ACCT_PINNED,
};
+static void __io_uring_cancel_task_requests(struct io_ring_ctx *ctx,
+ struct task_struct *task);
+
+static void destroy_fixed_file_ref_node(struct fixed_file_ref_node *ref_node);
+static struct fixed_file_ref_node *alloc_fixed_file_ref_node(
+ struct io_ring_ctx *ctx);
+
static void __io_complete_rw(struct io_kiocb *req, long res, long res2,
struct io_comp_state *cs);
static void io_cqring_fill_event(struct io_kiocb *req, long res);
spin_unlock_irq(&ctx->inflight_lock);
req->work.flags |= IO_WQ_WORK_FILES;
}
+ if (!(req->work.flags & IO_WQ_WORK_MM) &&
+ (def->work_flags & IO_WQ_WORK_MM)) {
+ if (id->mm != current->mm)
+ return false;
+ mmgrab(id->mm);
+ req->work.flags |= IO_WQ_WORK_MM;
+ }
return true;
}
{
const struct io_op_def *def = &io_op_defs[req->opcode];
struct io_ring_ctx *ctx = req->ctx;
- struct io_identity *id;
io_req_init_async(req);
- id = req->work.identity;
if (req->flags & REQ_F_FORCE_ASYNC)
req->work.flags |= IO_WQ_WORK_CONCURRENT;
req->work.flags |= IO_WQ_WORK_UNBOUND;
}
- /* ->mm can never change on us */
- if (!(req->work.flags & IO_WQ_WORK_MM) &&
- (def->work_flags & IO_WQ_WORK_MM)) {
- mmgrab(id->mm);
- req->work.flags |= IO_WQ_WORK_MM;
- }
-
/* if we fail grabbing identity, we must COW, regrab, and retry */
if (io_grab_identity(req))
return;
return io_wq_current_is_worker();
}
+static inline unsigned __io_cqring_events(struct io_ring_ctx *ctx)
+{
+ return ctx->cached_cq_tail - READ_ONCE(ctx->rings->cq.head);
+}
+
static void io_cqring_ev_posted(struct io_ring_ctx *ctx)
{
if (waitqueue_active(&ctx->wait))
eventfd_signal(ctx->cq_ev_fd, 1);
}
-static void io_cqring_mark_overflow(struct io_ring_ctx *ctx)
-{
- if (list_empty(&ctx->cq_overflow_list)) {
- clear_bit(0, &ctx->sq_check_overflow);
- clear_bit(0, &ctx->cq_check_overflow);
- ctx->rings->sq_flags &= ~IORING_SQ_CQ_OVERFLOW;
- }
-}
-
/* Returns true if there are no backlogged entries after the flush */
-static bool io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force,
- struct task_struct *tsk,
- struct files_struct *files)
+static bool __io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force,
+ struct task_struct *tsk,
+ struct files_struct *files)
{
struct io_rings *rings = ctx->rings;
struct io_kiocb *req, *tmp;
struct io_uring_cqe *cqe;
unsigned long flags;
+ bool all_flushed;
LIST_HEAD(list);
- if (!force) {
- if (list_empty_careful(&ctx->cq_overflow_list))
- return true;
- if ((ctx->cached_cq_tail - READ_ONCE(rings->cq.head) ==
- rings->cq_ring_entries))
- return false;
- }
+ if (!force && __io_cqring_events(ctx) == rings->cq_ring_entries)
+ return false;
spin_lock_irqsave(&ctx->completion_lock, flags);
-
- cqe = NULL;
list_for_each_entry_safe(req, tmp, &ctx->cq_overflow_list, compl.list) {
if (!io_match_task(req, tsk, files))
continue;
}
}
- io_commit_cqring(ctx);
- io_cqring_mark_overflow(ctx);
+ all_flushed = list_empty(&ctx->cq_overflow_list);
+ if (all_flushed) {
+ clear_bit(0, &ctx->sq_check_overflow);
+ clear_bit(0, &ctx->cq_check_overflow);
+ ctx->rings->sq_flags &= ~IORING_SQ_CQ_OVERFLOW;
+ }
+ io_commit_cqring(ctx);
