Skip to contentSkip to conversation

brown paper bag fix for SIGPOLL si_codes

GIT PULL Eric W. Biederman 2 participants 3 messages Sep 19, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Linus,

Please pull the for-linus branch from the git tree:

git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace.git for-linus

HEAD: 54640d238760a1a54dfebe039b49682522100186 fcntl: Don't set si_code to SI_SIGIO when sig == SIGPOLL

When sorting out the si_code ambiguity fcntl I accidentally overshot and included SIGPOLL as well. Ooops! This is my trivial fix for that.

Vince Weaver caught this when it landed in your tree with his perf_event_tests many of which started failing because the si_code changed.

From: "Eric W. Biederman" <[email protected]>
Date: Mon, 18 Sep 2017 22:51:14 -0500
Subject: [PATCH] fcntl: Don't set si_code to SI_SIGIO when sig == SIGPOLL

When fixing things to avoid ambiguous cases I had a thinko and included SIGPOLL/SIGIO in with all of the other signals that have signal specific si_codes. Which is completely wrong.

Fix that.

---
 fs/fcntl.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/fcntl.c +1 -1
index 0491da3b28c3..448a1119f0be 100644
@@ -749,7 +749,7 @@ static void send_sigio_to_task(struct task_struct *p,
 			 * specific si_codes.  In that case use SI_SIGIO instead
 			 * to remove the ambiguity.
 			 */
-			if (sig_specific_sicodes(signum))
+			if ((signum != SIGPOLL) && sig_specific_sicodes(signum))
 				si.si_code = SI_SIGIO;
 
 			/* Make sure we are called with one of the POLL_*
Signature · 1 line
2.14.1
#1 lore raw
Quoted 5 lines · Eric W. Biederman

When sorting out the si_code ambiguity fcntl I accidentally overshot and included SIGPOLL as well. Ooops! This is my trivial fix for that.

Vince Weaver caught this when it landed in your tree with his perf_event_tests many of which started failing because the si_code changed.

I've tested with this patch applied and can confirm all of my tests now pass again.

Thanks,

Vince

#2 lore raw
Quoted 10 lines · Vince Weaver
Quoted 5 lines · Eric W. Biederman

When sorting out the si_code ambiguity fcntl I accidentally overshot and included SIGPOLL as well. Ooops! This is my trivial fix for that.

Vince Weaver caught this when it landed in your tree with his perf_event_tests many of which started failing because the si_code changed.

I've tested with this patch applied and can confirm all of my tests now pass again.

And in my tiredness and distractedness doing this on my honeymoon I forgot to include:

Fixes d08477aa975e ("fcntl: Don't use ambiguous SIG_POLL si_codes")

Not that there are any backport issues since all of this is new with -rc1. But just in case someone needs the information for double checking me.

Eric