mm/memory-failure.c-failure: send right signal code to correct thread
authorTony Luck <tony.luck@intel.com>
Wed, 4 Jun 2014 23:10:59 +0000 (16:10 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 4 Jun 2014 23:54:13 +0000 (16:54 -0700)
commita70ffcac741d31a406c1d2b832ae43d658e7e1cf
tree656f7515976b7e5647918ba80c3b2df7a45ad9d0
parentd2f3102838d90ed6ed09a6154bdb2306f7cf1548
mm/memory-failure.c-failure: send right signal code to correct thread

When a thread in a multi-threaded application hits a machine check because
of an uncorrectable error in memory - we want to send the SIGBUS with
si.si_code = BUS_MCEERR_AR to that thread.  Currently we fail to do that
if the active thread is not the primary thread in the process.
collect_procs() just finds primary threads and this test:

if ((flags & MF_ACTION_REQUIRED) && t == current) {

will see that the thread we found isn't the current thread and so send a
si.si_code = BUS_MCEERR_AO to the primary (and nothing to the active
thread at this time).

We can fix this by checking whether "current" shares the same mm with the
process that collect_procs() said owned the page.  If so, we send the
SIGBUS to current (with code BUS_MCEERR_AR).

Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Reported-by: Otto Bruggeman <otto.g.bruggeman@intel.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: Chen Gong <gong.chen@linux.jf.intel.com>
Cc: <stable@vger.kernel.org> [3.2+]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/memory-failure.c