From 6ddd2a27948f0bd02a2ad001e8a6816898eba0dc Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 9 Jun 2008 16:59:53 +0200 Subject: x86: simplify idle selection default_idle is selected in cpu_idle(), when no other idle routine is selected. Select it in select_idle_routine() when mwait is not selected. Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- arch/x86/kernel/process.c | 18 +++++++----------- 1 file changed, 7 insertions(+), 11 deletions(-) (limited to 'arch/x86/kernel/process.c') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index ba370dc8685b..b3078f4ce25b 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -139,27 +139,23 @@ static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c) void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c) { - static int selected; - - if (selected) - return; #ifdef CONFIG_X86_SMP if (pm_idle == poll_idle && smp_num_siblings > 1) { printk(KERN_WARNING "WARNING: polling idle and HT enabled," " performance may degrade.\n"); } #endif + if (pm_idle) + return; + if (cpu_has(c, X86_FEATURE_MWAIT) && mwait_usable(c)) { /* - * Skip, if setup has overridden idle. * One CPU supports mwait => All CPUs supports mwait */ - if (!pm_idle) { - printk(KERN_INFO "using mwait in idle threads.\n"); - pm_idle = mwait_idle; - } - } - selected = 1; + printk(KERN_INFO "using mwait in idle threads.\n"); + pm_idle = mwait_idle; + } else + pm_idle = default_idle; } static int __init idle_setup(char *str) -- cgit v1.2.3 From 09fd4b4ef5bc7e5222c13acec1bee8cd252fb63f Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 9 Jun 2008 18:04:27 +0200 Subject: x86: use cpuid to check MWAIT support for C1 cpuid(0x05) provides extended information about MWAIT in EDX when bit 0 of ECX is set. Bit 4-7 of EDX determine whether MWAIT is supported for C1. C1E enabled CPUs have these bits set to 0. Based on an earlier patch from Andi Kleen. Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- arch/x86/kernel/process.c | 28 ++++++++++++++++++++-------- 1 file changed, 20 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel/process.c') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index b3078f4ce25b..fe415ba606e0 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -122,19 +122,31 @@ static void poll_idle(void) * * idle=mwait overrides this decision and forces the usage of mwait. */ + +#define MWAIT_INFO 0x05 +#define MWAIT_ECX_EXTENDED_INFO 0x01 +#define MWAIT_EDX_C1 0xf0 + static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c) { + u32 eax, ebx, ecx, edx; + if (force_mwait) return 1; - if (c->x86_vendor == X86_VENDOR_AMD) { - switch(c->x86) { - case 0x10: - case 0x11: - return 0; - } - } - return 1; + if (c->cpuid_level < MWAIT_INFO) + return 0; + + cpuid(MWAIT_INFO, &eax, &ebx, &ecx, &edx); + /* Check, whether EDX has extended info about MWAIT */ + if (!(ecx & MWAIT_ECX_EXTENDED_INFO)) + return 1; + + /* + * edx enumeratios MONITOR/MWAIT extensions. Check, whether + * C1 supports MWAIT + */ + return (edx & MWAIT_EDX_C1); } void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c) -- cgit v1.2.3 From 00dba56465228825ea806e3a7fc0aa6bba7bdc6c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 9 Jun 2008 18:35:28 +0200 Subject: x86: move more common idle functions/variables to process.c more unification. Should cause no change in functionality. Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- arch/x86/kernel/process.c | 70 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 70 insertions(+) (limited to 'arch/x86/kernel/process.c') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index fe415ba606e0..9fea14607dfe 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -45,6 +45,76 @@ void arch_task_cache_init(void) SLAB_PANIC, NULL); } +/* + * Idle related variables and functions + */ +unsigned long boot_option_idle_override = 0; +EXPORT_SYMBOL(boot_option_idle_override); + +/* + * Powermanagement idle function, if any.. + */ +void (*pm_idle)(void); +EXPORT_SYMBOL(pm_idle); + +#ifdef CONFIG_X86_32 +/* + * This halt magic was a workaround for ancient floppy DMA + * wreckage. It should be safe to remove. + */ +static int hlt_counter; +void disable_hlt(void) +{ + hlt_counter++; +} +EXPORT_SYMBOL(disable_hlt); + +void enable_hlt(void) +{ + hlt_counter--; +} +EXPORT_SYMBOL(enable_hlt); + +static inline int hlt_use_halt(void) +{ + return (!hlt_counter && boot_cpu_data.hlt_works_ok); +} +#else +static inline int hlt_use_halt(void) +{ + return 1; +} +#endif + +/* + * We use this if we don't have any better + * idle routine.. + */ +void default_idle(void) +{ + if (hlt_use_halt()) { + current_thread_info()->status &= ~TS_POLLING; + /* + * TS_POLLING-cleared state must be visible before we + * test NEED_RESCHED: + */ + smp_mb(); + + if (!need_resched()) + safe_halt(); /* enables interrupts racelessly */ + else + local_irq_enable(); + current_thread_info()->status |= TS_POLLING; + } else { + local_irq_enable(); + /* loop is done by the caller */ + cpu_relax(); + } +} +#ifdef CONFIG_APM_MODULE +EXPORT_SYMBOL(default_idle); +#endif + static void do_nothing(void *unused) { } -- cgit v1.2.3 From 127a237a1ff49fa5b8e00af91e841598aeea3513 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 27 Jun 2008 11:48:22 +0200 Subject: fix "smp_call_function: get rid of the unused nonatomic/retry argument" fix: arch/x86/kernel/process.c: In function 'cpu_idle_wait': arch/x86/kernel/process.c:64: error: too many arguments to function 'smp_call_function' Signed-off-by: Ingo Molnar --- arch/x86/kernel/process.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel/process.c') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index ba370dc8685b..2dad8fef391c 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -61,7 +61,7 @@ void cpu_idle_wait(void) { smp_mb(); /* kick all the CPUs so that they exit out of pm_idle */ - smp_call_function(do_nothing, NULL, 0, 1); + smp_call_function(do_nothing, NULL, 1); } EXPORT_SYMBOL_GPL(cpu_idle_wait); -- cgit v1.2.3 From aa276e1cafb3ce9d01d1e837bcd67e92616013ac Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 9 Jun 2008 19:15:00 +0200 Subject: x86, clockevents: add C1E aware idle function C1E on AMD machines is like C3 but without control from the OS. Up to now we disabled the local apic timer for those machines as it stops when the CPU goes into C1E. This excludes those machines from high resolution timers / dynamic ticks, which hurts especially X2 based laptops. The current boot time C1E detection has another, more serious flaw as well: some BIOSes do not enable C1E until the ACPI processor module is loaded. This causes systems to stop working after that point. To work nicely with C1E enabled machines we use a separate idle function, which checks on idle entry whether C1E was enabled in the Interrupt Pending Message MSR. This allows us to do timer broadcasting for C1E and covers the late enablement of C1E as well. Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- arch/x86/kernel/process.c | 66 +++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 66 insertions(+) (limited to 'arch/x86/kernel/process.c') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index 9fea14607dfe..68ad3539b143 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -6,6 +6,7 @@ #include #include #include +#include struct kmem_cache *task_xstate_cachep; @@ -219,6 +220,68 @@ static int __cpuinit mwait_usable(const struct cpuinfo_x86 *c) return (edx & MWAIT_EDX_C1); } +/* + * Check for AMD CPUs, which have potentially C1E support + */ +static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c) +{ + if (c->x86_vendor != X86_VENDOR_AMD) + return 0; + + if (c->x86 < 0x0F) + return 0; + + /* Family 0x0f models < rev F do not have C1E */ + if (c->x86 == 0x0f && c->x86_model < 0x40) + return 0; + + return 1; +} + +/* + * C1E aware idle routine. We check for C1E active in the interrupt + * pending message MSR. If we detect C1E, then we handle it the same + * way as C3 power states (local apic timer and TSC stop) + */ +static void c1e_idle(void) +{ + static cpumask_t c1e_mask = CPU_MASK_NONE; + static int c1e_detected; + + if (need_resched()) + return; + + if (!c1e_detected) { + u32 lo, hi; + + rdmsr(MSR_K8_INT_PENDING_MSG, lo, hi); + if (lo & K8_INTP_C1E_ACTIVE_MASK) { + c1e_detected = 1; + mark_tsc_unstable("TSC halt in C1E"); + printk(KERN_INFO "System has C1E enabled\n"); + } + } + + if (c1e_detected) { + int cpu = smp_processor_id(); + + if (!cpu_isset(cpu, c1e_mask)) { + cpu_set(cpu, c1e_mask); + /* Force broadcast so ACPI can not interfere */ + clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_FORCE, + &cpu); + printk(KERN_INFO "Switch to broadcast mode on CPU%d\n", + cpu); + } + clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_ENTER, &cpu); + default_idle(); + local_irq_disable(); + clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_EXIT, &cpu); + local_irq_enable(); + } else + default_idle(); +} + void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c) { #ifdef CONFIG_X86_SMP @@ -236,6 +299,9 @@ void __cpuinit select_idle_routine(const struct cpuinfo_x86 *c) */ printk(KERN_INFO "using mwait in idle threads.