From 36f5101a60de8f79c0d1ca06e50660bf5129e02c Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:51 -0800 Subject: [PATCH 01/16] x86: enable MAXSMP Impact: activates new off-stack cpumask code on MAXSMP (non-default) x86 configs Set MAXSMP to enable CONFIG_CPUMASK_OFFSTACK which moves cpumask's off the stack (and in structs) when using cpumask_var_t. Signed-off-by: Mike Travis Signed-off-by: Rusty Russell Acked-by: Ingo Molnar --- arch/x86/Kconfig | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index d99eeb7915c6..1fd44352f27c 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -591,16 +591,17 @@ config IOMMU_HELPER config MAXSMP bool "Configure Maximum number of SMP Processors and NUMA Nodes" - depends on X86_64 && SMP && BROKEN + depends on X86_64 && SMP && DEBUG_KERNEL && EXPERIMENTAL + select CPUMASK_OFFSTACK default n help Configure maximum number of CPUS and NUMA Nodes for this architecture. If unsure, say N. config NR_CPUS - int "Maximum number of CPUs (2-512)" if !MAXSMP - range 2 512 depends on SMP + int "Maximum number of CPUs" if SMP && !MAXSMP + range 2 512 if SMP && !MAXSMP default "4096" if MAXSMP default "32" if X86_NUMAQ || X86_SUMMIT || X86_BIGSMP || X86_ES7000 default "8" From e7986739a76cde5079da08809d8bbc6878387ae0 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:52 -0800 Subject: [PATCH 02/16] x86 smp: modify send_IPI_mask interface to accept cpumask_t pointers Impact: cleanup, change parameter passing * Change genapic interfaces to accept cpumask_t pointers where possible. * Modify external callers to use cpumask_t pointers in function calls. * Create new send_IPI_mask_allbutself which is the same as the send_IPI_mask functions but removes smp_processor_id() from list. This removes another common need for a temporary cpumask_t variable. * Functions that used a temp cpumask_t variable for: cpumask_t allbutme = cpu_online_map; cpu_clear(smp_processor_id(), allbutme); if (!cpus_empty(allbutme)) ... become: if (!cpus_equal(cpu_online_map, cpumask_of_cpu(cpu))) ... * Other minor code optimizations (like using cpus_clear instead of CPU_MASK_NONE, etc.) Applies to linux-2.6.tip/master. Signed-off-by: Mike Travis Signed-off-by: Rusty Russell Acked-by: Ingo Molnar --- arch/x86/include/asm/bigsmp/apic.h | 14 +- arch/x86/include/asm/bigsmp/ipi.h | 9 +- arch/x86/include/asm/es7000/apic.h | 38 ++--- arch/x86/include/asm/es7000/ipi.h | 9 +- arch/x86/include/asm/genapic_32.h | 9 +- arch/x86/include/asm/genapic_64.h | 11 +- arch/x86/include/asm/ipi.h | 21 ++- arch/x86/include/asm/mach-default/mach_apic.h | 17 +- arch/x86/include/asm/mach-default/mach_ipi.h | 18 +-- arch/x86/include/asm/numaq/apic.h | 6 +- arch/x86/include/asm/numaq/ipi.h | 9 +- arch/x86/include/asm/smp.h | 6 +- arch/x86/include/asm/summit/apic.h | 12 +- arch/x86/include/asm/summit/ipi.h | 9 +- arch/x86/kernel/apic.c | 6 +- arch/x86/kernel/crash.c | 5 +- arch/x86/kernel/genapic_flat_64.c | 76 +++++---- arch/x86/kernel/genx2apic_cluster.c | 60 +++++--- arch/x86/kernel/genx2apic_phys.c | 55 +++++-- arch/x86/kernel/genx2apic_uv_x.c | 43 ++++-- arch/x86/kernel/io_apic.c | 145 +++++++++--------- arch/x86/kernel/ipi.c | 26 +++- arch/x86/kernel/smp.c | 8 +- arch/x86/kernel/tlb_32.c | 2 +- arch/x86/kernel/tlb_64.c | 2 +- arch/x86/mach-generic/bigsmp.c | 5 +- arch/x86/mach-generic/es7000.c | 5 +- arch/x86/mach-generic/numaq.c | 5 +- arch/x86/mach-generic/summit.c | 5 +- arch/x86/xen/smp.c | 17 +- 30 files changed, 380 insertions(+), 273 deletions(-) diff --git a/arch/x86/include/asm/bigsmp/apic.h b/arch/x86/include/asm/bigsmp/apic.h index ce547f24a1cd..dc6225ca48ad 100644 --- a/arch/x86/include/asm/bigsmp/apic.h +++ b/arch/x86/include/asm/bigsmp/apic.h @@ -9,12 +9,12 @@ static inline int apic_id_registered(void) return (1); } -static inline cpumask_t target_cpus(void) +static inline const cpumask_t *target_cpus(void) { #ifdef CONFIG_SMP - return cpu_online_map; + return &cpu_online_map; #else - return cpumask_of_cpu(0); + return &cpumask_of_cpu(0); #endif } @@ -79,7 +79,7 @@ static inline int apicid_to_node(int logical_apicid) 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); return BAD_APICID; @@ -94,7 +94,7 @@ extern u8 cpu_2_logical_apicid[]; /* Mapping from cpu number to logical apicid */ static inline int cpu_to_logical_apicid(int cpu) { - if (cpu >= NR_CPUS) + if (cpu >= nr_cpu_ids) return BAD_APICID; return cpu_physical_id(cpu); } @@ -119,12 +119,12 @@ static inline int check_phys_apicid_present(int boot_cpu_physical_apicid) } /* As we are using single CPU as destination, pick only one CPU here */ -static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask) +static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) { int cpu; int apicid; - cpu = first_cpu(cpumask); + cpu = first_cpu(*cpumask); apicid = cpu_to_logical_apicid(cpu); return apicid; } diff --git a/arch/x86/include/asm/bigsmp/ipi.h b/arch/x86/include/asm/bigsmp/ipi.h index 9404c535b7ec..63553e9f22b2 100644 --- a/arch/x86/include/asm/bigsmp/ipi.h +++ b/arch/x86/include/asm/bigsmp/ipi.h @@ -1,9 +1,10 @@ #ifndef __ASM_MACH_IPI_H #define __ASM_MACH_IPI_H -void send_IPI_mask_sequence(cpumask_t mask, int vector); +void send_IPI_mask_sequence(const cpumask_t *mask, int vector); +void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); -static inline void send_IPI_mask(cpumask_t mask, int vector) +static inline void send_IPI_mask(const cpumask_t *mask, int vector) { send_IPI_mask_sequence(mask, vector); } @@ -14,12 +15,12 @@ static inline void send_IPI_allbutself(int vector) cpu_clear(smp_processor_id(), mask); if (!cpus_empty(mask)) - send_IPI_mask(mask, vector); + send_IPI_mask(&mask, vector); } static inline void send_IPI_all(int vector) { - send_IPI_mask(cpu_online_map, vector); + send_IPI_mask(&cpu_online_map, vector); } #endif /* __ASM_MACH_IPI_H */ diff --git a/arch/x86/include/asm/es7000/apic.h b/arch/x86/include/asm/es7000/apic.h index e24ef876915f..4cac0837bb40 100644 --- a/arch/x86/include/asm/es7000/apic.h +++ b/arch/x86/include/asm/es7000/apic.h @@ -9,14 +9,14 @@ static inline int apic_id_registered(void) return (1); } -static inline cpumask_t target_cpus_cluster(void) +static inline const cpumask_t *target_cpus_cluster(void) { - return CPU_MASK_ALL; + return &CPU_MASK_ALL; } -static inline cpumask_t target_cpus(void) +static inline const cpumask_t *target_cpus(void) { - return cpumask_of_cpu(smp_processor_id()); + return &cpumask_of_cpu(smp_processor_id()); } #define APIC_DFR_VALUE_CLUSTER (APIC_DFR_CLUSTER) @@ -80,9 +80,10 @@ extern int apic_version [MAX_APICS]; static inline void setup_apic_routing(void) { int apic = per_cpu(x86_bios_cpu_apicid, smp_processor_id()); - printk("Enabling APIC mode: %s. Using %d I/O APICs, target cpus %lx\n", + printk("Enabling APIC mode: %s. Using %d I/O APICs, target cpus %lx\n", (apic_version[apic] == 0x14) ? - "Physical Cluster" : "Logical Cluster", nr_ioapics, cpus_addr(target_cpus())[0]); + "Physical Cluster" : "Logical Cluster", + nr_ioapics, cpus_addr(*target_cpus())[0]); } static inline int multi_timer_check(int apic, int irq) @@ -100,7 +101,7 @@ static inline int cpu_present_to_apicid(int mps_cpu) { if (!mps_cpu) return boot_cpu_physical_apicid; - else if (mps_cpu < NR_CPUS) + else if (mps_cpu < nr_cpu_ids) return (int) per_cpu(x86_bios_cpu_apicid, mps_cpu); else return BAD_APICID; @@ -120,9 +121,9 @@ extern u8 cpu_2_logical_apicid[]; static