spin_unlock_irqrestore(&ctx->completion_lock, flags);
io_cqring_ev_posted(ctx);
io_put_req(req);
}
- return cqe != NULL;
+ return all_flushed;
+}
+
+static void io_cqring_overflow_flush(struct io_ring_ctx *ctx, bool force,
+ struct task_struct *tsk,
+ struct files_struct *files)
+{
+ if (test_bit(0, &ctx->cq_check_overflow)) {
+ /* iopoll syncs against uring_lock, not completion_lock */
+ if (ctx->flags & IORING_SETUP_IOPOLL)
+ mutex_lock(&ctx->uring_lock);
+ __io_cqring_overflow_flush(ctx, force, tsk, files);
+ if (ctx->flags & IORING_SETUP_IOPOLL)
+ mutex_unlock(&ctx->uring_lock);
+ }
}
static void __io_cqring_fill_event(struct io_kiocb *req, long res, long cflags)
static void __io_req_task_submit(struct io_kiocb *req)
{
struct io_ring_ctx *ctx = req->ctx;
+ bool fail;
- if (!__io_sq_thread_acquire_mm(ctx) &&
- !__io_sq_thread_acquire_files(ctx)) {
- mutex_lock(&ctx->uring_lock);
+ fail = __io_sq_thread_acquire_mm(ctx) ||
+ __io_sq_thread_acquire_files(ctx);
+ mutex_lock(&ctx->uring_lock);
+ if (!fail)
__io_queue_sqe(req, NULL);
- mutex_unlock(&ctx->uring_lock);
- } else {
+ else
__io_req_task_cancel(req, -EFAULT);
- }
+ mutex_unlock(&ctx->uring_lock);
}
static void io_req_task_submit(struct callback_head *cb)
io_free_req(req);
}
-static unsigned io_cqring_events(struct io_ring_ctx *ctx, bool noflush)
+static unsigned io_cqring_events(struct io_ring_ctx *ctx)
{
- struct io_rings *rings = ctx->rings;
-
- if (test_bit(0, &ctx->cq_check_overflow)) {
- /*
- * noflush == true is from the waitqueue handler, just ensure
- * we wake up the task, and the next invocation will flush the
- * entries. We cannot safely to it from here.
- */
- if (noflush)
- return -1U;
-
- io_cqring_overflow_flush(ctx, false, NULL, NULL);
- }
-
/* See comment at the top of this file */
smp_rmb();
- return ctx->cached_cq_tail - READ_ONCE(rings->cq.head);
+ return __io_cqring_events(ctx);
}
static inline unsigned int io_sqring_entries(struct io_ring_ctx *ctx)
* If we do, we can potentially be spinning for commands that
* already triggered a CQE (eg in error).
*/
- if (io_cqring_events(ctx, false))
+ if (test_bit(0, &ctx->cq_check_overflow))
+ __io_cqring_overflow_flush(ctx, false, NULL, NULL);
+ if (io_cqring_events(ctx))
break;
/*
iov[0].iov_len = kbuf->len;
return 0;
}
- if (!req->rw.len)
- return 0;
- else if (req->rw.len > 1)
+ if (req->rw.len != 1)
return -EINVAL;
#ifdef CONFIG_COMPAT
struct io_uring_task *tctx = req->task->io_uring;
unsigned long flags;
+ put_files_struct(req->work.identity->files);
+ put_nsproxy(req->work.identity->nsproxy);
spin_lock_irqsave(&ctx->inflight_lock, flags);
list_del(&req->inflight_entry);
- 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);
- put_nsproxy(req->work.identity->nsproxy);
req->work.flags &= ~IO_WQ_WORK_FILES;
+ if (atomic_read(&tctx->in_idle))
+ wake_up(&tctx->wait);
}
static void __io_clean_op(struct io_kiocb *req)
}
if (ret) {
+ struct io_ring_ctx *lock_ctx = NULL;
+
+ if (req->ctx->flags & IORING_SETUP_IOPOLL)
+ lock_ctx = req->ctx;
+
/*
- * 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.
