[Qemu-devel] [PATCH] cpus.c: Use pthread_sigmask() rather than sigprocmask()

[Qemu-devel] [PATCH] cpus.c: Use pthread_sigmask() rather than sigprocmask()

[Peter Maydell]

On Linux, sigprocmask() and pthread_sigmask() are in practice the
same thing (they only set the signal mask for the calling thread),
but the documentation states that the behaviour of sigprocmask() in a
multithreaded process is undefined. Use pthread_sigmask() instead
(which is what we do in almost all places in QEMU that alter the
signal mask already).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 cpus.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/cpus.c b/cpus.c
index cbeb1f6..1d0fc53 100644
--- a/cpus.c
+++ b/cpus.c
@@ -778,7 +778,7 @@ static void sigbus_reraise(void)
         raise(SIGBUS);
         sigemptyset(&set);
         sigaddset(&set, SIGBUS);
-        sigprocmask(SIG_UNBLOCK, &set, NULL);
+        pthread_sigmask(SIG_UNBLOCK, &set, NULL);
     }
     perror("Failed to re-raise SIGBUS!\n");
     abort();
-- 
1.9.1

Re: [Qemu-devel] [PATCH] cpus.c: Use pthread_sigmask() rather than sigprocmask()

[Peter Maydell]

On 16 May 2016 at 18:33, Peter Maydell <peter.maydell@linaro.org> wrote:
> On Linux, sigprocmask() and pthread_sigmask() are in practice the
> same thing (they only set the signal mask for the calling thread),
> but the documentation states that the behaviour of sigprocmask() in a
> multithreaded process is undefined. Use pthread_sigmask() instead
> (which is what we do in almost all places in QEMU that alter the
> signal mask already).

The only other sigprocmask() uses are in linux-user, apart from
a couple in net/tap.c, where they're used as part of forking and
spawning the helper process. I suspect this should be using
qemu_fork() instead of doing it all by hand, wrongly...

thanks
-- PMM

Re: [Qemu-devel] [PATCH] cpus.c: Use pthread_sigmask() rather than sigprocmask()

[Paolo Bonzini]

On 16/05/2016 19:37, Peter Maydell wrote:
> The only other sigprocmask() uses are in linux-user, apart from
> a couple in net/tap.c, where they're used as part of forking and
> spawning the helper process. I suspect this should be using
> qemu_fork() instead of doing it all by hand, wrongly...

In all fairness those predate qemu_fork(). :)

Your patch also changes a Linux-only function, but it's cleaner to user
pthread_sigmask so I've queued it for the next pull request.

Paolo