_add_tail(&req->inflight_entry, &ctx->iopoll_list); - if ((ctx->flags & IORING_SETUP_SQPOLL) && + /* + * If IORING_SETUP_SQPOLL is enabled, sqes are either handled in sq thread + * task context or in io worker task context. If current task context is + * sq thread, we don't need to check whether should wake up sq thread. + */ + if (in_async && (ctx->flags & IORING_SETUP_SQPOLL) && wq_has_sleeper(&ctx->sq_data->wait)) wake_up(&ctx->sq_data->wait); } -static void __io_state_file_put(struct io_submit_state *state) +static inline void __io_state_file_put(struct io_submit_state *state) { - if (state->has_refs) - fput_many(state->file, state->has_refs); - state->file = NULL; + fput_many(state->file, state->file_refs); + state->file_refs = 0; } static inline void io_state_file_put(struct io_submit_state *state) { - if (state->file) + if (state->file_refs) __io_state_file_put(state); } @@ -2698,19 +2784,19 @@ static struct file *__io_file_get(struct io_submit_state *state, int fd) if (!state) return fget(fd); - if (state->file) { + if (state->file_refs) { if (state->fd == fd) { - state->has_refs--; + state->file_refs--; return state->file; } __io_state_file_put(state); } state->file = fget_many(fd, state->ios_left); - if (!state->file) + if (unlikely(!state->file)) return NULL; state->fd = fd; - state->has_refs = state->ios_left - 1; + state->file_refs = state->ios_left - 1; return state->file; } @@ -3065,7 +3151,7 @@ static ssize_t io_iov_buffer_select(struct io_kiocb *req, struct iovec *iov, return __io_iov_buffer_select(req, iov, needs_lock); } -static ssize_t __io_import_iovec(int rw, struct io_kiocb *req, +static ssize_t io_import_iovec(int rw, struct io_kiocb *req, struct iovec **iovec, struct iov_iter *iter, bool needs_lock) { @@ -3094,7 +3180,7 @@ static ssize_t __io_import_iovec(int rw, struct io_kiocb *req, ret = import_single_range(rw, buf, sqe_len, *iovec, iter); *iovec = NULL; - return ret < 0 ? ret : sqe_len; + return ret; } if (req->flags & REQ_F_BUFFER_SELECT) { @@ -3111,18 +3197,6 @@ static ssize_t __io_import_iovec(int rw, struct io_kiocb *req, req->ctx->compat); } -static ssize_t io_import_iovec(int rw, struct io_kiocb *req, - struct iovec **iovec, struct iov_iter *iter, - bool needs_lock) -{ - struct io_async_rw *iorw = req->async_data; - - if (!iorw) - return __io_import_iovec(rw, req, iovec, iter, needs_lock); - *iovec = NULL; - return iov_iter_count(&iorw->iter); -} - static inline loff_t *io_kiocb_ppos(struct kiocb *kiocb) { return (kiocb->ki_filp->f_mode & FMODE_STREAM) ? NULL : &kiocb->ki_pos; @@ -3246,7 +3320,7 @@ static inline int io_rw_prep_async(struct io_kiocb *req, int rw) struct iovec *iov = iorw->fast_iov; ssize_t ret; - ret = __io_import_iovec(rw, req, &iov, &iorw->iter, false); + ret = io_import_iovec(rw, req, &iov, &iorw->iter, false); if (unlikely(ret < 0)) return ret; @@ -3379,17 +3453,17 @@ static int io_read(struct io_kiocb *req, bool force_nonblock, struct iov_iter __iter, *iter = &__iter; struct io_async_rw *rw = req->async_data; ssize_t io_size, ret, ret2; - size_t iov_count; bool no_async; - if (rw) + if (rw) { iter = &rw->iter; - - ret = io_import_iovec(READ, req, &iovec, iter, !force_nonblock); - if (ret < 0) - return ret; - iov_count = iov_iter_count(iter); - io_size = ret; + iovec = NULL; + } else { + ret = io_import_iovec(READ, req, &iovec, iter, !force_nonblock); + if (ret < 0) + return ret; + } + io_size = iov_iter_count(iter); req->result = io_size; ret = 0; @@ -3405,7 +3479,7 @@ static int io_read(struct io_kiocb *req, bool force_nonblock, if (no_async) goto copy_iov; - ret = rw_verify_area(READ, req->file, io_kiocb_ppos(kiocb), iov_count); + ret = rw_verify_area(READ, req->file, io_kiocb_ppos(kiocb), io_size); if (unlikely(ret)) goto out_free; @@ -3424,7 +3498,7 @@ static int io_read(struct io_kiocb *req, bool force_nonblock, if (req->file->f_flags & O_NONBLOCK) goto done; /* some cases will consume bytes even on error returns */ - iov_iter_revert(iter, iov_count - iov_iter_count(iter)); + iov_iter_revert(iter, io_size - iov_iter_count(iter)); ret = 0; goto copy_iov; } else if (ret < 0) { @@ -3507,17 +3581,17 @@ static int io_write(struct io_kiocb *req, bool force_nonblock, struct kiocb *kiocb = &req->rw.kiocb; struct iov_iter __iter, *iter = &__iter; struct io_async_rw *rw = req->async_data; - size_t iov_count; ssize_t ret, ret2, io_size; - if (rw) + if (rw) { iter = &rw->iter; - - ret = io_import_iovec(WRITE, req, &iovec, iter, !force_nonblock); - if (ret < 0) - return ret; - iov_count = iov_iter_count(iter); - io_size = ret; + iovec = NULL; + } else { + ret = io_import_iovec(WRITE, req, &iovec, iter, !force_nonblock); + if (ret < 0) + return ret; + } + io_size = iov_iter_count(iter); req->result = io_size; /* Ensure we clear previously set non-block flag */ @@ -3535,7 +3609,7 @@ static int io_write(struct io_kiocb *req, bool force_nonblock, (req->flags & REQ_F_ISREG)) goto copy_iov; - ret = rw_verify_area(WRITE, req->file, io_kiocb_ppos(kiocb), iov_count); + ret = rw_verify_area(WRITE, req->file, io_kiocb_ppos(kiocb), io_size); if (unlikely(ret)) goto out_free; @@ -3578,7 +3652,7 @@ done: } else { copy_iov: /* some cases will consume bytes even on error returns */ - iov_iter_revert(iter, iov_count - iov_iter_count(iter)); + iov_iter_revert(iter, io_size - iov_iter_count(iter)); ret = io_setup_async_rw(req, iovec, inline_vecs, iter, false); if (!ret) return -EAGAIN; @@ -3590,6 +3664,135 @@ out_free: return ret; } +static int io_renameat_prep(struct io_kiocb *req, + const struct io_uring_sqe *sqe) +{ + struct io_rename *ren = &req->rename; + const char __user *oldf, *newf; + + if (unlikely(req->flags & REQ_F_FIXED_FILE)) + return -EBADF; + + ren->old_dfd = READ_ONCE(sqe->fd); + oldf = u64_to_user_ptr(READ_ONCE(sqe->addr)); + newf = u64_to_user_ptr(READ_ONCE(sqe->addr2)); + ren->new_dfd = READ_ONCE(sqe->len); + ren->flags = READ_ONCE(sqe->rename_flags); + + ren->oldpath = getname(oldf); + if (IS_ERR(ren->oldpath)) + return PTR_ERR(ren->oldpath); + + ren->newpath = getname(newf); + if (IS_ERR(ren->newpath)) { + putname(ren->oldpath); + return PTR_ERR(ren->newpath); + } + + req->flags |= REQ_F_NEED_CLEANUP; + return 0; +} + +static int io_renameat(struct io_kiocb *req, bool force_nonblock) +{ + struct io_rename *ren = &req->rename; + int ret; + + if (force_nonblock) + return -EAGAIN; + + ret = do_renameat2(ren->old_dfd, ren->oldpath, ren->new_dfd, + ren->newpath, ren->flags); + + req->flags &= ~REQ_F_NEED_CLEANUP; + if (ret < 0) + req_set_fail_links(req); + io_req_complete(req, ret); + return 0; +} + +static int io_unlinkat_prep(struct io_kiocb *req, + const struct io_uring_sqe *sqe) +{ + struct io_unlink *un = &req->unlink; + const char __user *fname; + + if (unlikely(req->flags & REQ_F_FIXED_FILE)) + return -EBADF; + + un->dfd = READ_ONCE(sqe->fd); + + un->flags = READ_ONCE(sqe->unlink_flags); + if (un->flags & ~AT_REMOVEDIR) + return -EINVAL; + + fname = u64_to_user_ptr(READ_ONCE(sqe->addr)); + un->filename = getname(fname); + if (IS_ERR(un->filename)) + return PTR_ERR(un->filename); + + req->flags |= REQ_F_NEED_CLEANUP; + return 0; +} + +static int io_unlinkat(struct io_kiocb *req, bool force_nonblock) +{ + struct io_unlink *un = &req->unlink; + int ret; + + if (force_nonblock) + return -EAGAIN; + + if (un->flags & AT_REMOVEDIR) + ret = do_rmdir(un->dfd, un->filename); + else + ret = do_unlinkat(un->dfd, un->filename); + + req->flags &= ~REQ_F_NEED_CLEANUP; + if (ret < 0) + req_set_fail_links(req); + io_req_complete(req, ret); + return 0; +} + +static int io_shutdown_prep(struct io_kiocb *req, + const struct io_uring_sqe *sqe) +{ +#if defined(CONFIG_NET) + if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL)) + return -EINVAL; + if (sqe->ioprio || sqe->off || sqe->addr || sqe->rw_flags || + sqe->buf_index) + return -EINVAL; + + req->shutdown.how = READ_ONCE(sqe->len); + return 0; +#else + return -EOPNOTSUPP; +#endif +} + +static int io_shutdown(struct io_kiocb *req, bool force_nonblock) +{ +#if defined(CONFIG_NET) + struct socket *sock; + int ret; + + if (force_nonblock) + return -EAGAIN; + + sock = sock_from_file(req->file); + if (unlikely(!sock)) + return -ENOTSOCK; + + ret = __sys_shutdown_sock(sock, req->shutdown.how); + io_req_complete(req, ret); + return 0; +#else + return -EOPNOTSUPP; +#endif +} + static int __io_splice_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) { @@ -3804,7 +4007,7 @@ static int io_openat_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) { u64 flags, mode; - if (unlikely(req->ctx->flags & (IORING_SETUP_IOPOLL|IORING_SETUP_SQPOLL))) + if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL)) return -EINVAL; mode = READ_ONCE(sqe->len); flags = READ_ONCE(sqe->open_flags); @@ -3818,7 +4021,7 @@ static int io_openat2_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) size_t len; int ret; - if (unlikely(req->ctx->flags & (IORING_SETUP_IOPOLL|IORING_SETUP_SQPOLL))) + if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL)) return -EINVAL; how = u64_to_user_ptr(READ_ONCE(sqe->addr2)); len = READ_ONCE(sqe->len); @@ -3948,11 +4151,17 @@ static int io_remove_buffers(struct io_kiocb *req, bool force_nonblock, head = idr_find(&ctx->io_buffer_idr, p->bgid); if (head) ret = __io_remove_buffers(ctx, head, p->bgid, p->nbufs); - - io_ring_submit_lock(ctx, !force_nonblock); if (ret < 0) req_set_fail_links(req); - __io_req_complete(req, ret, 0, cs); + + /* need to hold the lock to complete IOPOLL requests */ + if (ctx->flags & IORING_SETUP_IOPOLL) { + __io_req_complete(req, ret, 0, cs); + io_ring_submit_unlock(ctx, !force_nonblock); + } else { + io_ring_submit_unlock(ctx, !force_nonblock); + __io_req_complete(req, ret, 0, cs); + } return 0; } @@ -4037,10 +4246,17 @@ static int io_provide_buffers(struct io_kiocb *req, bool force_nonblock, } } out: - io_ring_submit_unlock(ctx, !force_nonblock); if (ret < 0) req_set_fail_links(req); - __io_req_complete(req, ret, 0, cs); + + /* need to hold the lock to complete IOPOLL requests */ + if (ctx->flags & IORING_SETUP_IOPOLL) { + __io_req_complete(req, ret, 0, cs); + io_ring_submit_unlock(ctx, !force_nonblock); + } else { + io_ring_submit_unlock(ctx, !force_nonblock); + __io_req_complete(req, ret, 0, cs); + } return 0; } @@ -4212,7 +4428,7 @@ static int io_close_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) io_req_init_async(req); req->work.flags |= IO_WQ_WORK_NO_CANCEL; - if (unlikely(req->ctx->flags & (IORING_SETUP_IOPOLL|IORING_SETUP_SQPOLL))) + if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL)) return -EINVAL; if (sqe->ioprio || sqe->off || sqe->addr || sqe->len || sqe->rw_flags || sqe->buf_index) @@ -4694,7 +4910,7 @@ static int io_accept_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) { struct io_accept *accept = &req->accept; - if (unlikely(req->ctx->flags & (IORING_SETUP_IOPOLL|IORING_SETUP_SQPOLL))) + if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL)) return -EINVAL; if (sqe->ioprio || sqe->len || sqe->buf_index) return -EINVAL; @@ -4735,7 +4951,7 @@ static int io_connect_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) struct io_connect *conn = &req->connect; struct io_async_connect *io = req->async_data; - if (unlikely(req->ctx->flags & (IORING_SETUP_IOPOLL|IORING_SETUP_SQPOLL))) + if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL)) return -EINVAL; if (sqe->ioprio || sqe->len || sqe->buf_index || sqe->rw_flags) return -EINVAL; @@ -5270,7 +5486,8 @@ static bool io_poll_remove_one(struct io_kiocb *req) /* * Returns true if we found and killed one or more poll requests */ -static bool io_poll_remove_all(struct io_ring_ctx *ctx, struct task_struct *tsk) +static bool io_poll_remove_all(struct io_ring_ctx *ctx, struct task_struct *tsk, + struct files_struct *files) { struct hlist_node *tmp; struct io_kiocb *req; @@ -5282,7 +5499,7 @@ static bool io_poll_remove_all(struct io_ring_ctx *ctx, struct task_struct *tsk) list = &ctx->cancel_hash[i]; hlist_for_each_entry_safe(req, tmp, list, hash_node) { - if (io_task_match(req, tsk)) + if (io_match_task(req, tsk, files)) posted += io_poll_remove_one(req); } } @@ -5320,7 +5537,7 @@ static int io_poll_remove_prep(struct io_kiocb *req, sqe->poll_events) return -EINVAL; - req->poll.addr = READ_ONCE(sqe->addr); + req->poll_remove.addr = READ_ONCE(sqe->addr); return 0; } @@ -5331,12 +5548,10 @@ static int io_poll_remove_prep(struct io_kiocb *req, static int io_poll_remove(struct io_kiocb *req) { struct io_ring_ctx *ctx = req->ctx; - u64 addr; int ret; - addr = req->poll.addr; spin_lock_irq(&ctx->completion_lock); - ret = io_poll_cancel(ctx, addr); + ret = io_poll_cancel(ctx, req->poll_remove.addr); spin_unlock_irq(&ctx->completion_lock); if (ret < 0) @@ -5429,15 +5644,37 @@ static enum hrtimer_restart io_timeout_fn(struct hrtimer *timer) return HRTIMER_NORESTART; } -static int __io_timeout_cancel(struct io_kiocb *req) +static struct io_kiocb *io_timeout_extract(struct io_ring_ctx *ctx, + __u64 user_data) { - struct io_timeout_data *io = req->async_data; - int ret; + struct io_timeout_data *io; + struct io_kiocb *req; + int ret = -ENOENT; + list_for_each_entry(req, &ctx->timeout_list, timeout.list) { + if (user_data == req->user_data) { + ret = 0; + break; + } + } + + if (ret == -ENOENT) + return ERR_PTR(ret); + + io = req->async_data; ret = hrtimer_try_to_cancel(&io->timer); if (ret == -1) - return -EALREADY; + return ERR_PTR(-EALREADY); list_del_init(&req->timeout.list); + return req; +} + +static int io_timeout_cancel(struct io_ring_ctx *ctx, __u64 user_data) +{ + struct io_kiocb *req = io_timeout_extract(ctx, user_data); + + if (IS_ERR(req)) + return PTR_ERR(req); req_set_fail_links(req); io_cqring_fill_event(req, -ECANCELED); @@ -5445,35 +5682,48 @@ static int __io_timeout_cancel(struct io_kiocb *req) return 0; } -static int io_timeout_cancel(struct io_ring_ctx *ctx, __u64 user_data) +static int io_timeout_update(struct io_ring_ctx *ctx, __u64 user_data, + struct