\n"); pm_idle = mwait_idle; + } else if (check_c1e_idle(c)) { + printk(KERN_INFO "using C1E aware idle routine\n"); + pm_idle = c1e_idle; } else pm_idle = default_idle; } -- cgit v1.2.3 From 0beefa208bb3a9e581a60125703409ebe6f7fa53 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Tue, 17 Jun 2008 09:12:03 +0200 Subject: x86: add C1E aware idle function, fix On Tue, 17 Jun 2008, Rafael J. Wysocki wrote: > > BTW, with the C1E patches reverted I don't get the > WARNING: at /home/rafael/src/linux-next/kernel/smp.c:215 smp_call_function_single+0x3d/0xa2 > in the log. Thomas? The BROADCAST_FORCE notification uses smp_function_call and therefor must be run with interrupts enabled. While at it, add a comment for the BROADCAST_EXIT notifier as well. Reported-and-bisected-by: Rafael J. Wysocki Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- arch/x86/kernel/process.c | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel/process.c') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index 68ad3539b143..4061d63aabe7 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -267,17 +267,29 @@ static void c1e_idle(void) if (!cpu_isset(cpu, c1e_mask)) { cpu_set(cpu, c1e_mask); - /* Force broadcast so ACPI can not interfere */ + /* + * Force broadcast so ACPI can not interfere. Needs + * to run with interrupts enabled as it uses + * smp_function_call. + */ + local_irq_enable(); clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_FORCE, &cpu); printk(KERN_INFO "Switch to broadcast mode on CPU%d\n", cpu); + local_irq_disable(); } clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_ENTER, &cpu); + default_idle(); - local_irq_disable(); - clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_EXIT, &cpu); - local_irq_enable(); + + /* + * The switch back from broadcast mode needs to be + * called with interrupts disabled. + */ + local_irq_disable(); + clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_EXIT, &cpu); + local_irq_enable(); } else default_idle(); } -- cgit v1.2.3 From c1e3b377ad48febba6f91b8ae42c44ee4d4ab45e Mon Sep 17 00:00:00 2001 From: Zhao Yakui Date: Tue, 24 Jun 2008 17:58:53 +0800 Subject: ACPI: Create "idle=halt" bootparam "idle=halt" limits the idle loop to using the halt instruction. No MWAIT, no IO accesses, no C-states deeper than C1. If something is broken in the idle code, "idle=halt" is a less severe workaround than "idle=poll" which disables all power savings. Signed-off-by: Zhao Yakui Signed-off-by: Len Brown Signed-off-by: Andi Kleen --- arch/x86/kernel/process.c | 17 ++++++++++++++++- 1 file changed, 16 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel/process.c') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index 7dceea947232..7fc729498760 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -7,6 +7,10 @@ #include #include #include +#include + +unsigned long idle_halt; +EXPORT_SYMBOL(idle_halt); struct kmem_cache *task_xstate_cachep; @@ -325,7 +329,18 @@ static int __init idle_setup(char *str) pm_idle = poll_idle; } else if (!strcmp(str, "mwait")) force_mwait = 1; - else + else if (!strcmp(str, "halt")) { + /* + * When the boot option of idle=halt is added, halt is + * forced to be used for CPU idle. In such case CPU C2/C3 + * won't be used again. + * To continue to load the CPU idle driver, don't touch + * the boot_option_idle_override. + */ + pm_idle = default_idle; + idle_halt = 1; + return 0; + } else return -1; boot_option_idle_override = 1; -- cgit v1.2.3 From da5e09a1b3e5a9fc0b15a3feb64e921ccc55ba74 Mon Sep 17 00:00:00 2001 From: Zhao Yakui Date: Tue, 24 Jun 2008 18:01:09 +0800 Subject: ACPI : Create "idle=nomwait" bootparam "idle=nomwait" disables the use of the MWAIT instruction from both C1 (C1_FFH) and deeper (C2C3_FFH) C-states. When MWAIT is unavailable, the BIOS and OS generally negotiate to use the HALT instruction for C1, and use IO accesses for deeper C-states. This option is useful for power and performance comparisons, and also to work around BIOS bugs where broken MWAIT support is advertised. http://bugzilla.kernel.org/show_bug.cgi?id=10807 http://bugzilla.kernel.org/show_bug.cgi?id=10914 Signed-off-by: Zhao Yakui Signed-off-by: Li Shaohua Signed-off-by: Len Brown Signed-off-by: Andi Kleen --- arch/x86/kernel/process.c | 11 +++++++++++ 1 file changed, 11 insertions(+) (limited to 'arch/x86/kernel/process.c') diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index 7fc729498760..4d629c62f4f8 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -11,6 +11,8 @@ unsigned long idle_halt; EXPORT_SYMBOL(idle_halt); +unsigned long idle_nomwait; +EXPORT_SYMBOL(idle_nomwait); struct kmem_cache *task_xstate_cachep; @@ -340,6 +342,15 @@ static int __init idle_setup(char *str) pm_idle = default_idle; idle_halt = 1; return 0; + } else if (!strcmp(str, "nomwait")) { + /* + * If the boot option of "idle=nomwait" is added, + * it means that mwait will be disabled for CPU C2/C3 + * states. In such case it won't touch the variable + * of boot_option_idle_override. + */ + idle_nomwait = 1; + return 0; } else return -1; -- cgit v1.2.3