From ee6825c80e870fff1a370c718ec77022ade0889b Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 25 Mar 2016 15:52:34 +0100 Subject: x86/topology: Fix AMD core count It turns out AMD gets x86_max_cores wrong when there are compute units. The issue is that Linux assumes: nr_logical_cpus = nr_cores * nr_siblings But AMD reports its CU unit as 2 cores, but then sets num_smp_siblings to 2 as well. Boris: fixup ras/mce_amd_inj.c too, to compute the Node Base Core properly, according to the new nomenclature. Fixes: 1f12e32f4cd5 ("x86/topology: Create logical package id") Reported-by: Xiong Zhou Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Borislav Petkov Cc: Andreas Herrmann Cc: Andy Lutomirski Link: http://lkml.kernel.org/r/20160317095220.GO6344@twins.programming.kicks-ass.net Signed-off-by: Thomas Gleixner --- arch/x86/include/asm/smp.h | 1 + arch/x86/kernel/cpu/amd.c | 8 ++++---- arch/x86/ras/mce_amd_inj.c | 3 ++- 3 files changed, 7 insertions(+), 5 deletions(-) (limited to 'arch') diff --git a/arch/x86/include/asm/smp.h b/arch/x86/include/asm/smp.h index 20a3de5cb3b0..66b057306f40 100644 --- a/arch/x86/include/asm/smp.h +++ b/arch/x86/include/asm/smp.h @@ -155,6 +155,7 @@ static inline int wbinvd_on_all_cpus(void) wbinvd(); return 0; } +#define smp_num_siblings 1 #endif /* CONFIG_SMP */ extern unsigned disabled_cpus; diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 6e47e3a916f1..4d0087f94ee5 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -312,9 +312,9 @@ static void amd_get_topology(struct cpuinfo_x86 *c) node_id = ecx & 7; /* get compute unit information */ - smp_num_siblings = ((ebx >> 8) & 3) + 1; + cores_per_cu = smp_num_siblings = ((ebx >> 8) & 3) + 1; + c->x86_max_cores /= smp_num_siblings; c->compute_unit_id = ebx & 0xff; - cores_per_cu += ((ebx >> 8) & 3); } else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) { u64 value; @@ -329,8 +329,8 @@ static void amd_get_topology(struct cpuinfo_x86 *c) u32 cus_per_node; set_cpu_cap(c, X86_FEATURE_AMD_DCM); - cores_per_node = c->x86_max_cores / nodes_per_socket; - cus_per_node = cores_per_node / cores_per_cu; + cus_per_node = c->x86_max_cores / nodes_per_socket; + cores_per_node = cus_per_node * cores_per_cu; /* store NodeID, use llc_shared_map to store sibling info */ per_cpu(cpu_llc_id, cpu) = node_id; diff --git a/arch/x86/ras/mce_amd_inj.c b/arch/x86/ras/mce_amd_inj.c index 55d38cfa46c2..9e02dcaef683 100644 --- a/arch/x86/ras/mce_amd_inj.c +++ b/arch/x86/ras/mce_amd_inj.c @@ -20,6 +20,7 @@ #include #include +#include #include #include @@ -206,7 +207,7 @@ static u32 get_nbc_for_node(int node_id) struct cpuinfo_x86 *c = &boot_cpu_data; u32 cores_per_node; - cores_per_node = c->x86_max_cores / amd_get_nodes_per_socket(); + cores_per_node = (c->x86_max_cores * smp_num_siblings) / amd_get_nodes_per_socket(); return cores_per_node * node_id; } -- cgit v1.2.3 From 32b62f446827f696cc474a6d83cea93693c5ed49 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Fri, 25 Mar 2016 15:52:35 +0100 Subject: perf/x86/amd: Cleanup Fam10h NB event constraints Avoid allocating the AMD NB event constraints data structure when not needed. This gets rid of x86_max_cores usage and avoids allocating this on AMD Core Perfctr supporting hardware (which has separate MSRs for NB events). Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Borislav Petkov Cc: aherrmann@suse.com Cc: Rui Huang Cc: Borislav Petkov Cc: jencce.kernel@gmail.com Link: http://lkml.kernel.org/r/20160320124629.GY6375@twins.programming.kicks-ass.net Signed-off-by: Thomas Gleixner --- arch/x86/events/amd/core.c | 21 ++++++++++++++++++--- arch/x86/events/perf_event.h | 5 +++++ 2 files changed, 23 insertions(+), 3 deletions(-) (limited to 'arch') diff --git a/arch/x86/events/amd/core.c b/arch/x86/events/amd/core.c index 049ada8d4e9c..86a9bec18dab 100644 --- a/arch/x86/events/amd/core.c +++ b/arch/x86/events/amd/core.c @@ -369,7 +369,7 @@ static int amd_pmu_cpu_prepare(int cpu) WARN_ON_ONCE(cpuc->amd_nb); - if (boot_cpu_data.x86_max_cores < 2) + if (!x86_pmu.amd_nb_constraints) return NOTIFY_OK; cpuc->amd_nb = amd_alloc_nb(cpu); @@ -388,7 +388,7 @@ static void amd_pmu_cpu_starting(int cpu) cpuc->perf_ctr_virt_mask = AMD64_EVENTSEL_HOSTONLY; - if (boot_cpu_data.x86_max_cores < 2) + if (!x86_pmu.amd_nb_constraints) return; nb_id = amd_get_nb_id(cpu); @@ -414,7 +414,7 @@ static void amd_pmu_cpu_dead(int cpu) { struct cpu_hw_events *cpuhw; - if (boot_cpu_data.x86_max_cores < 2) + if (!x86_pmu.amd_nb_constraints) return; cpuhw = &per_cpu(cpu_hw_events, cpu); @@ -648,6 +648,8 @@ static __initconst const struct x86_pmu amd_pmu = { .cpu_prepare = amd_pmu_cpu_prepare, .cpu_starting = amd_pmu_cpu_starting, .cpu_dead = amd_pmu_cpu_dead, + + .amd_nb_constraints = 1, }; static int __init amd_core_pmu_init(void) @@ -674,6 +676,11 @@ static int __init amd_core_pmu_init(void) x86_pmu.eventsel = MSR_F15H_PERF_CTL; x86_pmu.perfctr = MSR_F15H_PERF_CTR; x86_pmu.num_counters = AMD64_NUM_COUNTERS_CORE; + /* + * AMD Core perfctr has separate MSRs for the NB events, see + * the amd/uncore.c driver. + */ + x86_pmu.amd_nb_constraints = 0; pr_cont("core perfctr, "); return 0; @@ -693,6 +700,14 @@ __init int amd_pmu_init(void) if (ret) return ret; + if (num_possible_cpus() == 1) { + /* + * No point in allocating data structures to serialize + * against other CPUs, when there is only the one CPU. + */ + x86_pmu.amd_nb_constraints = 0; + } + /* Events are common for all AMDs */ memcpy(hw_cache_event_ids, amd_hw_cache_event_ids, sizeof(hw_cache_event_ids)); diff --git a/arch/x86/events/perf_event.h b/arch/x86/events/perf_event.h index ba6ef18528c9..716d0482f5db 100644 --- a/arch/x86/events/perf_event.h +++ b/arch/x86/events/perf_event.h @@ -607,6 +607,11 @@ struct x86_pmu { */ atomic_t lbr_exclusive[x86_lbr_exclusive_max]; + /* + * AMD bits + */ + unsigned int amd_nb_constraints : 1; + /* * Extra registers for events */ -- cgit v1.2.3 From 8196dab4fc159943df6baaac04973bb1accb7100 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Fri, 25 Mar 2016 15:52:36 +0100 Subject: x86/cpu: Get rid of compute_unit_id It is cpu_core_id anyway. Signed-off-by: Borislav Petkov Link: http://lkml.kernel.org/r/1458917557-8757-3-git-send-email-bp@alien8.de Signed-off-by: Thomas Gleixner --- arch/x86/include/asm/processor.h | 2 -- arch/x86/kernel/amd_nb.c | 6 ++---- arch/x86/kernel/cpu/amd.c | 10 +++------- arch/x86/kernel/smpboot.c | 2 +- 4 files changed, 6 insertions(+), 14 deletions(-) (limited to 'arch') diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h index 983738ac014c..9264476f3d57 100644 --- a/arch/x86/include/asm/processor.h +++ b/arch/x86/include/asm/processor.h @@ -132,8 +132,6 @@ struct cpuinfo_x86 { u16 logical_proc_id; /* Core id: */ u16 cpu_core_id; - /* Compute unit id */ - u8 compute_unit_id; /* Index into per_cpu list: */ u16 cpu_index; u32 