+ * io_iopoll_complete() does not hold completion_lock to
+ * complete polled io, so here for polled io, we can not call
+ * io_req_complete() directly, otherwise there maybe concurrent
+ * access to cqring, defer_list, etc, which is not safe. Given
+ * that io_iopoll_complete() is always called under uring_lock,
+ * so here for polled io, we also get uring_lock to complete
+ * it.
*/
- if (req->ctx->flags & IORING_SETUP_IOPOLL) {
- struct kiocb *kiocb = &req->rw.kiocb;
+ if (lock_ctx)
+ mutex_lock(&lock_ctx->uring_lock);
- kiocb_done(kiocb, ret, NULL);
- } else {
- req_set_fail_links(req);
- io_req_complete(req, ret);
- }
+ req_set_fail_links(req);
+ io_req_complete(req, ret);
+
+ if (lock_ctx)
+ mutex_unlock(&lock_ctx->uring_lock);
}
return io_steal_work(req);
/* if we have a backlog and couldn't flush it all, return BUSY */
if (test_bit(0, &ctx->sq_check_overflow)) {
- if (!list_empty(&ctx->cq_overflow_list) &&
- !io_cqring_overflow_flush(ctx, false, NULL, NULL))
+ if (!__io_cqring_overflow_flush(ctx, false, NULL, NULL))
return -EBUSY;
}
unsigned nr_timeouts;
};
-static inline bool io_should_wake(struct io_wait_queue *iowq, bool noflush)
+static inline bool io_should_wake(struct io_wait_queue *iowq)
{
struct io_ring_ctx *ctx = iowq->ctx;
* started waiting. For timeouts, we always want to return to userspace,
* regardless of event count.
*/
- return io_cqring_events(ctx, noflush) >= iowq->to_wait ||
+ return io_cqring_events(ctx) >= iowq->to_wait ||
atomic_read(&ctx->cq_timeouts) != iowq->nr_timeouts;
}
struct io_wait_queue *iowq = container_of(curr, struct io_wait_queue,
wq);
- /* use noflush == true, as we can't safely rely on locking context */
- if (!io_should_wake(iowq, true))
- return -1;
-
- return autoremove_wake_function(curr, mode, wake_flags, key);
+ /*
+ * Cannot safely flush overflowed CQEs from here, ensure we wake up
+ * the task, and the next invocation will do it.
+ */
+ if (io_should_wake(iowq) || test_bit(0, &iowq->ctx->cq_check_overflow))
+ return autoremove_wake_function(curr, mode, wake_flags, key);
+ return -1;
}
static int io_run_task_work_sig(void)
int ret = 0;
do {
- if (io_cqring_events(ctx, false) >= min_events)
+ io_cqring_overflow_flush(ctx, false, NULL, NULL);
+ if (io_cqring_events(ctx) >= min_events)
return 0;
if (!io_run_task_work())
break;
iowq.nr_timeouts = atomic_read(&ctx->cq_timeouts);
trace_io_uring_cqring_wait(ctx, min_events);
do {
+ io_cqring_overflow_flush(ctx, false, NULL, NULL);
prepare_to_wait_exclusive(&ctx->wait, &iowq.wq,
TASK_INTERRUPTIBLE);
/* make sure we run task_work before checking for signals */
continue;
else if (ret < 0)
break;
- if (io_should_wake(&iowq, false))
+ if (io_should_wake(&iowq))
break;
+ if (test_bit(0, &ctx->cq_check_overflow))
+ continue;
if (uts) {
timeout = schedule_timeout(timeout);
if (timeout == 0) {
complete(&data->done);
}