timespec64 *ts, enum hrtimer_mode mode) { - struct io_kiocb *req; - int ret = -ENOENT; - - list_for_each_entry(req, &ctx->timeout_list, timeout.list) { - if (user_data == req->user_data) { - ret = 0; - break; - } - } + struct io_kiocb *req = io_timeout_extract(ctx, user_data); + struct io_timeout_data *data; - if (ret == -ENOENT) - return ret; + if (IS_ERR(req)) + return PTR_ERR(req); - return __io_timeout_cancel(req); + req->timeout.off = 0; /* noseq */ + data = req->async_data; + list_add_tail(&req->timeout.list, &ctx->timeout_list); + hrtimer_init(&data->timer, CLOCK_MONOTONIC, mode); + data->timer.function = io_timeout_fn; + hrtimer_start(&data->timer, timespec64_to_ktime(*ts), mode); + return 0; } static int io_timeout_remove_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) { + struct io_timeout_rem *tr = &req->timeout_rem; + if (unlikely(req->ctx->flags & IORING_SETUP_IOPOLL)) return -EINVAL; if (unlikely(req->flags & (REQ_F_FIXED_FILE | REQ_F_BUFFER_SELECT))) return -EINVAL; - if (sqe->ioprio || sqe->buf_index || sqe->len || sqe->timeout_flags) + if (sqe->ioprio || sqe->buf_index || sqe->len) return -EINVAL; - req->timeout_rem.addr = READ_ONCE(sqe->addr); + tr->addr = READ_ONCE(sqe->addr); + tr->flags = READ_ONCE(sqe->timeout_flags); + if (tr->flags & IORING_TIMEOUT_UPDATE) { + if (tr->flags & ~(IORING_TIMEOUT_UPDATE|IORING_TIMEOUT_ABS)) + return -EINVAL; + if (get_timespec64(&tr->ts, u64_to_user_ptr(sqe->addr2))) + return -EFAULT; + } else if (tr->flags) { + /* timeout removal doesn't support flags */ + return -EINVAL; + } + return 0; } @@ -5482,11 +5732,19 @@ static int io_timeout_remove_prep(struct io_kiocb *req, */ static int io_timeout_remove(struct io_kiocb *req) { + struct io_timeout_rem *tr = &req->timeout_rem; struct io_ring_ctx *ctx = req->ctx; int ret; spin_lock_irq(&ctx->completion_lock); - ret = io_timeout_cancel(ctx, req->timeout_rem.addr); + if (req->timeout_rem.flags & IORING_TIMEOUT_UPDATE) { + enum hrtimer_mode mode = (tr->flags & IORING_TIMEOUT_ABS) + ? HRTIMER_MODE_ABS : HRTIMER_MODE_REL; + + ret = io_timeout_update(ctx, tr->addr, &tr->ts, mode); + } else { + ret = io_timeout_cancel(ctx, tr->addr); + } io_cqring_fill_event(req, ret); io_commit_cqring(ctx); @@ -5766,6 +6024,12 @@ static int io_req_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe) return io_remove_buffers_prep(req, sqe); case IORING_OP_TEE: return io_tee_prep(req, sqe); + case IORING_OP_SHUTDOWN: + return io_shutdown_prep(req, sqe); + case IORING_OP_RENAMEAT: + return io_renameat_prep(req, sqe); + case IORING_OP_UNLINKAT: + return io_unlinkat_prep(req, sqe); } printk_once(KERN_WARNING "io_uring: unhandled opcode %d\n", @@ -5787,11 +6051,10 @@ static u32 io_get_sequence(struct io_kiocb *req) { struct io_kiocb *pos; struct io_ring_ctx *ctx = req->ctx; - u32 total_submitted, nr_reqs = 1; + u32 total_submitted, nr_reqs = 0; - if (req->flags & REQ_F_LINK_HEAD) - list_for_each_entry(pos, &req->link_list, link_list) - nr_reqs++; + io_for_each_link(pos, req) + nr_reqs++; total_submitted = ctx->cached_sq_head - ctx->cached_sq_dropped; return total_submitted - nr_reqs; @@ -5843,12 +6106,13 @@ static int io_req_defer(struct io_kiocb *req, const struct io_uring_sqe *sqe) static void io_req_drop_files(struct io_kiocb *req) { struct io_ring_ctx *ctx = req->ctx; + struct io_uring_task *tctx = req->task->io_uring; unsigned long flags; spin_lock_irqsave(&ctx->inflight_lock, flags); list_del(&req->inflight_entry); - if (waitqueue_active(&ctx->inflight_wait)) - wake_up(&ctx->inflight_wait); + if (atomic_read(&tctx->in_idle)) + wake_up(&tctx->wait); spin_unlock_irqrestore(&ctx->inflight_lock, flags); req->flags &= ~REQ_F_INFLIGHT; put_files_struct(req->work.identity->files); @@ -5903,6 +6167,13 @@ static void __io_clean_op(struct io_kiocb *req) if (req->open.filename) putname(req->open.filename); break; + case IORING_OP_RENAMEAT: + putname(req->rename.oldpath); + putname(req->rename.newpath); + break; + case IORING_OP_UNLINKAT: + putname(req->unlink.filename); + break; } req->flags &= ~REQ_F_NEED_CLEANUP; } @@ -6009,6 +6280,15 @@ static int io_issue_sqe(struct io_kiocb *req, bool force_nonblock, case IORING_OP_TEE: ret = io_tee(req, force_nonblock); break; + case IORING_OP_SHUTDOWN: + ret = io_shutdown(req, force_nonblock); + break; + case IORING_OP_RENAMEAT: + ret = io_renameat(req, force_nonblock); + break; + case IORING_OP_UNLINKAT: + ret = io_unlinkat(req, force_nonblock); + break; default: ret = -EINVAL; break; @@ -6025,7 +6305,7 @@ static int io_issue_sqe(struct io_kiocb *req, bool force_nonblock, if (in_async) mutex_lock(&ctx->uring_lock); - io_iopoll_req_issued(req); + io_iopoll_req_issued(req, in_async); if (in_async) mutex_unlock(&ctx->uring_lock); @@ -6065,8 +6345,19 @@ static struct io_wq_work *io_wq_submit_work(struct io_wq_work *work) } if (ret) { - req_set_fail_links(req); - io_req_complete(req, ret); + /* + * io_iopoll_complete() does not hold completion_lock to complete + * polled io, so here for polled io, just mark it done and still let + * io_iopoll_complete() complete it. + */ + if (req->ctx->flags & IORING_SETUP_IOPOLL) { + struct kiocb *kiocb = &req->rw.kiocb; + + kiocb_done(kiocb, ret, NULL); + } else { + req_set_fail_links(req); + io_req_complete(req, ret); + } } return io_steal_work(req); @@ -6092,10 +6383,7 @@ static struct file *io_file_get(struct io_submit_state *state, return NULL; fd = array_index_nospec(fd, ctx->nr_user_files); file = io_file_from_index(ctx, fd); - if (file) { - req->fixed_file_refs = &ctx->file_data->node->refs; - percpu_ref_get(req->fixed_file_refs); - } + io_set_resource_node(req); } else { trace_io_uring_file_get(ctx, fd); file = __io_file_get(state, fd); @@ -6104,45 +6392,26 @@ static struct file *io_file_get(struct io_submit_state *state, return file; } -static int io_req_set_file(struct io_submit_state *state, struct io_kiocb *req, - int fd) -{ - bool fixed; - - fixed = (req->flags & REQ_F_FIXED_FILE) != 0; - if (unlikely(!fixed && io_async_submit(req->ctx))) - return -EBADF; - - req->file = io_file_get(state, req, fd, fixed); - if (req->file || io_op_defs[req->opcode].needs_file_no_error) - return 0; - return -EBADF; -} - static enum hrtimer_restart io_link_timeout_fn(struct hrtimer *timer) { struct io_timeout_data *data = container_of(timer, struct io_timeout_data, timer); - struct io_kiocb *req = data->req; + struct io_kiocb *prev, *req = data->req; struct io_ring_ctx *ctx = req->ctx; - struct io_kiocb *prev = NULL; unsigned long flags; spin_lock_irqsave(&ctx->completion_lock, flags); + prev = req->timeout.head; + req->timeout.head = NULL; /* * We don't expect the list to be empty, that will only happen if we * race with the completion of the linked work. */ - if (!list_empty(&req->link_list)) { - prev = list_entry(req->link_list.prev, struct io_kiocb, - link_list); - if (refcount_inc_not_zero(&prev->refs)) - list_del_init(&req->link_list); - else - prev = NULL; - } - + if (prev && refcount_inc_not_zero(&prev->refs)) + io_remove_next_linked(prev); + else + prev = NULL; spin_unlock_irqrestore(&ctx->completion_lock, flags); if (prev) { @@ -6158,10 +6427,10 @@ static enum hrtimer_restart io_link_timeout_fn(struct hrtimer *timer) static void __io_queue_linked_timeout(struct io_kiocb *req) { /* - * If the list is now empty, then our linked request finished before - * we got a chance to setup the timer + * If the back reference is NULL, then our linked request finished + * before we got a chance to setup the timer */ - if (!list_empty(&req->link_list)) { + if (req->timeout.head) { struct io_timeout_data *data = req->async_data; data->timer.function = io_link_timeout_fn; @@ -6184,18 +6453,13 @@ static void io_queue_linked_timeout(struct io_kiocb *req) static struct io_kiocb *io_prep_linked_timeout(struct io_kiocb *req) { - struct io_kiocb *nxt; + struct io_kiocb *nxt = req->link; - if (!