unsigned int drain_next: 1;
unsigned int eventfd_async: 1;
unsigned int restricted: 1;
- unsigned int sqo_dead: 1;
unsigned int sqo_exec: 1;
/*
struct io_rings *rings;
- /*
- * For SQPOLL usage
- */
- struct task_struct *sqo_task;
-
/* Only used for accounting purposes */
struct mm_struct *mm_account;
REQ_F_POLLED_BIT,
REQ_F_BUFFER_SELECTED_BIT,
REQ_F_NO_FILE_TABLE_BIT,
- REQ_F_WORK_INITIALIZED_BIT,
REQ_F_LTIMEOUT_ACTIVE_BIT,
REQ_F_COMPLETE_INLINE_BIT,
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),
- /* io_wq_work is initialized */
- REQ_F_WORK_INITIALIZED = BIT(REQ_F_WORK_INITIALIZED_BIT),
/* linked timeout is active, i.e. prepared by link's head */
REQ_F_LTIMEOUT_ACTIVE = BIT(REQ_F_LTIMEOUT_ACTIVE_BIT),
/* completion is deferred through io_comp_state */
req->flags |= REQ_F_FAIL_LINK;
}
-static inline void __io_req_init_async(struct io_kiocb *req)
-{
- memset(&req->work, 0, sizeof(req->work));
- req->flags |= REQ_F_WORK_INITIALIZED;
-}
-
-/*
- * Note: must call io_req_init_async() for the first time you
- * touch any members of io_wq_work.
- */
-static inline void io_req_init_async(struct io_kiocb *req)
-{
- if (req->flags & REQ_F_WORK_INITIALIZED)
- return;
-
- __io_req_init_async(req);
-}
-
static void io_ring_ctx_ref_free(struct percpu_ref *ref)
{
struct io_ring_ctx *ctx = container_of(ref, struct io_ring_ctx, refs);
static void io_req_clean_work(struct io_kiocb *req)
{
- if (!(req->flags & REQ_F_WORK_INITIALIZED))
- return;
-
- if (req->work.creds) {
- put_cred(req->work.creds);
- req->work.creds = NULL;
- }
if (req->flags & REQ_F_INFLIGHT) {
struct io_ring_ctx *ctx = req->ctx;
struct io_uring_task *tctx = req->task->io_uring;
if (atomic_read(&tctx->in_idle))
wake_up(&tctx->wait);
}
-
- req->flags &= ~REQ_F_WORK_INITIALIZED;
}
static void io_req_track_inflight(struct io_kiocb *req)
struct io_ring_ctx *ctx = req->ctx;
if (!(req->flags & REQ_F_INFLIGHT)) {
- io_req_init_async(req);
req->flags |= REQ_F_INFLIGHT;
spin_lock_irq(&ctx->inflight_lock);
const struct io_op_def *def = &io_op_defs[req->opcode];
struct io_ring_ctx *ctx = req->ctx;
- io_req_init_async(req);
-
if (req->flags & REQ_F_FORCE_ASYNC)
req->work.flags |= IO_WQ_WORK_CONCURRENT;
if (def->unbound_nonreg_file)
req->work.flags |= IO_WQ_WORK_UNBOUND;
}
- if (!req->work.creds)
- req->work.creds = get_current_cred();
}
static void io_prep_async_link(struct io_kiocb *req)
return __io_req_find_next(req);
}
+static void ctx_flush_and_put(struct io_ring_ctx *ctx)
+{
+ if (!ctx)
+ return;
+ if (ctx->submit_state.comp.nr) {
+ mutex_lock(&ctx->uring_lock);
+ io_submit_flush_completions(&ctx->submit_state.comp, ctx);
+ mutex_unlock(&ctx->uring_lock);
+ }
+ percpu_ref_put(&ctx->refs);
+}
+
static bool __tctx_task_work(struct io_uring_task *tctx)
{
struct io_ring_ctx *ctx = NULL;
node = list.first;
while (node) {
struct io_wq_work_node *next = node->next;
- struct io_ring_ctx *this_ctx;
struct io_kiocb *req;
req = container_of(node, struct io_kiocb, io_task_work.node);
- this_ctx = req->ctx;
- req->task_work.func(&req->task_work);
- node = next;
-
