summaryrefslogtreecommitdiff
path: root/io_uring
diff options
context:
space:
mode:
Diffstat (limited to 'io_uring')
-rw-r--r--io_uring/cancel.c9
-rw-r--r--io_uring/fdinfo.c12
-rw-r--r--io_uring/io-wq.c6
-rw-r--r--io_uring/io_uring.c49
-rw-r--r--io_uring/io_uring.h15
-rw-r--r--io_uring/poll.c48
-rw-r--r--io_uring/rw.c6
7 files changed, 103 insertions, 42 deletions
diff --git a/io_uring/cancel.c b/io_uring/cancel.c
index 2291a53cdabd..b4f5dfacc0c3 100644
--- a/io_uring/cancel.c
+++ b/io_uring/cancel.c
@@ -288,24 +288,23 @@ int io_sync_cancel(struct io_ring_ctx *ctx, void __user *arg)
ret = __io_sync_cancel(current->io_uring, &cd, sc.fd);
+ mutex_unlock(&ctx->uring_lock);
if (ret != -EALREADY)
break;
- mutex_unlock(&ctx->uring_lock);
ret = io_run_task_work_sig(ctx);
- if (ret < 0) {
- mutex_lock(&ctx->uring_lock);
+ if (ret < 0)
break;
- }
ret = schedule_hrtimeout(&timeout, HRTIMER_MODE_ABS);
- mutex_lock(&ctx->uring_lock);
if (!ret) {
ret = -ETIME;
break;
}
+ mutex_lock(&ctx->uring_lock);
} while (1);
finish_wait(&ctx->cq_wait, &wait);
+ mutex_lock(&ctx->uring_lock);
if (ret == -ENOENT || ret > 0)
ret = 0;
diff --git a/io_uring/fdinfo.c b/io_uring/fdinfo.c
index 2e04850a657b..882bd56b01ed 100644
--- a/io_uring/fdinfo.c
+++ b/io_uring/fdinfo.c
@@ -170,12 +170,11 @@ static __cold void __io_uring_show_fdinfo(struct io_ring_ctx *ctx,
xa_for_each(&ctx->personalities, index, cred)
io_uring_show_cred(m, index, cred);
}
- if (has_lock)
- mutex_unlock(&ctx->uring_lock);
seq_puts(m, "PollList:\n");
for (i = 0; i < (1U << ctx->cancel_table.hash_bits); i++) {
struct io_hash_bucket *hb = &ctx->cancel_table.hbs[i];
+ struct io_hash_bucket *hbl = &ctx->cancel_table_locked.hbs[i];
struct io_kiocb *req;
spin_lock(&hb->lock);
@@ -183,8 +182,17 @@ static __cold void __io_uring_show_fdinfo(struct io_ring_ctx *ctx,
seq_printf(m, " op=%d, task_works=%d\n", req->opcode,
task_work_pending(req->task));
spin_unlock(&hb->lock);
+
+ if (!has_lock)
+ continue;
+ hlist_for_each_entry(req, &hbl->list, hash_node)
+ seq_printf(m, " op=%d, task_works=%d\n", req->opcode,
+ task_work_pending(req->task));
}
+ if (has_lock)
+ mutex_unlock(&ctx->uring_lock);
+
seq_puts(m, "CqOverflowList:\n");
spin_lock(&ctx->completion_lock);
list_for_each_entry(ocqe, &ctx->cq_overflow_list, list) {
diff --git a/io_uring/io-wq.c b/io_uring/io-wq.c
index 6f1d0e5df23a..411bb2d1acd4 100644
--- a/io_uring/io-wq.c
+++ b/io_uring/io-wq.c
@@ -1230,6 +1230,12 @@ static void io_wq_cancel_tw_create(struct io_wq *wq)
worker = container_of(cb, struct io_worker, create_work);
io_worker_cancel_cb(worker);
+ /*
+ * Only the worker continuation helper has worker allocated and
+ * hence needs freeing.
