From 9b545c04abd4f7246a3bde040efde587abebb23c Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Mon, 4 Feb 2019 14:23:30 -0800 Subject: perf/x86/kvm: Avoid unnecessary work in guest filtering KVM added a workaround for PEBS events leaking into guests with commit: 26a4f3c08de4 ("perf/x86: disable PEBS on a guest entry.") This uses the VT entry/exit list to add an extra disable of the PEBS_ENABLE MSR. Intel also added a fix for this issue to microcode updates on Haswell/Broadwell/Skylake. It turns out using the MSR entry/exit list makes VM exits significantly slower. The list is only needed for disabling PEBS, because the GLOBAL_CTRL change gets optimized by KVM into changing the VMCS. Check for the microcode updates that have the microcode fix for leaking PEBS, and disable the extra entry/exit list entry for PEBS_ENABLE. In addition we always clear the GLOBAL_CTRL for the PEBS counter while running in the guest, which is enough to make them never fire at the wrong side of the host/guest transition. The overhead for VM exits with the filtering active with the patch is reduced from 8% to 4%. The microcode patch has already been merged into future platforms. This patch is one-off thing. The quirks is used here. For other old platforms which doesn't have microcode patch and quirks, extra disable of the PEBS_ENABLE MSR is still required. Signed-off-by: Andi Kleen Signed-off-by: Kan Liang Signed-off-by: Peter Zijlstra (Intel) Cc: Alexander Shishkin Cc: Arnaldo Carvalho de Melo Cc: David Ahern Cc: Jiri Olsa Cc: Linus Torvalds Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Thomas Gleixner Cc: Vince Weaver Cc: bp@alien8.de Link: https://lkml.kernel.org/r/1549319013-4522-2-git-send-email-kan.liang@linux.intel.com Signed-off-by: Ingo Molnar --- arch/x86/events/perf_event.h | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'arch/x86/events/perf_event.h') diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h index 78d7b7031bfc..dea716e1f713 100644 --- a/arch/x86/events/perf_event.h +++ b/arch/x86/events/perf_event.h @@ -601,13 +601,14 @@ struct x86_pmu { /* * Intel DebugStore bits */ - unsigned int bts :1, - bts_active :1, - pebs :1, - pebs_active :1, - pebs_broken :1, - pebs_prec_dist :1, - pebs_no_tlb :1; + unsigned int bts :1, + bts_active :1, + pebs :1, + pebs_active :1, + pebs_broken :1, + pebs_prec_dist :1, + pebs_no_tlb :1, + pebs_no_isolation :1; int pebs_record_size; int pebs_buffer_size; void (*drain_pebs)(struct pt_regs *regs); -- cgit v1.2.3-55-g7522 From d01b1f96a82e5dd7841a1d39db3abfdaf95f70ab Mon Sep 17 00:00:00 2001 From: Peter Zijlstra (Intel) Date: Tue, 5 Mar 2019 22:23:15 +0100 Subject: perf/x86/intel: Make cpuc allocations consistent The cpuc data structure allocation is different between fake and real cpuc's; use the same code to init/free both. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Thomas Gleixner --- arch/x86/events/core.c | 13 +++++-------- arch/x86/events/intel/core.c | 29 ++++++++++++++++++----------- arch/x86/events/perf_event.h | 11 ++++++++--- 3 files changed, 31 insertions(+), 22 deletions(-) (limited to 'arch/x86/events/perf_event.h') diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c index b684f0294f35..e2b1447192a8 100644 --- a/arch/x86/events/core.c +++ b/arch/x86/events/core.c @@ -1995,7 +1995,7 @@ static int x86_pmu_commit_txn(struct pmu *pmu) */ static void free_fake_cpuc(struct cpu_hw_events *cpuc) { - kfree(cpuc->shared_regs); + intel_cpuc_finish(cpuc); kfree(cpuc); } @@ -2007,14 +2007,11 @@ static struct cpu_hw_events *allocate_fake_cpuc(void) cpuc = kzalloc(sizeof(*cpuc), GFP_KERNEL); if (!cpuc) return ERR_PTR(-ENOMEM); - - /* only needed, if we have extra_regs */ - if (x86_pmu.extra_regs) { - cpuc->shared_regs = allocate_shared_regs(cpu); - if (!cpuc->shared_regs) - goto error; - } cpuc->is_fake = 1; + + if (intel_cpuc_prepare(cpuc, cpu)) + goto error; + return cpuc; error: free_fake_cpuc(cpuc); diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 730978dff63f..dc39f2482a4f 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -3398,7 +3398,7 @@ ssize_t intel_event_sysfs_show(char *page, u64 config) return x86_event_sysfs_show(page, config, event); } -struct intel_shared_regs *allocate_shared_regs(int cpu) +static struct intel_shared_regs *allocate_shared_regs(int cpu) { struct intel_shared_regs *regs; int i; @@ -3430,10 +3430,9 @@ static struct intel_excl_cntrs *allocate_excl_cntrs(int cpu) return c; } -static int intel_pmu_cpu_prepare(int cpu) -{ - struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu); +int intel_cpuc_prepare(struct cpu_hw_events *cpuc, int cpu) +{ if (x86_pmu.extra_regs || x86_pmu.lbr_sel_map) { cpuc->shared_regs = allocate_shared_regs(cpu); if (!cpuc->shared_regs) @@ -3443,7 +3442,7 @@ static int intel_pmu_cpu_prepare(int cpu) if (x86_pmu.flags & PMU_FL_EXCL_CNTRS) { size_t sz = X86_PMC_IDX_MAX * sizeof(struct event_constraint); - cpuc->constraint_list = kzalloc(sz, GFP_KERNEL); + cpuc->constraint_list = kzalloc_node(sz, GFP_KERNEL, cpu_to_node(cpu)); if (!cpuc->constraint_list) goto err_shared_regs; @@ -3468,6 +3467,11 @@ err: return -ENOMEM; } +static int intel_pmu_cpu_prepare(int cpu) +{ + return intel_cpuc_prepare(&per_cpu(cpu_hw_events, cpu), cpu); +} + static void flip_smm_bit(void *data) { unsigned long set = *(unsigned long *)data; @@ -3542,9 +3546,8 @@ static void intel_pmu_cpu_starting(int cpu) } } -static void free_excl_cntrs(int cpu) +static void free_excl_cntrs(struct cpu_hw_events *cpuc) { - struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu); struct intel_excl_cntrs *c; c = cpuc->excl_cntrs; @@ -3565,9 +3568,8 @@ static void intel_pmu_cpu_dying(int cpu) disable_counter_freeze(); } -static void intel_pmu_cpu_dead(int cpu) +void intel_cpuc_finish(struct cpu_hw_events *cpuc) { - struct cpu_hw_events *cpuc = &per_cpu(cpu_hw_events, cpu); struct intel_shared_regs *pc; pc = cpuc->shared_regs; @@ -3577,7 +3579,12 @@ static void intel_pmu_cpu_dead(int cpu) cpuc->shared_regs = NULL; } - free_excl_cntrs(cpu); + free_excl_cntrs(cpuc); +} + +static void intel_pmu_cpu_dead(int cpu) +{ + intel_cpuc_finish(&per_cpu(cpu_hw_events, cpu)); } static void intel_pmu_sched_task(struct perf_event_context *ctx, @@ -4715,7 +4722,7 @@ static __init int fixup_ht_bug(void) hardlockup_detector_perf_restart(); for_each_online_cpu(c) - free_excl_cntrs(c); + free_excl_cntrs(&per_cpu(cpu_hw_events, c)); cpus_read_unlock(); pr_info("PMU erratum BJ122, BV98, HSD29 workaround disabled, HT off\n"); diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h index d46fd6754d92..964212b80abb 100644 --- a/arch/x86/events/perf_event.h +++ b/arch/x86/events/perf_event.h @@ -889,7 +889,8 @@ struct event_constraint * x86_get_event_constraints(struct cpu_hw_events *cpuc, int idx, struct perf_event *event); -struct intel_shared_regs *allocate_shared_regs(int cpu); +extern int intel_cpuc_prepare(struct cpu_hw_events *cpuc, int cpu); +extern void intel_cpuc_finish(struct cpu_hw_events *cpuc); int intel_pmu_init(void); @@ -1025,9 +1026,13 @@ static inline int intel_pmu_init(void) return 0; } -static inline struct intel_shared_regs *allocate_shared_regs(int cpu) +static inline int intel_cpuc_prepare(struct cpu_hw_event *cpuc, int cpu) +{ + return 0; +} + +static inline void intel_cpuc_finish(struct cpu_hw_event *cpuc) { - return NULL; } static inline int is_ht_workaround_enabled(void) -- cgit v1.2.3-55-g7522 From 400816f60c543153656ac74eaf7f36f6b7202378 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra (Intel) Date: Tue, 5 Mar 2019 22:23:18 +0100 Subject: perf/x86/intel: Implement support for TSX Force Abort Skylake (and later) will receive a microcode update to address a TSX errata. This microcode will, on execution of a TSX instruction (speculative or not) use (clobber) PMC3. This update will also provide a new MSR to change this behaviour along with a CPUID bit to enumerate the presence of this new MSR. When the MSR gets set; the microcode will no longer use PMC3 but will Force Abort every TSX transaction (upon executing COMMIT). When TSX Force Abort (TFA) is allowed (default); the MSR gets set when PMC3 gets scheduled and cleared when, after scheduling, PMC3 is unused. When TFA is not allowed; clear PMC3 from all constraints such that it will not get used. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Thomas Gleixner --- arch/x86/events/intel/core.c | 74 ++++++++++++++++++++++++++++++++++++++++++-- arch/x86/events/perf_event.h | 6 ++++ 2 files changed, 77 insertions(+), 3 deletions(-) (limited to 'arch/x86/events/perf_event.h') diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index 9d69e107e19c..dadb8f7e5a0d 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -1999,6 +1999,39 @@ static void intel_pmu_nhm_enable_all(int added) intel_pmu_enable_all(added); } +static void intel_set_tfa(struct cpu_hw_events *cpuc, bool on) +{ + u64 val = on ? MSR_TFA_RTM_FORCE_ABORT : 0; + + if (cpuc->tfa_shadow != val) { + cpuc->tfa_shadow = val; + wrmsrl(MSR_TSX_FORCE_ABORT, val); + } +} + +static void intel_tfa_commit_scheduling(struct cpu_hw_events *cpuc, int idx, int cntr) +{ + /* + * We're going to use PMC3, make sure TFA is set before we touch it. + */ + if (cntr == 3 && !cpuc->is_fake) + intel_set_tfa(cpuc, true); +} + +static void intel_tfa_pmu_enable_all(int added) +{ + struct cpu_hw_events *cpuc = this_cpu_ptr(&cpu_hw_events); + + /* + * If we find PMC3 is no longer used when we enable the PMU, we can + * clear TFA. + */ + if (!test_bit(3, cpuc->active_mask)) + intel_set_tfa(cpuc, false); + + intel_pmu_enable_all(added); +} + static void enable_counter_freeze(void) { update_debugctlmsr(get_debugctlmsr() | @@ -3354,6 +3387,26 @@ glp_get_event_constraints(struct cpu_hw_events *cpuc, int idx, return c; } +static bool allow_tsx_force_abort = true; + +static struct event_constraint * +tfa_get_event_constraints(struct cpu_hw_events *cpuc, int idx, + struct perf_event *event) +{ + struct event_constraint *c = hsw_get_event_constraints(cpuc, idx, event); + + /* + * Without TFA we must not use PMC3. + */ + if (!allow_tsx_force_abort && test_bit(3, c->idxmsk)) { + c = dyn_constraint(cpuc, c, idx); + c->idxmsk64 &= ~(1ULL << 3); + c->weight--; + } + + return c; +} + /* * Broadwell: * @@ -3448,13 +3501,15 @@ int intel_cpuc_prepare(struct cpu_hw_events *cpuc, int cpu) goto err; } - if (x86_pmu.flags & PMU_FL_EXCL_CNTRS) { + if (x86_pmu.flags & (PMU_FL_EXCL_CNTRS | PMU_FL_TFA)) { size_t sz = X86_PMC_IDX_MAX * sizeof(struct event_constraint); cpuc->constraint_list = kzalloc_node(sz, GFP_KERNEL, cpu_to_node(cpu)); if (!cpuc->constraint_list) goto err_shared_regs; + } + if (x86_pmu.flags & PMU_FL_EXCL_CNTRS) { cpuc->excl_cntrs = allocate_excl_cntrs(cpu); if (!cpuc->excl_cntrs) goto err_constraint_list; @@ -3564,9 +3619,10 @@ static void free_excl_cntrs(struct