perf/x86/amd: support capturing LBR from software events
Subsystems (3)
Andrii Nakryiko
maintainerPATCH v2 1/1 perf/x86/amd: support capturing LBR from software events
[0] added ability to capture LBR (Last Branch Records) on Intel CPUs from inside BPF program at pretty much any arbitrary point. This is extremely useful capability that allows to figure out otherwise hard-to-debug problems, because LBR is now available based on some application-defined conditions, not just hardware-supported events.
retsnoop ([1]) is one such tool that takes a huge advantage of this functionality and has proved to be an extremely useful tool in practice.
Now, AMD Zen4 CPUs got support for similar LBR functionality, but necessary wiring inside the kernel is not yet setup. This patch seeks to rectify this and follows a similar approach to the original patch [0] for Intel CPUs.
Given LBR can be set up to capture any indirect jumps, it's critical to minimize indirect jumps on the way to requesting LBR from BPF program, so we split amd_pmu_lbr_disable_all() into a wrapper with some generic conditions vs always-inlined __amd_pmu_lbr_disable() called directly from BPF subsystem (through perf_snapshot_branch_stack static call).
Now that it's possible to capture LBR on AMD CPU from BPF at arbitrary point, there is no reason to artificially limit this feature to sampling events. So corresponding check is removed. AFAIU, there is no correctness implications of doing this (and it was possible to bypass this check by just setting perf_event's sample_period to 1 anyways, so it doesn't guard all that much).
This was tested on AMD Bergamo CPU and worked well when utilized from the aforementioned retsnoop tool.
[0] https://lore.kernel.org/bpf/[email protected]/
[1] https://github.com/anakryiko/retsnoop
--- arch/x86/events/amd/core.c | 29 ++++++++++++++++++++++++++++- arch/x86/events/amd/lbr.c | 11 +---------- arch/x86/events/perf_event.h | 11 +++++++++++ 3 files changed, 40 insertions(+), 11 deletions(-)
arch/x86/events/amd/core.c +28 -1
@@ -618,7 +618,7 @@ static void amd_pmu_cpu_dead(int cpu) } } -static inline void amd_pmu_set_global_ctl (u64 ctl) +static __always_inline void amd_pmu_set_global_ctl(u64 ctl) {wrmsrl (MSR_AMD64_PERF_CNTR_GLOBAL_CTL , ctl); }
@@ -878,6 +878,29 @@ static int amd_pmu_handle_irq(struct pt_regs *regs)return amd_pmu_adjust_nmi_window (handled); } +static int amd_pmu_v2_snapshot_branch_stack (struct perf_branch_entry *entries,unsigned int cnt) +{ +struct cpu_hw_events *cpuc; +unsigned long flags; + +/* must not have branches... */ +local_irq_save (flags); +amd_pmu_core_disable_all (); +__amd_pmu_lbr_disable (); +/* ... until here */ + + cpuc =this_cpu_ptr (&cpu_hw_events); + +amd_pmu_lbr_read (); + cnt = min_t(unsigned int, cnt, x86_pmu.lbr_nr); +memcpy (entries, cpuc->lbr_entries,sizeof (struct perf_branch_entry ) * cnt); + +amd_pmu_v2_enable_all (0 ); +local_irq_restore (flags); + +return cnt; +} +static int amd_pmu_v2_handle_irq (struct pt_regs *regs) {struct cpu_hw_events *cpuc =this_cpu_ptr (&cpu_hw_events);
@@ -1414,6 +1437,10 @@ static int __init amd_core_pmu_init(void)static_call_update (amd_pmu_branch_reset, amd_pmu_lbr_reset);static_call_update (amd_pmu_branch_add, amd_pmu_lbr_add);static_call_update (amd_pmu_branch_del, amd_pmu_lbr_del); + +/* only support branch_stack snapshot on perfmon v2 */ +if (x86_pmu.handle_irq == amd_pmu_v2_handle_irq) +static_call_update (perf_snapshot_branch_stack, amd_pmu_v2_snapshot_branch_stack); }else if (!amd_brs_init ()) {/* * BRS requires special event constraints and flushing on ctxsw.
