| Message ID | 20250409134057.198671-3-axboe@kernel.dk (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | Cancel and wait for all requests on exit | expand |
On Wed, Apr 09, 2025 at 07:35:20AM -0600, Jens Axboe wrote: > There are two types of work here: > > 1) Fallback work, if the task is exiting > 2) The exit side cancelations > > and both of them may do the final fput() of a file. When this happens, > fput() will schedule delayed work. This slows down exits when io_uring I was a bit surprised by this because it means that all those __fput()s are done with kthread credentials which is a bit surprising (but harmless afaict). > needs to wait for that work to finish. It is possible to flush this via > flush_delayed_fput(), but that's a big hammer as other unrelated files > could be involved, and from other tasks as well. > > Add two io_uring helpers to temporarily clear PF_NO_TASKWORK for the > worker threads, and run any queued task_work before setting the flag > again. Then we can ensure we only flush related items that received > their final fput as part of work cancelation and flushing. Ok, so the only change is that this isn't offloaded to the global delayed fput workqueue but to the task work that you're running off of your kthread helpers. > > For now these are io_uring private, but could obviously be made > generically available, should there be a need to do so. > > Signed-off-by: Jens Axboe <axboe@kernel.dk> > --- > io_uring/io_uring.c | 21 +++++++++++++++++++++ > 1 file changed, 21 insertions(+) > > diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c > index c6209fe44cb1..bff99e185217 100644 > --- a/io_uring/io_uring.c > +++ b/io_uring/io_uring.c > @@ -238,6 +238,20 @@ static inline void io_req_add_to_cache(struct io_kiocb *req, struct io_ring_ctx > wq_stack_add_head(&req->comp_list, &ctx->submit_state.free_list); > } > > +static __cold void io_kworker_tw_start(void) > +{ > + if (WARN_ON_ONCE(!(current->flags & PF_NO_TASKWORK))) > + return; > + current->flags &= ~PF_NO_TASKWORK; > +} > + > +static __cold void io_kworker_tw_end(void) > +{ > + while (task_work_pending(current)) > + task_work_run(); > + current->flags |= PF_NO_TASKWORK; > +} > + > static __cold void io_ring_ctx_ref_free(struct percpu_ref *ref) > { > struct io_ring_ctx *ctx = container_of(ref, struct io_ring_ctx, refs); > @@ -253,6 +267,8 @@ static __cold void io_fallback_req_func(struct work_struct *work) > struct io_kiocb *req, *tmp; > struct io_tw_state ts = {}; > > + io_kworker_tw_start(); > + > percpu_ref_get(&ctx->refs); > mutex_lock(&ctx->uring_lock); > llist_for_each_entry_safe(req, tmp, node, io_task_work.node) > @@ -260,6 +276,7 @@ static __cold void io_fallback_req_func(struct work_struct *work) > io_submit_flush_completions(ctx); > mutex_unlock(&ctx->uring_lock); > percpu_ref_put(&ctx->refs); > + io_kworker_tw_end(); > } > > static int io_alloc_hash_table(struct io_hash_table *table, unsigned bits) > @@ -2876,6 +2893,8 @@ static __cold void io_ring_exit_work(struct work_struct *work) > struct io_tctx_node *node; > int ret; > > + io_kworker_tw_start(); > + > /* > * If we're doing polled IO and end up having requests being > * submitted async (out-of-line), then completions can come in while > @@ -2932,6 +2951,8 @@ static __cold void io_ring_exit_work(struct work_struct *work) > */ > } while (!wait_for_completion_interruptible_timeout(&ctx->ref_comp, interval)); > > + io_kworker_tw_end(); > + > init_completion(&exit.completion); > init_task_work(&exit.task_work, io_tctx_exit_cb); > exit.ctx = ctx; > -- > 2.49.0 >