(req->flags & REQ_F_LINK_HEAD)) - return NULL; - if (req->flags & REQ_F_LINK_TIMEOUT) - return NULL; - - nxt = list_first_entry_or_null(&req->link_list, struct io_kiocb, - link_list); - if (!nxt || nxt->opcode != IORING_OP_LINK_TIMEOUT) + if (!nxt || (req->flags & REQ_F_LINK_TIMEOUT) || + nxt->opcode != IORING_OP_LINK_TIMEOUT) return NULL; + nxt->timeout.head = req; nxt->flags |= REQ_F_LTIMEOUT_ACTIVE; req->flags |= REQ_F_LINK_TIMEOUT; return nxt; @@ -6301,8 +6565,13 @@ static inline void io_queue_link_head(struct io_kiocb *req, io_queue_sqe(req, NULL, cs); } +struct io_submit_link { + struct io_kiocb *head; + struct io_kiocb *last; +}; + static int io_submit_sqe(struct io_kiocb *req, const struct io_uring_sqe *sqe, - struct io_kiocb **link, struct io_comp_state *cs) + struct io_submit_link *link, struct io_comp_state *cs) { struct io_ring_ctx *ctx = req->ctx; int ret; @@ -6314,8 +6583,8 @@ static int io_submit_sqe(struct io_kiocb *req, const struct io_uring_sqe *sqe, * submitted sync once the chain is complete. If none of those * conditions are true (normal request), then just queue it. */ - if (*link) { - struct io_kiocb *head = *link; + if (link->head) { + struct io_kiocb *head = link->head; /* * Taking sequential execution of a link, draining both sides @@ -6335,12 +6604,13 @@ static int io_submit_sqe(struct io_kiocb *req, const struct io_uring_sqe *sqe, return ret; } trace_io_uring_link(ctx, req, head); - list_add_tail(&req->link_list, &head->link_list); + link->last->link = req; + link->last = req; /* last request of a link, enqueue the link */ if (!(req->flags & (REQ_F_LINK | REQ_F_HARDLINK))) { io_queue_link_head(head, cs); - *link = NULL; + link->head = NULL; } } else { if (unlikely(ctx->drain_next)) { @@ -6348,13 +6618,11 @@ static int io_submit_sqe(struct io_kiocb *req, const struct io_uring_sqe *sqe, ctx->drain_next = 0; } if (req->flags & (REQ_F_LINK | REQ_F_HARDLINK)) { - req->flags |= REQ_F_LINK_HEAD; - INIT_LIST_HEAD(&req->link_list); - ret = io_req_defer_prep(req, sqe); if (unlikely(ret)) req->flags |= REQ_F_FAIL_LINK; - *link = req; + link->head = req; + link->last = req; } else { io_queue_sqe(req, sqe, cs); } @@ -6370,7 +6638,8 @@ static void io_submit_state_end(struct io_submit_state *state) { if (!list_empty(&state->comp.list)) io_submit_flush_completions(&state->comp); - blk_finish_plug(&state->plug); + if (state->plug_started) + blk_finish_plug(&state->plug); io_state_file_put(state); if (state->free_reqs) kmem_cache_free_bulk(req_cachep, state->free_reqs, state->reqs); @@ -6382,12 +6651,12 @@ static void io_submit_state_end(struct io_submit_state *state) static void io_submit_state_start(struct io_submit_state *state, struct io_ring_ctx *ctx, unsigned int max_ios) { - blk_start_plug(&state->plug); + state->plug_started = false; state->comp.nr = 0; INIT_LIST_HEAD(&state->comp.list); state->comp.ctx = ctx; state->free_reqs = 0; - state->file = NULL; + state->file_refs = 0; state->ios_left = max_ios; } @@ -6482,6 +6751,8 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req, req->file = NULL; req->ctx = ctx; req->flags = 0; + req->link = NULL; + req->fixed_file_refs = NULL; /* one is dropped after submission, the other at completion */ refcount_set(&req->refs, 2); req->task = current; @@ -6490,7 +6761,7 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req, if (unlikely(req->opcode >= IORING_OP_LAST)) return -EINVAL; - if (unlikely(io_sq_thread_acquire_mm(ctx, req))) + if (unlikely(io_sq_thread_acquire_mm_files(ctx, req))) return -EFAULT; sqe_flags = READ_ONCE(sqe->flags); @@ -6523,10 +6794,26 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req, /* same numerical values with corresponding REQ_F_*, safe to copy */ req->flags |= sqe_flags; - if (!io_op_defs[req->opcode].needs_file) - return 0; + /* + * Plug now if we have more than 1 IO left after this, and the target + * is potentially a read/write to block based storage. + */ + if (!state->plug_started && state->ios_left > 1 && + io_op_defs[req->opcode].plug) { + blk_start_plug(&state->plug); + state->plug_started = true; + } + + ret = 0; + if (io_op_defs[req->opcode].needs_file) { + bool fixed = req->flags & REQ_F_FIXED_FILE; + + req->file = io_file_get(state, req, READ_ONCE(sqe->fd), fixed); + if (unlikely(!req->file && + !io_op_defs[req->opcode].needs_file_no_error)) + ret = -EBADF; + } - ret = io_req_set_file(state, req, READ_ONCE(sqe->fd)); state->ios_left--; return ret; } @@ -6534,7 +6821,7 @@ static int io_init_req(struct io_ring_ctx *ctx, struct io_kiocb *req, static int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr) { struct io_submit_state state; - struct io_kiocb *link = NULL; + struct io_submit_link link; int i, submitted = 0; /* if we have a backlog and couldn't flush it all, return BUSY */ @@ -6554,6 +6841,7 @@ static int io_submit_sqes(struct io_ring_ctx *ctx, unsigned int nr) refcount_add(nr, ¤t->usage); io_submit_state_start(&state, ctx, nr); + link.head = NULL; for (i = 0; i < nr; i++) { const struct io_uring_sqe *sqe; @@ -6599,8 +6887,8 @@ fail_req: percpu_counter_sub(&tctx->inflight, unused); put_task_struct_many(current, unused); } - if (link) - io_queue_link_head(link, &state.comp); + if (link.head) + io_queue_link_head(link.head, &state.comp); io_submit_state_end(&state); /* Commit SQ ring head once we've consumed and submitted all SQEs */ @@ -6624,111 +6912,45 @@ static inline void io_ring_clear_wakeup_flag(struct io_ring_ctx *ctx) spin_unlock_irq(&ctx->completion_lock); } -static int io_sq_wake_function(struct wait_queue_entry *wqe, unsigned mode, - int sync, void *key) +static int __io_sq_thread(struct io_ring_ctx *ctx, bool cap_entries) { - struct io_ring_ctx *ctx = container_of(wqe, struct io_ring_ctx, sqo_wait_entry); - int ret; - - ret = autoremove_wake_function(wqe, mode, sync, key); - if (ret) { - unsigned long flags; - - spin_lock_irqsave(&ctx->completion_lock, flags); - ctx->rings->sq_flags &= ~IORING_SQ_NEED_WAKEUP; - spin_unlock_irqrestore(&ctx->completion_lock, flags); - } - return ret; -} - -enum sq_ret { - SQT_IDLE = 