+static void io_sqe_files_set_node(struct fixed_file_data *file_data,
+ struct fixed_file_ref_node *ref_node)
+{
+ spin_lock_bh(&file_data->lock);
+ file_data->node = ref_node;
+ list_add_tail(&ref_node->node, &file_data->ref_list);
+ spin_unlock_bh(&file_data->lock);
+ percpu_ref_get(&file_data->refs);
+}
+
static int io_sqe_files_unregister(struct io_ring_ctx *ctx)
{
struct fixed_file_data *data = ctx->file_data;
- struct fixed_file_ref_node *ref_node = NULL;
+ struct fixed_file_ref_node *backup_node, *ref_node = NULL;
unsigned nr_tables, i;
+ int ret;
if (!data)
return -ENXIO;
+ backup_node = alloc_fixed_file_ref_node(ctx);
+ if (!backup_node)
+ return -ENOMEM;
spin_lock_bh(&data->lock);
ref_node = data->node;
/* wait for all refs nodes to complete */
flush_delayed_work(&ctx->file_put_work);
- wait_for_completion(&data->done);
+ do {
+ ret = wait_for_completion_interruptible(&data->done);
+ if (!ret)
+ break;
+ ret = io_run_task_work_sig();
+ if (ret < 0) {
+ percpu_ref_resurrect(&data->refs);
+ reinit_completion(&data->done);
+ io_sqe_files_set_node(data, backup_node);
+ return ret;
+ }
+ } while (1);
__io_sqe_files_unregister(ctx);
nr_tables = DIV_ROUND_UP(ctx->nr_user_files, IORING_MAX_FILES_TABLE);
kfree(data);
ctx->file_data = NULL;
ctx->nr_user_files = 0;
+ destroy_fixed_file_ref_node(backup_node);
return 0;
}
ref_node = kzalloc(sizeof(*ref_node), GFP_KERNEL);
if (!ref_node)
- return ERR_PTR(-ENOMEM);
+ return NULL;
if (percpu_ref_init(&ref_node->refs, io_file_data_ref_zero,
0, GFP_KERNEL)) {
kfree(ref_node);
- return ERR_PTR(-ENOMEM);
+ return NULL;
}
INIT_LIST_HEAD(&ref_node->node);
INIT_LIST_HEAD(&ref_node->file_list);
}
ref_node = alloc_fixed_file_ref_node(ctx);
- if (IS_ERR(ref_node)) {
+ if (!ref_node) {
io_sqe_files_unregister(ctx);
- return PTR_ERR(ref_node);
+ return -ENOMEM;
}
- file_data->node = ref_node;
- spin_lock_bh(&file_data->lock);
- list_add_tail(&ref_node->node, &file_data->ref_list);
- spin_unlock_bh(&file_data->lock);
- percpu_ref_get(&file_data->refs);
+ io_sqe_files_set_node(file_data, ref_node);
return ret;
out_fput:
for (i = 0; i < ctx->nr_user_files; i++) {
return -EINVAL;
ref_node = alloc_fixed_file_ref_node(ctx);
- if (IS_ERR(ref_node))
- return PTR_ERR(ref_node);
+ if (!ref_node)
+ return -ENOMEM;
done = 0;
fds = u64_to_user_ptr(up->fds);
if (needs_switch) {
percpu_ref_kill(&data->node->refs);
- spin_lock_bh(&data->lock);
- list_add_tail(&ref_node->node, &data->ref_list);
- data->node = ref_node;
- spin_unlock_bh(&data->lock);
- percpu_ref_get(&ctx->file_data->refs);
+ io_sqe_files_set_node(data, ref_node);
} else
destroy_fixed_file_ref_node(ref_node);
smp_rmb();
if (!io_sqring_full(ctx))
mask |= EPOLLOUT | EPOLLWRNORM;
- if (io_cqring_events(ctx, false))
+ io_cqring_overflow_flush(ctx, false, NULL, NULL);
+ if (io_cqring_events(ctx))
mask |= EPOLLIN | EPOLLRDNORM;
return mask;
* as nobody else will be looking for them.