arch/x86/events/amd/lbr.c +1 -10
@@ -310,10 +310,6 @@ int amd_pmu_lbr_hw_config(struct perf_event *event) {int ret =0 ; -/* LBR is not recommended in counting mode */ -if (!is_sampling_event (event)) -return -EINVAL ; - ret =amd_pmu_lbr_setup_filter (event);if (!ret) event->attach_state |=PERF_ATTACH_SCHED_CB ;
@@ -412,16 +408,11 @@ void amd_pmu_lbr_enable_all(void)void amd_pmu_lbr_disable_all (void ) {struct cpu_hw_events *cpuc =this_cpu_ptr (&cpu_hw_events); -u64 dbg_ctl, dbg_extn_cfg;if (!cpuc->lbr_users || !x86_pmu.lbr_nr)return ; -rdmsrl (MSR_AMD_DBG_EXTN_CFG , dbg_extn_cfg); -rdmsrl (MSR_IA32_DEBUGCTLMSR , dbg_ctl); - -wrmsrl (MSR_AMD_DBG_EXTN_CFG , dbg_extn_cfg & ~DBG_EXTN_CFG_LBRV2EN ); -wrmsrl (MSR_IA32_DEBUGCTLMSR , dbg_ctl & ~DEBUGCTLMSR_FREEZE_LBRS_ON_PMI ); +__amd_pmu_lbr_disable (); }__init int amd_pmu_lbr_init (void )
arch/x86/events/perf_event.h +11 -0
@@ -1329,6 +1329,17 @@ void amd_pmu_lbr_enable_all(void);void amd_pmu_lbr_disable_all (void );int amd_pmu_lbr_hw_config (struct perf_event *event); +static __always_inline void __amd_pmu_lbr_disable(void) +{ +u64 dbg_ctl, dbg_extn_cfg; + +rdmsrl (MSR_AMD_DBG_EXTN_CFG , dbg_extn_cfg); +rdmsrl (MSR_IA32_DEBUGCTLMSR , dbg_ctl); + +wrmsrl (MSR_AMD_DBG_EXTN_CFG , dbg_extn_cfg & ~DBG_EXTN_CFG_LBRV2EN ); +wrmsrl (MSR_IA32_DEBUGCTLMSR , dbg_ctl & ~DEBUGCTLMSR_FREEZE_LBRS_ON_PMI ); +} +#ifdef CONFIG_PERF_EVENTS_AMD_BRS #define AMD_FAM19H_BRS_EVENT 0xc4/* RETIRED_TAKEN_BRANCH_INSTRUCTIONS */
Signature · 1 line
2.43.0
Quoted 12 lines · Andrii Nakryiko
Now that it's possible to capture LBR on AMD CPU from BPF at arbitrary point, there is no reason to artificially limit this feature to sampling events. So corresponding check is removed. AFAIU, there is no correctness implications of doing this (and it was possible to bypass this check by just setting perf_event's
sample_periodto 1 anyways, so it doesn't guard all that much).arch/x86/events/amd/lbr.c +0
-0
arch/x86/events/amd/lbr.c int amd_pmu_lbr_hw_config(struct perf_event *event) · line 316
Quoted 7 lines
arch/x86/events/amd/lbr.c +0
-4@@ -310,10 +310,6 @@ int amd_pmu_lbr_hw_config(struct perf_event *event) {int ret =0 ; -/* LBR is not recommended in counting mode */ -if (!is_sampling_event (event)) -return -EINVAL ; -
Could you please split out this change into a separate patch?
Thanks,
Ingo
Quoted 28 lines · Andrii Nakryiko
[0] added ability to capture LBR (Last Branch Records) on Intel CPUs from inside BPF program at pretty much any arbitrary point. This is extremely useful capability that allows to figure out otherwise hard-to-debug problems, because LBR is now available based on some application-defined conditions, not just hardware-supported events.
retsnoop ([1]) is one such tool that takes a huge advantage of this functionality and has proved to be an extremely useful tool in practice.
Now, AMD Zen4 CPUs got support for similar LBR functionality, but necessary wiring inside the kernel is not yet setup. This patch seeks to rectify this and follows a similar approach to the original patch [0] for Intel CPUs.