1, - SQT_SPIN = 2, - SQT_DID_WORK = 4, -}; - -static enum sq_ret __io_sq_thread(struct io_ring_ctx *ctx, - unsigned long start_jiffies, bool cap_entries) -{ - unsigned long timeout = start_jiffies + ctx->sq_thread_idle; - struct io_sq_data *sqd = ctx->sq_data; unsigned int to_submit; int ret = 0; -again: - if (!list_empty(&ctx->iopoll_list)) { + to_submit = io_sqring_entries(ctx); + /* if we're handling multiple rings, cap submit size for fairness */ + if (cap_entries && to_submit > 8) + to_submit = 8; + + if (!list_empty(&ctx->iopoll_list) || to_submit) { unsigned nr_events = 0; mutex_lock(&ctx->uring_lock); - if (!list_empty(&ctx->iopoll_list) && !need_resched()) + if (!list_empty(&ctx->iopoll_list)) io_do_iopoll(ctx, &nr_events, 0); + + if (to_submit && likely(!percpu_ref_is_dying(&ctx->refs))) + ret = io_submit_sqes(ctx, to_submit); mutex_unlock(&ctx->uring_lock); } - to_submit = io_sqring_entries(ctx); - - /* - * If submit got -EBUSY, flag us as needing the application - * to enter the kernel to reap and flush events. - */ - if (!to_submit || ret == -EBUSY || need_resched()) { - /* - * Drop cur_mm before scheduling, we can't hold it for - * long periods (or over schedule()). Do this before - * adding ourselves to the waitqueue, as the unuse/drop - * may sleep. - */ - io_sq_thread_drop_mm(); - - /* - * We're polling. If we're within the defined idle - * period, then let us spin without work before going - * to sleep. The exception is if we got EBUSY doing - * more IO, we should wait for the application to - * reap events and wake us up. - */ - if (!list_empty(&ctx->iopoll_list) || need_resched() || - (!time_after(jiffies, timeout) && ret != -EBUSY && - !percpu_ref_is_dying(&ctx->refs))) - return SQT_SPIN; + if (!io_sqring_full(ctx) && wq_has_sleeper(&ctx->sqo_sq_wait)) + wake_up(&ctx->sqo_sq_wait); - prepare_to_wait(&sqd->wait, &ctx->sqo_wait_entry, - TASK_INTERRUPTIBLE); + return ret; +} - /* - * While doing polled IO, before going to sleep, we need - * to check if there are new reqs added to iopoll_list, - * it is because reqs may have been punted to io worker - * and will be added to iopoll_list later, hence check - * the iopoll_list again. - */ - if ((ctx->flags & IORING_SETUP_IOPOLL) && - !list_empty_careful(&ctx->iopoll_list)) { - finish_wait(&sqd->wait, &ctx->sqo_wait_entry); - goto again; - } +static void io_sqd_update_thread_idle(struct io_sq_data *sqd) +{ + struct io_ring_ctx *ctx; + unsigned sq_thread_idle = 0; - to_submit = io_sqring_entries(ctx); - if (!to_submit || ret == -EBUSY) - return SQT_IDLE; + list_for_each_entry(ctx, &sqd->ctx_list, sqd_list) { + if (sq_thread_idle < ctx->sq_thread_idle) + sq_thread_idle = ctx->sq_thread_idle; } - finish_wait(&sqd->wait, &ctx->sqo_wait_entry); - io_ring_clear_wakeup_flag(ctx); - - /* if we're handling multiple rings, cap submit size for fairness */ - if (cap_entries && to_submit > 8) - to_submit = 8; - - mutex_lock(&ctx->uring_lock); - if (likely(!percpu_ref_is_dying(&ctx->refs))) - ret = io_submit_sqes(ctx, to_submit); - mutex_unlock(&ctx->uring_lock); - - if (!io_sqring_full(ctx) && wq_has_sleeper(&ctx->sqo_sq_wait)) - wake_up(&ctx->sqo_sq_wait); - - return SQT_DID_WORK; + sqd->sq_thread_idle = sq_thread_idle; } static void io_sqd_init_new(struct io_sq_data *sqd) @@ -6737,39 +6959,56 @@ static void io_sqd_init_new(struct io_sq_data *sqd) while (!list_empty(&sqd->ctx_new_list)) { ctx = list_first_entry(&sqd->ctx_new_list, struct io_ring_ctx, sqd_list); - init_wait(&ctx->sqo_wait_entry); - ctx->sqo_wait_entry.func = io_sq_wake_function; list_move_tail(&ctx->sqd_list, &sqd->ctx_list); complete(&ctx->sq_thread_comp); } + + io_sqd_update_thread_idle(sqd); } static int io_sq_thread(void *data) { struct cgroup_subsys_state *cur_css = NULL; + struct files_struct *old_files = current->files; + struct nsproxy *old_nsproxy = current->nsproxy; const struct cred *old_cred = NULL; struct io_sq_data *sqd = data; struct io_ring_ctx *ctx; - unsigned long start_jiffies; + unsigned long timeout = 0; + DEFINE_WAIT(wait); + + task_lock(current); + current->files = NULL; + current->nsproxy = NULL; + task_unlock(current); - start_jiffies = jiffies; while (!kthread_should_stop()) { - enum sq_ret ret = 0; - bool cap_entries; + int ret; + bool cap_entries, sqt_spin, needs_sched; /* * Any changes to the sqd lists are synchronized through the * kthread parking. This synchronizes the thread vs users, * the users are synchronized on the sqd->ctx_lock. */ - if (kthread_should_park()) + if (kthread_should_park()) { kthread_parkme(); + /* + * When sq thread is unparked, in case the previous park operation + * comes from io_put_sq_data(), which means that sq thread is going + * to be stopped, so here needs to have a check. + */ + if (kthread_should_stop()) + break; + } - if (unlikely(!list_empty(&sqd->ctx_new_list))) + if (unlikely(!list_empty(&sqd->ctx_new_list))) { io_sqd_init_new(sqd); + timeout = jiffies + sqd->sq_thread_idle; + } + sqt_spin = false; cap_entries = !list_is_singular(&sqd->ctx_list); - list_for_each_entry(ctx, &sqd->ctx_list, sqd_list) { if (current->cred != ctx->creds) { if (old_cred) @@ -6782,24 +7021,49 @@ static int io_sq_thread(void *data) current->sessionid = ctx->sessionid; #endif - ret |= __io_sq_thread(ctx, start_jiffies, cap_entries); + ret = __io_sq_thread(ctx, cap_entries); + if (!sqt_spin && (ret > 0 || !list_empty(&ctx->iopoll_list))) + sqt_spin = true; - io_sq_thread_drop_mm(); + io_sq_thread_drop_mm_files(); } - if (ret & SQT_SPIN) { + if (sqt_spin || !time_after(jiffies, timeout)) { io_run_task_work(); cond_resched(); - } else if (ret == SQT_IDLE) { - if (kthread_should_park()) - continue; + if (sqt_spin) + timeout = jiffies + sqd->sq_thread_idle; + continue; + } + + if (kthread_should_park()) + continue; + + needs_sched = true; + prepare_to_wait(&sqd->wait, &wait, TASK_INTERRUPTIBLE); + list_for_each_entry(ctx, &sqd->ctx_list, sqd_list) { + if ((ctx->flags & IORING_SETUP_IOPOLL) && + !list_empty_careful(&ctx->iopoll_list)) { + needs_sched = false; + break; + } + if (io_sqring_entries(ctx)) { + needs_sched = false; + break; + } + } + + if (needs_sched) { list_for_each_entry(ctx, &sqd->ctx_list, sqd_list) io_ring_set_wakeup_flag(ctx); + schedule(); - start_jiffies = jiffies; list_for_each_entry(ctx, &sqd->ctx_list, sqd_list) io_ring_clear_wakeup_flag(ctx); } + + finish_wait(&sqd->wait, &wait); + timeout = jiffies + sqd->sq_thread_idle; } io_run_task_work(); @@ -6809,6 +7073,11 @@ static int io_sq_thread(void *data) if (old_cred) revert_creds(old_cred); + task_lock(current); + current->files = old_files; + current->nsproxy = old_nsproxy; + task_unlock(current); + kthread_parkme(); return 0; @@ -6863,7 +7132,8 @@ static int io_run_task_work_sig(void) * application must reap them itself, as they reside on the shared cq ring. */ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events, - const sigset_t __user *sig, size_t sigsz) + const sigset_t __user *sig, size_t sigsz, + struct __kernel_timespec __user *uts) { struct io_wait_queue iowq = { .wq = { @@ -6875,6 +7145,8 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events, .to_wait = min_events, }; struct io_rings *rings = ctx->rings; + struct timespec64 ts; + signed long timeout = 0; int ret = 0; do { @@ -6897,6 +7169,12 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events, return ret; } + if (uts) { + if (get_timespec64(&ts, uts)) + return -EFAULT; + timeout = timespec64_to_jiffies(&ts); + } + iowq.nr_timeouts = atomic_read(&ctx->cq_timeouts); trace_io_uring_cqring_wait(ctx, min_events); do { @@ -6910,7 +7188,15 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events, break; if (io_should_wake(&iowq, false)) break; - schedule(); + if (uts) { + timeout = schedule_timeout(timeout); + if (timeout == 0) { + ret = -ETIME; + break; + } + } else { + schedule(); + } } while (1); finish_wait(&ctx->wait, &iowq.wq); @@ -6959,9 +7245,9 @@ static int io_sqe_files_unregister(struct io_ring_ctx *ctx) if (!data) return -ENXIO; - spin_lock(&data->lock); + spin_lock_bh(&data->lock); ref_node = data->node; - spin_unlock(&data->lock); + spin_unlock_bh(&data->lock); if (ref_node) percpu_ref_kill(&ref_node->refs); @@ -7084,12 +7370,11 @@ static void io_sq_thread_stop(struct io_ring_ctx *ctx) mutex_lock(&sqd->ctx_lock); list_del(&ctx->sqd_list); + io_sqd_update_thread_idle(sqd); mutex_unlock(&sqd->ctx_lock); - if (sqd->thread) { - finish_wait(&sqd->wait, &ctx->sqo_wait_entry); + if (sqd->thread) io_sq_thread_unpark(sqd); - } io_put_sq_data(sqd); ctx->sq_data = NULL; @@ -7344,7 +7629,7 @@ static void io_file_data_ref_zero(struct percpu_ref *ref) data = ref_node->file_data; ctx = data->ctx; - spin_lock(&data->lock); + spin_lock_bh(&data->lock); ref_node->done = true; while (!list_empty(&data->ref_list)) { @@ -7356,7 +7641,7 @@ static void io_file_data_ref_zero(struct percpu_ref *ref) list_del(&ref_node->node); first_add |= llist_add(&ref_node->llist, &ctx->file_put_llist); } - spin_unlock(&data->lock); + spin_unlock_bh(&data->lock); if (percpu_ref_is_dying(&data->refs)) delay = 0; @@ -7479,9 +7764,9 @@ static int io_sqe_files_register(struct io_ring_ctx *ctx, void __user *arg, } file_data->node = ref_node; - spin_lock(&file_data->lock); + spin_lock_bh(&file_data->lock); list_add_tail(&ref_node->node, &file_data->ref_list); - spin_unlock(&file_data->lock); + spin_unlock_bh(&file_data->lock); percpu_ref_get(&file_data->refs); return ret; out_fput: @@ -7638,10 +7923,10 @@ static int __io_sqe_files_update(struct io_ring_ctx *ctx, if (needs_switch) { percpu_ref_kill(&data->node->refs); - spin_lock(&data->lock); + spin_lock_bh(&data->lock); list_add_tail(&ref_node->node, &data->ref_list); data->node = ref_node; - spin_unlock(&data->lock); + spin_unlock_bh(&data->lock); percpu_ref_get(&ctx->file_data->refs); } else destroy_fixed_file_ref_node(ref_node); @@ -7769,7 +8054,7 @@ static int io_sq_offload_create(struct io_ring_ctx *ctx, struct io_sq_data *sqd; ret = -EPERM; - if (!capable(CAP_SYS_ADMIN)) + if (!capable(CAP_SYS_ADMIN) && !capable(CAP_SYS_NICE)) goto err; sqd = io_get_sq_data(p); @@ -8355,8 +8640,6 @@ static void io_ring_exit_work(struct work_struct *work) * as nobody else will be looking for them. */ do { - if (ctx->rings) - io_cqring_overflow_flush(ctx, true, NULL, NULL); io_iopoll_try_reap_events(ctx); } while (!wait_for_completion_timeout(&ctx->ref_comp, HZ/20)); io_ring_ctx_free(ctx); @@ -8366,17 +8649,17 @@ static void io_ring_ctx_wait_and_kill(struct io_ring_ctx *ctx) { mutex_lock(&ctx->uring_lock); percpu_ref_kill(&ctx->refs); + if (ctx->rings) + io_cqring_overflow_flush(ctx, true, NULL, NULL); mutex_unlock(&ctx->uring_lock); - io_kill_timeouts(ctx, NULL); - io_poll_remove_all(ctx, NULL); + io_kill_timeouts(ctx, NULL, NULL); + io_poll_remove_all(ctx, NULL, NULL); if (ctx->io_wq) io_wq_cancel_all(ctx->io_wq); /* if we failed setting up the ctx, we might not have any rings */ - if (ctx->rings) - io_cqring_overflow_flush(ctx, true, NULL, NULL); io_iopoll_try_reap_events(ctx); idr_for_each(&ctx->personality_idr, io_remove_personalities, ctx); @@ -8407,120 +8690,31 @@ static int io_uring_release(struct inode *inode, struct file *file) return 0; } -static bool io_wq_files_match(struct io_wq_work *work, void *data) -{ - struct files_struct *files = data; - - return !files || ((work->flags & IO_WQ_WORK_FILES) && - work->identity->files == files); -} - -/* - * Returns true if 'preq' is the link parent of 'req' - */ -static bool io_match_link(struct io_kiocb *preq, struct io_kiocb *req) -{ - struct io_kiocb *link; - - if (!(preq->flags & REQ_F_LINK_HEAD)) - return false; - - list_for_each_entry(link, &preq->link_list, link_list) { - if (link == req) - return true; - } - - return false; -} - -/* - * We're looking to cancel 'req' because it's holding on to our files, but - * 'req' could be a link to another request. See if it is, and cancel that - * parent request if so. - */ -static bool io_poll_remove_link(struct io_ring_ctx *ctx, struct io_kiocb *req) -{ - struct hlist_node *tmp; - struct io_kiocb *preq; - bool found = false; - int i; - - spin_lock_irq(&ctx->completion_lock); - for (i = 0; i < (1U << ctx->cancel_hash_bits); i++) { - struct hlist_head *list; - - list = &ctx->cancel_hash[i]; - hlist_for_each_entry_safe(preq, tmp, list, hash_node) { - found = io_match_link(preq, req); - if (found) { - io_poll_remove_one(preq); - break; - } - } - } - spin_unlock_irq(&ctx->completion_lock); - return found; -} - -static bool io_timeout_remove_link(struct io_ring_ctx *ctx, - struct io_kiocb *req) -{ - struct io_kiocb *preq; - bool found = false; - - spin_lock_irq(&ctx->completion_lock); - list_for_each_entry(preq, &ctx->timeout_list, timeout.list) { - found = io_match_link(preq, req); - if (found) { - __io_timeout_cancel(preq); - break; - } - } - spin_unlock_irq(&ctx->completion_lock); - return found; -} +struct io_task_cancel { + struct task_struct *task; + struct files_struct *files; +}; -static bool io_cancel_link_cb(struct io_wq_work *work, void *data) +static bool io_cancel_task_cb(struct io_wq_work *work, void *data) { struct io_kiocb *req = container_of(work, struct io_kiocb, work); + struct io_task_cancel *cancel = data; bool ret; - if (req->flags & REQ_F_LINK_TIMEOUT) { + if (cancel->files && (req->flags & REQ_F_LINK_TIMEOUT)) { unsigned long flags; struct io_ring_ctx *ctx = req->ctx; /* protect against races with linked timeouts */ spin_lock_irqsave(&ctx->completion_lock, flags); - ret = io_match_link(req, data); + ret = io_match_task(req, cancel->task, cancel->files); spin_unlock_irqrestore(&ctx->completion_lock, flags); } else { - ret = io_match_link(req, data); + ret = io_match_task(req, cancel->task, cancel->files); } return ret; } -static void io_attempt_cancel(struct io_ring_ctx *ctx, struct io_kiocb *req) -{ - enum io_wq_cancel cret; - - /* cancel this particular work, if it's running */ - cret = io_wq_cancel_work(ctx->io_wq, &req->work); - if (cret != IO_WQ_CANCEL_NOTFOUND) - return; - - /* find links that hold this pending, cancel those */ - cret = io_wq_cancel_cb(ctx->io_wq, io_cancel_link_cb, req, true); - if (cret != IO_WQ_CANCEL_NOTFOUND) - return; - - /* if we have a poll link holding this pending, cancel that */ - if (io_poll_remove_link(ctx, req)) - return; - - /* final option, timeout link is holding this req pending */ - io_timeout_remove_link(ctx, req); -} - static void io_cancel_defer_files(struct io_ring_ctx *ctx, struct task_struct *task, struct files_struct *files) @@ -8530,8 +8724,7 @@ static void io_cancel_defer_files(struct io_ring_ctx *ctx, spin_lock_irq(&ctx->completion_lock); list_for_each_entry_reverse(de, &ctx->defer_list, list) { - if (io_task_match(de->req, task) && - io_match_files(de->req, files)) { + if (io_match_task(de->req, task, files)) { list_cut_position(&list, &ctx->defer_list, &de->list); break; } @@ -8548,72 +8741,52 @@ static void io_cancel_defer_files(struct io_ring_ctx *ctx, } } -/* - * Returns true if we found and killed one or more files pinning requests - */ -static bool io_uring_cancel_files(struct io_ring_ctx *ctx, +static void io_uring_cancel_files(struct io_ring_ctx *ctx, + struct task_struct *task, struct files_struct *files) { - if (list_empty_careful(&ctx->inflight_list)) - return false; - - /* cancel all at once, should be faster than doing it one by one*/ - io_wq_cancel_cb(ctx->io_wq, io_wq_files_match, files, true); - while (!list_empty_careful(&ctx->inflight_list)) { - struct io_kiocb *cancel_req = NULL, *req; + struct io_task_cancel cancel = { .task = task, .files = files }; + struct io_kiocb *req; DEFINE_WAIT(wait); + bool found = false; spin_lock_irq(&ctx->inflight_lock); list_for_each_entry(req, &ctx->inflight_list, inflight_entry) { - if (files && (req->work.flags & IO_WQ_WORK_FILES) && + if (req->task != task || req->work.identity->files != files) continue; - /* req is being completed, ignore */ - if (!refcount_inc_not_zero(&req->refs)) - continue; - cancel_req = req; + found = true; break; } - if (cancel_req) - prepare_to_wait(&ctx->inflight_wait, &wait, - TASK_UNINTERRUPTIBLE); + if (found) + prepare_to_wait(&task->io_uring->wait, &wait, + TASK_UNINTERRUPTIBLE); spin_unlock_irq(&ctx->inflight_lock); /* We need to keep going until we don't find a matching req */ - if (!cancel_req) + if (!found) break; - /* cancel this request, or head link requests */ - io_attempt_cancel(ctx, cancel_req); - io_put_req(cancel_req); + + io_wq_cancel_cb(ctx->io_wq, io_cancel_task_cb, &cancel, true); + io_poll_remove_all(ctx, task, files); + io_kill_timeouts(ctx, task, files); /* cancellations _may_ trigger task work */ io_run_task_work(); schedule(); - finish_wait(&ctx->inflight_wait, &wait); + finish_wait(&task->io_uring->wait, &wait); } - - return true; } -static bool io_cancel_task_cb(struct io_wq_work *work, void *data) +static void __io_uring_cancel_task_requests(struct io_ring_ctx *ctx, + struct task_struct *task) { - struct io_kiocb *req = container_of(work, struct io_kiocb, work); - struct task_struct *task = data; - - return io_task_match(req, task); -} - -static bool __io_uring_cancel_task_requests(struct io_ring_ctx *ctx, - struct task_struct *task, - struct files_struct *files) -{ - bool ret; - - ret = io_uring_cancel_files(ctx, files); - if (!files) { + while (1) { + struct io_task_cancel cancel = { .task = task, .files = NULL, }; enum io_wq_cancel cret; + bool ret = false; - cret = io_wq_cancel_cb(ctx->io_wq, io_cancel_task_cb, task, true); + cret = io_wq_cancel_cb(ctx->io_wq, io_cancel_task_cb, &cancel, true); if (cret != IO_WQ_CANCEL_NOTFOUND) ret = true; @@ -8625,11 +8798,13 @@ static bool __io_uring_cancel_task_requests(struct io_ring_ctx *ctx, } } - ret |= io_poll_remove_all(ctx, task); - ret |= io_kill_timeouts(ctx, task); + ret |= io_poll_remove_all(ctx, task, NULL); + ret |= io_kill_timeouts(ctx, task, NULL); + if (!ret) + break; + io_run_task_work(); + cond_resched(); } - - return ret; } /* @@ -8648,17 +8823,15 @@ static void io_uring_cancel_task_requests(struct io_ring_ctx *ctx, io_sq_thread_park(ctx->sq_data); } - if (files) - io_cancel_defer_files(ctx, NULL, files); - else - io_cancel_defer_files(ctx, task, NULL); - + io_cancel_defer_files(ctx, task, files); + io_ring_submit_lock(ctx, (ctx->flags & IORING_SETUP_IOPOLL)); io_cqring_overflow_flush(ctx, true, task, files); + io_ring_submit_unlock(ctx, (ctx->flags & IORING_SETUP_IOPOLL)); - while (__io_uring_cancel_task_requests(ctx, task, files)) { - io_run_task_work(); - cond_resched(); - } + if (!files) + __io_uring_cancel_task_requests(ctx, task); + else + io_uring_cancel_files(ctx, task, files); if ((ctx->flags & IORING_SETUP_SQPOLL) && ctx->sq_data) { atomic_dec(&task->io_uring->in_idle); @@ -8916,9 +9089,39 @@ static void io_sqpoll_wait_sq(struct io_ring_ctx *ctx) finish_wait(&ctx->sqo_sq_wait, &wait); } +static int io_get_ext_arg(unsigned flags, const void __user *argp, size_t *argsz, + struct __kernel_timespec __user **ts, + const sigset_t __user **sig) +{ + struct io_uring_getevents_arg arg; + + /* + * If EXT_ARG isn't set, then we have no timespec and the argp pointer + * is just a pointer to the sigset_t. + */ + if (!(flags & IORING_ENTER_EXT_ARG)) { + *sig = (const sigset_t __user *) argp; + *ts = NULL; + return 0; + } + + /* + * EXT_ARG is set - ensure we agree on the size of it and copy in our + * timespec and sigset_t pointers if good. + */ + if (*argsz != sizeof(arg)) + return -EINVAL; + if (copy_from_user(&arg, argp, sizeof(arg))) + return -EFAULT; + *sig = u64_to_user_ptr(arg.sigmask); + *argsz = arg.sigmask_sz; + *ts = u64_to_user_ptr(arg.ts); + return 0; +} + SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit, - u32, min_complete, u32, flags, const sigset_t __user *, sig, - size_t, sigsz) + u32, min_complete, u32, flags, const void __user *, argp, + size_t, argsz) { struct io_ring_ctx *ctx; long ret = -EBADF; @@ -8928,7 +9131,7 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit, io_run_task_work(); if (flags & ~(IORING_ENTER_GETEVENTS | IORING_ENTER_SQ_WAKEUP | - IORING_ENTER_SQ_WAIT)) + IORING_ENTER_SQ_WAIT | IORING_ENTER_EXT_ARG)) return -EINVAL; f = fdget(fd); @@ -8955,8 +9158,10 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit, */ ret = 0; if (ctx->flags & IORING_SETUP_SQPOLL) { + io_ring_submit_lock(ctx, (ctx->flags & IORING_SETUP_IOPOLL)); if (!list_empty_careful(&ctx->cq_overflow_list)) io_cqring_overflow_flush(ctx, false, NULL, NULL); + io_ring_submit_unlock(ctx, (ctx->flags & IORING_SETUP_IOPOLL)); if (flags & IORING_ENTER_SQ_WAKEUP) wake_up(&ctx->sq_data->wait); if (flags & IORING_ENTER_SQ_WAIT) @@ -8974,6 +9179,13 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit, goto out; } if (flags & IORING_ENTER_GETEVENTS) { + const sigset_t __user *sig; + struct __kernel_timespec __user *ts; + + ret = io_get_ext_arg(flags, argp, &argsz, &ts, &sig); + if (unlikely(ret)) + goto out; + min_complete = min(min_complete, ctx->cq_entries); /* @@ -8986,7 +9198,7 @@ SYSCALL_DEFINE6(io_uring_enter, unsigned int, fd, u32, to_submit, !