*/
do {
- io_iopoll_try_reap_events(ctx);
+ __io_uring_cancel_task_requests(ctx, NULL);
} while (!wait_for_completion_timeout(&ctx->ref_comp, HZ/20));
io_ring_ctx_free(ctx);
}
+static bool io_cancel_ctx_cb(struct io_wq_work *work, void *data)
+{
+ struct io_kiocb *req = container_of(work, struct io_kiocb, work);
+
+ return req->ctx == data;
+}
+
static void io_ring_ctx_wait_and_kill(struct io_ring_ctx *ctx)
{
mutex_lock(&ctx->uring_lock);
/* if force is set, the ring is going away. always drop after that */
ctx->cq_overflow_flushed = 1;
if (ctx->rings)
- io_cqring_overflow_flush(ctx, true, NULL, NULL);
+ __io_cqring_overflow_flush(ctx, true, NULL, NULL);
mutex_unlock(&ctx->uring_lock);
io_kill_timeouts(ctx, NULL, NULL);
io_poll_remove_all(ctx, NULL, NULL);
if (ctx->io_wq)
- io_wq_cancel_all(ctx->io_wq);
+ io_wq_cancel_cb(ctx->io_wq, io_cancel_ctx_cb, ctx, true);
/* if we failed setting up the ctx, we might not have any rings */
io_iopoll_try_reap_events(ctx);
enum io_wq_cancel cret;
bool ret = false;
- cret = io_wq_cancel_cb(ctx->io_wq, io_cancel_task_cb, &cancel, true);
- if (cret != IO_WQ_CANCEL_NOTFOUND)
- ret = true;
+ if (ctx->io_wq) {
+ cret = io_wq_cancel_cb(ctx->io_wq, io_cancel_task_cb,
+ &cancel, true);
+ ret |= (cret != IO_WQ_CANCEL_NOTFOUND);
+ }
/* SQPOLL thread does its own polling */
if (!(ctx->flags & IORING_SETUP_SQPOLL)) {
ret |= io_poll_remove_all(ctx, task, NULL);
ret |= io_kill_timeouts(ctx, task, NULL);
+ ret |= io_run_task_work();
if (!ret)
break;
- io_run_task_work();
cond_resched();
}
}
}
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));
if (!files)
__io_uring_cancel_task_requests(ctx, task);
static int io_uring_add_task_file(struct io_ring_ctx *ctx, struct file *file)
{
struct io_uring_task *tctx = current->io_uring;
+ int ret;
if (unlikely(!tctx)) {
- int ret;
-
ret = io_uring_alloc_task_context(current);
if (unlikely(ret))
return ret;
if (!old) {
get_file(file);
- xa_store(&tctx->xa, (unsigned long)file, file, GFP_KERNEL);
+ ret = xa_err(xa_store(&tctx->xa, (unsigned long)file,
+ file, GFP_KERNEL));
+ if (ret) {
+ fput(file);
+ return ret;
+ }
}
tctx->last = file;
}
io_uring_del_task_file(file);
}
+static void io_uring_remove_task_files(struct io_uring_task *tctx)
+{
+ struct file *file;
+ unsigned long index;
+
+ xa_for_each(&tctx->xa, index, file)
+ io_uring_del_task_file(file);
+}
+
void __io_uring_files_cancel(struct files_struct *files)
{
struct io_uring_task *tctx = current->io_uring;
/* make sure overflow events are dropped */
atomic_inc(&tctx->in_idle);
-
- xa_for_each(&tctx->xa, index, file) {
- struct io_ring_ctx *ctx = file->private_data;
-
- io_uring_cancel_task_requests(ctx, files);
- if (files)
- io_uring_del_task_file(file);
- }
-
+ xa_for_each(&tctx->xa, index, file)
+ io_uring_cancel_task_requests(file->private_data, files);
atomic_dec(&tctx->in_idle);