Given LBR can be set up to capture any indirect jumps, it's critical to minimize indirect jumps on the way to requesting LBR from BPF program, so we split
amd_pmu_lbr_disable_all()into a wrapper with some generic conditions vs always-inlined__amd_pmu_lbr_disable()called directly from BPF subsystem (throughperf_snapshot_branch_stackstatic call).Now that it's possible to capture LBR on AMD CPU from BPF at arbitrary point, there is no reason to artificially limit this feature to sampling events. So corresponding check is removed. AFAIU, there is no correctness implications of doing this (and it was possible to bypass this check by just setting perf_event's
sample_periodto 1 anyways, so it doesn't guard all that much).
Quoted 7 lines
This was tested on AMD Bergamo CPU and worked well when utilized from the aforementioned retsnoop tool.
[0] https://lore.kernel.org/bpf/[email protected]/
[1] https://github.com/anakryiko/retsnoopSigned-off-by Andrii Nakryiko <[email protected]>
Changes look good to me.
Quoted 115 lines
--- arch/x86/events/amd/core.c | 29 ++++++++++++++++++++++++++++- arch/x86/events/amd/lbr.c | 11 +---------- arch/x86/events/perf_event.h | 11 +++++++++++ 3 files changed, 40 insertions(+), 11 deletions(-)arch/x86/events/amd/core.c +28
-1@@ -618,7 +618,7 @@ static void amd_pmu_cpu_dead(int cpu) } } -static inline void amd_pmu_set_global_ctl (u64 ctl) +static __always_inline void amd_pmu_set_global_ctl(u64 ctl) {wrmsrl (MSR_AMD64_PERF_CNTR_GLOBAL_CTL , ctl); }@@ -878,6 +878,29 @@ static int amd_pmu_handle_irq(struct pt_regs *regs)return amd_pmu_adjust_nmi_window (handled); } +static int amd_pmu_v2_snapshot_branch_stack (struct perf_branch_entry *entries,unsigned int cnt) +{ +struct cpu_hw_events *cpuc; +unsigned long flags; + +/* must not have branches... */ +local_irq_save (flags); +amd_pmu_core_disable_all (); +__amd_pmu_lbr_disable (); +/* ... until here */ + + cpuc =this_cpu_ptr (&cpu_hw_events); + +amd_pmu_lbr_read (); + cnt = min_t(unsigned int, cnt, x86_pmu.lbr_nr); +memcpy (entries, cpuc->lbr_entries,sizeof (struct perf_branch_entry ) * cnt); + +amd_pmu_v2_enable_all (0 ); +local_irq_restore (flags); + +return cnt; +} +static int amd_pmu_v2_handle_irq (struct pt_regs *regs) {struct cpu_hw_events *cpuc =this_cpu_ptr (&cpu_hw_events);@@ -1414,6 +1437,10 @@ static int __init amd_core_pmu_init(void)static_call_update (amd_pmu_branch_reset, amd_pmu_lbr_reset);static_call_update (amd_pmu_branch_add, amd_pmu_lbr_add);static_call_update (amd_pmu_branch_del, amd_pmu_lbr_del); + +/* only support branch_stack snapshot on perfmon v2 */ +if (x86_pmu.handle_irq == amd_pmu_v2_handle_irq) +static_call_update (perf_snapshot_branch_stack, amd_pmu_v2_snapshot_branch_stack); }else if (!amd_brs_init ()) {/* * BRS requires special event constraints and flushing on ctxsw. arch/x86/events/amd/lbr.c +1
