| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Pull io_uring update from Jens Axboe:
"On top of the usual cleanups and fixes, the bigger items in here are:
- zcrx work, most of it centered around adding dynamic area
provisioning, plus a bunch of prep and cleanups leading up to it:
scale refilling with large pages, coalesce same-niov RQEs on
refill, separate the RQ head/tail cache lines and cache the RQ
tail, and rework the area creation locking.
- Fix the futex inflight accounting so that only private futex waits
are marked inflight, and don't mark wake requests as inflight at
all.
- Drop the custom iov copy in the buffer select prep and msg header
copy paths, using the generic helpers instead.
- Fix a folio size overflow in io_vec_fill_bvec(), and account the
pages a compound region really uses in the memmap path.
- Fix an iovec leak in uring_cmd when the async cmd isn't recycled,
skip the blocking task work for io_uring_cmd_issue_blocking(), and
don't skip completion for a synchronous multishot cmd
- Defer eventfd signaling when queued from a wakeup handler
- Fix io-wq worker accounting when canceling creation callbacks
- Annotate remote tasks for kcoverage"
* tag 'for-7.3/io_uring-20260819' of git://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux: (30 commits)
io_uring: Add missing include for ITER_SOURCE and ITER_DEST
io_uring/uring_cmd: don't skip completion for a synchronous multishot cmd
io_uring/memmap: account the pages a compound region really uses
io_uring/zcrx: add dynamic area provisioning
io_uring/zcrx: lock area creation with pp_lock
io_uring/zcrx: keep array of areas
io_uring/zcrx: move freelist lock to struct zcrx
io_uring/zcrx: unmap under netdev lock
io_uring/zcrx: split dmabuf unmap and release
io_uring/zcrx: don't pass ifq_reg to area creation
io_uring/zcrx: add helper for deriving area token
io_uring/zcrx: don't reload skb_shinfo
io_urint/zcrx: narrow var scope in io_zcrx_recv_skb()
io_uring/zcrx: constify area_reg on import
io_uring/zcrx: coalesce same-niov RQEs on refill
io_uring/zcrx: cache RQ tail
io_uring/zcrx: add RQ iterator
io_uring/zcrx: move RQ head/tail to separate cache lines
io_uring/zcrx: scale refilling with large pages
io_uring/io-wq: fix worker accounting when canceling creation callbacks
...
|
|
io_req_local_work_add() signals the CQ ring eventfd inline when it is the
one to push the first entry onto ->work_list. For DEFER_TASKRUN rings that
add is frequently done from a waitqueue wakeup handler, where an
arbitrary waitqueue lock is held.
eventfd_signal_mask() only refuses to recurse when current->in_eventfd
is set, but that bit is set by eventfd_signal_mask() itself. If the wake
chain starts somewhere else, signal goes out inline and can feed back
into epoll.
Add IOU_F_TWQ_IN_WAKE, set it on the task_work add done from the three
waitqueue callbacks, and use it to force io_eventfd_signal() down the
existing call_rcu_hurry() deferral instead of signaling inline.
Fixes: 21a091b970cd ("io_uring: signal registered eventfd to process deferred task work")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/all/20260813133843.2933127-1-4ncienth@gmail.com/
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
This reverts commit ed64f5c546b3d5e3a4840f6c055448ce90edf56c.
Since commit:
648790e09527 ("io_uring: restore RCU read section in io_req_local_work_add()")
io_ctx_mark_taskrun() is only ever called with the RCU read lock
already held, like previously. Hence's there's no need for this commit
anymore, which grabbed the RCU read lock inside io_ctx_mark_taskrun().
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The task-work refactor that moved io_req_local_work_add() out of
io_uring.c into the new io_uring/tw.c dropped the whole-body
guard(rcu)() that used to cover the function body.
For DEFER_TASKRUN rings the ring teardown still relies on that RCU read
section pairing with its grace period:
/* pairs with RCU read section in io_req_local_work_add() */
if (ctx->flags & IORING_SETUP_DEFER_TASKRUN)
synchronize_rcu();
io_ring_ctx_free(ctx);
io_req_local_work_add() keeps dereferencing ctx after mpscq_push() has
published the request to the work list (ctx->cq_wait_nr, and
ctx->submitter_task in the final wake_up_state()), without holding a ctx
reference across that window. The RCU read section was the only thing
guaranteeing an in-flight adder had finished touching ctx before
io_ring_ctx_free() ran; synchronize_rcu() only waits for readers that
are actually inside an RCU read-side critical section. With the guard
gone the grace period no longer pairs with anything on the add side, so
ctx can be freed and reused while io_req_local_work_add() is still using
it.