+ */
+ if (cb->func == create_worker_cont)
+ kfree(worker);
}
}
diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
index ff2bbac1a10f..2ac1cd8d23ea 100644
--- a/io_uring/io_uring.c
+++ b/io_uring/io_uring.c
@@ -677,16 +677,20 @@ static void __io_cqring_overflow_flush(struct io_ring_ctx *ctx)
io_cq_unlock_post(ctx);
}
+static void io_cqring_do_overflow_flush(struct io_ring_ctx *ctx)
+{
+ /* iopoll syncs against uring_lock, not completion_lock */
+ if (ctx->flags & IORING_SETUP_IOPOLL)
+ mutex_lock(&ctx->uring_lock);
+ __io_cqring_overflow_flush(ctx);
+ if (ctx->flags & IORING_SETUP_IOPOLL)
+ mutex_unlock(&ctx->uring_lock);
+}
+
static void io_cqring_overflow_flush(struct io_ring_ctx *ctx)
{
- if (test_bit(IO_CHECK_CQ_OVERFLOW_BIT, &ctx->check_cq)) {
- /* iopoll syncs against uring_lock, not completion_lock */
- if (ctx->flags & IORING_SETUP_IOPOLL)
- mutex_lock(&ctx->uring_lock);
- __io_cqring_overflow_flush(ctx);
- if (ctx->flags & IORING_SETUP_IOPOLL)
- mutex_unlock(&ctx->uring_lock);
- }
+ if (test_bit(IO_CHECK_CQ_OVERFLOW_BIT, &ctx->check_cq))
+ io_cqring_do_overflow_flush(ctx);
}
void __io_put_task(struct task_struct *task, int nr)
@@ -727,6 +731,8 @@ static bool io_cqring_event_overflow(struct io_ring_ctx *ctx, u64 user_data,
size_t ocq_size = sizeof(struct io_overflow_cqe);
bool is_cqe32 = (ctx->flags & IORING_SETUP_CQE32);
+ lockdep_assert_held(&ctx->completion_lock);
+
if (is_cqe32)
ocq_size += sizeof(struct io_uring_cqe);
@@ -816,9 +822,6 @@ static bool io_fill_cqe_aux(struct io_ring_ctx *ctx, u64 user_data, s32 res,
{
struct io_uring_cqe *cqe;
- if (!ctx->task_complete)
- lockdep_assert_held(&ctx->completion_lock);
-
ctx->cq_extra++;
/*
@@ -1232,13 +1235,18 @@ static void io_req_local_work_add(struct io_kiocb *req)
{
struct io_ring_ctx *ctx = req->ctx;
- if (!llist_add(&req->io_task_work.node, &ctx->work_llist))
+ percpu_ref_get(&ctx->refs);
+
+ if (!llist_add(&req->io_task_work.node, &ctx->work_llist)) {
+ percpu_ref_put(&ctx->refs);
return;
+ }
/* need it for the following io_cqring_wake() */
smp_mb__after_atomic();
if (unlikely(atomic_read(&req->task->io_uring->in_idle))) {
io_move_task_work_from_local(ctx);
+ percpu_ref_put(&ctx->refs);
return;
}
@@ -1248,6 +1256,7 @@ static void io_req_local_work_add(struct io_kiocb *req)
if (ctx->has_evfd)
io_eventfd_signal(ctx);
__io_cqring_wake(ctx);
+ percpu_ref_put(&ctx->refs);
}
void __io_req_task_work_add(struct io_kiocb *req, bool allow_local)
@@ -2461,7 +2470,7 @@ int io_run_task_work_sig(struct io_ring_ctx *ctx)
/* when returns >0, the caller should retry */
static inline int io_cqring_wait_schedule(struct io_ring_ctx *ctx,
struct io_wait_queue *iowq,
- ktime_t timeout)
+ ktime_t *timeout)
{
int ret;
unsigned long check_cq;
@@ -2479,7 +2488,7 @@ static inline int io_cqring_wait_schedule(struct io_ring_ctx *ctx,
if (check_cq & BIT(IO_CHECK_CQ_DROPPED_BIT))
return -EBADR;
}
- if (!schedule_hrtimeout(&timeout, HRTIMER_MODE_ABS))
+ if (!schedule_hrtimeout(timeout, HRTIMER_MODE_ABS))
return -ETIME;
/*
@@ -2549,10 +2558,13 @@ static int io_cqring_wait(struct io_ring_ctx *ctx, int min_events,
trace_io_uring_cqring_wait(ctx, min_events);
do {
