From 393d68fb9929817cde7ab31c82d66fcb28ad35fc Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Fri, 26 Dec 2008 22:23:38 +1030 Subject: cpumask: x86: Introduce cpumask_of_{node,pcibus} to replace {node,pcibus}_to_cpumask Impact: New APIs The old node_to_cpumask/node_to_pcibus returned a cpumask_t: these return a pointer to a struct cpumask. Part of removing cpumasks from the stack. Also makes __pcibus_to_node take a const pointer. Signed-off-by: Rusty Russell Acked-by: Ingo Molnar --- arch/x86/include/asm/pci.h | 10 ++++++++-- arch/x86/include/asm/topology.h | 35 +++++++++++++++++++++++------------ arch/x86/kernel/setup_percpu.c | 8 ++++---- 3 files changed, 35 insertions(+), 18 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/pci.h b/arch/x86/include/asm/pci.h index 875b38edf193..52d80d3d94f3 100644 --- a/arch/x86/include/asm/pci.h +++ b/arch/x86/include/asm/pci.h @@ -98,9 +98,9 @@ static inline void early_quirks(void) { } #ifdef CONFIG_NUMA /* Returns the node based on pci bus */ -static inline int __pcibus_to_node(struct pci_bus *bus) +static inline int __pcibus_to_node(const struct pci_bus *bus) { - struct pci_sysdata *sd = bus->sysdata; + const struct pci_sysdata *sd = bus->sysdata; return sd->node; } @@ -109,6 +109,12 @@ static inline cpumask_t __pcibus_to_cpumask(struct pci_bus *bus) { return node_to_cpumask(__pcibus_to_node(bus)); } + +static inline const struct cpumask * +cpumask_of_pcibus(const struct pci_bus *bus) +{ + return cpumask_of_node(__pcibus_to_node(bus)); +} #endif #endif /* _ASM_X86_PCI_H */ diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h index ff386ff50ed7..45da5dc50fc8 100644 --- a/arch/x86/include/asm/topology.h +++ b/arch/x86/include/asm/topology.h @@ -61,13 +61,19 @@ static inline int cpu_to_node(int cpu) * * Side note: this function creates the returned cpumask on the stack * so with a high NR_CPUS count, excessive stack space is used. The - * node_to_cpumask_ptr function should be used whenever possible. + * cpumask_of_node function should be used whenever possible. */ static inline cpumask_t node_to_cpumask(int node) { return node_to_cpumask_map[node]; } +/* Returns a bitmask of CPUs on Node 'node'. */ +static inline const struct cpumask *cpumask_of_node(int node) +{ + return &node_to_cpumask_map[node]; +} + #else /* CONFIG_X86_64 */ /* Mappings between node number and cpus on that node. */ @@ -82,7 +88,7 @@ DECLARE_EARLY_PER_CPU(int, x86_cpu_to_node_map); #ifdef CONFIG_DEBUG_PER_CPU_MAPS extern int cpu_to_node(int cpu); extern int early_cpu_to_node(int cpu); -extern const cpumask_t *_node_to_cpumask_ptr(int node); +extern const cpumask_t *cpumask_of_node(int node); extern cpumask_t node_to_cpumask(int node); #else /* !CONFIG_DEBUG_PER_CPU_MAPS */ @@ -103,7 +109,7 @@ static inline int early_cpu_to_node(int cpu) } /* Returns a pointer to the cpumask of CPUs on Node 'node'. */ -static inline const cpumask_t *_node_to_cpumask_ptr(int node) +static inline const cpumask_t *cpumask_of_node(int node) { return &node_to_cpumask_map[node]; } @@ -116,12 +122,15 @@ static inline cpumask_t node_to_cpumask(int node) #endif /* !CONFIG_DEBUG_PER_CPU_MAPS */ -/* Replace default node_to_cpumask_ptr with optimized version */ +/* + * Replace default node_to_cpumask_ptr with optimized version + * Deprecated: use "const struct cpumask *mask = cpumask_of_node(node)" + */ #define node_to_cpumask_ptr(v, node) \ - const cpumask_t *v = _node_to_cpumask_ptr(node) + const cpumask_t *v = cpumask_of_node(node) #define node_to_cpumask_ptr_next(v, node) \ - v = _node_to_cpumask_ptr(node) + v = cpumask_of_node(node) #endif /* CONFIG_X86_64 */ @@ -187,7 +196,7 @@ extern int __node_distance(int, int); #define cpu_to_node(cpu) 0 #define early_cpu_to_node(cpu) 0 -static inline const cpumask_t *_node_to_cpumask_ptr(int node) +static inline const cpumask_t *cpumask_of_node(int node) { return &cpu_online_map; } @@ -200,12 +209,15 @@ static inline int node_to_first_cpu(int node) return first_cpu(cpu_online_map); } -/* Replace default node_to_cpumask_ptr with optimized version */ +/* + * Replace default node_to_cpumask_ptr with optimized version + * Deprecated: use "const struct cpumask *mask = cpumask_of_node(node)" + */ #define node_to_cpumask_ptr(v, node) \ - const cpumask_t *v = _node_to_cpumask_ptr(node) + const cpumask_t *v = cpumask_of_node(node) #define node_to_cpumask_ptr_next(v, node) \ - v = _node_to_cpumask_ptr(node) + v = cpumask_of_node(node) #endif #include @@ -214,8 +226,7 @@ static inline int node_to_first_cpu(int node) /* Returns the number of the first CPU on Node 'node'. */ static inline int node_to_first_cpu(int node) { - node_to_cpumask_ptr(mask, node); - return first_cpu(*mask); + return cpumask_first(cpumask_of_node(node)); } #endif diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 1c2084291f97..8e8b1193add5 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -334,25 +334,25 @@ static const cpumask_t cpu_mask_none; /* * Returns a pointer to the bitmask of CPUs on Node 'node'. */ -const cpumask_t *_node_to_cpumask_ptr(int node) +const cpumask_t *cpumask_of_node(int node) { if (node_to_cpumask_map == NULL) { printk(KERN_WARNING - "_node_to_cpumask_ptr(%d): no node_to_cpumask_map!\n", + "cpumask_of_node(%d): no node_to_cpumask_map!\n", node); dump_stack(); return (const cpumask_t *)&cpu_online_map; } if (node >= nr_node_ids) { printk(KERN_WARNING - "_node_to_cpumask_ptr(%d): node > nr_node_ids(%d)\n", + "cpumask_of_node(%d): node > nr_node_ids(%d)\n", node, nr_node_ids); dump_stack(); return &cpu_mask_none; } return &node_to_cpumask_map[node]; } -EXPORT_SYMBOL(_node_to_cpumask_ptr); +EXPORT_SYMBOL(cpumask_of_node); /* * Returns a bitmask of CPUs on Node 'node'. -- cgit v1.2.3 From 030bb203e01db12e3f2866799f4f03a114d06349 Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Fri, 26 Dec 2008 22:23:41 +1030 Subject: cpumask: cpu_coregroup_mask(): x86 Impact: New API Like cpu_coregroup_map, but returns a (const) pointer. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis Cc: Ingo Molnar --- arch/x86/include/asm/topology.h | 1 + arch/x86/kernel/smpboot.c | 11 ++++++++--- 2 files changed, 9 insertions(+), 3 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h index 45da5dc50fc8..168203c0c316 100644 --- a/arch/x86/include/asm/topology.h +++ b/arch/x86/include/asm/topology.h @@ -231,6 +231,7 @@ static inline int node_to_first_cpu(int node) #endif extern cpumask_t cpu_coregroup_map(int cpu); +extern const struct cpumask *cpu_coregroup_mask(int cpu); #ifdef ENABLE_TOPO_DEFINES #define topology_physical_package_id(cpu) (cpu_data(cpu).phys_proc_id) diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 468c2f9d47ae..d5274b6b088e 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -497,7 +497,7 @@ void __cpuinit set_cpu_sibling_map(int cpu) } /* maps the cpu to the sched domain representing multi-core */ -cpumask_t cpu_coregroup_map(int cpu) +const struct cpumask *cpu_coregroup_mask(int cpu) { struct cpuinfo_x86 *c = &cpu_data(cpu); /* @@ -505,9 +505,14 @@ cpumask_t cpu_coregroup_map(int cpu) * And for power savings, we return cpu_core_map */ if (sched_mc_power_savings || sched_smt_power_savings) - return per_cpu(cpu_core_map, cpu); + return &per_cpu(cpu_core_map, cpu); else - return c->llc_shared_map; + return &c->llc_shared_map; +} + +cpumask_t cpu_coregroup_map(int cpu) +{ + return *cpu_coregroup_mask(cpu); } static void impress_friends(void) -- cgit v1.2.3 From 6092848a2a23b660150a38bc06f59d75838d70c8 Mon Sep 17 00:00:00 2001 From: Sergio Luis Date: Sun, 28 Dec 2008 04:12:26 -0300 Subject: x86: mark get_cpu_leaves() with __cpuinit annotation Impact: fix section mismatch warning Commit b2bb85549134c005e997e5a7ed303bda6a1ae738 ("x86: Remove cpumask games in x86/kernel/cpu/intel_cacheinfo.c") introduced get_cpu_leaves(), which references __cpuinit cpuid4_cache_lookup(). Mark get_cpu_leaves() with a __cpuinit annotation. Signed-off-by: Sergio Luis Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/intel_cacheinfo.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c index fb7f946cb65e..7bd00a565672 100644 --- a/arch/x86/kernel/cpu/intel_cacheinfo.c +++ b/arch/x86/kernel/cpu/intel_cacheinfo.c @@ -534,7 +534,7 @@ static void __cpuinit free_cache_attributes(unsigned int cpu) per_cpu(cpuid4_info, cpu) = NULL; } -static void get_cpu_leaves(void *_retval) +static void __cpuinit get_cpu_leaves(void *_retval) { int j, *retval = _retval, cpu = smp_processor_id(); -- cgit v1.2.3 From 730cf27246225d56ca1603b2f3c4fdbf882d4e51 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Wed, 31 Dec 2008 18:08:45 -0800 Subject: x86: enable cpus display of kernel_max and offlined cpus Impact: enables /sys/devices/system/cpu/{kernel_max,offline} user interface By setting total_cpus, the drivers/base/cpu.c will display the values of kernel_max (NR_CPUS-1) and the offlined cpu map. Signed-off-by: Mike Travis Acked-by: Rusty Russell Signed-off-by: Ingo Molnar --- arch/x86/kernel/smpboot.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'arch/x86') diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 9e177a4077ee..f49c26bd7e2d 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -1298,6 +1298,8 @@ __init void prefill_possible_map(void) else possible = setup_possible_cpus; + total_cpus = max_t(int, possible, num_processors + disabled_cpus); + if (possible > CONFIG_NR_CPUS) { printk(KERN_WARNING "%d Processors exceeds NR_CPUS limit of %d\n", -- cgit v1.2.3 From 9628937d5b37169151c5f6bbd40919c6ac958a46 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Wed, 31 Dec 2008 18:08:46 -0800 Subject: x86: cleanup some remaining usages of NR_CPUS where s/b nr_cpu_ids Impact: Reduce future system panics due to cpumask operations using NR_CPUS Insure that code does not look at bits >= nr_cpu_ids as when cpumasks are allocated based on nr_cpu_ids, these extra bits will not be defined. Also some other minor updates: * change in to use cpu accessor function set_cpu_present() instead of directly accessing cpu_present_map w/cpu_clear() [arch/x86/kernel/reboot.c] * use cpumask_of() instead of &cpumask_of_cpu() [arch/x86/kernel/reboot.c] * optimize some cpu_mask_to_apicid_and functions. Signed-off-by: Mike Travis Acked-by: Rusty Russell Signed-off-by: Ingo Molnar --- arch/x86/include/asm/es7000/apic.h | 32 +++------------------------- arch/x86/include/asm/lguest.h | 2 +- arch/x86/include/asm/numaq/apic.h | 4 ++-- arch/x86/include/asm/summit/apic.h | 42 +++++++------------------------------ arch/x86/kernel/acpi/boot.c | 2 +- arch/x86/kernel/apic.c | 4 ++-- arch/x86/kernel/cpu/common.c | 2 +- arch/x86/kernel/cpuid.c | 2 +- arch/x86/kernel/msr.c | 2 +- arch/x86/kernel/reboot.c | 4 ++-- arch/x86/kernel/smpboot.c | 2 +- arch/x86/mach-voyager/voyager_smp.c | 7 +++---- 12 files changed, 26 insertions(+), 79 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/es7000/apic.h b/arch/x86/include/asm/es7000/apic.h index 51ac1230294e..bc53d5ef1386 100644 --- a/arch/x86/include/asm/es7000/apic.h +++ b/arch/x86/include/asm/es7000/apic.h @@ -157,7 +157,7 @@ cpu_mask_to_apicid_cluster(const struct cpumask *cpumask) num_bits_set = cpumask_weight(cpumask); /* Return id to all */ - if (num_bits_set == NR_CPUS) + if (num_bits_set == nr_cpu_ids) return 0xFF; /* * The cpus in the mask must all be on the