On Thu, Nov 17, 2016 at 05:47:51PM +0000, Mark Rutland wrote: > On Thu, Nov 17, 2016 at 05:45:41PM +0100, Andrew Jones wrote: > > Hi Mark, > > > > On Wed, Nov 16, 2016 at 05:45:44PM +0000, Mark Rutland wrote: > > > So I don't think we should try to distinguish ARMv8-A AArch32 from > > > ARMv7-A. We should test individual features, or if that's not possible, > > > group them in the same bucket. > > > > Perhaps I was too quick to look for a general way to approach the first > > place I saw the need, which is > > > > check_cntr(read_pmccntr()); > > if (is_aarch64()) > > check_cntr(read_pmccntr64()); > > > > Is PMCCNTR the only "weird" difference? > > Define "weird". ;) > > I believe the extension of PMCCNTR to 64 bits is part of PMUv3. So you > should be able to check ID_DFR0.PerfMon == 0b0011. For overflow you'll > also need to configure PMCR.LC. Thanks Mark, checking for ID_DFR0.PerfMon == 0b0011 for this case does indeed look like the right thing to do. Wei, when you respin the PMU test can you add the additional 64-bit read, when a PMUv3 is present? Thanks, drew > > Thanks, > Mark. > -- > To unsubscribe from this list: send the line "unsubscribe kvm" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe kvm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
