[PATCH] Memory fencing problem in pthread cancellation

[Jeff Law]

On 01/15/2013 12:20 PM, Richard Henderson wrote:
> On 01/14/2013 07:13 PM, Jeff Law wrote:
>> I suspect the asm memory clobber is historical, probably introduced to
>> work around issues with pthread_cancel_init being inlined and the
>> compiler moving things around too aggressively.  One could make an
>> argument it should be removed.
>
> It's definitely redundant with the atomic_read_barrier you added.
Removed locally, will appear in next patch iteration.
jeff