inline int cpu_to_logical_apicid(int cpu) { #ifdef CONFIG_SMP - if (cpu >= NR_CPUS) - return BAD_APICID; - return (int)cpu_2_logical_apicid[cpu]; + if (cpu >= nr_cpu_ids) + return BAD_APICID; + return (int)cpu_2_logical_apicid[cpu]; #else return logical_smp_processor_id(); #endif @@ -146,14 +147,15 @@ static inline int check_phys_apicid_present(int cpu_physical_apicid) return (1); } -static inline unsigned int cpu_mask_to_apicid_cluster(cpumask_t cpumask) +static inline unsigned int +cpu_mask_to_apicid_cluster(const struct cpumask *cpumask) { int num_bits_set; int cpus_found = 0; int cpu; int apicid; - num_bits_set = cpus_weight(cpumask); + num_bits_set = cpumask_weight(cpumask); /* Return id to all */ if (num_bits_set == NR_CPUS) return 0xFF; @@ -161,10 +163,10 @@ static inline unsigned int cpu_mask_to_apicid_cluster(cpumask_t cpumask) * 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 = first_cpu(cpumask); + cpu = cpumask_first(cpumask); apicid = cpu_to_logical_apicid(cpu); while (cpus_found < num_bits_set) { - if (cpu_isset(cpu, cpumask)) { + if (cpumask_test_cpu(cpu, cpumask)) { int new_apicid = cpu_to_logical_apicid(cpu); if (apicid_cluster(apicid) != apicid_cluster(new_apicid)){ @@ -179,14 +181,14 @@ static inline unsigned int cpu_mask_to_apicid_cluster(cpumask_t cpumask) return apicid; } -static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask) +static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) { int num_bits_set; int cpus_found = 0; int cpu; int apicid; - num_bits_set = cpus_weight(cpumask); + num_bits_set = cpus_weight(*cpumask); /* Return id to all */ if (num_bits_set == NR_CPUS) return cpu_to_logical_apicid(0); @@ -194,10 +196,10 @@ static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask) * 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 = first_cpu(cpumask); + cpu = first_cpu(*cpumask); apicid = cpu_to_logical_apicid(cpu); while (cpus_found < num_bits_set) { - if (cpu_isset(cpu, cpumask)) { + if (cpu_isset(cpu, *cpumask)) { int new_apicid = cpu_to_logical_apicid(cpu); if (apicid_cluster(apicid) != apicid_cluster(new_apicid)){ diff --git a/arch/x86/include/asm/es7000/ipi.h b/arch/x86/include/asm/es7000/ipi.h index 632a955fcc0a..1a8507265f91 100644 --- a/arch/x86/include/asm/es7000/ipi.h +++ b/arch/x86/include/asm/es7000/ipi.h @@ -1,9 +1,10 @@ #ifndef __ASM_ES7000_IPI_H #define __ASM_ES7000_IPI_H -void send_IPI_mask_sequence(cpumask_t mask, int vector); +void send_IPI_mask_sequence(const cpumask_t *mask, int vector); +void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); -static inline void send_IPI_mask(cpumask_t mask, int vector) +static inline void send_IPI_mask(const cpumask_t *mask, int vector) { send_IPI_mask_sequence(mask, vector); } @@ -13,12 +14,12 @@ static inline void send_IPI_allbutself(int vector) cpumask_t mask = cpu_online_map; cpu_clear(smp_processor_id(), mask); if (!cpus_empty(mask)) - send_IPI_mask(mask, vector); + send_IPI_mask(&mask, vector); } static inline void send_IPI_all(int vector) { - send_IPI_mask(cpu_online_map, vector); + send_IPI_mask(&cpu_online_map, vector); } #endif /* __ASM_ES7000_IPI_H */ diff --git a/arch/x86/include/asm/genapic_32.h b/arch/x86/include/asm/genapic_32.h index 0ac17d33a8c7..b21ed21c574d 100644 --- a/arch/x86/include/asm/genapic_32.h +++ b/arch/x86/include/asm/genapic_32.h @@ -24,7 +24,7 @@ struct genapic { int (*probe)(void); int (*apic_id_registered)(void); - cpumask_t (*target_cpus)(void); + const cpumask_t *(*target_cpus)(void); int int_delivery_mode; int int_dest_mode; int ESR_DISABLE; @@ -57,12 +57,13 @@ struct genapic { unsigned (*get_apic_id)(unsigned long x); unsigned long apic_id_mask; - unsigned