On 4/11/25 7:55 AM, Christian Brauner wrote: > On Wed, Apr 09, 2025 at 07:35:20AM -0600, Jens Axboe wrote: >> There are two types of work here: >> >> 1) Fallback work, if the task is exiting >> 2) The exit side cancelations >> >> and both of them may do the final fput() of a file. When this happens, >> fput() will schedule delayed work. This slows down exits when io_uring > > I was a bit surprised by this because it means that all those __fput()s > are done with kthread credentials which is a bit surprising (but > harmless afaict). Sure hope it is, because that's already what happens off the delayed fput that it'd otherwise go through! >> needs to wait for that work to finish. It is possible to flush this via >> flush_delayed_fput(), but that's a big hammer as other unrelated files >> could be involved, and from other tasks as well. >> >> Add two io_uring helpers to temporarily clear PF_NO_TASKWORK for the >> worker threads, and run any queued task_work before setting the flag >> again. Then we can ensure we only flush related items that received >> their final fput as part of work cancelation and flushing. > > Ok, so the only change is that this isn't offloaded to the global > delayed fput workqueue but to the task work that you're running off of > your kthread helpers. Exactly
diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c index c6209fe44cb1..bff99e185217 100644 --- a/io_uring/io_uring.c +++ b/io_uring/io_uring.c @@ -238,6 +238,20 @@ static inline void io_req_add_to_cache(struct io_kiocb *req, struct io_ring_ctx wq_stack_add_head(&req->comp_list, &ctx->submit_state.free_list); } +static __cold void io_kworker_tw_start(void) +{ + if (WARN_ON_ONCE(!(current->flags & PF_NO_TASKWORK))) + return; + current->flags &= ~PF_NO_TASKWORK; +} + +static __cold void io_kworker_tw_end(void) +{ + while (task_work_pending(current)) + task_work_run(); + current->flags |= PF_NO_TASKWORK; +} + static __cold void io_ring_ctx_ref_free(struct percpu_ref *ref) { struct io_ring_ctx *ctx = container_of(ref, struct io_ring_ctx, refs); @@ -253,6 +267,8 @@ static __cold void io_fallback_req_func(struct work_struct *work) struct io_kiocb *req, *tmp; struct io_tw_state ts = {}; + io_kworker_tw_start(); + percpu_ref_get(&ctx->refs); mutex_lock(&ctx->uring_lock); llist_for_each_entry_safe(req, tmp, node, io_task_work.node) @@ -260,6 +276,7 @@ static __cold void io_fallback_req_func(struct work_struct *work) io_submit_flush_completions(ctx); mutex_unlock(&ctx->uring_lock); percpu_ref_put(&ctx->refs); + io_kworker_tw_end(); } static int io_alloc_hash_table(struct io_hash_table *table, unsigned bits) @@ -2876,6 +2893,8 @@ static __cold void io_ring_exit_work(struct work_struct *work) struct io_tctx_node *node; int ret; + io_kworker_tw_start(); + /* * If we're doing polled IO and end up having requests being * submitted async (out-of-line), then completions can come in while @@ -2932,6 +2951,8 @@ static __cold void io_ring_exit_work(struct work_struct *work) */ } while (!wait_for_completion_interruptible_timeout(&ctx->ref_comp, interval)); + io_kworker_tw_end(); + init_completion(&exit.completion); init_task_work(&exit.task_work, io_tctx_exit_cb); exit.ctx = ctx;
There are two types of work here: 1) Fallback work, if the task is exiting 2) The exit side cancelations and both of them may do the final fput() of a file. When this happens, fput() will schedule delayed work. This slows down exits when io_uring needs to wait for that work to finish. It is possible to flush this via flush_delayed_fput(), but that's a big hammer as other unrelated files could be involved, and from other tasks as well. Add two io_uring helpers to temporarily clear PF_NO_TASKWORK for the worker threads, and run any queued task_work before setting the flag again. Then we can ensure we only flush related items that received their final fput as part of work cancelation and flushing. For now these are io_uring private, but could obviously be made generically available, should there be a need to do so. Signed-off-by: Jens Axboe <axboe@kernel.dk> --- io_uring/io_uring.c | 21 +++++++++++++++++++++ 1 file changed, 21 insertions(+)