From 9d260ebc09a0ad6b5c73e17676df42c7bc75ff64 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Wed, 16 Dec 2009 15:43:55 +0100 Subject: x86, amd: Get multi-node CPU info from NodeId MSR instead of PCI config space Use NodeId MSR to get NodeId and number of nodes per processor. Signed-off-by: Andreas Herrmann LKML-Reference: <20091216144355.GB28798@alberich.amd.com> Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/amd.c | 53 ++++++++++++++--------------------------------- 1 file changed, 15 insertions(+), 38 deletions(-) (limited to 'arch/x86/kernel/cpu') diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index 8dc3ea145c9..e485825130d 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -254,59 +254,36 @@ static int __cpuinit nearby_node(int apicid) /* * Fixup core topology information for AMD multi-node processors. - * Assumption 1: Number of cores in each internal node is the same. - * Assumption 2: Mixed systems with both single-node and dual-node - * processors are not supported. + * Assumption: Number of cores in each internal node is the same. */ #ifdef CONFIG_X86_HT static void __cpuinit amd_fixup_dcm(struct cpuinfo_x86 *c) { -#ifdef CONFIG_PCI - u32 t, cpn; - u8 n, n_id; + unsigned long long value; + u32 nodes, cores_per_node; int cpu = smp_processor_id(); + if (!cpu_has(c, X86_FEATURE_NODEID_MSR)) + return; + /* fixup topology information only once for a core */ if (cpu_has(c, X86_FEATURE_AMD_DCM)) return; - /* check for multi-node processor on boot cpu */ - t = read_pci_config(0, 24, 3, 0xe8); - if (!(t & (1 << 29))) + rdmsrl(MSR_FAM10H_NODE_ID, value); + + nodes = ((value >> 3) & 7) + 1; + if (nodes == 1) return; set_cpu_cap(c, X86_FEATURE_AMD_DCM); + cores_per_node = c->x86_max_cores / nodes; - /* cores per node: each internal node has half the number of cores */ - cpn = c->x86_max_cores >> 1; + /* store NodeID, use llc_shared_map to store sibling info */ + per_cpu(cpu_llc_id, cpu) = value & 7; - /* even-numbered NB_id of this dual-node processor */ - n = c->phys_proc_id << 1; - - /* - * determine internal node id and assign cores fifty-fifty to - * each node of the dual-node processor - */ - t = read_pci_config(0, 24 + n, 3, 0xe8); - n = (t>>30) & 0x3; - if (n == 0) { - if (c->cpu_core_id < cpn) - n_id = 0; - else - n_id = 1; - } else { - if (c->cpu_core_id < cpn) - n_id = 1; - else - n_id = 0; - } - - /* compute entire NodeID, use llc_shared_map to store sibling info */ - per_cpu(cpu_llc_id, cpu) = (c->phys_proc_id << 1) + n_id; - - /* fixup core id to be in range from 0 to cpn */ - c->cpu_core_id = c->cpu_core_id % cpn; -#endif + /* fixup core id to be in range from 0 to (cores_per_node - 1) */ + c->cpu_core_id = c->cpu_core_id % cores_per_node; } #endif -- cgit v1.2.3-70-g09d2 From a4636818f8e0991f32d9528f39cf4f3d6a7d30a3 Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Thu, 17 Dec 2009 11:43:29 -0600 Subject: cpumask: rename tsk_cpumask to tsk_cpus_allowed Noone uses this wrapper yet, and Ingo asked that it be kept consistent with current task_struct usage. (One user crept in via linux-next: fixed) Signed-off-by: Rusty Russell Cc: Tejun Heo --- arch/x86/kernel/cpu/cpufreq/powernow-k8.c | 2 +- include/linux/sched.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel/cpu') diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index a9df9441a9a..f125e5c551c 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -1136,7 +1136,7 @@ static int powernowk8_target(struct cpufreq_policy *pol, if (!alloc_cpumask_var(&oldmask, GFP_KERNEL)) return -ENOMEM; - cpumask_copy(oldmask, tsk_cpumask(current)); + cpumask_copy(oldmask, tsk_cpus_allowed(current)); set_cpus_allowed_ptr(current, cpumask_of(pol->cpu)); if (smp_processor_id() != pol->cpu) { diff --git a/include/linux/sched.h b/include/linux/sched.h index 244c287a5ac..4d7adb282bd 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1555,7 +1555,7 @@ struct task_struct { }; /* Future-safe accessor for struct task_struct's cpus_allowed. */ -#define tsk_cpumask(tsk) (&(tsk)->cpus_allowed) +#define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed) /* * Priority of a process goes from 0..MAX_PRIO-1, valid RT -- cgit v1.2.3-70-g09d2 From 6c56ccecf05fafe100ab4ea94f6fccbf5ff00db7 Mon Sep 17 00:00:00 2001 From: "Pallipadi, Venkatesh" Date: Thu, 17 Dec 2009 12:27:02 -0800 Subject: x86: Reenable TSC sync check at boot, even with NONSTOP_TSC Commit 83ce4009 did the following change If the TSC is constant and non-stop, also set it reliable. But, there seems to be few systems that will end up