> -----Original Message----- > From: Jean-Philippe Brucker [mailto:jean-philippe@xxxxxxxxxx] > Sent: 24 September 2020 11:14 > To: Shameerali Kolothum Thodi <shameerali.kolothum.thodi@xxxxxxxxxx> > Cc: iommu@xxxxxxxxxxxxxxxxxxxxxxxxxx; linux-arm-kernel@xxxxxxxxxxxxxxxxxxx; > linux-mm@xxxxxxxxx; fenghua.yu@xxxxxxxxx; catalin.marinas@xxxxxxx; > Suzuki K Poulose <suzuki.poulose@xxxxxxx>; robin.murphy@xxxxxxx; > zhangfei.gao@xxxxxxxxxx; will@xxxxxxxxxx > Subject: Re: [PATCH v10 10/13] iommu/arm-smmu-v3: Check for SVA features > > Hi Shameer, > > On Mon, Sep 21, 2020 at 08:59:39AM +0000, Shameerali Kolothum Thodi > wrote: > > > +bool arm_smmu_sva_supported(struct arm_smmu_device *smmu) > > > +{ > > > + unsigned long reg, fld; > > > + unsigned long oas; > > > + unsigned long asid_bits; > > > + u32 feat_mask = ARM_SMMU_FEAT_BTM | > > > ARM_SMMU_FEAT_COHERENCY; > > > > Why is BTM mandated for SVA? I couldn't find this requirement in SMMU spec > > (Sorry if I missed it or this got discussed earlier). But if performance is the > only concern here, > > is it better just to allow it with a warning rather than limiting SMMUs without > BTM? > > It's a performance concern and requires to support multiple > configurations, but the spec allows it. Are there SMMUs without BTM that > need it? Ok. Thanks for clarifying. May be better to add a comment here. Our platforms do support BTM, but I had a strange case where the UEFI didn't enable DVM but SMMU reported BTM and was causing random failures due to lack of explicit tlbi on mm invalidation. Anyway that doesn't count here :) Thanks, Shameer > Thanks, > Jean
