[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [Patch] x86: enforce strict memory order for x2apic



Yes. Now applied as c/s 18541 including explanation of why mb() inside
send_IPI_mask_x2apic() is sufficient (it's subtle!).

 -- Keir

On 25/9/08 09:40, "Tian, Kevin" <kevin.tian@xxxxxxxxx> wrote:

> Please forget this one. I made mistake since we'll assume
> send_IPI_mask as a barrier now. :-)
> 
> Thanks,
> Kevin
> 
>> From: Tian, Kevin
>> Sent: 2008年9月25日 16:39
>> 
>>> From: Keir Fraser [mailto:keir.fraser@xxxxxxxxxxxxx]
>>> Sent: 2008年9月25日 16:36
>>> 
>>> On 25/9/08 09:29, "Tian, Kevin" <kevin.tian@xxxxxxxxx> wrote:
>>> 
>>>>> On the other hand you're penalizing LAPIC systems with
>>> this, too. I'm
>>>>> not certain that's better than fixing the (incorrect)
>>>>> assumptions just in
>>>>> the x2APIC case.
>>>> 
>>>> I agree. I'll revise the patch to add a 'mb()' in x2APIC instead,
>>>> though that may bring a bit overhead to cases when fencing
>>>> is not required, like event check IPI.
>>> 
>>> To be precise, mb() at top of send_IPI_mask_x2apic(), and
>>> wmb() may as well
>>> go from on_selected_cpus() as we can assume send_IPI_mask() is
>>> a barrier.
>>> 
>>> I'll do the patch for this myself.
>>> 
>> 
>> Yes, and thanks. Also please add a wmb() in flush_area_mask,
>> though it's a nop but semanticly desired like for on_selected_cpus.
> 



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.