with TSC warp across sockets, depending on how the cpus come out of reset. Skipping TSC sync test on such systems may result in time inconsistency later. So, reenable TSC sync test even on constant and non-stop TSC systems. Set, sched_clock_stable to 1 by default and reset it in mark_tsc_unstable, if TSC sync fails. This change still gives perf benefit mentioned in 83ce4009 for systems where TSC is reliable. Signed-off-by: Venkatesh Pallipadi Acked-by: Suresh Siddha LKML-Reference: <20091217202702.GA18015@linux-os.sc.intel.com> Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/intel.c | 1 - arch/x86/kernel/tsc.c | 1 + 2 files changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel/cpu') diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index 9c31e8b09d2..879666f4d87 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -70,7 +70,6 @@ static void __cpuinit early_init_intel(struct cpuinfo_x86 *c) if (c->x86_power & (1 << 8)) { set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); set_cpu_cap(c, X86_FEATURE_NONSTOP_TSC); - set_cpu_cap(c, X86_FEATURE_TSC_RELIABLE); sched_clock_stable = 1; } diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index cd982f48e23..597683aa5ba 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -763,6 +763,7 @@ void mark_tsc_unstable(char *reason) { if (!tsc_unstable) { tsc_unstable = 1; + sched_clock_stable = 0; printk(KERN_INFO "Marking TSC unstable due to %s\n", reason); /* Change only the rating, when not registered */ if (clocksource_tsc.mult) -- cgit v1.2.3-70-g09d2 From 4a28395d72a956f2dad24e343d06bc08c9afb89a Mon Sep 17 00:00:00 2001 From: Andrew Morton Date: Mon, 21 Dec 2009 16:19:58 -0800 Subject: arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c: avoid cross-CPU interrupts by using smp_call_function_any() Presently acpi-cpufreq will perform the MSR read on the first CPU in the mask. That's inefficient if that CPU differs from the current CPU. Because we have to perform a cross-CPU call, but we could have run the rdmsr on the current CPU. So switch to using the new smp_call_function_any(), which will perform the call on the current CPU if that CPU is present in the mask (it is). Cc: "Zhang, Yanmin" Cc: Dave Jones Cc: Ingo Molnar Cc: Jaswinder Singh Rajput Cc: Len Brown Cc: Mike Galbraith Cc: Rusty Russell Cc: Thomas Gleixner Cc: Venkatesh Pallipadi Cc: Zhao Yakui Signed-off-by: Andrew Morton Signed-off-by: Len Brown --- arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel/cpu') diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index f28decf8dde..1b1920fa7c8 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -190,9 +190,11 @@ static void do_drv_write(void *_cmd) static void drv_read(struct drv_cmd *cmd) { + int err; cmd->val = 0; - smp_call_function_single(cpumask_any(cmd->mask), do_drv_read, cmd, 1); + err = smp_call_function_any(cmd->mask, do_drv_read, cmd, 1); + WARN_ON_ONCE(err); /* smp_call_function_any() was buggy? */ } static void drv_write(struct drv_cmd *cmd) -- cgit v1.2.3-70-g09d2 From 48b5ba9cc98d676712da29d9931f1c88e5185ff2 Mon Sep 17 00:00:00 2001 From: Frederic Weisbecker Date: Thu, 31 Dec 2009 05:53:02 +0100 Subject: perf: Pass appropriate frame pointer to dump_trace() Pass the frame pointer from the regs of the interrupted path to dump_trace() while processing the stack trace. Currently, dump_trace() takes the current bp and starts the callchain from dump_trace() itself. This is wasteful because we need to walk through the entire NMI/DEBUG stack before retrieving the interrupted point. We can fix that by just using the frame pointer from the captured regs. It points exactly where we want to start. Signed-off-by: Frederic Weisbecker Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras LKML-Reference: <1262235183-5320-1-git-send-regression-fweisbec@gmail.com> Signed-off-by: Ingo Molnar Cc: Peter Zijlstra Cc: Arnaldo Carvalho de Melo Cc: Paul Mackerras --- arch/x86/kernel/cpu/perf_event.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel/cpu') diff --git a/arch/x86/kernel/cpu/perf_event.c b/arch/x86/kernel/cpu/perf_event.c index c223b7e895d..d616c06e99b 100644 --- a/arch/x86/kernel/cpu/perf_event.c +++ b/arch/x86/kernel/cpu/perf_event.c @@ -2347,7 +2347,7 @@ perf_callchain_kernel(struct pt_regs *regs, struct perf_callchain_entry *entry) callchain_store(entry, PERF_CONTEXT_KERNEL); callchain_store(entry, regs->ip); - dump_trace(NULL, regs, NULL, 0, &backtrace_ops, entry); + dump_trace(NULL, regs, NULL, regs->bp, &backtrace_ops, entry); } /* -- cgit v1.2.3-70-g09d2