From: Suresh Siddha Subject: x86: use cpuid vector 0xb when available for detecting cpu topology References: fate #303948 and fate #303984 Patch-Mainline: queued for .28 Commit-ID: bbb65d2d365efe9951290e61678dcf81ec60add4 Signed-off-by: Thomas Renninger cpuid leaf 0xb provides extended topology enumeration. This interface provides the 32-bit x2APIC id of the logical processor and it also provides a new mechanism to detect SMT and core siblings (which provides increased addressability). Signed-off-by: Suresh Siddha Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/addon_cpuid_features.c | 86 +++++++++++++++++++++++++++++ arch/x86/kernel/cpu/common_64.c | 3 + arch/x86/kernel/cpu/intel.c | 11 +++ arch/x86/kernel/cpu/intel_64.c | 5 + include/asm-x86/cpufeature.h | 2 include/asm-x86/processor.h | 1 6 files changed, 105 insertions(+), 3 deletions(-) --- a/arch/x86/kernel/cpu/addon_cpuid_features.c +++ b/arch/x86/kernel/cpu/addon_cpuid_features.c @@ -7,6 +7,8 @@ #include #include +#include + struct cpuid_bit { u16 feature; u8 reg; @@ -48,6 +50,90 @@ void __cpuinit init_scattered_cpuid_feat } } +/* leaf 0xb SMT level */ +#define SMT_LEVEL 0 + +/* leaf 0xb sub-leaf types */ +#define INVALID_TYPE 0 +#define SMT_TYPE 1 +#define CORE_TYPE 2 + +#define LEAFB_SUBTYPE(ecx) (((ecx) >> 8) & 0xff) +#define BITS_SHIFT_NEXT_LEVEL(eax) ((eax) & 0x1f) +#define LEVEL_MAX_SIBLINGS(ebx) ((ebx) & 0xffff) + +/* + * Check for extended topology enumeration cpuid leaf 0xb and if it + * exists, use it for populating initial_apicid and cpu topology + * detection. + */ +void __cpuinit detect_extended_topology(struct cpuinfo_x86 *c) +{ + unsigned int eax, ebx, ecx, edx, sub_index; + unsigned int ht_mask_width, core_plus_mask_width; + unsigned int core_select_mask, core_level_siblings; + + if (c->cpuid_level < 0xb) + return; + + cpuid_count(0xb, SMT_LEVEL, &eax, &ebx, &ecx, &edx); + + /* + * check if the cpuid leaf 0xb is actually implemented. + */ + if (ebx == 0 || (LEAFB_SUBTYPE(ecx) != SMT_TYPE)) + return; + + set_cpu_cap(c, X86_FEATURE_XTOPOLOGY); + + /* + * initial apic id, which also represents 32-bit extended x2apic id. + */ + c->initial_apicid = edx; + + /* + * Populate HT related information from sub-leaf level 0. + */ + core_level_siblings = smp_num_siblings = LEVEL_MAX_SIBLINGS(ebx); + core_plus_mask_width = ht_mask_width = BITS_SHIFT_NEXT_LEVEL(eax); + + sub_index = 1; + do { + cpuid_count(0xb, sub_index, &eax, &ebx, &ecx, &edx); + + /* + * Check for the Core type in the implemented sub leaves. + */ + if (LEAFB_SUBTYPE(ecx) == CORE_TYPE) { + core_level_siblings = LEVEL_MAX_SIBLINGS(ebx); + core_plus_mask_width = BITS_SHIFT_NEXT_LEVEL(eax); + break; + } + + sub_index++; + } while (LEAFB_SUBTYPE(ecx) != INVALID_TYPE); + + core_select_mask = (~(-1 << core_plus_mask_width)) >> ht_mask_width; + +#ifdef CONFIG_X86_32 + c->cpu_core_id = phys_pkg_id(c->initial_apicid, ht_mask_width) + & core_select_mask; + c->phys_proc_id = phys_pkg_id(c->initial_apicid, core_plus_mask_width); +#else + c->cpu_core_id = phys_pkg_id(ht_mask_width) & core_select_mask; + c->phys_proc_id = phys_pkg_id(core_plus_mask_width); +#endif + c->x86_max_cores = (core_level_siblings / smp_num_siblings); + + + printk(KERN_INFO "CPU: Physical Processor ID: %d\n", + c->phys_proc_id); + if (c->x86_max_cores > 1) + printk(KERN_INFO "CPU: Processor Core ID: %d\n", + c->cpu_core_id); + return; +} + #ifdef CONFIG_X86_PAT void __cpuinit validate_pat_support(struct cpuinfo_x86 *c) { --- a/arch/x86/kernel/cpu/common_64.c +++ b/arch/x86/kernel/cpu/common_64.c @@ -128,6 +128,9 @@ void __cpuinit detect_ht(struct cpuinfo_ u32 eax, ebx, ecx, edx; int index_msb, core_bits; + if (cpu_has(c, X86_FEATURE_XTOPOLOGY)) + return; + cpuid(1, &eax, &ebx, &ecx, &edx); --- a/arch/x86/kernel/cpu/intel_64.c +++ b/arch/x86/kernel/cpu/intel_64.c @@ -93,7 +93,10 @@ static void __cpuinit init_intel(struct if (c->x86 == 6) set_cpu_cap(c, X86_FEATURE_REP_GOOD); set_cpu_cap(c, X86_FEATURE_LFENCE_RDTSC); - c->x86_max_cores = intel_num_cpu_cores(c); + + detect_extended_topology(c); + if (!cpu_has(c, X86_FEATURE_XTOPOLOGY)) + c->x86_max_cores = intel_num_cpu_cores(c); srat_detect_node(); } --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -196,9 +196,16 @@ static void __cpuinit init_intel(struct if (p) strcpy(c->x86_model_id, p); - c->x86_max_cores = num_cpu_cores(c); + detect_extended_topology(c); - detect_ht(c); + if (!cpu_has(c, X86_FEATURE_XTOPOLOGY)) { + /* + * let's use the legacy cpuid vector 0x1 and 0x4 for topology + * detection. + */ + c->x86_max_cores = num_cpu_cores(c); + detect_ht(c); + } /* Work around errata */ Intel_errata_workarounds(c); --- a/include/asm-x86/cpufeature.h +++ b/include/asm-x86/cpufeature.h @@ -82,6 +82,8 @@ #define X86_FEATURE_11AP (3*32+19) /* Bad local APIC aka 11AP */ #define X86_FEATURE_NOPL (3*32+20) /* The NOPL (0F 1F) instructions */ #define X86_FEATURE_AMDC1E (3*32+21) /* AMD C1E detected */ +#define X86_FEATURE_XTOPOLOGY (3*32+22) /* cpu topology enum extensions */ + /* Intel-defined CPU features, CPUID level 0x00000001 (ecx), word 4 */ #define X86_FEATURE_XMM3 (4*32+ 0) /* Streaming SIMD Extensions-3 */ --- a/include/asm-x86/processor.h +++ b/include/asm-x86/processor.h @@ -161,6 +161,7 @@ extern void init_scattered_cpuid_feature extern unsigned int init_intel_cacheinfo(struct cpuinfo_x86 *c); extern unsigned short num_cache_leaves; +extern void detect_extended_topology(struct cpuinfo_x86 *c); #if defined(CONFIG_X86_HT) || defined(CONFIG_X86_64) extern void detect_ht(struct cpuinfo_x86 *c); #else