From: Oleg Nesterov <oleg@redhat.com>
To: Linus Torvalds <torvalds@linux-foundation.org>,
Nick Piggin <npiggin@gmail.com>,
Peter Zijlstra <peterz@infradead.org>,
Mel Gorman <mgorman@techsingularity.net>, Jan Kara <jack@suse.cz>,
Davidlohr Bueso <dave@stgolabs.net>,
Andi Kleen <ak@linux.intel.com>
Cc: Lukas Czerner <lczerner@redhat.com>, linux-kernel@vger.kernel.org
Subject: wait_on_page_bit_common(TASK_KILLABLE, EXCLUSIVE) can miss wakeup?
Date: Wed, 24 Jun 2020 18:11:42 +0200 [thread overview]
Message-ID: <20200624161142.GA12184@redhat.com> (raw)
Suppose that 2 threads T1 and T2 call __lock_page_killable() and sleep in
wait_on_page_bit_common() -> io_schedule().
T1 is killed, it does test_and_set_bit_lock() but the page is still locked.
unlock_page() calls __wake_up_common(nr_exclusive = 1), this wakes T1 up.
T2 is not woken.
T1 checks signal_pending_state() and returns EINTR.
T2 will sleep until another thread does lock/unlock ?
----------------------------------------------------------------------------
I noticed this by accident, I am hunting for another / unrelated bug. I did
git-blame and iiuc the commit a8b169afbf06a ("Avoid page waitqueue race leaving
possible page locker waiting") tried to fix the problem but see above, I don't
understand how can it help.
Don't we need something like below or I am totally confused?
Oleg.
--- x/mm/filemap.c
+++ x/mm/filemap.c
@@ -1131,14 +1131,23 @@ static inline int wait_on_page_bit_commo
wait_page.bit_nr = bit_nr;
for (;;) {
+ int intr = 0;
+
spin_lock_irq(&q->lock);
- if (likely(list_empty(&wait->entry))) {
- __add_wait_queue_entry_tail(q, wait);
- SetPageWaiters(page);
- }
+ // see the comment prepare_to_wait_event()
+ if (signal_pending_state(state, current)) {
+ list_del_init(&wait->entry);
+ intr = 1;
+ } else {
+ if (likely(list_empty(&wait->entry))) {
+ // HMM. head/tail depending on EXCLUSIVE ???
+ __add_wait_queue_entry_tail(q, wait);
+ SetPageWaiters(page);
+ }
- set_current_state(state);
+ set_current_state(state);
+ }
spin_unlock_irq(&q->lock);
@@ -1146,7 +1155,7 @@ static inline int wait_on_page_bit_commo
if (behavior == DROP)
put_page(page);
- if (likely(bit_is_set))
+ if (!intr && likely(bit_is_set))
io_schedule();
if (behavior == EXCLUSIVE) {
@@ -1157,7 +1166,7 @@ static inline int wait_on_page_bit_commo
break;
}
- if (signal_pending_state(state, current)) {
+ if (intr) {
ret = -EINTR;
break;
}
next reply other threads:[~2020-06-24 16:11 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-24 16:11 Oleg Nesterov [this message]
2020-06-24 16:20 ` wait_on_page_bit_common(TASK_KILLABLE, EXCLUSIVE) can miss wakeup? Oleg Nesterov
2020-06-24 16:36 ` Linus Torvalds
2020-06-26 15:43 ` Peter Zijlstra
2020-06-28 5:39 ` Linus Torvalds
2020-06-28 13:18 ` Peter Zijlstra
2020-06-29 3:28 ` Nicholas Piggin
2020-06-29 13:16 ` Nicholas Piggin
2020-06-29 16:36 ` Linus Torvalds
2020-06-30 2:12 ` Nicholas Piggin
2020-06-29 14:02 ` Oleg Nesterov
2020-06-30 2:08 ` Nicholas Piggin
2020-06-30 6:17 ` Oleg Nesterov
2020-06-30 9:08 ` Nicholas Piggin
2020-06-30 10:53 ` Oleg Nesterov
2020-06-30 11:36 ` Oleg Nesterov
2020-06-30 11:50 ` Oleg Nesterov
2020-06-30 18:02 ` Linus Torvalds
2020-06-30 18:29 ` Oleg Nesterov
2020-06-30 18:57 ` Linus Torvalds
2020-06-29 15:13 ` Oleg Nesterov
2020-06-24 16:22 ` Linus Torvalds
2020-06-24 16:43 ` Oleg Nesterov
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=20200624161142.GA12184@redhat.com \
--to=oleg@redhat.com \
--cc=ak@linux.intel.com \
--cc=dave@stgolabs.net \
--cc=jack@suse.cz \
--cc=lczerner@redhat.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mgorman@techsingularity.net \
--cc=npiggin@gmail.com \
--cc=peterz@infradead.org \
--cc=torvalds@linux-foundation.org \
/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;
as well as URLs for NNTP newsgroup(s).