apic cluster. If are not @@ -190,7 +190,7 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) num_bits_set = cpus_weight(*cpumask); /* Return id to all */ - if (num_bits_set == NR_CPUS) + if (num_bits_set == nr_cpu_ids) return cpu_to_logical_apicid(0); /* * The cpus in the mask must all be on the apic cluster. If are not @@ -218,9 +218,6 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *inmask, const struct cpumask *andmask) { - int num_bits_set; - int cpus_found = 0; - int cpu; int apicid = cpu_to_logical_apicid(0); cpumask_var_t cpumask; @@ -229,31 +226,8 @@ static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *inmask, cpumask_and(cpumask, inmask, andmask); cpumask_and(cpumask, cpumask, cpu_online_mask); + apicid = cpu_mask_to_apicid(cpumask); - num_bits_set = cpumask_weight(cpumask); - /* Return id to all */ - if (num_bits_set == NR_CPUS) - goto exit; - /* - * The cpus in the mask must all be on the apic cluster. If are not - * on the same apicid cluster return default value of TARGET_CPUS. - */ - cpu = cpumask_first(cpumask); - apicid = cpu_to_logical_apicid(cpu); - while (cpus_found < num_bits_set) { - if (cpumask_test_cpu(cpu, cpumask)) { - int new_apicid = cpu_to_logical_apicid(cpu); - if (apicid_cluster(apicid) != - apicid_cluster(new_apicid)){ - printk ("%s: Not a valid mask!\n", __func__); - return cpu_to_logical_apicid(0); - } - apicid = new_apicid; - cpus_found++; - } - cpu++; - } -exit: free_cpumask_var(cpumask); return apicid; } diff --git a/arch/x86/include/asm/lguest.h b/arch/x86/include/asm/lguest.h index d28a507cef39..1caf57628b9c 100644 --- a/arch/x86/include/asm/lguest.h +++ b/arch/x86/include/asm/lguest.h @@ -15,7 +15,7 @@ #define SHARED_SWITCHER_PAGES \ DIV_ROUND_UP(end_switcher_text - start_switcher_text, PAGE_SIZE) /* Pages for switcher itself, then two pages per cpu */ -#define TOTAL_SWITCHER_PAGES (SHARED_SWITCHER_PAGES + 2 * NR_CPUS) +#define TOTAL_SWITCHER_PAGES (SHARED_SWITCHER_PAGES + 2 * nr_cpu_ids) /* We map at -4M for ease of mapping into the guest (one PTE page). */ #define SWITCHER_ADDR 0xFFC00000 diff --git a/arch/x86/include/asm/numaq/apic.h b/arch/x86/include/asm/numaq/apic.h index c80f00d29965..bf37bc49bd8e 100644 --- a/arch/x86/include/asm/numaq/apic.h +++ b/arch/x86/include/asm/numaq/apic.h @@ -63,8 +63,8 @@ static inline physid_mask_t ioapic_phys_id_map(physid_mask_t phys_map) extern u8 cpu_2_logical_apicid[]; static inline int cpu_to_logical_apicid(int cpu) { - if (cpu >= NR_CPUS) - return BAD_APICID; + if (cpu >= nr_cpu_ids) + return BAD_APICID; return (int)cpu_2_logical_apicid[cpu]; } diff --git a/arch/x86/include/asm/summit/apic.h b/arch/x86/include/asm/summit/apic.h index 99327d1be49f..4bb5fb34f030 100644 --- a/arch/x86/include/asm/summit/apic.h +++ b/arch/x86/include/asm/summit/apic.h @@ -52,7 +52,7 @@ static inline void init_apic_ldr(void) int i; /* Create logical APIC IDs by counting CPUs already in cluster. */ - for (count = 0, i = NR_CPUS; --i >= 0; ) { + for (count = 0, i = nr_cpu_ids; --i >= 0; ) { lid = cpu_2_logical_apicid[i]; if (lid != BAD_APICID && apicid_cluster(lid) == my_cluster) ++count; @@ -97,8 +97,8 @@ static inline int apicid_to_node(int logical_apicid) static inline int cpu_to_logical_apicid(int cpu) { #ifdef CONFIG_SMP - if (cpu >= NR_CPUS) - return BAD_APICID; + if (cpu >= nr_cpu_ids) + return BAD_APICID; return (int)cpu_2_logical_apicid[cpu]; #else return logical_smp_processor_id(); @@ -107,7 +107,7 @@ static inline int cpu_to_logical_apicid(int cpu) static inline int cpu_present_to_apicid(int mps_cpu) { - if (mps_cpu < NR_CPUS) + if (mps_cpu < nr_cpu_ids) return (int)per_cpu(x86_bios_cpu_apicid, mps_cpu); else return BAD_APICID; @@ -146,7 +146,7 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) num_bits_set = cpus_weight(*cpumask); /* Return id to all */ - if (num_bits_set == NR_CPUS) + if (num_bits_set >= nr_cpu_ids) return (int) 0xFF; /* * The cpus in the mask must all be on the apic cluster. If are not @@ -173,42 +173,16 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *inmask, const struct cpumask *andmask) { - int num_bits_set; - int cpus_found = 0; - int cpu; - int apicid = 0xFF; + int apicid = cpu_to_logical_apicid(0); cpumask_var_t cpumask; if (!alloc_cpumask_var(&cpumask, GFP_ATOMIC)) - return (int) 0xFF; + return apicid; cpumask_and(cpumask, inmask, andmask); cpumask_and(cpumask, cpumask, cpu_online_mask); + apicid = cpu_mask_to_apicid(cpumask); - num_bits_set = cpumask_weight(cpumask); - /* Return id to all */ - if (num_bits_set == nr_cpu_ids) - goto exit; - /* - * The cpus in the mask must all be on the apic cluster. If are not - * on the same apicid cluster return default value of TARGET_CPUS. - */ - cpu = cpumask_first(cpumask); - apicid = cpu_to_logical_apicid(cpu); - while (cpus_found < num_bits_set) { - if (cpumask_test_cpu(cpu, cpumask)) { - int new_apicid = cpu_to_logical_apicid(cpu); - if (apicid_cluster(apicid) != - apicid_cluster(new_apicid)){ - printk ("%s: Not a valid mask!