From: Pavel Begunkov <[email protected]>
To: Jens Axboe <[email protected]>, [email protected]
Subject: [PATCH 2/2] io_uring: fix complete_post races for linked req
Date: Tue, 9 Mar 2021 00:37:59 +0000 [thread overview]
Message-ID: <5672a62f3150ee7c55849f40c0037655c4f2840f.1615250156.git.asml.silence@gmail.com> (raw)
In-Reply-To: <[email protected]>
Calling io_queue_next() after spin_unlock in io_req_complete_post()
races with the other side extracting and reusing this request. Hand
coded parts of io_req_find_next() considering that io_disarm_next()
and io_req_task_queue() have (and safe) to be called with
completion_lock held.
It already does io_commit_cqring() and io_cqring_ev_posted(), so just
reuse it for post io_disarm_next().
Signed-off-by: Pavel Begunkov <[email protected]>
---
fs/io_uring.c | 21 ++++++++++++++-------
1 file changed, 14 insertions(+), 7 deletions(-)
diff --git a/fs/io_uring.c b/fs/io_uring.c
index b4fa6fb371c5..87a71fda5745 100644
--- a/fs/io_uring.c
+++ b/fs/io_uring.c
@@ -986,6 +986,7 @@ static const struct io_op_def io_op_defs[] = {
[IORING_OP_UNLINKAT] = {},
};
+static bool io_disarm_next(struct io_kiocb *req);
static void io_uring_del_task_file(unsigned long index);
static void io_uring_try_cancel_requests(struct io_ring_ctx *ctx,
struct task_struct *task,
@@ -1526,15 +1527,14 @@ static void io_cqring_fill_event(struct io_kiocb *req, long res)
__io_cqring_fill_event(req, res, 0);
}
-static inline void io_req_complete_post(struct io_kiocb *req, long res,
- unsigned int cflags)
+static void io_req_complete_post(struct io_kiocb *req, long res,
+ unsigned int cflags)
{
struct io_ring_ctx *ctx = req->ctx;
unsigned long flags;
spin_lock_irqsave(&ctx->completion_lock, flags);
__io_cqring_fill_event(req, res, cflags);
- io_commit_cqring(ctx);
/*
* If we're the last reference to this request, add to our locked
* free_list cache.
@@ -1542,19 +1542,26 @@ static inline void io_req_complete_post(struct io_kiocb *req, long res,
if (refcount_dec_and_test(&req->refs)) {
struct io_comp_state *cs = &ctx->submit_state.comp;
+ if (req->flags & (REQ_F_LINK | REQ_F_HARDLINK)) {
+ if (req->flags & (REQ_F_LINK_TIMEOUT | REQ_F_FAIL_LINK))
+ io_disarm_next(req);
+ if (req->link) {
+ io_req_task_queue(req->link);
+ req->link = NULL;
+ }
+ }
io_dismantle_req(req);
io_put_task(req->task, 1);
list_add(&req->compl.list, &cs->locked_free_list);
cs->locked_free_nr++;
} else
req = NULL;
+ io_commit_cqring(ctx);
spin_unlock_irqrestore(&ctx->completion_lock, flags);
-
io_cqring_ev_posted(ctx);
- if (req) {
- io_queue_next(req);
+
+ if (req)
percpu_ref_put(&ctx->refs);
- }
}
static void io_req_complete_state(struct io_kiocb *req, long res,
--
2.24.0
next prev parent reply other threads:[~2021-03-09 0:43 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-09 0:37 [PATCH 0/2] io_req_complete_post() fix Pavel Begunkov
2021-03-09 0:37 ` [PATCH 1/2] io_uring: add io_disarm_next() helper Pavel Begunkov
2021-03-09 0:37 ` Pavel Begunkov [this message]
2021-03-09 15:01 ` [PATCH 0/2] io_req_complete_post() fix Jens Axboe
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=5672a62f3150ee7c55849f40c0037655c4f2840f.1615250156.git.asml.silence@gmail.com \
[email protected] \
[email protected] \
[email protected] \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox