diff options
| author | Emilio G. Cota <cota@braap.org> | 2016-05-24 16:06:13 -0400 |
|---|---|---|
| committer | Paolo Bonzini <pbonzini@redhat.com> | 2016-05-29 09:11:11 +0200 |
| commit | c983895258a771f8a5e4a53950bfb7fd2216651c (patch) | |
| tree | b383a83403d84c2d8247143783f3f700772dcaf8 /include | |
| parent | 56ebe02203f033a8399f7f6ea6972225ed87101c (diff) | |
| download | qemu-c983895258a771f8a5e4a53950bfb7fd2216651c.zip qemu-c983895258a771f8a5e4a53950bfb7fd2216651c.tar.gz qemu-c983895258a771f8a5e4a53950bfb7fd2216651c.tar.bz2 | |
atomics: emit an smp_read_barrier_depends() barrier only for Alpha and Thread Sanitizer
For correctness, smp_read_barrier_depends() is only required to
emit a barrier on Alpha hosts. However, we are currently emitting
a consume fence unconditionally, and most compilers currently treat
consume and acquire fences as equivalent.
Fix it by keeping the consume fence if we're compiling with Thread
Sanitizer, since this might help prevent false warnings. Otherwise,
only emit the barrier for Alpha hosts. Note that we still guarantee
that smp_read_barrier_depends() is a compiler barrier.
Signed-off-by: Emilio G. Cota <cota@braap.org>
Message-Id: <1464120374-8950-3-git-send-email-cota@braap.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Diffstat (limited to 'include')
| -rw-r--r-- | include/qemu/atomic.h | 11 |
1 files changed, 11 insertions, 0 deletions
diff --git a/include/qemu/atomic.h b/include/qemu/atomic.h index 5bc4d6c..96db6e9 100644 --- a/include/qemu/atomic.h +++ b/include/qemu/atomic.h @@ -36,7 +36,18 @@ #define smp_wmb() ({ barrier(); __atomic_thread_fence(__ATOMIC_RELEASE); barrier(); }) #define smp_rmb() ({ barrier(); __atomic_thread_fence(__ATOMIC_ACQUIRE); barrier(); }) +/* Most compilers currently treat consume and acquire the same, but really + * no processors except Alpha need a barrier here. Leave it in if + * using Thread Sanitizer to avoid warnings, otherwise optimize it away. + */ +#if defined(__SANITIZE_THREAD__) #define smp_read_barrier_depends() ({ barrier(); __atomic_thread_fence(__ATOMIC_CONSUME); barrier(); }) +#elsif defined(__alpha__) +#define smp_read_barrier_depends() asm volatile("mb":::"memory") +#else +#define smp_read_barrier_depends() barrier() +#endif + /* Weak atomic operations prevent the compiler moving other * loads/stores past the atomic operation load/store. However there is |