summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorSean Christopherson <seanjc@google.com>2023-11-09 18:28:51 -0800
committerSean Christopherson <seanjc@google.com>2024-02-01 09:35:48 -0800
commit004a0aa56edea3effc97bc90a620da1bcde5c63c (patch)
tree379a33ec1a69534c15209580c9debb48457ab0fd
parentb31880ca2f41dc2196e31d97e498b0fa884c2b2a (diff)
downloadlinux-stable-004a0aa56edea3effc97bc90a620da1bcde5c63c.tar.gz
linux-stable-004a0aa56edea3effc97bc90a620da1bcde5c63c.tar.bz2
linux-stable-004a0aa56edea3effc97bc90a620da1bcde5c63c.zip
KVM: x86/pmu: Snapshot and clear reprogramming bitmap before reprogramming
Refactor the handling of the reprogramming bitmap to snapshot and clear to-be-processed bits before doing the reprogramming, and then explicitly set bits for PMCs that need to be reprogrammed (again). This will allow adding a macro to iterate over all valid PMCs without having to add special handling for the reprogramming bit, which (a) can have bits set for non-existent PMCs and (b) needs to clear such bits to avoid wasting cycles in perpetuity. Note, the existing behavior of clearing bits after reprogramming does NOT have a race with kvm_vm_ioctl_set_pmu_event_filter(). Setting a new PMU filter synchronizes SRCU _before_ setting the bitmap, i.e. guarantees that the vCPU isn't in the middle of reprogramming with a stale filter prior to setting the bitmap. Link: https://lore.kernel.org/r/20231110022857.1273836-5-seanjc@google.com Signed-off-by: Sean Christopherson <seanjc@google.com>
-rw-r--r--arch/x86/include/asm/kvm_host.h1
-rw-r--r--arch/x86/kvm/pmu.c52
2 files changed, 30 insertions, 23 deletions
diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
index b5b2d0fde579..ad5319a503f0 100644
--- a/arch/x86/include/asm/kvm_host.h
+++ b/arch/x86/include/asm/kvm_host.h
@@ -536,6 +536,7 @@ struct kvm_pmc {
#define KVM_PMC_MAX_FIXED 3
#define MSR_ARCH_PERFMON_FIXED_CTR_MAX (MSR_ARCH_PERFMON_FIXED_CTR0 + KVM_PMC_MAX_FIXED - 1)
#define KVM_AMD_PMC_MAX_GENERIC 6
+
struct kvm_pmu {
u8 version;
unsigned nr_arch_gp_counters;
diff --git a/arch/x86/kvm/pmu.c b/arch/x86/kvm/pmu.c
index 0873937c90bc..81a0c6719863 100644
--- a/arch/x86/kvm/pmu.c
+++ b/arch/x86/kvm/pmu.c
@@ -444,7 +444,7 @@ static bool pmc_event_is_allowed(struct kvm_pmc *pmc)
check_pmu_event_filter(pmc);
}
-static void reprogram_counter(struct kvm_pmc *pmc)
+static int reprogram_counter(struct kvm_pmc *pmc)
{
struct kvm_pmu *pmu = pmc_to_pmu(pmc);
u64 eventsel = pmc->eventsel;
@@ -455,7 +455,7 @@ static void reprogram_counter(struct kvm_pmc *pmc)
emulate_overflow = pmc_pause_counter(pmc);
if (!pmc_event_is_allowed(pmc))
- goto reprogram_complete;
+ return 0;
if (emulate_overflow)
__kvm_perf_overflow(pmc, false);
@@ -476,43 +476,49 @@ static void reprogram_counter(struct kvm_pmc *pmc)
}
if (pmc->current_config == new_config && pmc_resume_counter(pmc))
- goto reprogram_complete;
+ return 0;
pmc_release_perf_event(pmc);
pmc->current_config = new_config;
- /*
- * If reprogramming fails, e.g. due to contention, leave the counter's
- * regprogram bit set, i.e. opportunistically try again on the next PMU
- * refresh. Don't make a new request as doing so can stall the guest
- * if reprogramming repeatedly fails.
- */
- if (pmc_reprogram_counter(pmc, PERF_TYPE_RAW,
- (eventsel & pmu->raw_event_mask),
- !(eventsel & ARCH_PERFMON_EVENTSEL_USR),
- !(eventsel & ARCH_PERFMON_EVENTSEL_OS),
- eventsel & ARCH_PERFMON_EVENTSEL_INT))
- return;
-
-reprogram_complete:
- clear_bit(pmc->idx, (unsigned long *)&pmc_to_pmu(pmc)->reprogram_pmi);
+ return pmc_reprogram_counter(pmc, PERF_TYPE_RAW,
+ (eventsel & pmu->raw_event_mask),
+ !(eventsel & ARCH_PERFMON_EVENTSEL_USR),
+ !(eventsel & ARCH_PERFMON_EVENTSEL_OS),
+ eventsel & ARCH_PERFMON_EVENTSEL_INT);
}
void kvm_pmu_handle_event(struct kvm_vcpu *vcpu)
{
+ DECLARE_BITMAP(bitmap, X86_PMC_IDX_MAX);
struct kvm_pmu *pmu = vcpu_to_pmu(vcpu);
int bit;
- for_each_set_bit(bit, pmu->reprogram_pmi, X86_PMC_IDX_MAX) {
+ bitmap_copy(bitmap, pmu->reprogram_pmi, X86_PMC_IDX_MAX);
+
+ /*
+ * The reprogramming bitmap can be written asynchronously by something
+ * other than the task that holds vcpu->mutex, take care to clear only
+ * the bits that will actually processed.
+ */
+ BUILD_BUG_ON(sizeof(bitmap) != sizeof(atomic64_t));
+ atomic64_andnot(*(s64 *)bitmap, &pmu->__reprogram_pmi);
+
+ for_each_set_bit(bit, bitmap, X86_PMC_IDX_MAX) {
struct kvm_pmc *pmc = kvm_pmc_idx_to_pmc(pmu, bit);
- if (unlikely(!pmc)) {
- clear_bit(bit, pmu->reprogram_pmi);
+ if (unlikely(!pmc))
continue;
- }
- reprogram_counter(pmc);
+ /*
+ * If reprogramming fails, e.g. due to contention, re-set the
+ * regprogram bit set, i.e. opportunistically try again on the
+ * next PMU refresh. Don't make a new request as doing so can
+ * stall the guest if reprogramming repeatedly fails.
+ */
+ if (reprogram_counter(pmc))
+ set_bit(pmc->idx, pmu->reprogram_pmi);
}
/*