microcode; diff --git a/arch/x86/kernel/amd_nb.c b/arch/x86/kernel/amd_nb.c index 29fa475ec518..a147e676fc7b 100644 --- a/arch/x86/kernel/amd_nb.c +++ b/arch/x86/kernel/amd_nb.c @@ -170,15 +170,13 @@ int amd_get_subcaches(int cpu) { struct pci_dev *link = node_to_amd_nb(amd_get_nb_id(cpu))->link; unsigned int mask; - int cuid; if (!amd_nb_has_feature(AMD_NB_L3_PARTITIONING)) return 0; pci_read_config_dword(link, 0x1d4, &mask); - cuid = cpu_data(cpu).compute_unit_id; - return (mask >> (4 * cuid)) & 0xf; + return (mask >> (4 * cpu_data(cpu).cpu_core_id)) & 0xf; } int amd_set_subcaches(int cpu, unsigned long mask) @@ -204,7 +202,7 @@ int amd_set_subcaches(int cpu, unsigned long mask) pci_write_config_dword(nb->misc, 0x1b8, reg & ~0x180000); } - cuid = cpu_data(cpu).compute_unit_id; + cuid = cpu_data(cpu).cpu_core_id; mask <<= 4 * cuid; mask |= (0xf ^ (1 << cuid)) << 26; diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 4d0087f94ee5..7b76eb67a9b3 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -300,7 +300,6 @@ static int nearby_node(int apicid) #ifdef CONFIG_SMP static void amd_get_topology(struct cpuinfo_x86 *c) { - u32 cores_per_cu = 1; u8 node_id; int cpu = smp_processor_id(); @@ -312,9 +311,9 @@ static void amd_get_topology(struct cpuinfo_x86 *c) node_id = ecx & 7; /* get compute unit information */ - cores_per_cu = smp_num_siblings = ((ebx >> 8) & 3) + 1; + smp_num_siblings = ((ebx >> 8) & 3) + 1; c->x86_max_cores /= smp_num_siblings; - c->compute_unit_id = ebx & 0xff; + c->cpu_core_id = ebx & 0xff; } else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) { u64 value; @@ -325,19 +324,16 @@ static void amd_get_topology(struct cpuinfo_x86 *c) /* fixup multi-node processor information */ if (nodes_per_socket > 1) { - u32 cores_per_node; u32 cus_per_node; set_cpu_cap(c, X86_FEATURE_AMD_DCM); cus_per_node = c->x86_max_cores / nodes_per_socket; - cores_per_node = cus_per_node * cores_per_cu; /* store NodeID, use llc_shared_map to store sibling info */ per_cpu(cpu_llc_id, cpu) = node_id; /* core id has to be in the [0 .. cores_per_node - 1] range */ - c->cpu_core_id %= cores_per_node; - c->compute_unit_id %= cus_per_node; + c->cpu_core_id %= cus_per_node; } } #endif diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index b2c99f811c3f..a2065d3b3b39 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -422,7 +422,7 @@ static bool match_smt(struct cpuinfo_x86 *c, struct cpuinfo_x86 *o) if (c->phys_proc_id == o->phys_proc_id && per_cpu(cpu_llc_id, cpu1) == per_cpu(cpu_llc_id, cpu2) && - c->compute_unit_id == o->compute_unit_id) + c->cpu_core_id == o->cpu_core_id) return topology_sane(c, o, "smt"); } else if (c->phys_proc_id == o->phys_proc_id && -- cgit v1.2.3 From 4a6772f514891eaacf26bcb7c2c808c557d23c6f Mon Sep 17 00:00:00 2001 From: Vladimir Zapolskiy Date: Sat, 26 Mar 2016 20:47:00 +0200 Subject: x86/cpufreq: Remove duplicated TDP MSR macro definitions The list of CPU model specific registers contains two copies of TDP registers, remove the one, which is out of numerical order in the list. Fixes: 6a35fc2d6c22 ("cpufreq: intel_pstate: get P1 from TAR when available") Signed-off-by: Vladimir Zapolskiy Cc: Len Brown Cc: "Rafael J. Wysocki" Cc: Kristen Carlson Accardi Cc: Srinivas Pandruvada Link: http://lkml.kernel.org/r/1459018020-24577-1-git-send-email-vladimir_zapolskiy@mentor.com Signed-off-by: Thomas Gleixner --- arch/x86/include/asm/msr-index.h | 8 +------- 1 file changed, 1 insertion(+), 7 deletions(-) (limited to 'arch') diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h index 2da46ac16e37..426e946ed0c0 100644 --- a/arch/x86/include/asm/msr-index.h +++ b/arch/x86/include/asm/msr-index.h @@ -190,6 +190,7 @@ #define MSR_PP1_ENERGY_STATUS 0x00000641 #define MSR_PP1_POLICY 0x00000642 +/* Config TDP MSRs */ #define MSR_CONFIG_TDP_NOMINAL 0x00000648 #define MSR_CONFIG_TDP_LEVEL_1 0x00000649 #define MSR_CONFIG_TDP_LEVEL_2 0x0000064A @@ -210,13 +211,6 @@ #define MSR_GFX_PERF_LIMIT_REASONS 0x000006B0 #define MSR_RING_PERF_LIMIT_REASONS 0x000006B1 -/* Config TDP MSRs */ -#define MSR_CONFIG_TDP_NOMINAL 0x00000648 -#define MSR_CONFIG_TDP_LEVEL1 0x00000649 -#define MSR_CONFIG_TDP_LEVEL2 0x0000064A -#define MSR_CONFIG_TDP_CONTROL 0x0000064B -#define MSR_TURBO_ACTIVATION_RATIO 0x0000064C - /* Hardware P state interface */ #define MSR_PPERF 0x0000064e #define MSR_PERF_LIMIT_REASONS 0x0000064f -- cgit v1.2.3 From 5f870a3f7188065e13efafe1faeb01c136173bc4 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Mon, 28 Mar 2016 20:20:17 +0200 Subject: x86/thread_info: Merge two !__ASSEMBLY__ sections We have #ifndef __ASSEMBLY__ ... #endif #ifndef __ASSEMBLY__ ... #endif Merge the two. No functionality change. Signed-off-by: Borislav Petkov Link: http://lkml.kernel.org/r/1459189217-25532-1-git-send-email-bp@alien8.de Signed-off-by: Thomas Gleixner --- arch/x86/include/asm/thread_info.h | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'arch') diff --git a/arch/x86/include/asm/thread_info.h b/arch/x86/include/asm/thread_info.h index 82866697fcf1..ffae84df8a93 100644 --- a/arch/x86/include/asm/thread_info.h +++ b/arch/x86/include/asm/thread_info.h @@ -276,11 +276,9 @@ static inline bool is_ia32_task(void) */ #define force_iret() set_thread_flag(TIF_NOTIFY_RESUME) -#endif /* !__ASSEMBLY__ */ - -#ifndef __ASSEMBLY__ extern void arch_task_cache_init(void); extern int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src); extern void arch_release_task_struct(struct task_struct *tsk); -#endif +#endif /* !__ASSEMBLY__ */ + #endif /* _ASM_X86_THREAD_INFO_H */ -- cgit v1.2.3 From 34a4cceb78e48c75d1b48b25352a3f3b2cc2b2da Mon Sep 17 00:00:00 2001 From: Huang Rui Date: Fri, 25 Mar 2016 10:08:40 +0800 Subject: x86/cpu: Add advanced power management bits Bit 11 of CPUID 8000_0007 edx is processor feedback interface. Bit 12 of CPUID 8000_0007 edx is accumulated power. Print proper names in proc/cpuinfo Reported-and-tested-by: Borislav Petkov Signed-off-by: Huang Rui Cc: Tony Li Cc: Fenghua Yu Cc: Tony Luck Cc: Peter Zijlstra Cc: "Rafael J. Wysocki" Cc: Andy Lutomirski Cc: Fengguang Wu Cc: Sherry Hurwitz Cc: Borislav Petkov Cc: "Len Brown" Link: http://lkml.kernel.org/r/1458871720-3209-1-git-send-email-ray.huang@amd.com Signed-off-by: Thomas Gleixner --- arch/x86/kernel/cpu/powerflags.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch') diff --git a/arch/x86/kernel/cpu/powerflags.c b/arch/x86/kernel/cpu/powerflags.c index 31f0f335ed22..1dd8294fd730 100644 --- a/arch/x86/kernel/cpu/powerflags.c +++ b/arch/x86/kernel/cpu/powerflags.c @@ -18,4 +18,6 @@ const char *const x86_power_flags[32] = { "", /* tsc invariant mapped to constant_tsc */ "cpb", /* core performance boost */ "eff_freq_ro", /* Readonly aperf/mperf */ + "proc_feedback", /* processor feedback interface */ + "acc_power", /* accumulated power mechanism */ }; -- cgit v1.2.3