Fixes: d46ab2c98aba ("io_uring: switch local task_work to a mpscq")
Signed-off-by: Woraphat Khiaodaeng <worapat.kd2@gmail.com>
Link: https://patch.msgid.link/20260709035100.2269-1-worapat.kd2@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Commit de7341ffe49e ("io_uring: switch normal task_work to a mpscq")
added a use of system_unbound_wq, which is deprecated in favor of
system_dfl_wq added by commit 128ea9f6ccfb ("workqueue: Add
system_percpu_wq and system_dfl_wq"). An upcoming warning in the
workqueue tree flags this with:
workqueue: work func io_tctx_fallback_work enqueued on deprecated workqueue. Use system_{percpu|dfl}_wq instead.
Switch to system_dfl_wq to clear up the warning.
Fixes: de7341ffe49e ("io_uring: switch normal task_work to a mpscq")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Link: https://patch.msgid.link/20260616-io_uring-fix-wq-warning-v1-1-cfc9d934eedb@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The normal task_work path used a tw_pending bit to ensure the callback
was only added once: the mpscq drains incrementally, so a single
tctx_task_work() run can take the queue through empty -> non-empty
several times, and each transition would otherwise re-add the already
pending callback_head. This corrupts the task_work list, and is what
tw_pending protects again.
This can go away, if we stop running the task_work as soon as the queue
empties.
Suggested-by: Caleb Sander Mateos <csander@purestorage.com>
Reviewed-by: Caleb Sander Mateos <csander@purestorage.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
With the tctx fallback running its entries directly, the per-ctx
fallback work has a single user left: moving local (DEFER_TASKRUN)
task_work entries out of a ring that is going away. Both of its call
sites are process context and don't hold ->uring_lock, the same
conditions the deferred fallback work itself ran under - so run the
entries in cancel mode right there instead, and rename the helper to
io_cancel_local_task_work() to match what it now does.
With that, ->fallback_llist, ->fallback_work, io_fallback_req_func()
and __io_fallback_tw() can all go away, along with the fallback work
flushing in the ring exit and cancel paths. Requests that get
orphaned by an exiting task now run via the tctx fallback work, which
the ring exit side implicitly waits on through the ctx refs those
requests hold.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The fallback work drains the tctx queue only to redistribute the entries
into the per-ctx fallback lists, bouncing them through a second
(per-ctx) work item before they finally run. That made sense when the
producer side did the draining and could be in any context, but the
fallback work is a regular process context kworker: it can just run the
entries itself. Reuse the normal run loop - if run from the fallback
kernel thread, ts.cancel will get set, and the work terminated.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Like the local task_work list, the normal (tctx) task_work list is an
llist, and hence needs the O(n) llist_reverse_order() pass before
running entries in queue order. On top of that, capped runs - sqpoll
processing IORING_TW_CAP_ENTRIES_VALUE entries at a time - need the
claimed-but-unprocessed leftovers carried in a separate retry_list,
as they can't be pushed back to the shared list.
Switch tctx->task_list to a mpscq, like what was done for the
DEFER_TASKRUN paths as well.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The local (DEFER_TASKRUN) task_work list is an llist, which is LIFO
ordered, and hence __io_run_local_work() has to restore the right
running order with an O(n) llist_reverse_order() pass first. On top of
that, a batch that gets capped by max_events needs the leftover entries
parked on a separate ->retry_llist, as they can't be pushed back to the
shared list.
Switch it to the FIFO mpscq. Adds are wait-free instead of a cmpxchg
retry loop, entries are popped in queue order with no reversal pass,
capping a run simply leaves the remainder on the queue, and
->retry_llist goes away entirely. The consumer cursor, ->work_head,
lives with the rest of the ->uring_lock protected state rather than
next to the queue, so that popping entries doesn't dirty the producer
side cacheline.
