Make sure SIGKILL gets proper respect
authorLinus Torvalds <torvalds@g5.osdl.org>
Fri, 23 Sep 2005 20:22:21 +0000 (13:22 -0700)
committerLinus Torvalds <torvalds@g5.osdl.org>
Fri, 23 Sep 2005 20:22:21 +0000 (13:22 -0700)
commit188a1eafa03aaa5e5fe6f53e637e704cd2c31c7c
tree33d9480c792724c10e4e6fb681fc69c477cf1ff0
parentede1327ea4ca8019ec6df24b3e837def091c26b8
Make sure SIGKILL gets proper respect

Bhavesh P. Davda <bhavesh@avaya.com> noticed that SIGKILL wouldn't
properly kill a process under just the right cicumstances: a stopped
task that already had another signal queued would get the SIGKILL
queued onto the shared queue, and there it would remain until SIGCONT.

This simplifies the signal acceptance logic, and fixes the bug in the
process.

Losely based on an earlier patch by Bhavesh.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
kernel/signal.c