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

Re: [Xen-devel] [PATCH] x86: don't use VA for cache flush when also flushing TLB

Jan Beulich wrote on 2014-05-28:
>>>> On 28.05.14 at 06:35, <yang.z.zhang@xxxxxxxxx> wrote:
>> Besides, I am thinking should we flush cache unconditionally, e.g., if
>> all agents that able to access memory have the snooping capability,
>> then software is not required to flush cache always.(If I am wrong,
>> please correct me and ignore the following changes). So is there any
>> problem with this change? @@ -145,7 +145,7 @@ void
>> flush_area_local(const void *va, unsigned int
> flags)
>>          }
>>      }
>> -    if ( flags & FLUSH_CACHE )
>> +    if ( flags & FLUSH_CACHE && !(iommu_enable && iommu_snoop))
>>      {
>>          unsigned long i, sz = 0;
> At least the iommu_snoop part was discussed at length a couple of weeks back:
> I don't see how this relates to cache flushing requirements. I also
> don't see how IOMMU presence/absence would matter here - the behavior
> should clearly be identical between systems with and without IOMMU.

Ok. I remember there is a discussion around this before. I will check it.

> Jan

Best regards,

Xen-devel mailing list



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