For low amounts of task_work, this ends up being a bit more efficient
than the existing scheme. As an example of that, doing multishot
receives for 8 clients has the following task_work overhead:
1.02% sock-test [kernel.kallsyms] [k] io_req_local_work_add
0.88% sock-test [kernel.kallsyms] [k] __io_run_local_work_loop
0.60% sock-test [kernel.kallsyms] [k] llist_reverse_order
0.14% sock-test [kernel.kallsyms] [k] __io_run_local_work
2.64% at ~46Gb/sec
and after this change:
1.08% sock-test [kernel.kallsyms] [k] io_req_local_work_add
1.03% sock-test [kernel.kallsyms] [k] __io_run_local_work
2.11% at ~53Gb/sec
which has less overhead even though that test run was faster. For a case
of having 1024 clients on a single ring:
2.22% sock-test [kernel.kallsyms] [k] llist_reverse_order
0.84% sock-test [kernel.kallsyms] [k] __io_run_local_work_loop
0.42% sock-test [kernel.kallsyms] [k] io_req_local_work_add
0.02% sock-test [kernel.kallsyms] [k] __io_run_local_work
3.50% at ~24Gb/sec
we start to see the llist reversing taking a considerable amount of
time, and the total add+run task_work overhead is around 3.5%. After
the change:
0.90% sock-test [kernel.kallsyms] [k] __io_run_local_work
0.42% sock-test [kernel.kallsyms] [k] io_req_local_work_add
1.32% at ~26Gb/sec
most of that overhead is gone, and performance is better as well.
Caleb Sander Mateos <csander@purestorage.com> reports that it improves
the performance of a ublk 4kb workload by 4% [1], while testing v1 of
this patchset.
[1] https://lore.kernel.org/io-uring/CADUfDZr-MMYBaP-e+y9+xuRhuiunO2sBTUCmwZyd7AgT8sVtiQ@mail.gmail.com/
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Not required right now, as io_req_local_work_add() already calls this
helper with the RCU read lock held. But in preparation for that not
being the case, grab it locally.
Reviewed-by: Caleb Sander Mateos <csander@purestorage.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
The DEFER_TASKRUN local task work paths all run under ctx->uring_lock,
which serializes them with each other and with the rest of the ring's
hot paths. io_move_task_work_from_local() is the exception - it's called
from io_ring_exit_work() on a kworker without holding the lock and from
the iopoll cancelation side right after dropping it.
->work_llist is fine with this, as it's only ever updated via the
expected paths. But the ->retry_llist is updated while runing, and hence
it could potentially race between normal task_work running and the
task-has-exited shutdown path.
Simply grab ->uring_lock while moving the local work to the fallback
list for exit purposes, which nicely serializes it across both the
normal additions and the exit prune path.
Cc: stable@vger.kernel.org
Fixes: f46b9cdb22f7 ("io_uring: limit local tw done")
Reported-by: Robert Femmer <robert.femmer@x41-dsec.de>
Reported-by: Christian Reitter <invd@inhq.net>
Reported-by: Michael Rodler <michael.rodler@x41-dsec.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
There are a few of these, where flags are read outside of the
uring_lock, yet it's harmless to race on them.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Bitfields cannot be set and checked atomically, and this makes it more
clear that these are indeed in shared storage and must be checked and
set in a sane fashion. This is in preparation for annotating a few of
the known racy, but harmless, flags checking.
No intended functional changes in this patch.
Reviewed-by: Gabriel Krisman Bertazi <krisman@suse.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
If DEFER_TASKRUN | SETUP_TASKRUN is used and task work is added while
the ring is being resized, it's possible for the OR'ing of
IORING_SQ_TASKRUN to happen in the small window of swapping into the
new rings and the old rings being freed.
Prevent this by adding a 2nd ->rings pointer, ->rings_rcu, which is
protected by RCU. The task work flags manipulation is inside RCU
already, and if the resize ring freeing is done post an RCU synchronize,
then there's no need to add locking to the fast path of task work
additions.
Note: this is only done for DEFER_TASKRUN, as that's the only setup mode
that supports ring resizing. If this ever changes, then they too need to
use the io_ctx_mark_taskrun() helper.
Link: https://lore.kernel.org/io-uring/20260309062759.482210-1-naup96721@gmail.com/
Cc: stable@vger.kernel.org
Fixes: 79cfe9e59c2a ("io_uring/register: add IORING_REGISTER_RESIZE_RINGS")
Reported-by: Hao-Yu Yang <naup96721@gmail.com>
Suggested-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Move the completion queue waiting and scheduling code out of io_uring.c
into a dedicated wait.c file. This further removes code out of the
main io_uring C and header file, and into a topical new file.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
|
Move the task work handling code out of io_uring.c into a new tw.c file.
This includes the local work, normal work, and fallback work handling
infrastructure.
The associated tw.h header contains io_should_terminate_tw() as a static
inline helper, along with the necessary function declarations.
Signed-off-by: Jens Axboe <axboe@kernel.dk>
|
Linus Torvalds
Jens Axboe
Woraphat Khiaodaeng
Nathan Chancellor