\n", __func__); - return 0xFF; - } - apicid = apicid | new_apicid; - cpus_found++; - } - cpu++; - } -exit: free_cpumask_var(cpumask); return apicid; } diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 65d0b72777ea..fd24c55e4ae2 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -598,7 +598,7 @@ EXPORT_SYMBOL(acpi_map_lsapic); int acpi_unmap_lsapic(int cpu) { per_cpu(x86_cpu_to_apicid, cpu) = -1; - cpu_clear(cpu, cpu_present_map); + set_cpu_present(cpu, false); num_processors--; return (0); diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 6b7f824db160..99589245fd8d 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -140,7 +140,7 @@ static int lapic_next_event(unsigned long delta, struct clock_event_device *evt); static void lapic_timer_setup(enum clock_event_mode mode, struct clock_event_device *evt); -static void lapic_timer_broadcast(const cpumask_t *mask); +static void lapic_timer_broadcast(const struct cpumask *mask); static void apic_pm_activate(void); /* @@ -453,7 +453,7 @@ static void lapic_timer_setup(enum clock_event_mode mode, /* * Local APIC timer broadcast function */ -static void lapic_timer_broadcast(const cpumask_t *mask) +static void lapic_timer_broadcast(const struct cpumask *mask) { #ifdef CONFIG_SMP send_IPI_mask(mask, LOCAL_TIMER_VECTOR); diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 42e0853030cb..3f95a40f718a 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -355,7 +355,7 @@ void __cpuinit detect_ht(struct cpuinfo_x86 *c) printk(KERN_INFO "CPU: Hyper-Threading is disabled\n"); } else if (smp_num_siblings > 1) { - if (smp_num_siblings > NR_CPUS) { + if (smp_num_siblings > nr_cpu_ids) { printk(KERN_WARNING "CPU: Unsupported number of siblings %d", smp_num_siblings); smp_num_siblings = 1; diff --git a/arch/x86/kernel/cpuid.c b/arch/x86/kernel/cpuid.c index 72cefd1e649b..62a3c23bd703 100644 --- a/arch/x86/kernel/cpuid.c +++ b/arch/x86/kernel/cpuid.c @@ -121,7 +121,7 @@ static int cpuid_open(struct inode *inode, struct file *file) lock_kernel(); cpu = iminor(file->f_path.dentry->d_inode); - if (cpu >= NR_CPUS || !cpu_online(cpu)) { + if (cpu >= nr_cpu_ids || !cpu_online(cpu)) { ret = -ENXIO; /* No such CPU */ goto out; } diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c index 82a7c7ed6d45..726266695b2c 100644 --- a/arch/x86/kernel/msr.c +++ b/arch/x86/kernel/msr.c @@ -136,7 +136,7 @@ static int msr_open(struct inode *inode, struct file *file) lock_kernel(); cpu = iminor(file->f_path.dentry->d_inode); - if (cpu >= NR_CPUS || !cpu_online(cpu)) { + if (cpu >= nr_cpu_ids || !cpu_online(cpu)) { ret = -ENXIO; /* No such CPU */ goto out; } diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c index ba7b9a0e6063..de4a9d643bee 100644 --- a/arch/x86/kernel/reboot.c +++ b/arch/x86/kernel/reboot.c @@ -449,7 +449,7 @@ void native_machine_shutdown(void) #ifdef CONFIG_X86_32 /* See if there has been given a command line override */ - if ((reboot_cpu != -1) && (reboot_cpu < NR_CPUS) && + if ((reboot_cpu != -1) && (reboot_cpu < nr_cpu_ids) && cpu_online(reboot_cpu)) reboot_cpu_id = reboot_cpu; #endif @@ -459,7 +459,7 @@ void native_machine_shutdown(void) reboot_cpu_id = smp_processor_id(); /* Make certain I only run on the appropriate processor */ - set_cpus_allowed_ptr(current, &cpumask_of_cpu(reboot_cpu_id)); + set_cpus_allowed_ptr(current, cpumask_of(reboot_cpu_id)); /* O.K Now that I'm on the appropriate processor, * stop all of the others. diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index f49c26bd7e2d..6bd4d9b73870 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -1154,7 +1154,7 @@ static void __init smp_cpu_index_default(void) for_each_possible_cpu(i) { c = &cpu_data(i); /* mark all to hotplug */ - c->cpu_index = NR_CPUS; + c->cpu_index = nr_cpu_ids; } } diff --git a/arch/x86/mach-voyager/voyager_smp.c b/arch/x86/mach-voyager/voyager_smp.c index a5bc05492b1e..9840b7ec749a 100644 --- a/arch/x86/mach-voyager/voyager_smp.c +++ b/arch/x86/mach-voyager/voyager_smp.c @@ -357,9 +357,8 @@ void __init find_smp_config(void) printk("VOYAGER SMP: Boot cpu is %d\n", boot_cpu_id); /* initialize the CPU structures (moved from smp_boot_cpus) */ - for (i = 0; i < NR_CPUS; i++) { + for (i = 0; i < nr_cpu_ids; i++) cpu_irq_affinity[i] = ~0; - } cpu_online_map = cpumask_of_cpu(boot_cpu_id); /* The boot CPU must be extended */ @@ -1227,7 +1226,7 @@ int setup_profiling_timer(unsigned int multiplier) * new values until the next timer interrupt in which they do process * accounting. */ - for (i = 0; i < NR_CPUS; ++i) + for (i = 0; i < nr_cpu_ids; ++i) per_cpu(prof_multiplier, i) = multiplier; return 0; @@ -1257,7 +1256,7 @@ void __init voyager_smp_intr_init(void) int i; /* initialize the per cpu irq mask to all disabled */ - for (i = 0; i < NR_CPUS; i++) + for (i = 0; i < nr_cpu_ids; i++) vic_irq_mask[i] = 0xFFFF; VIC_SET_GATE(VIC_CPI_LEVEL0, vic_cpi_interrupt); -- cgit v1.2.3 From ee943a82b697456f9d2ac46f1e6d230beedb4b6c Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 31 Dec 2008 18:08:47 -0800 Subject: x86: use cpumask_var_t in acpi/boot.c Impact: reduce stack size, use new API. Replace cpumask_t with cpumask_var_t. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/boot.c | 29 ++++++++++++++++++++++------- 1 file changed, 22 insertions(+), 7 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index fd24c55e4ae2..29dc0c89d4af 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -538,9 +538,10 @@ static int __cpuinit _acpi_map_lsapic(acpi_handle handle, int *pcpu) struct acpi_buffer buffer = { ACPI_ALLOCATE_BUFFER, NULL }; union acpi_object *obj; struct acpi_madt_local_apic *lapic; - cpumask_t tmp_map, new_map; + cpumask_var_t tmp_map, new_map; u8 physid; int cpu; + int retval = -ENOMEM; if (ACPI_FAILURE(acpi_evaluate_object(handle, "_MAT", NULL, &buffer))) return -EINVAL; @@ -569,23 +570,37 @@ static int __cpuinit _acpi_map_lsapic(acpi_handle handle, int *pcpu) buffer.length = ACPI_ALLOCATE_BUFFER; buffer.pointer = NULL; - tmp_map = cpu_present_map; + if (!alloc_cpumask_var(&tmp_map, GFP_KERNEL)) + goto out; + + if (!alloc_cpumask_var(&new_map, GFP_KERNEL)) + goto free_tmp_map; + + cpumask_copy(tmp_map, cpu_present_mask); acpi_register_lapic(physid, lapic->lapic_flags & ACPI_MADT_ENABLED); /* * If mp_register_lapic successfully generates a new