[alsa-devel] [PATCH] mixer: Fix memory leak for more than 16 file descriptor case
--- I saw this when digging through the source code, wrote up a test case, and verified with valgrind. I don't have that many sound cards so I modified mixer.c to always take the conditional. alloca seemed like the least effort way to fix this. Other thoughts were to test pfds != spfds and free, but given how little bytes would be allocated putting on the stack seems reasonable.
I also created the github pull request 20. https://github.com/alsa-project/alsa-lib/pull/20
src/mixer/mixer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/mixer/mixer.c b/src/mixer/mixer.c index 82056474..b1af9945 100644 --- a/src/mixer/mixer.c +++ b/src/mixer/mixer.c @@ -767,7 +767,7 @@ int snd_mixer_wait(snd_mixer_t *mixer, int timeout) if (count < 0) return count; if ((unsigned int) count > sizeof(spfds) / sizeof(spfds[0])) { - pfds = malloc(count * sizeof(*pfds)); + pfds = alloca(count * sizeof(*pfds)); if (!pfds) return -ENOMEM; err = snd_mixer_poll_descriptors(mixer, pfds,
On Mon, 06 Jan 2020 22:41:29 +0100, David Fries wrote:
I saw this when digging through the source code, wrote up a test case, and verified with valgrind. I don't have that many sound cards so I modified mixer.c to always take the conditional. alloca seemed like the least effort way to fix this. Other thoughts were to test pfds != spfds and free, but given how little bytes would be allocated putting on the stack seems reasonable.
Looks good to me.
Could you give your sign-off? It's not strictly needed for alsa-lib, but we prefer having it.
thanks,
Takashi
I also created the github pull request 20. https://github.com/alsa-project/alsa-lib/pull/20
src/mixer/mixer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/mixer/mixer.c b/src/mixer/mixer.c index 82056474..b1af9945 100644 --- a/src/mixer/mixer.c +++ b/src/mixer/mixer.c @@ -767,7 +767,7 @@ int snd_mixer_wait(snd_mixer_t *mixer, int timeout) if (count < 0) return count; if ((unsigned int) count > sizeof(spfds) / sizeof(spfds[0])) {
pfds = malloc(count * sizeof(*pfds));
pfds = alloca(count * sizeof(*pfds));-- 2.11.0
Alsa-devel mailing list Alsa-devel@alsa-project.org https://mailman.alsa-project.org/mailman/listinfo/alsa-devel
I realized that after I sent it, and I wasn't sure if that was going to be the solution anyway. I sent another patch and added you as reviewed by if that's okay.
On Tue, Jan 07, 2020 at 02:03:48PM +0100, Takashi Iwai wrote:
On Mon, 06 Jan 2020 22:41:29 +0100, David Fries wrote:
I saw this when digging through the source code, wrote up a test case, and verified with valgrind. I don't have that many sound cards so I modified mixer.c to always take the conditional. alloca seemed like the least effort way to fix this. Other thoughts were to test pfds != spfds and free, but given how little bytes would be allocated putting on the stack seems reasonable.
Looks good to me.
Could you give your sign-off? It's not strictly needed for alsa-lib, but we prefer having it.
thanks,
Takashi
I also created the github pull request 20. https://github.com/alsa-project/alsa-lib/pull/20
src/mixer/mixer.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/mixer/mixer.c b/src/mixer/mixer.c index 82056474..b1af9945 100644 --- a/src/mixer/mixer.c +++ b/src/mixer/mixer.c @@ -767,7 +767,7 @@ int snd_mixer_wait(snd_mixer_t *mixer, int timeout) if (count < 0) return count; if ((unsigned int) count > sizeof(spfds) / sizeof(spfds[0])) {
pfds = malloc(count * sizeof(*pfds));
pfds = alloca(count * sizeof(*pfds));-- 2.11.0
Alsa-devel mailing list Alsa-devel@alsa-project.org https://mailman.alsa-project.org/mailman/listinfo/alsa-devel
participants (2)
-
David Fries -
Takashi Iwai