(ctx->flags & IORING_SETUP_SQPOLL)) { ret = io_iopoll_check(ctx, min_complete); } else { - ret = io_cqring_wait(ctx, min_complete, sig, sigsz); + ret = io_cqring_wait(ctx, min_complete, sig, argsz, ts); } } @@ -9354,7 +9566,8 @@ static int io_uring_create(unsigned entries, struct io_uring_params *p, p->features = IORING_FEAT_SINGLE_MMAP | IORING_FEAT_NODROP | IORING_FEAT_SUBMIT_STABLE | IORING_FEAT_RW_CUR_POS | IORING_FEAT_CUR_PERSONALITY | IORING_FEAT_FAST_POLL | - IORING_FEAT_POLL_32BITS; + IORING_FEAT_POLL_32BITS | IORING_FEAT_SQPOLL_NONFIXED | + IORING_FEAT_EXT_ARG; if (copy_to_user(params, p, sizeof(*p))) { ret = -EFAULT; diff --git a/fs/namei.c b/fs/namei.c index d4a6dd772303..03d0e11e4f36 100644 --- a/fs/namei.c +++ b/fs/namei.c @@ -4346,8 +4346,8 @@ out: } EXPORT_SYMBOL(vfs_rename); -static int do_renameat2(int olddfd, const char __user *oldname, int newdfd, - const char __user *newname, unsigned int flags) +int do_renameat2(int olddfd, struct filename *from, int newdfd, + struct filename *to, unsigned int flags) { struct dentry *old_dentry, *new_dentry; struct dentry *trap; @@ -4355,32 +4355,30 @@ static int do_renameat2(int olddfd, const char __user *oldname, int newdfd, struct qstr old_last, new_last; int old_type, new_type; struct inode *delegated_inode = NULL; - struct filename *from; - struct filename *to; unsigned int lookup_flags = 0, target_flags = LOOKUP_RENAME_TARGET; bool should_retry = false; - int error; + int error = -EINVAL; if (flags & ~(RENAME_NOREPLACE | RENAME_EXCHANGE | RENAME_WHITEOUT)) - return -EINVAL; + goto put_both; if ((flags & (RENAME_NOREPLACE | RENAME_WHITEOUT)) && (flags & RENAME_EXCHANGE)) - return -EINVAL; + goto put_both; if (flags & RENAME_EXCHANGE) target_flags = 0; retry: - from = filename_parentat(olddfd, getname(oldname), lookup_flags, - &old_path, &old_last, &old_type); + from = filename_parentat(olddfd, from, lookup_flags, &old_path, + &old_last, &old_type); if (IS_ERR(from)) { error = PTR_ERR(from); - goto exit; + goto put_new; } - to = filename_parentat(newdfd, getname(newname), lookup_flags, - &new_path, &new_last, &new_type); + to = filename_parentat(newdfd, to, lookup_flags, &new_path, &new_last, + &new_type); if (IS_ERR(to)) { error = PTR_ERR(to); goto exit1; @@ -4473,34 +4471,40 @@ exit2: if (retry_estale(error, lookup_flags)) should_retry = true; path_put(&new_path); - putname(to); exit1: path_put(&old_path); - putname(from); if (should_retry) { should_retry = false; lookup_flags |= LOOKUP_REVAL; goto retry; } -exit: +put_both: + if (!IS_ERR(from)) + putname(from); +put_new: + if (!IS_ERR(to)) + putname(to); return error; } SYSCALL_DEFINE5(renameat2, int, olddfd, const char __user *, oldname, int, newdfd, const char __user *, newname, unsigned int, flags) { - return do_renameat2(olddfd, oldname, newdfd, newname, flags); + return do_renameat2(olddfd, getname(oldname), newdfd, getname(newname), + flags); } SYSCALL_DEFINE4(renameat, int, olddfd, const char __user *, oldname, int, newdfd, const char __user *, newname) { - return do_renameat2(olddfd, oldname, newdfd, newname, 0); + return do_renameat2(olddfd, getname(oldname), newdfd, getname(newname), + 0); } SYSCALL_DEFINE2(rename, const char __user *, oldname, const char __user *, newname) { - return do_renameat2(AT_FDCWD, oldname, AT_FDCWD, newname, 0); + return do_renameat2(AT_FDCWD, getname(oldname), AT_FDCWD, + getname(newname), 0); } int readlink_copy(char __user *buffer, int buflen, const char *link) diff --git a/include/linux/socket.h b/include/linux/socket.h index e9cb30d8cbfb..385894b4a8bb 100644 --- a/include/linux/socket.h +++ b/include/linux/socket.h @@ -436,6 +436,7 @@ extern int __sys_getpeername(int fd, struct sockaddr __user *usockaddr, int __user *usockaddr_len); extern int __sys_socketpair(int family, int type, int protocol, int __user *usockvec); +extern int __sys_shutdown_sock(struct socket *sock, int how); extern int __sys_shutdown(int fd, int how); extern struct ns_common *get_net_ns(struct ns_common *ns); diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h index 17ac91876544..df0c3c74609e 100644 --- a/include/linux/syscalls.h +++ b/include/linux/syscalls.h @@ -317,7 +317,7 @@ asmlinkage long sys_io_uring_setup(u32 entries, struct io_uring_params __user *p); asmlinkage long sys_io_uring_enter(unsigned int fd, u32 to_submit, u32 min_complete, u32 flags, - const sigset_t __user *sig, size_t sigsz); + const void __user *argp, size_t argsz); asmlinkage long sys_io_uring_register(unsigned int fd, unsigned int op, void __user *arg, unsigned int nr_args); diff --git a/include/uapi/linux/io_uring.h b/include/uapi/linux/io_uring.h index 98d8e06dea22..d31a2a1e8ef9 100644 --- a/include/uapi/linux/io_uring.h +++ b/include/uapi/linux/io_uring.h @@ -42,6 +42,8 @@ struct io_uring_sqe { __u32 statx_flags; __u32 fadvise_advice; __u32 splice_flags; + __u32 rename_flags; + __u32 unlink_flags; }; __u64 user_data; /* data to be passed back at completion time */ union { @@ -132,6 +134,9 @@ enum { IORING_OP_PROVIDE_BUFFERS, IORING_OP_REMOVE_BUFFERS, IORING_OP_TEE, + IORING_OP_SHUTDOWN, + IORING_OP_RENAMEAT, + IORING_OP_UNLINKAT, /* this goes last, obviously */ IORING_OP_LAST, @@ -146,6 +151,7 @@ enum { * sqe->timeout_flags */ #define IORING_TIMEOUT_ABS (1U << 0) +#define IORING_TIMEOUT_UPDATE (1U << 1) /* * sqe->splice_flags @@ -226,6 +232,7 @@ struct io_cqring_offsets { #define IORING_ENTER_GETEVENTS (1U << 0) #define IORING_ENTER_SQ_WAKEUP (1U << 1) #define IORING_ENTER_SQ_WAIT (1U << 2) +#define IORING_ENTER_EXT_ARG (1U << 3) /* * Passed in for io_uring_setup(2). Copied back with updated info on success @@ -253,6 +260,8 @@ struct io_uring_params { #define IORING_FEAT_CUR_PERSONALITY (1U << 4) #define IORING_FEAT_FAST_POLL (1U << 5) #define IORING_FEAT_POLL_32BITS (1U << 6) +#define IORING_FEAT_SQPOLL_NONFIXED (1U << 7) +#define IORING_FEAT_EXT_ARG (1U << 8) /* * io_uring_register(2) opcodes and arguments @@ -329,4 +338,11 @@ enum { IORING_RESTRICTION_LAST }; +struct io_uring_getevents_arg { + __u64 sigmask; + __u32 sigmask_sz; + __u32 pad; + __u64 ts; +}; + #endif diff --git a/net/socket.c b/net/socket.c index 9a240b45bdf3..33e8b6c4e1d3 100644 --- a/net/socket.c +++ b/net/socket.c @@ -2175,6 +2175,17 @@ SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, * Shutdown a socket. */ +int __sys_shutdown_sock(struct socket *sock, int how) +{ + int err; + + err = security_socket_shutdown(sock, how); + if (!err) + err = sock->ops->shutdown(sock, how); + + return err; +} + int __sys_shutdown(int fd, int how) { int err, fput_needed; @@ -2182,9 +2193,7 @@ int __sys_shutdown(int fd, int how) sock = sockfd_lookup_light(fd, &err, &fput_needed); if (sock != NULL) { - err = security_socket_shutdown(sock, how); - if (!err) - err = sock->ops->shutdown(sock, how); + err = __sys_shutdown_sock(sock, how); fput_light(sock->file, fput_needed); } return err; |