logical cpu * number, then the following will get us exactly what was mapped */ - cpus_andnot(new_map, cpu_present_map, tmp_map); - if (cpus_empty(new_map)) { + cpumask_andnot(new_map, cpu_present_mask, tmp_map); + if (cpumask_empty(new_map)) { printk ("Unable to map lapic to logical cpu number\n"); - return -EINVAL; + retval = -EINVAL; + goto free_new_map; } - cpu = first_cpu(new_map); + cpu = cpumask_first(new_map); *pcpu = cpu; - return 0; + retval = 0; + +free_new_map: + free_cpumask_var(new_map); +free_tmp_map: + free_cpumask_var(tmp_map); +out: + return retval; } /* wrapper to silence section mismatch warning */ -- cgit v1.2.3 From 2fdf66b491ac706657946442789ec644cc317e1a Mon Sep 17 00:00:00 2001 From: Rusty Russell Date: Wed, 31 Dec 2008 18:08:47 -0800 Subject: cpumask: convert shared_cpu_map in acpi_processor* structs to cpumask_var_t Impact: Reduce memory usage, use new API. This is part of an effort to reduce structure sizes for machines configured with large NR_CPUS. cpumask_t gets replaced by cpumask_var_t, which is either struct cpumask[1] (small NR_CPUS) or struct cpumask * (large NR_CPUS). (Changes to powernow-k* by .) Signed-off-by: Rusty Russell Signed-off-by: Mike Travis Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c | 27 ++++++++-- arch/x86/kernel/cpu/cpufreq/powernow-k7.c | 9 ++++ arch/x86/kernel/cpu/cpufreq/powernow-k8.c | 24 +++++---- drivers/acpi/processor_core.c | 14 ++++-- drivers/acpi/processor_perflib.c | 28 ++++++----- drivers/acpi/processor_throttling.c | 80 +++++++++++++++++++----------- include/acpi/processor.h | 4 +- 7 files changed, 128 insertions(+), 58 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index 88ea02dcb622..d0a001093b2d 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -517,6 +517,17 @@ acpi_cpufreq_guess_freq(struct acpi_cpufreq_data *data, unsigned int cpu) } } +static void free_acpi_perf_data(void) +{ + unsigned int i; + + /* Freeing a NULL pointer is OK, and alloc_percpu zeroes. */ + for_each_possible_cpu(i) + free_cpumask_var(per_cpu_ptr(acpi_perf_data, i) + ->shared_cpu_map); + free_percpu(acpi_perf_data); +} + /* * acpi_cpufreq_early_init - initialize ACPI P-States library * @@ -527,6 +538,7 @@ acpi_cpufreq_guess_freq(struct acpi_cpufreq_data *data, unsigned int cpu) */ static int __init acpi_cpufreq_early_init(void) { + unsigned int i; dprintk("acpi_cpufreq_early_init\n"); acpi_perf_data = alloc_percpu(struct acpi_processor_performance); @@ -534,6 +546,15 @@ static int __init acpi_cpufreq_early_init(void) dprintk("Memory allocation error for acpi_perf_data.\n"); return -ENOMEM; } + for_each_possible_cpu(i) { + if (!alloc_cpumask_var(&per_cpu_ptr(acpi_perf_data, i) + ->shared_cpu_map, GFP_KERNEL)) { + + /* Freeing a NULL pointer is OK: alloc_percpu zeroes. */ + free_acpi_perf_data(); + return -ENOMEM; + } + } /* Do initialization in ACPI core */ acpi_processor_preregister_performance(acpi_perf_data); @@ -604,9 +625,9 @@ static int acpi_cpufreq_cpu_init(struct cpufreq_policy *policy) */ if (policy->shared_type == CPUFREQ_SHARED_TYPE_ALL || policy->shared_type == CPUFREQ_SHARED_TYPE_ANY) { - policy->cpus = perf->shared_cpu_map; + cpumask_copy(&policy->cpus, perf->shared_cpu_map); } - policy->related_cpus = perf->shared_cpu_map; + cpumask_copy(&policy->related_cpus, perf->shared_cpu_map); #ifdef CONFIG_SMP dmi_check_system(sw_any_bug_dmi_table); @@ -795,7 +816,7 @@ static int __init acpi_cpufreq_init(void) ret = cpufreq_register_driver(&acpi_cpufreq_driver); if (ret) - free_percpu(acpi_perf_data); + free_acpi_perf_data(); return ret; } diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k7.c b/arch/x86/kernel/cpu/cpufreq/powernow-k7.c index 7c7d56b43136..1b446d79a8fd 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k7.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k7.c @@ -310,6 +310,12 @@ static int powernow_acpi_init(void) goto err0; } + if (!alloc_cpumask_var(&acpi_processor_perf->shared_cpu_map, + GFP_KERNEL)) { + retval = -ENOMEM; + goto err05; + } + if (acpi_processor_register_performance(acpi_processor_perf, 0)) { retval = -EIO; goto err1; @@ -412,6 +418,8 @@ static int powernow_acpi_init(void) err2: acpi_processor_unregister_performance(acpi_processor_perf, 0); err1: + free_cpumask_var(acpi_processor_perf->shared_cpu_map); +err05: kfree(acpi_processor_perf); err0: printk(KERN_WARNING PFX "ACPI perflib can not be used in this platform\n"); @@ -652,6 +660,7 @@ static int powernow_cpu_exit (struct cpufreq_policy *policy) { #ifdef CONFIG_X86_POWERNOW_K7_ACPI if (acpi_processor_perf) { acpi_processor_unregister_performance(acpi_processor_perf, 0); + free_cpumask_var(acpi_processor_perf->shared_cpu_map); kfree(acpi_processor_perf); } #endif diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index 7f05f44b97e9..c3c9adbaa26f 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -766,7 +766,7 @@ static void powernow_k8_acpi_pst_values(struct powernow_k8_data *data, unsigned static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) { struct cpufreq_frequency_table *powernow_table; - int ret_val; + int ret_val = -ENODEV; if (acpi_processor_register_performance(&data->acpi_data, data->cpu)) { dprintk("register performance failed: bad ACPI data\n"); @@ -815,6 +815,13 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) /* notify BIOS that we exist */ acpi_processor_notify_smm(THIS_MODULE); + if (!alloc_cpumask_var(&data->acpi_data.shared_cpu_map, GFP_KERNEL)) { + printk(KERN_ERR PFX + "unable to alloc powernow_k8_data cpumask\n"); + ret_val = -ENOMEM; + goto err_out_mem; + } + return 0; err_out_mem: @@ -826,7 +833,7 @@ err_out: /* data->acpi_data.state_count