mm: Place preemption point in do_mlockall() loop
authorPaul E. McKenney <paulmck@linux.vnet.ibm.com>
Wed, 25 Sep 2013 01:29:11 +0000 (18:29 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 25 Sep 2013 02:44:40 +0000 (19:44 -0700)
commit22356f447ceb8d97a4885792e7d9e4607f712e1b
tree5d19a3ce69f94a92e0e303054f9ed44a826b8b19
parenta153e67bda3639a46edac6205610ae63c0fdea4c
mm: Place preemption point in do_mlockall() loop

There is a loop in do_mlockall() that lacks a preemption point, which
means that the following can happen on non-preemptible builds of the
kernel. Dave Jones reports:

 "My fuzz tester keeps hitting this.  Every instance shows the non-irq
  stack came in from mlockall.  I'm only seeing this on one box, but
  that has more ram (8gb) than my other machines, which might explain
  it.

    INFO: rcu_preempt self-detected stall on CPU { 3}  (t=6500 jiffies g=470344 c=470343 q=0)
    sending NMI to all CPUs:
    NMI backtrace for cpu 3
    CPU: 3 PID: 29664 Comm: trinity-child2 Not tainted 3.11.0-rc1+ #32
    Call Trace:
      lru_add_drain_all+0x15/0x20
      SyS_mlockall+0xa5/0x1a0
      tracesys+0xdd/0xe2"

This commit addresses this problem by inserting the required preemption
point.

Reported-by: Dave Jones <davej@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: KOSAKI Motohiro <kosaki.motohiro@gmail.com>
Cc: Michel Lespinasse <walken@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/mlock.c