cpu_hw_events *cpuc) if (c->core_id == -1 || --c->refcnt == 0) kfree(c); cpuc->excl_cntrs = NULL; - kfree(cpuc->constraint_list); - cpuc->constraint_list = NULL; } + + kfree(cpuc->constraint_list); + cpuc->constraint_list = NULL; } static void intel_pmu_cpu_dying(int cpu) @@ -4086,8 +4142,11 @@ static struct attribute *intel_pmu_caps_attrs[] = { NULL }; +DEVICE_BOOL_ATTR(allow_tsx_force_abort, 0644, allow_tsx_force_abort); + static struct attribute *intel_pmu_attrs[] = { &dev_attr_freeze_on_smi.attr, + NULL, /* &dev_attr_allow_tsx_force_abort.attr.attr */ NULL, }; @@ -4580,6 +4639,15 @@ __init int intel_pmu_init(void) tsx_attr = hsw_tsx_events_attrs; intel_pmu_pebs_data_source_skl( boot_cpu_data.x86_model == INTEL_FAM6_SKYLAKE_X); + + if (boot_cpu_has(X86_FEATURE_TSX_FORCE_ABORT)) { + x86_pmu.flags |= PMU_FL_TFA; + x86_pmu.get_event_constraints = tfa_get_event_constraints; + x86_pmu.enable_all = intel_tfa_pmu_enable_all; + x86_pmu.commit_scheduling = intel_tfa_commit_scheduling; + intel_pmu_attrs[1] = &dev_attr_allow_tsx_force_abort.attr.attr; + } + pr_cont("Skylake events, "); name = "skylake"; break; diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h index 964212b80abb..a345d079f876 100644 --- a/arch/x86/events/perf_event.h +++ b/arch/x86/events/perf_event.h @@ -242,6 +242,11 @@ struct cpu_hw_events { struct intel_excl_cntrs *excl_cntrs; int excl_thread_id; /* 0 or 1 */ + /* + * SKL TSX_FORCE_ABORT shadow + */ + u64 tfa_shadow; + /* * AMD specific bits */ @@ -681,6 +686,7 @@ do { \ #define PMU_FL_EXCL_CNTRS 0x4 /* has exclusive counter requirements */ #define PMU_FL_EXCL_ENABLED 0x8 /* exclusive counter active */ #define PMU_FL_PEBS_ALL 0x10 /* all events are valid PEBS events */ +#define PMU_FL_TFA 0x20 /* deal with TSX force abort */ #define EVENT_VAR(_id) event_attr_##_id #define EVENT_PTR(_id) &event_attr_##_id.attr.attr -- cgit v1.2.3-55-g7522 From f764c58b7faa26f5714e6907f892abc2bc0de4f8 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 15 Mar 2019 09:14:10 +0100 Subject: perf/x86: Fixup typo in stub functions Guenter reported a build warning for CONFIG_CPU_SUP_INTEL=n: > With allmodconfig-CONFIG_CPU_SUP_INTEL, this patch results in: > > In file included from arch/x86/events/amd/core.c:8:0: > arch/x86/events/amd/../perf_event.h:1036:45: warning: ‘struct cpu_hw_event’ declared inside parameter list will not be visible outside of this definition or declaration > static inline int intel_cpuc_prepare(struct cpu_hw_event *cpuc, int cpu) While harmless (an unsed pointer is an unused pointer, no matter the type) it needs fixing. Reported-by: Guenter Roeck Signed-off-by: Peter Zijlstra (Intel) Cc: Greg Kroah-Hartman Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Cc: stable@vger.kernel.org Fixes: d01b1f96a82e ("perf/x86/intel: Make cpuc allocations consistent") Link: http://lkml.kernel.org/r/20190315081410.GR5996@hirez.programming.kicks-ass.net Signed-off-by: Ingo Molnar --- arch/x86/events/perf_event.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/events/perf_event.h') diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h index b04ae6c8775e..a75955741c50 100644 --- a/arch/x86/events/perf_event.h +++ b/arch/x86/events/perf_event.h @@ -1033,12 +1033,12 @@ static inline int intel_pmu_init(void) return 0; } -static inline int intel_cpuc_prepare(struct cpu_hw_event *cpuc, int cpu) +static inline int intel_cpuc_prepare(struct cpu_hw_events *cpuc, int cpu) { return 0; } -static inline void intel_cpuc_finish(struct cpu_hw_event *cpuc) +static inline void intel_cpuc_finish(struct cpu_hw_events *cpuc) { } -- cgit v1.2.3-55-g7522