]> git.baikalelectronics.ru Git - kernel.git/commit
mm: do not rely on preempt_count in print_vma_addr
authorMichal Hocko <mhocko@suse.com>
Thu, 16 Nov 2017 01:38:59 +0000 (17:38 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Thu, 16 Nov 2017 02:21:07 +0000 (18:21 -0800)
commit0a7f682d04652fd91de0cbc1ea3d7aa1e45611c7
tree8b66408158a876f9854166795e124fb407d10ce4
parentfcdaf842bd8f538a88059ce0243bc2822ed1b0e0
mm: do not rely on preempt_count in print_vma_addr

The preempt count check on print_vma_addr has been added by commit
e8bff74afbdb ("x86: fix "BUG: sleeping function called from invalid
context" in print_vma_addr()") and it relied on the elevated preempt
count from preempt_conditional_sti because preempt_count check doesn't
work on non preemptive kernels by default.

The code has evolved though and commit d99e1bd175f4 ("x86/entry/traps:
Refactor preemption and interrupt flag handling") has replaced
preempt_conditional_sti by an explicit preempt_disable which is noop on
!PREEMPT so the check in print_vma_addr is broken.

Fix the issue by using trylock on mmap_sem rather than chacking the
preempt count.  The allocation we are relying on has to be GFP_NOWAIT as
well.  There is a chance that we won't dump the vma state if the lock is
contended or the memory short but this is acceptable outcome and much
less fragile than the not working preemption check or tricks around it.

Link: http://lkml.kernel.org/r/20171106134031.g6dbelg55mrbyc6i@dhcp22.suse.cz
Fixes: d99e1bd175f4 ("x86/entry/traps: Refactor preemption and interrupt flag handling")
Signed-off-by: Michal Hocko <mhocko@suse.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Acked-by: Yang Shi <yang.s@alibaba-inc.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/memory.c