The following commit has been merged into the perf/urgent branch of tip: Commit-ID: 24919fdea6f8b31d7cdf32ac291bc5dd0b023878 Gitweb: https://git.kernel.org/tip/24919fdea6f8b31d7cdf32ac291bc5dd0b023878 Author: Kan Liang <kan.liang@xxxxxxxxxxxxxxx> AuthorDate: Wed, 31 Aug 2022 07:27:02 -07:00 Committer: Peter Zijlstra <peterz@xxxxxxxxxxxxx> CommitterDate: Thu, 01 Sep 2022 11:19:41 +02:00 perf/x86/intel: Fix unchecked MSR access error for Alder Lake N For some Alder Lake N machine, the below unchecked MSR access error may be triggered. [ 0.088017] rcu: Hierarchical SRCU implementation. [ 0.088017] unchecked MSR access error: WRMSR to 0x38f (tried to write 0x0001000f0000003f) at rIP: 0xffffffffb5684de8 (native_write_msr+0x8/0x30) [ 0.088017] Call Trace: [ 0.088017] <TASK> [ 0.088017] __intel_pmu_enable_all.constprop.46+0x4a/0xa0 The Alder Lake N only has e-cores. The X86_FEATURE_HYBRID_CPU flag is not set. The perf cannot retrieve the correct CPU type via get_this_hybrid_cpu_type(). The model specific get_hybrid_cpu_type() is hardcode to p-core. The wrong CPU type is given to the PMU of the Alder Lake N. Since Alder Lake N isn't in fact a hybrid CPU, remove ALDERLAKE_N from the rest of {ALDER,RAPTOP}LAKE and create a non-hybrid PMU setup. The differences between Gracemont and the previous Tremont are, - Number of GP counters - Load and store latency Events - PEBS event_constraints - Instruction Latency support - Data source encoding - Memory access latency encoding Fixes: c2a960f7c574 ("perf/x86: Add new Alder Lake and Raptor Lake support") Reported-by: Jianfeng Gao <jianfeng.gao@xxxxxxxxx> Suggested-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx> Signed-off-by: Kan Liang <kan.liang@xxxxxxxxxxxxxxx> Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx> Link: https://lkml.kernel.org/r/20220831142702.153110-1-kan.liang@xxxxxxxxxxxxxxx --- arch/x86/events/intel/core.c | 40 ++++++++++++++++++++++++++++++++++- arch/x86/events/intel/ds.c | 9 ++++++-- arch/x86/events/perf_event.h | 2 ++- 3 files changed, 48 insertions(+), 3 deletions(-) diff --git a/arch/x86/events/intel/core.c b/arch/x86/events/intel/core.c index cb98a05..e8eb936 100644 --- a/arch/x86/events/intel/core.c +++ b/arch/x86/events/intel/core.c @@ -2102,6 +2102,15 @@ static struct extra_reg intel_tnt_extra_regs[] __read_mostly = { EVENT_EXTRA_END }; +EVENT_ATTR_STR(mem-loads, mem_ld_grt, "event=0xd0,umask=0x5,ldlat=3"); +EVENT_ATTR_STR(mem-stores, mem_st_grt, "event=0xd0,umask=0x6"); + +static struct attribute *grt_mem_attrs[] = { + EVENT_PTR(mem_ld_grt), + EVENT_PTR(mem_st_grt), + NULL +}; + static struct extra_reg intel_grt_extra_regs[] __read_mostly = { /* must define OFFCORE_RSP_X first, see intel_fixup_er() */ INTEL_UEVENT_EXTRA_REG(0x01b7, MSR_OFFCORE_RSP_0, 0x3fffffffffull, RSP_0), @@ -5974,6 +5983,36 @@ __init int intel_pmu_init(void) name = "Tremont"; break; + case INTEL_FAM6_ALDERLAKE_N: + x86_pmu.mid_ack = true; + memcpy(hw_cache_event_ids, glp_hw_cache_event_ids, + sizeof(hw_cache_event_ids)); + memcpy(hw_cache_extra_regs, tnt_hw_cache_extra_regs, + sizeof(hw_cache_extra_regs)); + hw_cache_event_ids[C(ITLB)][C(OP_READ)][C(RESULT_ACCESS)] = -1; + + x86_pmu.event_constraints = intel_slm_event_constraints; + x86_pmu.pebs_constraints = intel_grt_pebs_event_constraints; + x86_pmu.extra_regs = intel_grt_extra_regs; + + x86_pmu.pebs_aliases = NULL; + x86_pmu.pebs_prec_dist = true; + x86_pmu.pebs_block = true; + x86_pmu.lbr_pt_coexist = true; + x86_pmu.flags |= PMU_FL_HAS_RSP_1; + x86_pmu.flags |= PMU_FL_INSTR_LATENCY; + + intel_pmu_pebs_data_source_grt(); + x86_pmu.pebs_latency_data = adl_latency_data_small; + x86_pmu.get_event_constraints = tnt_get_event_constraints; + x86_pmu.limit_period = spr_limit_period; + td_attr = tnt_events_attrs; + mem_attr = grt_mem_attrs; + extra_attr = nhm_format_attr; + pr_cont("Gracemont events, "); + name = "gracemont"; + break; + case INTEL_FAM6_WESTMERE: case INTEL_FAM6_WESTMERE_EP: case INTEL_FAM6_WESTMERE_EX: @@ -6316,7 +6355,6 @@ __init int intel_pmu_init(void) case INTEL_FAM6_ALDERLAKE: case INTEL_FAM6_ALDERLAKE_L: - case INTEL_FAM6_ALDERLAKE_N: case INTEL_FAM6_RAPTORLAKE: case INTEL_FAM6_RAPTORLAKE_P: /* diff --git a/arch/x86/events/intel/ds.c b/arch/x86/events/intel/ds.c index de1f55d..ac973c6 100644 --- a/arch/x86/events/intel/ds.c +++ b/arch/x86/events/intel/ds.c @@ -110,13 +110,18 @@ void __init intel_pmu_pebs_data_source_skl(bool pmem) __intel_pmu_pebs_data_source_skl(pmem, pebs_data_source); } -static void __init intel_pmu_pebs_data_source_grt(u64 *data_source) +static void __init __intel_pmu_pebs_data_source_grt(u64 *data_source) { data_source[0x05] = OP_LH | P(LVL, L3) | LEVEL(L3) | P(SNOOP, HIT); data_source[0x06] = OP_LH | P(LVL, L3) | LEVEL(L3) | P(SNOOP, HITM); data_source[0x08] = OP_LH | P(LVL, L3) | LEVEL(L3) | P(SNOOPX, FWD); } +void __init intel_pmu_pebs_data_source_grt(void) +{ + __intel_pmu_pebs_data_source_grt(pebs_data_source); +} + void __init intel_pmu_pebs_data_source_adl(void) { u64 *data_source; @@ -127,7 +132,7 @@ void __init intel_pmu_pebs_data_source_adl(void) data_source = x86_pmu.hybrid_pmu[X86_HYBRID_PMU_ATOM_IDX].pebs_data_source; memcpy(data_source, pebs_data_source, sizeof(pebs_data_source)); - intel_pmu_pebs_data_source_grt(data_source); + __intel_pmu_pebs_data_source_grt(data_source); } static u64 precise_store_data(u64 status) diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h index ba3d24a..266143a 100644 --- a/arch/x86/events/perf_event.h +++ b/arch/x86/events/perf_event.h @@ -1516,6 +1516,8 @@ void intel_pmu_pebs_data_source_skl(bool pmem); void intel_pmu_pebs_data_source_adl(void); +void intel_pmu_pebs_data_source_grt(void); + int intel_pmu_setup_lbr_filter(struct perf_event *event); void intel_pt_interrupt(void);
![]() |