informs us at ->exit() whether ACPI was used */ data->acpi_data.state_count = 0; - return -ENODEV; + return ret_val; } static int fill_powernow_table_pstate(struct powernow_k8_data *data, struct cpufreq_frequency_table *powernow_table) @@ -929,6 +936,7 @@ static void powernow_k8_cpu_exit_acpi(struct powernow_k8_data *data) { if (data->acpi_data.state_count) acpi_processor_unregister_performance(&data->acpi_data, data->cpu); + free_cpumask_var(data->acpi_data.shared_cpu_map); } #else @@ -1134,7 +1142,8 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) data->cpu = pol->cpu; data->currpstate = HW_PSTATE_INVALID; - if (powernow_k8_cpu_init_acpi(data)) { + rc = powernow_k8_cpu_init_acpi(data); + if (rc) { /* * Use the PSB BIOS structure. This is only availabe on * an UP version, and is deprecated by AMD. @@ -1152,20 +1161,17 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) "ACPI maintainers and complain to your BIOS " "vendor.\n"); #endif - kfree(data); - return -ENODEV; + goto err_out; } if (pol->cpu != 0) { printk(KERN_ERR FW_BUG PFX "No ACPI _PSS objects for " "CPU other than CPU0. Complain to your BIOS " "vendor.\n"); - kfree(data); - return -ENODEV; + goto err_out; } rc = find_psb_table(data); if (rc) { - kfree(data); - return -ENODEV; + goto err_out; } } diff --git a/drivers/acpi/processor_core.c b/drivers/acpi/processor_core.c index 34948362f41d..0cc2fd31e376 100644 --- a/drivers/acpi/processor_core.c +++ b/drivers/acpi/processor_core.c @@ -826,6 +826,11 @@ static int acpi_processor_add(struct acpi_device *device) if (!pr) return -ENOMEM; + if (!alloc_cpumask_var(&pr->throttling.shared_cpu_map, GFP_KERNEL)) { + kfree(pr); + return -ENOMEM; + } + pr->handle = device->handle; strcpy(acpi_device_name(device), ACPI_PROCESSOR_DEVICE_NAME); strcpy(acpi_device_class(device), ACPI_PROCESSOR_CLASS); @@ -845,10 +850,8 @@ static int acpi_processor_remove(struct acpi_device *device, int type) pr = acpi_driver_data(device); - if (pr->id >= nr_cpu_ids) { - kfree(pr); - return 0; - } + if (pr->id >= nr_cpu_ids) + goto free; if (type == ACPI_BUS_REMOVAL_EJECT) { if (acpi_processor_handle_eject(pr)) @@ -873,6 +876,9 @@ static int acpi_processor_remove(struct acpi_device *device, int type) per_cpu(processors, pr->id) = NULL; per_cpu(processor_device_array, pr->id) = NULL; + +free: + free_cpumask_var(pr->throttling.shared_cpu_map); kfree(pr); return 0; diff --git a/drivers/acpi/processor_perflib.c b/drivers/acpi/processor_perflib.c index 0d7b772bef50..846e227592d4 100644 --- a/drivers/acpi/processor_perflib.c +++ b/drivers/acpi/processor_perflib.c @@ -588,12 +588,15 @@ int acpi_processor_preregister_performance( int count, count_target; int retval = 0; unsigned int i, j; - cpumask_t covered_cpus; + cpumask_var_t covered_cpus; struct acpi_processor *pr; struct acpi_psd_package *pdomain; struct acpi_processor *match_pr; struct acpi_psd_package *match_pdomain; + if (!alloc_cpumask_var(&covered_cpus, GFP_KERNEL)) + return -ENOMEM; + mutex_lock(&performance_mutex); retval = 0; @@ -617,7 +620,7 @@ int acpi_processor_preregister_performance( } pr->performance = percpu_ptr(performance, i); - cpu_set(i, pr->performance->shared_cpu_map); + cpumask_set_cpu(i, pr->performance->shared_cpu_map); if (acpi_processor_get_psd(pr)) { retval = -EINVAL; continue; @@ -650,18 +653,18 @@ int acpi_processor_preregister_performance( } } - cpus_clear(covered_cpus); + cpumask_clear(covered_cpus); for_each_possible_cpu(i) { pr = per_cpu(processors, i); if (!pr) continue; - if (cpu_isset(i, covered_cpus)) + if (cpumask_test_cpu(i, covered_cpus)) continue; pdomain = &(pr->performance->domain_info); - cpu_set(i, pr->performance->shared_cpu_map); - cpu_set(i, covered_cpus); + cpumask_set_cpu(i, pr->performance->shared_cpu_map); + cpumask_set_cpu(i, covered_cpus); if (pdomain->num_processors <= 1) continue; @@ -699,8 +702,8 @@ int acpi_processor_preregister_performance( goto err_ret; } - cpu_set(j, covered_cpus); - cpu_set(j, pr->performance->shared_cpu_map); + cpumask_set_cpu(j, covered_cpus); + cpumask_set_cpu(j, pr->performance->shared_cpu_map); count++; } @@ -718,8 +721,8 @@ int acpi_processor_preregister_performance( match_pr->performance->shared_type = pr->performance->shared_type; - match_pr->performance->shared_cpu_map = - pr->performance->shared_cpu_map; + cpumask_copy(match_pr->performance->shared_cpu_map, + pr->performance->shared_cpu_map); } } @@ -731,14 +734,15 @@ err_ret: /* Assume no coordination on any error parsing domain info */ if (retval) { - cpus_clear(pr->performance->shared_cpu_map); - cpu_set(i, pr->performance->shared_cpu_map); + cpumask_clear(pr->performance->shared_cpu_map); + cpumask_set_cpu(i, pr->performance->shared_cpu_map); pr->performance->shared_type = CPUFREQ_SHARED_TYPE_ALL; } pr->performance = NULL; /* Will be set for real in register */ } mutex_unlock(&performance_mutex); + free_cpumask_var(covered_cpus); return retval; } EXPORT_SYMBOL(acpi_processor_preregister_performance); diff --git a/drivers/acpi/processor_throttling.c b/drivers/acpi/processor_throttling.c index a0c38c94a8a0..d27838171f4a 100644 --- a/drivers/acpi/processor_throttling.c +++ b/drivers/acpi/processor_throttling.c @@ -61,11 +61,14 @@ static int acpi_processor_update_tsd_coord(void) int count, count_target; int retval = 0; unsigned int i, j; - cpumask_t covered_cpus; + cpumask_var_t covered_cpus; struct acpi_processor *pr, *match_pr; struct acpi_tsd_package *pdomain, *match_pdomain; struct acpi_processor_throttling *pthrottling, *match_pthrottling; + if (!alloc_cpumask_var(&covered_cpus, GFP_KERNEL)) + return -ENOMEM; + /* * Now that we have _TSD data from all CPUs, lets setup