>-----Original Message----- >From: Liu, Yi L <[email protected]> >Subject: Re: [PATCH v2] intel_iommu: Keep PASID cache consistent with PT >fast path > >On 11/5/26 17:07, Duan, Zhenzhong wrote: >> Hi Clement, >> >> Good finding! >> >> It looks not easy to refactor all these as >> vtd_switch_address_space() is called in many callsites, >> not only fast path. >> >> I see in vtd_switch_address_space() vtd_as_pt_enabled() >> is called twice, that's obviously unnecessary. >> >> I'm not sure if it's possible to cache CE and PE before >> all vtd_switch_address_space() so it only needs to check >> cache for PT. > >I had considered it as well during reviewing the prior patches. It might >be worth thinking about caching both CE and PE in the AddressSpace and >gating their life circle by the CC/PC invalidation request from guest. >In other paths, we can just refer to the cached CE and PE. But this >means we rely on guest flushes cache timely when there is CE/PE >modification. This is spec defined behavior, so might be fine?
Yes, it should work for linux guest, I'm not clear if it breaks OS that doesn't follow spec in that path:) Thanks Zhenzhong