- io_cqring_overflow_flush(ctx);
+ if (test_bit(IO_CHECK_CQ_OVERFLOW_BIT, &ctx->check_cq)) {
+ finish_wait(&ctx->cq_wait, &iowq.wq);
+ io_cqring_do_overflow_flush(ctx);
+ }
prepare_to_wait_exclusive(&ctx->cq_wait, &iowq.wq,
TASK_INTERRUPTIBLE);
- ret = io_cqring_wait_schedule(ctx, &iowq, timeout);
+ ret = io_cqring_wait_schedule(ctx, &iowq, &timeout);
if (__io_cqring_events_user(ctx) >= min_events)
break;
cond_resched();
@@ -4013,8 +4025,6 @@ static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
return -EEXIST;
if (ctx->restricted) {
- if (opcode >= IORING_REGISTER_LAST)
- return -EINVAL;
opcode = array_index_nospec(opcode, IORING_REGISTER_LAST);
if (!test_bit(opcode, ctx->restrictions.register_op))
return -EACCES;
@@ -4170,6 +4180,9 @@ SYSCALL_DEFINE4(io_uring_register, unsigned int, fd, unsigned int, opcode,
long ret = -EBADF;
struct fd f;
+ if (opcode >= IORING_REGISTER_LAST)
+ return -EINVAL;
+
f = fdget(fd);
if (!f.file)
return -EBADF;
diff --git a/io_uring/io_uring.h b/io_uring/io_uring.h
index e9f0d41ebb99..ab4b2a1c3b7e 100644
--- a/io_uring/io_uring.h
+++ b/io_uring/io_uring.h
@@ -79,6 +79,19 @@ bool __io_alloc_req_refill(struct io_ring_ctx *ctx);
bool io_match_task_safe(struct io_kiocb *head, struct task_struct *task,
bool cancel_all);
+#define io_lockdep_assert_cq_locked(ctx) \
+ do { \
+ if (ctx->flags & IORING_SETUP_IOPOLL) { \
+ lockdep_assert_held(&ctx->uring_lock); \
+ } else if (!ctx->task_complete) { \
+ lockdep_assert_held(&ctx->completion_lock); \
+ } else if (ctx->submitter_task->flags & PF_EXITING) { \
+ lockdep_assert(current_work()); \
+ } else { \
+ lockdep_assert(current == ctx->submitter_task); \
+ } \
+ } while (0)
+
static inline void io_req_task_work_add(struct io_kiocb *req)
{
__io_req_task_work_add(req, true);
@@ -92,6 +105,8 @@ void io_cq_unlock_post(struct io_ring_ctx *ctx);
static inline struct io_uring_cqe *io_get_cqe_overflow(struct io_ring_ctx *ctx,
bool overflow)
{
+ io_lockdep_assert_cq_locked(ctx);
+
if (likely(ctx->cqe_cached < ctx->cqe_sentinel)) {
struct io_uring_cqe *cqe = ctx->cqe_cached;
diff --git a/io_uring/poll.c b/io_uring/poll.c
index ee7da6150ec4..32e5fc8365e6 100644
--- a/io_uring/poll.c
+++ b/io_uring/poll.c
@@ -223,21 +223,22 @@ enum {
IOU_POLL_DONE = 0,
IOU_POLL_NO_ACTION = 1,
IOU_POLL_REMOVE_POLL_USE_RES = 2,
+ IOU_POLL_REISSUE = 3,
};
/*
* All poll tw should go through this. Checks for poll events, manages
* references, does rewait, etc.
*
- * Returns a negative error on failure. IOU_POLL_NO_ACTION when no action require,
- * which is either spurious wakeup or multishot CQE is served.
- * IOU_POLL_DONE when it's done with the request, then the mask is stored in req->cqe.res.
- * IOU_POLL_REMOVE_POLL_USE_RES indicates to remove multishot poll and that the result
- * is stored in req->cqe.
+ * Returns a negative error on failure. IOU_POLL_NO_ACTION when no action
+ * require, which is either spurious wakeup or multishot CQE is served.
+ * IOU_POLL_DONE when it's done with the request, then the mask is stored in
+ * req->cqe.res. IOU_POLL_REMOVE_POLL_USE_RES indicates to remove multishot
+ * poll and that the result is stored in req->cqe.
