[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] [PATCH v24 04/15] x86/VPMU: Interface for setting PMU mode and flags
On 06/11/2015 11:23 PM, Tian, Kevin wrote: From: Boris Ostrovsky [mailto:boris.ostrovsky@xxxxxxxxxx] + + case XENPMU_feature_set: + if ( pmu_params.val & ~XENPMU_FEATURE_INTEL_BTS ) + return -EINVAL; + + spin_lock(&vpmu_lock); + + if ( vpmu_count == 0 ) + vpmu_features = pmu_params.val; + else + { + printk(XENLOG_WARNING "VPMU: Cannot change features while" + " active VPMUs exist\n"); + ret = -EBUSY; + }what about setting same features as existing in vpmu_features? we should do same check as done in mode setting.Not sure I follow you. There is only one feature currently that we support --- BTS. And trying to set any other feature will result in -EINVAL. What is wrong with trying to set the same bit twice? (except for being pointless)My point is whether you want to allow setting same bit twice when active PMUs exist. From above code it's disallowed w/ check on vpmu_count. However in earlier code handling vpmu_mode setting, you actually allow setting same mode twice: + if ( (vpmu_count == 0) || (vpmu_mode == pmu_params.val) || + ((vpmu_mode ^ pmu_params.val) == + (XENPMU_MODE_SELF | XENPMU_MODE_HV)) ) + vpmu_mode = pmu_params.val; So I thought you may keep the same policy to allow setting vpmu_features with same bit twice too. Oh, I see. You are suggesting changing XENPMU_feature_set to if ( (vpmu_count == 0) || (vpmu_features == pmu_params.val) ) vpmu_features = pmu_params.val;or dropping '|| (vpmu_mode == pmu_params.val)' from XENPMU_mode_set for consistency's sake? I suppose I can do that, yes. -boris _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxx http://lists.xen.org/xen-devel
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |