[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 12 Feb 2023 21:23:21 +0300
From: Alexey Izbyshev <izbyshev@...ras.ru>
To: musl@...ts.openwall.com
Subject: Re: [PATCH] mq_notify: fix close/recv race on failure path
On 2023-02-12 03:32, Rich Felker wrote:
diff --git a/src/mq/mq_notify.c b/src/mq/mq_notify.c
index a42888d2..8eac71ed 100644
--- a/src/mq/mq_notify.c
+++ b/src/mq/mq_notify.c
@@ -10,6 +10,8 @@
struct args {
sem_t sem;
int sock;
+ mqd_t mqd;
+ int err;
const struct sigevent *sev;
};
@@ -21,8 +23,21 @@ static void *start(void *p)
int s = args->sock;
void (*func)(union sigval) = args->sev->sigev_notify_function;
union sigval val = args->sev->sigev_value;
+ struct sigevent sev2;
+ static const char zeros[32];
+ int err = 0;
+ sev2.sigev_notify = SIGEV_THREAD;
+ sev2.sigev_signo = s;
+ sev2.sigev_value.sival_ptr = (void *)&zeros;
+
+ err = 0;
This assignment is redundant.
Maybe this hunk could be simplified by getting rid of err and simply
doing "args->err = -__syscall(SYS_mq_notify, args->mqd, &sev2)".
Except for this nit, the patches look good to me, thanks!
Alexey
+ if (syscall(SYS_mq_notify, args->mqd, &sev2) < 0)
+ err = errno;
+ args->err = err;
sem_post(&args->sem);
+ if (err) return 0;
+
n = recv(s, buf, sizeof(buf), MSG_NOSIGNAL|MSG_WAITALL);
close(s);
if (n==sizeof buf && buf[sizeof buf - 1] == 1)
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
