queue-4.19/bpf-fix-preempt_enable_no_resched-abuse.patch

   1 From df6f03a6d7b7bff9a695fed7530c1bed0ec29611 Mon Sep 17 00:00:00 2001
   2 From: Peter Zijlstra <peterz@infradead.org>
   3 Date: Tue, 23 Apr 2019 21:55:59 +0200
   4 Subject: bpf: Fix preempt_enable_no_resched() abuse
   5
   6 [ Upstream commit 0edd6b64d1939e9e9168ff27947995bb7751db5d ]
   7
   8 Unless the very next line is schedule(), or implies it, one must not use
   9 preempt_enable_no_resched(). It can cause a preemption to go missing and
  10 thereby cause arbitrary delays, breaking the PREEMPT=y invariant.
  11
  12 Cc: Roman Gushchin <guro@fb.com>
  13 Cc: Alexei Starovoitov <ast@kernel.org>
  14 Cc: Daniel Borkmann <daniel@iogearbox.net>
  15 Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
  16 Signed-off-by: Alexei Starovoitov <ast@kernel.org>
  17 Signed-off-by: Sasha Levin <sashal@kernel.org>
  18 ---
  19  include/linux/bpf.h | 2 +-
  20  1 file changed, 1 insertion(+), 1 deletion(-)
  21
  22 diff --git a/include/linux/bpf.h b/include/linux/bpf.h
  23 index 523481a3471b5..fd95f2efe5f32 100644
  24 --- a/include/linux/bpf.h
  25 +++ b/include/linux/bpf.h
  26 @@ -400,7 +400,7 @@ int bpf_prog_array_copy(struct bpf_prog_array __rcu *old_array,
  27                 }                                       \
  28  _out:                                                  \
  29                 rcu_read_unlock();                      \
  30 -               preempt_enable_no_resched();            \
  31 +               preempt_enable();                       \
  32                 _ret;                                   \
  33          })
  34
  35 --
  36 2.20.1
  37