On 2017/3/31 6:00, Julien Grall wrote:
>
>
> On 30/03/2017 22:49, Stefano Stabellini wrote:
>> On Thu, 30 Mar 2017, Wei Chen wrote:
>>> + /*
>>> + * If the SErrors option is "FORWARD", we have to prevent forwarding
>>> + * serror to wrong vCPU. So before context switch, we have to use the
>>> + * synchronize_serror to guarantee that the pending serror would be
>>> + * caught by current vCPU.
>>> + *
>>> + * The SKIP_CTXT_SWITCH_SERROR_SYNC will be set to cpu_hwcaps when the
>>> + * SErrors option is NOT "FORWARD".
>>> + */
>>> + asm volatile(ALTERNATIVE("bl synchronize_serror",
>>> + "nop",
>>> + SKIP_CTXT_SWITCH_SERROR_SYNC));
>>
>>
>> This good, but here you need to add:
>>
>> barrier();
>>
>> because the compiler is free to reorder even asm volatile instructions
>> (it could move the asm volatile after __context_switch theoretically).
>
> ... or it could moved before hand because there are no barrier... What
> you want to use is asm volatile(ALTERNATIVE(...) : : : "memory");
>
I would cover this consideration in the change of #16 in next version.
> Cheers,
>
--
Regards,
Wei Chen
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
