md: don't use flush_signals in userspace processes
authorMikulas Patocka <mpatocka@redhat.com>
Wed, 7 Jun 2017 23:05:31 +0000 (19:05 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 27 Jul 2017 22:03:26 +0000 (15:03 -0700)
commit94e0ff076aacf541ef392a2b0aa493ed71f48614
tree44155b5286f833262e811064210698cceacd030f
parent936c77c8dbca5a534617fb6a03baedb3c9535d1f
md: don't use flush_signals in userspace processes

commit f9c79bc05a2a91f4fba8bfd653579e066714b1ec upstream.

The function flush_signals clears all pending signals for the process. It
may be used by kernel threads when we need to prepare a kernel thread for
responding to signals. However using this function for an userspaces
processes is incorrect - clearing signals without the program expecting it
can cause misbehavior.

The raid1 and raid5 code uses flush_signals in its request routine because
it wants to prepare for an interruptible wait. This patch drops
flush_signals and uses sigprocmask instead to block all signals (including
SIGKILL) around the schedule() call. The signals are not lost, but the
schedule() call won't respond to them.

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Acked-by: NeilBrown <neilb@suse.com>
Signed-off-by: Shaohua Li <shli@fb.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/md/raid1.c
drivers/md/raid5.c