- if (!ctx) {
- ctx = this_ctx;
- } else if (ctx != this_ctx) {
- mutex_lock(&ctx->uring_lock);
- io_submit_flush_completions(&ctx->submit_state.comp, ctx);
- mutex_unlock(&ctx->uring_lock);
- ctx = this_ctx;
+ if (req->ctx != ctx) {
+ ctx_flush_and_put(ctx);
+ ctx = req->ctx;
+ percpu_ref_get(&ctx->refs);
}
- }
- if (ctx && ctx->submit_state.comp.nr) {
- mutex_lock(&ctx->uring_lock);
- io_submit_flush_completions(&ctx->submit_state.comp, ctx);
- mutex_unlock(&ctx->uring_lock);
+ req->task_work.func(&req->task_work);
+ node = next;
}
+ ctx_flush_and_put(ctx);
return list.first != NULL;
}
/* ctx stays valid until unlock, even if we drop all ours ctx->refs */
mutex_lock(&ctx->uring_lock);
- if (!ctx->sqo_dead && !(current->flags & PF_EXITING) && !current->in_execve)
+ if (!(current->flags & PF_EXITING) && !current->in_execve)
__io_queue_sqe(req);
else
__io_req_task_cancel(req, -EFAULT);
return false;
return !io_setup_async_rw(req, iovec, inline_vecs, &iter, false);
}
-#endif
-static bool io_rw_reissue(struct io_kiocb *req)
+static bool io_rw_should_reissue(struct io_kiocb *req)
{
-#ifdef CONFIG_BLOCK
umode_t mode = file_inode(req->file)->i_mode;
+ struct io_ring_ctx *ctx = req->ctx;
if (!S_ISBLK(mode) && !S_ISREG(mode))
return false;
- if ((req->flags & REQ_F_NOWAIT) || io_wq_current_is_worker())
+ if ((req->flags & REQ_F_NOWAIT) || (io_wq_current_is_worker() &&
+ !(ctx->flags & IORING_SETUP_IOPOLL)))
return false;
/*
* If ref is dying, we might be running poll reap from the exit work.
* Don't attempt to reissue from that path, just let it fail with
* -EAGAIN.
*/
- if (percpu_ref_is_dying(&req->ctx->refs))
+ if (percpu_ref_is_dying(&ctx->refs))
+ return false;
+ return true;
+}
+#endif
+
+static bool io_rw_reissue(struct io_kiocb *req)
+{
+#ifdef CONFIG_BLOCK
+ if (!io_rw_should_reissue(req))
return false;
lockdep_assert_held(&req->ctx->uring_lock);
{
struct io_kiocb *req = container_of(kiocb, struct io_kiocb, rw.kiocb);
+#ifdef CONFIG_BLOCK
+ /* Rewind iter, if we have one. iopoll path resubmits as usual */
+ if (res == -EAGAIN && io_rw_should_reissue(req)) {
+ struct io_async_rw *rw = req->async_data;
+
+ if (rw)
+ iov_iter_revert(&rw->iter,
+ req->result - iov_iter_count(&rw->iter));
+ else if (!io_resubmit_prep(req))
+ res = -EIO;
+ }
+#endif
+
if (kiocb->ki_flags & IOCB_WRITE)
kiocb_end_write(req);
ret = io_iter_do_read(req, iter);
if (ret == -EIOCBQUEUED) {
+ if (req->async_data)
+ iov_iter_revert(iter, io_size - iov_iter_count(iter));
goto out_free;
} else if (ret == -EAGAIN) {
/* IOPOLL retry should happen for io-wq threads */
/* no retry on NONBLOCK nor RWF_NOWAIT */
if (ret2 == -EAGAIN && (req->flags & REQ_F_NOWAIT))
goto done;
+ if (ret2 == -EIOCBQUEUED && req->async_data)
+ iov_iter_revert(iter, io_size - iov_iter_count(iter));
if (!force_nonblock || ret2 != -EAGAIN) {
/* IOPOLL retry should happen for io-wq threads */
if ((req->ctx->flags & IORING_SETUP_IOPOLL) && ret2 == -EAGAIN)
* Splice operation will be punted aync, and here need to
* modify io_wq_work.flags, so initialize io_wq_work firstly.
