Relax si_code check in rt_sigqueueinfo and rt_tgsigqueueinfo
authorRoland Dreier <roland@purestorage.com>
Mon, 28 Mar 2011 21:13:35 +0000 (14:13 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 28 Mar 2011 22:45:44 +0000 (15:45 -0700)
commit243b422af9ea9af4ead07a8ad54c90d4f9b6081a
treebdcd77b51cce8bd2ed5fcc41c6b430a8d6a9d74f
parent5847098cd896c92819800e17e983bf6530035209
Relax si_code check in rt_sigqueueinfo and rt_tgsigqueueinfo

Commit da48524eb206 ("Prevent rt_sigqueueinfo and rt_tgsigqueueinfo
from spoofing the signal code") made the check on si_code too strict.
There are several legitimate places where glibc wants to queue a
negative si_code different from SI_QUEUE:

 - This was first noticed with glibc's aio implementation, which wants
   to queue a signal with si_code SI_ASYNCIO; the current kernel
   causes glibc's tst-aio4 test to fail because rt_sigqueueinfo()
   fails with EPERM.

 - Further examination of the glibc source shows that getaddrinfo_a()
   wants to use SI_ASYNCNL (which the kernel does not even define).
   The timer_create() fallback code wants to queue signals with SI_TIMER.

As suggested by Oleg Nesterov <oleg@redhat.com>, loosen the check to
forbid only the problematic SI_TKILL case.

Reported-by: Klaus Dittrich <kladit@arcor.de>
Acked-by: Julien Tinnes <jln@google.com>
Cc: <stable@kernel.org>
Signed-off-by: Roland Dreier <roland@purestorage.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
kernel/signal.c