-10@@ -310,10 +310,6 @@ int amd_pmu_lbr_hw_config(struct perf_event *event) {int ret =0 ; -/* LBR is not recommended in counting mode */ -if (!is_sampling_event (event)) -return -EINVAL ; - ret =amd_pmu_lbr_setup_filter (event);if (!ret) event->attach_state |=PERF_ATTACH_SCHED_CB ;@@ -412,16 +408,11 @@ void amd_pmu_lbr_enable_all(void)void amd_pmu_lbr_disable_all (void ) {struct cpu_hw_events *cpuc =this_cpu_ptr (&cpu_hw_events); -u64 dbg_ctl, dbg_extn_cfg;if (!cpuc->lbr_users || !x86_pmu.lbr_nr)return ; -rdmsrl (MSR_AMD_DBG_EXTN_CFG , dbg_extn_cfg); -rdmsrl (MSR_IA32_DEBUGCTLMSR , dbg_ctl); - -wrmsrl (MSR_AMD_DBG_EXTN_CFG , dbg_extn_cfg & ~DBG_EXTN_CFG_LBRV2EN ); -wrmsrl (MSR_IA32_DEBUGCTLMSR , dbg_ctl & ~DEBUGCTLMSR_FREEZE_LBRS_ON_PMI ); +__amd_pmu_lbr_disable (); }__init int amd_pmu_lbr_init (void )arch/x86/events/perf_event.h +11
-0@@ -1329,6 +1329,17 @@ void amd_pmu_lbr_enable_all(void);void amd_pmu_lbr_disable_all (void );int amd_pmu_lbr_hw_config (struct perf_event *event); +static __always_inline void __amd_pmu_lbr_disable(void) +{ +u64 dbg_ctl, dbg_extn_cfg; + +rdmsrl (MSR_AMD_DBG_EXTN_CFG , dbg_extn_cfg); +rdmsrl (MSR_IA32_DEBUGCTLMSR , dbg_ctl); + +wrmsrl (MSR_AMD_DBG_EXTN_CFG , dbg_extn_cfg & ~DBG_EXTN_CFG_LBRV2EN ); +wrmsrl (MSR_IA32_DEBUGCTLMSR , dbg_ctl & ~DEBUGCTLMSR_FREEZE_LBRS_ON_PMI ); +} +#ifdef CONFIG_PERF_EVENTS_AMD_BRS #define AMD_FAM19H_BRS_EVENT 0xc4/* RETIRED_TAKEN_BRANCH_INSTRUCTIONS */
Quoted 33 lines · Sandipan Das
Quoted 31 lines · Andrii Nakryiko
[0] added ability to capture LBR (Last Branch Records) on Intel CPUs from inside BPF program at pretty much any arbitrary point. This is extremely useful capability that allows to figure out otherwise hard-to-debug problems, because LBR is now available based on some application-defined conditions, not just hardware-supported events.
retsnoop ([1]) is one such tool that takes a huge advantage of this functionality and has proved to be an extremely useful tool in practice.
Now, AMD Zen4 CPUs got support for similar LBR functionality, but necessary wiring inside the kernel is not yet setup. This patch seeks to rectify this and follows a similar approach to the original patch [0] for Intel CPUs.
Given LBR can be set up to capture any indirect jumps, it's critical to minimize indirect jumps on the way to requesting LBR from BPF program, so we split
amd_pmu_lbr_disable_all()into a wrapper with some generic conditions vs always-inlined__amd_pmu_lbr_disable()called directly from BPF subsystem (throughperf_snapshot_branch_stackstatic call).Now that it's possible to capture LBR on AMD CPU from BPF at arbitrary point, there is no reason to artificially limit this feature to sampling events. So corresponding check is removed. AFAIU, there is no correctness implications of doing this (and it was possible to bypass this check by just setting perf_event's
sample_periodto 1 anyways, so it doesn't guard all that much).This was tested on AMD Bergamo CPU and worked well when utilized from the aforementioned retsnoop tool.
Quoted 8 lines
Quoted 4 lines
[0] https://lore.kernel.org/bpf/[email protected]/
[1] https://github.com/anakryiko/retsnoopSigned-off-by Andrii Nakryiko <[email protected]>Changes look good to me.
Reviewed-by Sandipan Das <[email protected]>
Thanks for your review! Will split out the check removal, as Ingo asked, and will post v3, thanks!
Quoted 7 lines
Quoted 7 lines
--- arch/x86/events/amd/core.c | 29 ++++++++++++++++++++++++++++- arch/x86/events/amd/lbr.c | 11 +---------- arch/x86/events/perf_event.h | 11 +++++++++++ 3 files changed, 40 insertions(+), 11 deletions(-)
[...]