*/
- io_req_init_async(req);
req->work.flags |= IO_WQ_WORK_UNBOUND;
}
pt->error = -EINVAL;
return;
}
+ /* double add on the same waitqueue head, ignore */
+ if (poll->head == head)
+ return;
poll = kmalloc(sizeof(*poll), GFP_ATOMIC);
if (!poll) {
pt->error = -ENOMEM;
static int io_issue_sqe(struct io_kiocb *req, unsigned int issue_flags)
{
struct io_ring_ctx *ctx = req->ctx;
+ const struct cred *creds = NULL;
int ret;
+ if (req->work.personality) {
+ const struct cred *new_creds;
+
+ if (!(issue_flags & IO_URING_F_NONBLOCK))
+ mutex_lock(&ctx->uring_lock);
+ new_creds = idr_find(&ctx->personality_idr, req->work.personality);
+ if (!(issue_flags & IO_URING_F_NONBLOCK))
+ mutex_unlock(&ctx->uring_lock);
+ if (!new_creds)
+ return -EINVAL;
+ creds = override_creds(new_creds);
+ }
+
switch (req->opcode) {
case IORING_OP_NOP:
ret = io_nop(req, issue_flags);
break;
}
+ if (creds)
+ revert_creds(creds);
+
if (ret)
return ret;
static void __io_queue_sqe(struct io_kiocb *req)
{
struct io_kiocb *linked_timeout = io_prep_linked_timeout(req);
- const struct cred *old_creds = NULL;
int ret;
- if ((req->flags & REQ_F_WORK_INITIALIZED) && req->work.creds &&
- req->work.creds != current_cred())
- old_creds = override_creds(req->work.creds);
-
ret = io_issue_sqe(req, IO_URING_F_NONBLOCK|IO_URING_F_COMPLETE_DEFER);
- if (old_creds)
- revert_creds(old_creds);
-
/*
* We async punt it if the file wasn't marked NOWAIT, or if the file
* doesn't support non-blocking read/write attempts
{
struct io_submit_state *state;
unsigned int sqe_flags;
- int id, ret = 0;
+ int ret = 0;
req->opcode = READ_ONCE(sqe->opcode);
/* same numerical values with corresponding REQ_F_*, safe to copy */
!io_op_defs[req->opcode].buffer_select)
return -EOPNOTSUPP;
- id = READ_ONCE(sqe->personality);
- if (id) {
- __io_req_init_async(req);
- req->work.creds = idr_find(&ctx->personality_idr, id);
- if (unlikely(!req->work.creds))
- return -EINVAL;
- get_cred(req->work.creds);
- }
-
+ req->work.list.next = NULL;
+ req->work.flags = 0;
+ req->work.personality = READ_ONCE(sqe->personality);
state = &ctx->submit_state;
/*
if (!list_empty(&ctx->iopoll_list))
io_do_iopoll(ctx, &nr_events, 0);
- if (to_submit && !ctx->sqo_dead &&
- likely(!percpu_ref_is_dying(&ctx->refs)))
+ if (to_submit && likely(!percpu_ref_is_dying(&ctx->refs)))
ret = io_submit_sqes(ctx, to_submit);
mutex_unlock(&ctx->uring_lock);
}
clear_bit(IO_SQ_THREAD_SHOULD_STOP, &sqd->state);
reinit_completion(&sqd->completion);
- ctx->sqo_dead = ctx->sqo_exec = 0;
+ ctx->sqo_exec = 0;
sqd->task_pid = current->pid;
current->flags |= PF_IO_WORKER;
ret = io_wq_fork_thread(io_sq_thread, sqd);
{
struct io_sq_data *sqd = ctx->sq_data;
+ ctx->flags &= ~IORING_SETUP_R_DISABLED;
if (ctx->flags & IORING_SETUP_SQPOLL)
complete(&sqd->startup);
}
static bool io_run_ctx_fallback(struct io_ring_ctx *ctx)
{
- struct callback_head *work, *head, *next;
+ struct callback_head *work, *next;
bool executed = false;
do {
- do {
- head = NULL;
- work = READ_ONCE(ctx->exit_task_work);
- } while (cmpxchg(&ctx->exit_task_work, work, head) != work);
-
+ work = xchg(&ctx->exit_task_work, NULL);
if (!work)
break;
*/
do {
io_uring_try_cancel_requests(ctx, NULL, NULL);
- io_run_ctx_fallback(ctx);
} while (!wait_for_completion_timeout(&ctx->ref_comp, HZ/20));
io_ring_ctx_free(ctx);
}
{
mutex_lock(&ctx->uring_lock);
percpu_ref_kill(&ctx->refs);
-
- if (WARN_ON_ONCE((ctx->flags & IORING_SETUP_SQPOLL) && !ctx->sqo_dead))
- ctx->sqo_dead = 1;
-
/* if force is set, the ring is going away. always drop after that */
ctx->cq_overflow_flushed = 1;
if (ctx->rings)
}
}
-static void io_disable_sqo_submit(struct io_ring_ctx *ctx)
-{
- mutex_lock(&ctx->uring_lock);
- ctx->sqo_dead = 1;
- mutex_unlock(&ctx->uring_lock);
-
- /* make sure callers enter the ring to get error */
- if (ctx->rings)