*/
static int io_poll_check_events(struct io_kiocb *req, bool *locked)
{
- int v, ret;
+ int v;
/* req->task == current here, checking PF_EXITING is safe */
if (unlikely(req->task->flags & PF_EXITING))
@@ -276,10 +277,15 @@ static int io_poll_check_events(struct io_kiocb *req, bool *locked)
if (!req->cqe.res) {
struct poll_table_struct pt = { ._key = req->apoll_events };
req->cqe.res = vfs_poll(req->file, &pt) & req->apoll_events;
+ /*
+ * We got woken with a mask, but someone else got to
+ * it first. The above vfs_poll() doesn't add us back
+ * to the waitqueue, so if we get nothing back, we
+ * should be safe and attempt a reissue.
+ */
+ if (unlikely(!req->cqe.res))
+ return IOU_POLL_REISSUE;
}
-
- if ((unlikely(!req->cqe.res)))
- continue;
if (req->apoll_events & EPOLLONESHOT)
return IOU_POLL_DONE;
@@ -294,7 +300,7 @@ static int io_poll_check_events(struct io_kiocb *req, bool *locked)
return IOU_POLL_REMOVE_POLL_USE_RES;
}
} else {
- ret = io_poll_issue(req, locked);
+ int ret = io_poll_issue(req, locked);
if (ret == IOU_STOP_MULTISHOT)
return IOU_POLL_REMOVE_POLL_USE_RES;
if (ret < 0)
@@ -330,6 +336,9 @@ static void io_poll_task_func(struct io_kiocb *req, bool *locked)
poll = io_kiocb_to_cmd(req, struct io_poll);
req->cqe.res = mangle_poll(req->cqe.res & poll->events);
+ } else if (ret == IOU_POLL_REISSUE) {
+ io_req_task_submit(req, locked);
+ return;
} else if (ret != IOU_POLL_REMOVE_POLL_USE_RES) {
req->cqe.res = ret;
req_set_fail(req);
@@ -342,7 +351,7 @@ static void io_poll_task_func(struct io_kiocb *req, bool *locked)
if (ret == IOU_POLL_REMOVE_POLL_USE_RES)
io_req_task_complete(req, locked);
- else if (ret == IOU_POLL_DONE)
+ else if (ret == IOU_POLL_DONE || ret == IOU_POLL_REISSUE)
io_req_task_submit(req, locked);
else
io_req_defer_failed(req, ret);
@@ -533,6 +542,14 @@ static bool io_poll_can_finish_inline(struct io_kiocb *req,
return pt->owning || io_poll_get_ownership(req);
}
+static void io_poll_add_hash(struct io_kiocb *req)
+{
+ if (req->flags & REQ_F_HASH_LOCKED)
+ io_poll_req_insert_locked(req);
+ else
+ io_poll_req_insert(req);
+}
+
/*
* Returns 0 when it's handed over for polling. The caller owns the requests if
* it returns non-zero, but otherwise should not touch it. Negative values
@@ -591,18 +608,17 @@ static int __io_arm_poll_handler(struct io_kiocb *req,
if (mask &&
((poll->events & (EPOLLET|EPOLLONESHOT)) == (EPOLLET|EPOLLONESHOT))) {
- if (!io_poll_can_finish_inline(req, ipt))
+ if (!io_poll_can_finish_inline(req, ipt)) {
+ io_poll_add_hash(req);
return 0;
+ }
io_poll_remove_entries(req);
ipt->result_mask = mask;
/* no one else has access to the req, forget about the ref */
return 1;
}
- if (req->flags & REQ_F_HASH_LOCKED)
- io_poll_req_insert_locked(req);
- else
- io_poll_req_insert(req);
+ io_poll_add_hash(req);
if (mask && (poll->events & EPOLLET) &&
io_poll_can_finish_inline(req, ipt)) {
diff --git a/io_uring/rw.c b/io_uring/rw.c
index 8227af2e1c0f..9c3ddd46a1ad 100644
--- a/io_uring/rw.c
+++ b/io_uring/rw.c
@@ -1062,7 +1062,11 @@ int io_do_iopoll(struct io_ring_ctx *ctx, bool force_nonspin)
continue;
req->cqe.flags = io_put_kbuf(req, 0);
- io_fill_cqe_req(req->ctx, req);
+ if (unlikely(!__io_fill_cqe_req(ctx, req))) {
+ spin_lock(&ctx->completion_lock);
+ io_req_cqe_overflow(req);
+ spin_unlock(&ctx->completion_lock);
+ }
}
if (unlikely(!nr_events))
generated by cgit v1.2.3-70-g09d2 (git 2.47.1) at 2024-12-26 06:59:39 +0000