+
+ if (files)
+ io_uring_remove_task_files(tctx);
}
static s64 tctx_inflight(struct io_uring_task *tctx)
if (inflight != tctx_inflight(tctx))
continue;
schedule();
+ finish_wait(&tctx->wait, &wait);
} while (1);
- finish_wait(&tctx->wait, &wait);
atomic_dec(&tctx->in_idle);
+
+ io_uring_remove_task_files(tctx);
}
static int io_uring_flush(struct file *file, void *data)
*/
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));
+ io_cqring_overflow_flush(ctx, false, NULL, NULL);
+
if (flags & IORING_ENTER_SQ_WAKEUP)
wake_up(&ctx->sq_data->wait);
if (flags & IORING_ENTER_SQ_WAIT)
return 0;
}
+static int io_uring_install_fd(struct io_ring_ctx *ctx, struct file *file)
+{
+ int ret, fd;
+
+ fd = get_unused_fd_flags(O_RDWR | O_CLOEXEC);
+ if (fd < 0)
+ return fd;
+
+ ret = io_uring_add_task_file(ctx, file);
+ if (ret) {
+ put_unused_fd(fd);
+ return ret;
+ }
+ fd_install(fd, file);
+ return fd;
+}
+
/*
* Allocate an anonymous fd, this is what constitutes the application
* visible backing of an io_uring instance. The application mmaps this
* fd to gain access to the SQ/CQ ring details. If UNIX sockets are enabled,
* we have to tie this fd to a socket for file garbage collection purposes.
*/
-static int io_uring_get_fd(struct io_ring_ctx *ctx)
+static struct file *io_uring_get_file(struct io_ring_ctx *ctx)
{
struct file *file;
+#if defined(CONFIG_UNIX)
int ret;
- int fd;
-#if defined(CONFIG_UNIX)
ret = sock_create_kern(&init_net, PF_UNIX, SOCK_RAW, IPPROTO_IP,
&ctx->ring_sock);
if (ret)
- return ret;
+ return ERR_PTR(ret);
#endif
- ret = get_unused_fd_flags(O_RDWR | O_CLOEXEC);
- if (ret < 0)
- goto err;
- fd = ret;
-
file = anon_inode_getfile("[io_uring]", &io_uring_fops, ctx,
O_RDWR | O_CLOEXEC);
- if (IS_ERR(file)) {
- put_unused_fd(fd);
- ret = PTR_ERR(file);
- goto err;
- }
-
#if defined(CONFIG_UNIX)
- ctx->ring_sock->file = file;
-#endif
- ret = io_uring_add_task_file(ctx, file);
- if (ret) {
- fput(file);
- put_unused_fd(fd);
- goto err;
+ if (IS_ERR(file)) {
+ sock_release(ctx->ring_sock);
+ ctx->ring_sock = NULL;
+ } else {
+ ctx->ring_sock->file = file;
}
- fd_install(fd, file);
- return fd;
-err:
-#if defined(CONFIG_UNIX)
- sock_release(ctx->ring_sock);
- ctx->ring_sock = NULL;
#endif
- return ret;
+ return file;
}
static int io_uring_create(unsigned entries, struct io_uring_params *p,
{
struct user_struct *user = NULL;
struct io_ring_ctx *ctx;
+ struct file *file;
bool limit_mem;
int ret;
goto err;
}
+ file = io_uring_get_file(ctx);
+ if (IS_ERR(file)) {
+ ret = PTR_ERR(file);
+ goto err;
+ }
+
/*
* Install ring fd as the very last thing, so we don't risk someone
* having closed it before we finish setup
*/
- ret = io_uring_get_fd(ctx);
- if (ret < 0)
- goto err;
+ ret = io_uring_install_fd(ctx, file);
+ if (ret < 0) {
+ /* 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;