- io_ring_set_wakeup_flag(ctx);
-}
-
/*
* We need to iteratively cancel requests, in case a request has dependent
* hard links. These persist even for failure of cancelations, hence keep
bool did_park = false;
if ((ctx->flags & IORING_SETUP_SQPOLL) && ctx->sq_data) {
- io_disable_sqo_submit(ctx);
+ /* never started, nothing to cancel */
+ if (ctx->flags & IORING_SETUP_R_DISABLED) {
+ io_sq_offload_start(ctx);
+ return;
+ }
did_park = io_sq_thread_park(ctx->sq_data);
if (did_park) {
task = ctx->sq_data->thread;
fput(file);
return ret;
}
-
- /* one and only SQPOLL file note, held by sqo_task */
- WARN_ON_ONCE((ctx->flags & IORING_SETUP_SQPOLL) &&
- current != ctx->sqo_task);
}
tctx->last = file;
}
if (!sqd)
return;
- io_disable_sqo_submit(ctx);
if (!io_sq_thread_park(sqd))
return;
tctx = ctx->sq_data->thread->io_uring;
/* make sure overflow events are dropped */
atomic_inc(&tctx->in_idle);
- /* trigger io_disable_sqo_submit() */
if (tctx->sqpoll) {
struct file *file;
unsigned long index;
__io_uring_free(current);
}
-static int io_uring_flush(struct file *file, void *data)
-{
- struct io_uring_task *tctx = current->io_uring;
- struct io_ring_ctx *ctx = file->private_data;
-
- /* Ignore helper thread files exit */
- if (current->flags & PF_IO_WORKER)
- return 0;
-
- if (fatal_signal_pending(current) || (current->flags & PF_EXITING)) {
- io_uring_cancel_task_requests(ctx, NULL);
- io_req_caches_free(ctx);
- }
-
- io_run_ctx_fallback(ctx);
-
- if (!tctx)
- return 0;
-
- /* we should have cancelled and erased it before PF_EXITING */
- WARN_ON_ONCE((current->flags & PF_EXITING) &&
- xa_load(&tctx->xa, (unsigned long)file));
-
- /*
- * fput() is pending, will be 2 if the only other ref is our potential
- * task file note. If the task is exiting, drop regardless of count.
- */
- if (atomic_long_read(&file->f_count) != 2)
- return 0;
-
- if (ctx->flags & IORING_SETUP_SQPOLL) {
- /* there is only one file note, which is owned by sqo_task */
- WARN_ON_ONCE(ctx->sqo_task != current &&
- xa_load(&tctx->xa, (unsigned long)file));
- /* sqo_dead check is for when this happens after cancellation */
- WARN_ON_ONCE(ctx->sqo_task == current && !ctx->sqo_dead &&
- !xa_load(&tctx->xa, (unsigned long)file));
-
- io_disable_sqo_submit(ctx);
- }
-
- if (!(ctx->flags & IORING_SETUP_SQPOLL) || ctx->sqo_task == current)
- io_uring_del_task_file(file);
- return 0;
-}
-
static void *io_uring_validate_mmap_request(struct file *file,
loff_t pgoff, size_t sz)
{
do {
if (!io_sqring_full(ctx))
break;
-
prepare_to_wait(&ctx->sqo_sq_wait, &wait, TASK_INTERRUPTIBLE);
- if (unlikely(ctx->sqo_dead)) {
- ret = -EOWNERDEAD;
- goto out;
- }
-
if (!io_sqring_full(ctx))
break;
-
schedule();
} while (!signal_pending(current));
finish_wait(&ctx->sqo_sq_wait, &wait);
-out:
return ret;
}
ctx->sqo_exec = 0;
}
ret = -EOWNERDEAD;
- if (unlikely(ctx->sqo_dead))
- goto out;
if (flags & IORING_ENTER_SQ_WAKEUP)
wake_up(&ctx->sq_data->wait);
if (flags & IORING_ENTER_SQ_WAIT) {
static const struct file_operations io_uring_fops = {
.release = io_uring_release,
- .flush = io_uring_flush,
.mmap = io_uring_mmap,
#ifndef CONFIG_MMU
.get_unmapped_area = io_uring_nommu_get_unmapped_area,
ctx->compat = in_compat_syscall();
if (!capable(CAP_IPC_LOCK))
ctx->user = get_uid(current_user());
- ctx->sqo_task = current;
/*
* This is just grabbed for accounting purposes. When a process exits,
*/
ret = io_uring_install_fd(ctx, file);
if (ret < 0) {
- io_disable_sqo_submit(ctx);
/* fput will clean it up */
fput(file);
return ret;
trace_io_uring_create(ret, ctx, p->sq_entries, p->cq_entries, p->flags);
return ret;
err:
- io_disable_sqo_submit(ctx);
io_ring_ctx_wait_and_kill(ctx);
return ret;
}
if (ctx->restrictions.registered)
ctx->restricted = 1;
- ctx->flags &= ~IORING_SETUP_R_DISABLED;
-
io_sq_offload_start(ctx);
-
return 0;
}