T-state * coordination between all CPUs. @@ -91,19 +94,19 @@ static int acpi_processor_update_tsd_coord(void) if (retval) goto err_ret; - cpus_clear(covered_cpus); + cpumask_clear(covered_cpus); for_each_possible_cpu(i) { pr = per_cpu(processors, i); if (!pr) continue; - if (cpu_isset(i, covered_cpus)) + if (cpumask_test_cpu(i, covered_cpus)) continue; pthrottling = &pr->throttling; pdomain = &(pthrottling->domain_info); - cpu_set(i, pthrottling->shared_cpu_map); - cpu_set(i, covered_cpus); + cpumask_set_cpu(i, pthrottling->shared_cpu_map); + cpumask_set_cpu(i, covered_cpus); /* * If the number of processor in the TSD domain is 1, it is * unnecessary to parse the coordination for this CPU. @@ -144,8 +147,8 @@ static int acpi_processor_update_tsd_coord(void) goto err_ret; } - cpu_set(j, covered_cpus); - cpu_set(j, pthrottling->shared_cpu_map); + cpumask_set_cpu(j, covered_cpus); + cpumask_set_cpu(j, pthrottling->shared_cpu_map); count++; } for_each_possible_cpu(j) { @@ -165,12 +168,14 @@ static int acpi_processor_update_tsd_coord(void) * If some CPUS have the same domain, they * will have the same shared_cpu_map. */ - match_pthrottling->shared_cpu_map = - pthrottling->shared_cpu_map; + cpumask_copy(match_pthrottling->shared_cpu_map, + pthrottling->shared_cpu_map); } } err_ret: + free_cpumask_var(covered_cpus); + for_each_possible_cpu(i) { pr = per_cpu(processors, i); if (!pr) @@ -182,8 +187,8 @@ err_ret: */ if (retval) { pthrottling = &(pr->throttling); - cpus_clear(pthrottling->shared_cpu_map); - cpu_set(i, pthrottling->shared_cpu_map); + cpumask_clear(pthrottling->shared_cpu_map); + cpumask_set_cpu(i, pthrottling->shared_cpu_map); pthrottling->shared_type = DOMAIN_COORD_TYPE_SW_ALL; } } @@ -567,7 +572,7 @@ static int acpi_processor_get_tsd(struct acpi_processor *pr) pthrottling = &pr->throttling; pthrottling->tsd_valid_flag = 1; pthrottling->shared_type = pdomain->coord_type; - cpu_set(pr->id, pthrottling->shared_cpu_map); + cpumask_set_cpu(pr->id, pthrottling->shared_cpu_map); /* * If the coordination type is not defined in ACPI spec, * the tsd_valid_flag will be clear and coordination type @@ -826,7 +831,7 @@ static int acpi_processor_get_throttling_ptc(struct acpi_processor *pr) static int acpi_processor_get_throttling(struct acpi_processor *pr) { - cpumask_t saved_mask; + cpumask_var_t saved_mask; int ret; if (!pr) @@ -834,14 +839,20 @@ static int acpi_processor_get_throttling(struct acpi_processor *pr) if (!pr->flags.throttling) return -ENODEV; + + if (!alloc_cpumask_var(&saved_mask, GFP_KERNEL)) + return -ENOMEM; + /* * Migrate task to the cpu pointed by pr. */ - saved_mask = current->cpus_allowed; - set_cpus_allowed_ptr(current, &cpumask_of_cpu(pr->id)); + cpumask_copy(saved_mask, ¤t->cpus_allowed); + /* FIXME: use work_on_cpu() */ + set_cpus_allowed_ptr(current, cpumask_of(pr->id)); ret = pr->throttling.acpi_processor_get_throttling(pr); /* restore the previous state */ - set_cpus_allowed_ptr(current, &saved_mask); + set_cpus_allowed_ptr(current, saved_mask); + free_cpumask_var(saved_mask); return ret; } @@ -986,13 +997,13 @@ static int acpi_processor_set_throttling_ptc(struct acpi_processor *pr, int acpi_processor_set_throttling(struct acpi_processor *pr, int state) { - cpumask_t saved_mask; + cpumask_var_t saved_mask; int ret = 0; unsigned int i; struct acpi_processor *match_pr; struct acpi_processor_throttling *p_throttling; struct throttling_tstate t_state; - cpumask_t online_throttling_cpus; + cpumask_var_t online_throttling_cpus; if (!pr) return -EINVAL; @@ -1003,17 +1014,25 @@ int acpi_processor_set_throttling(struct acpi_processor *pr, int state) if ((state < 0) || (state > (pr->throttling.state_count - 1))) return -EINVAL; - saved_mask = current->cpus_allowed; + if (!alloc_cpumask_var(&saved_mask, GFP_KERNEL)) + return -ENOMEM; + + if (!alloc_cpumask_var(&online_throttling_cpus, GFP_KERNEL)) { + free_cpumask_var(saved_mask); + return -ENOMEM; + } + + cpumask_copy(saved_mask, ¤t->cpus_allowed); t_state.target_state = state; p_throttling = &(pr->throttling); - cpus_and(online_throttling_cpus, cpu_online_map, - p_throttling->shared_cpu_map); + cpumask_and(online_throttling_cpus, cpu_online_mask, + p_throttling->shared_cpu_map); /* * The throttling notifier will be called for every * affected cpu in order to get one proper T-state. * The notifier event is THROTTLING_PRECHANGE. */ - for_each_cpu_mask_nr(i, online_throttling_cpus) { + for_each_cpu(i, online_throttling_cpus) { t_state.cpu = i; acpi_processor_throttling_notifier(THROTTLING_PRECHANGE, &t_state); @@ -1025,7 +1044,8 @@ int acpi_processor_set_throttling(struct acpi_processor *pr, int state) * it can be called only for the cpu pointed by pr. */ if (p_throttling->shared_type == DOMAIN_COORD_TYPE_SW_ANY) { - set_cpus_allowed_ptr(current, &cpumask_of_cpu(pr->id)); + /* FIXME: use work_on_cpu() */ + set_cpus_allowed_ptr(current, cpumask_of(pr->id)); ret = p_throttling->acpi_processor_set_throttling(pr, t_state.target_state); } else { @@ -1034,7 +1054,7 @@ int acpi_processor_set_throttling(struct acpi_processor *pr, int state) * it is necessary to set T-state for every affected * cpus. */ - for_each_cpu_mask_nr(i, online_throttling_cpus) { + for_each_cpu(i, online_throttling_cpus) { match_pr = per_cpu(processors, i); /* * If the pointer is invalid, we will report the @@ -1056,7 +1076,8 @@ int acpi_processor_set_throttling(struct acpi_processor *pr, int state) continue; } t_state.cpu = i; - set_cpus_allowed_ptr(current, &cpumask_of_cpu(i)); + /* FIXME: use work_on_cpu() */ + set_cpus_allowed_ptr(current, cpumask_of(i)); ret = match_pr->throttling. acpi_processor_set_throttling( match_pr, t_state.target_state); @@ -1068,13 +1089,16 @@ int acpi_processor_set_throttling(struct acpi_processor *pr, int state) * affected cpu to update the T-states. * The notifier event is THROTTLING_POSTCHANGE */ - for_each_cpu_mask_nr(i, online_throttling_cpus) { + for_each_cpu(i, online_throttling_cpus) { t_state.cpu = i; acpi_processor_throttling_notifier(THROTTLING_POSTCHANGE, &t_state); } /* restore the previous state */ - set_cpus_allowed_ptr(current, &saved_mask); + /* FIXME: use work_on_cpu() */ + set_cpus_allowed_ptr(current, saved_mask); + free_cpumask_var(online_throttling_cpus); + free_cpumask_var(saved_mask); return ret; } @@ -1120,7 +1144,7 @@ int acpi_processor_get_throttling_info(struct acpi_processor *pr) if (acpi_processor_get_tsd(pr)) { pthrottling = &pr->throttling; pthrottling->tsd_valid_flag = 0; - cpu_set(pr->id, pthrottling->shared_cpu_map); + cpumask_set_cpu(pr->id, pthrottling->shared_cpu_map); pthrottling->shared_type = DOMAIN_COORD_TYPE_SW_ALL; } diff --git a/include/acpi/processor.h b/include/acpi/processor.h index 3795590e152a..0574add2a1e3 100644 --- a/include/acpi/processor.h +++ b/include/acpi/processor.h @@ -127,7 +127,7 @@ struct acpi_processor_performance { unsigned int state_count; struct acpi_processor_px *states; struct acpi_psd_package domain_info; - cpumask_t shared_cpu_map; + cpumask_var_t shared_cpu_map; unsigned int shared_type; }; @@ -172,7 +172,7 @@ struct acpi_processor_throttling { unsigned int state_count; struct acpi_processor_tx_tss *states_tss; struct acpi_tsd_package domain_info; - cpumask_t shared_cpu_map; + cpumask_var_t shared_cpu_map; int (*acpi_processor_get_throttling) (struct acpi_processor * pr); int (*acpi_processor_set_throttling) (struct acpi_processor * pr, int state); -- cgit v1.2.3 From 80855f7361eb68205e6bc1981928629d9b02d5c9 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Wed, 31 Dec 2008 18:08:47 -0800 Subject: cpumask: use alloc_cpumask_var_node where appropriate Impact: Reduce inter-node memory traffic. Reduces inter-node memory traffic (offloading the global system bus) by allocating referenced struct cpumasks on the same node as the referring struct. Signed-off-by: Mike Travis Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c | 5 +++-- arch/x86/kernel/io_apic.c | 6 +++--- 2 files changed, 6 insertions(+), 5 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c index d0a001093b2d..28102ad1a363 100644 --- a/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c +++ b/arch/x86/kernel/cpu/cpufreq/acpi-cpufreq.c @@ -547,8 +547,9 @@ static int __init acpi_cpufreq_early_init(void) return -ENOMEM; } for_each_possible_cpu(i) { - if (!alloc_cpumask_var(&per_cpu_ptr(acpi_perf_data, i) - ->shared_cpu_map, GFP_KERNEL)) { + if (!alloc_cpumask_var_node( + &per_cpu_ptr(acpi_perf_data, i)->shared_cpu_map, + GFP_KERNEL, cpu_to_node(i))) { /* Freeing a NULL pointer is OK: alloc_percpu zeroes. */ free_acpi_perf_data(); diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 3e070bb961d7..a25c3f76b8ac 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -212,11 +212,11 @@ static struct irq_cfg *get_one_free_irq_cfg(int cpu) cfg = kzalloc_node(sizeof(*cfg), GFP_ATOMIC, node); if (cfg) { - /* FIXME: needs alloc_cpumask_var_node() */ - if (!alloc_cpumask_var(&cfg->domain, GFP_ATOMIC)) { + if (!alloc_cpumask_var_node(&cfg->domain, GFP_ATOMIC, node)) { kfree(cfg); cfg = NULL; - } else if (!alloc_cpumask_var(&cfg->old_domain, GFP_ATOMIC)) { + } else if (!alloc_cpumask_var_node(&cfg->old_domain, + GFP_ATOMIC, node)) { free_cpumask_var(cfg->domain); kfree(cfg); cfg = NULL; -- cgit v1.2.3 From ab14398abd195af91a744c320a52a1bce814dd1e Mon Sep 17 00:00:00 2001 From: Cyrill Gorcunov Date: Fri, 2 Jan 2009 21:51:32 +0300 Subject: x86: setup_per_cpu_areas() cleanup Impact: cleanup __alloc_bootmem and __alloc_bootmem_node do panic for us in case of fail so no need for additional checks here. Also lets use pr_*() macros for printing. Signed-off-by: Cyrill Gorcunov Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup_percpu.c | 25 ++++++++----------------- 1 file changed, 8 insertions(+), 17 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 49f3f709ee1f..a4b619c33106 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -153,12 +153,10 @@ void __init setup_per_cpu_areas(void) align = max_t(unsigned long, PAGE_SIZE, align); size = roundup(old_size, align); - printk(KERN_INFO - "NR_CPUS:%d nr_cpumask_bits:%d nr_cpu_ids:%d nr_node_ids:%d\n", + pr_info("NR_CPUS:%d nr_cpumask_bits:%d nr_cpu_ids:%d nr_node_ids:%d\n", NR_CPUS, nr_cpumask_bits, nr_cpu_ids, nr_node_ids); - printk(KERN_INFO "PERCPU: Allocating %zd bytes of per cpu data\n", - size); + pr_info("PERCPU: Allocating %zd bytes of per cpu data\n", size); for_each_possible_cpu(cpu) { #ifndef CONFIG_NEED_MULTIPLE_NODES @@ -169,22 +167,15 @@ void __init setup_per_cpu_areas(void) if (!node_online(node) || !NODE_DATA(node)) { ptr = __alloc_bootmem(size, align, __pa(MAX_DMA_ADDRESS)); - printk(KERN_INFO - "cpu %d has no node %d or node-local memory\n", + pr_info("cpu %d has no node %d or node-local memory\n", cpu, node); - if (ptr) - printk(KERN_DEBUG - "per cpu data for cpu%d at %016lx\n", - cpu, __pa(ptr)); - } - else { + pr_debug("per cpu data for cpu%d at %016lx\n", + cpu, __pa(ptr)); + } else { ptr = __alloc_bootmem_node(NODE_DATA(node), size, align, __pa(MAX_DMA_ADDRESS)); - if (ptr) - printk(KERN_DEBUG - "per cpu data for cpu%d on node%d " - "at %016lx\n", - cpu, node, __pa(ptr)); + pr_debug("per cpu data for cpu%d on node%d at %016lx\n", + cpu, node, __pa(ptr)); } #endif per_cpu_offset(cpu) = ptr - __per_cpu_start; -- cgit v1.2.3