int (*cpu_mask_to_apicid)(cpumask_t cpumask); - cpumask_t (*vector_allocation_domain)(int cpu); + unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask); + void (*vector_allocation_domain)(int cpu, cpumask_t *retmask); #ifdef CONFIG_SMP /* ipi */ - void (*send_IPI_mask)(cpumask_t mask, int vector); + void (*send_IPI_mask)(const cpumask_t *mask, int vector); + void (*send_IPI_mask_allbutself)(const cpumask_t *mask, int vector); void (*send_IPI_allbutself)(int vector); void (*send_IPI_all)(int vector); #endif diff --git a/arch/x86/include/asm/genapic_64.h b/arch/x86/include/asm/genapic_64.h index 2cae011668b7..a020e7d35a40 100644 --- a/arch/x86/include/asm/genapic_64.h +++ b/arch/x86/include/asm/genapic_64.h @@ -1,6 +1,8 @@ #ifndef _ASM_X86_GENAPIC_64_H #define _ASM_X86_GENAPIC_64_H +#include + /* * Copyright 2004 James Cleverdon, IBM. * Subject to the GNU Public License, v.2 @@ -18,16 +20,17 @@ struct genapic { u32 int_delivery_mode; u32 int_dest_mode; int (*apic_id_registered)(void); - cpumask_t (*target_cpus)(void); - cpumask_t (*vector_allocation_domain)(int cpu); + const cpumask_t *(*target_cpus)(void); + void (*vector_allocation_domain)(int cpu, cpumask_t *retmask); void (*init_apic_ldr)(void); /* ipi */ - void (*send_IPI_mask)(cpumask_t mask, int vector); + void (*send_IPI_mask)(const cpumask_t *mask, int vector); + void (*send_IPI_mask_allbutself)(const cpumask_t *mask, int vector); void (*send_IPI_allbutself)(int vector); void (*send_IPI_all)(int vector); void (*send_IPI_self)(int vector); /* */ - unsigned int (*cpu_mask_to_apicid)(cpumask_t cpumask); + unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask); unsigned int (*phys_pkg_id)(int index_msb); unsigned int (*get_apic_id)(unsigned long x); unsigned long (*set_apic_id)(unsigned int id); diff --git a/arch/x86/include/asm/ipi.h b/arch/x86/include/asm/ipi.h index f89dffb28aa9..24b6e613edfa 100644 --- a/arch/x86/include/asm/ipi.h +++ b/arch/x86/include/asm/ipi.h @@ -117,7 +117,7 @@ static inline void __send_IPI_dest_field(unsigned int mask, int vector, native_apic_mem_write(APIC_ICR, cfg); } -static inline void send_IPI_mask_sequence(cpumask_t mask, int vector) +static inline void send_IPI_mask_sequence(const cpumask_t *mask, int vector) { unsigned long flags; unsigned long query_cpu; @@ -128,11 +128,28 @@ static inline void send_IPI_mask_sequence(cpumask_t mask, int vector) * - mbligh */ local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, mask) { + for_each_cpu_mask_nr(query_cpu, *mask) { __send_IPI_dest_field(per_cpu(x86_cpu_to_apicid, query_cpu), vector, APIC_DEST_PHYSICAL); } local_irq_restore(flags); } +static inline void send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +{ + unsigned long flags; + unsigned int query_cpu; + unsigned int this_cpu = smp_processor_id(); + + /* See Hack comment above */ + + local_irq_save(flags); + for_each_cpu_mask_nr(query_cpu, *mask) + if (query_cpu != this_cpu) + __send_IPI_dest_field( + per_cpu(x86_cpu_to_apicid, query_cpu), + vector, APIC_DEST_PHYSICAL); + local_irq_restore(flags); +} + #endif /* _ASM_X86_IPI_H */ diff --git a/arch/x86/include/asm/mach-default/mach_apic.h b/arch/x86/include/asm/mach-default/mach_apic.h index 6cb3a467e067..c18896b0508c 100644 --- a/arch/x86/include/asm/mach-default/mach_apic.h +++ b/arch/x86/include/asm/mach-default/mach_apic.h @@ -8,12 +8,12 @@ #define APIC_DFR_VALUE (APIC_DFR_FLAT) -static inline cpumask_t target_cpus(void) +static inline const cpumask_t *target_cpus(void) { #ifdef CONFIG_SMP - return cpu_online_map; + return &cpu_online_map; #else - return cpumask_of_cpu(0); + return &cpumask_of_cpu(0); #endif } @@ -61,9 +61,9 @@ static inline int apic_id_registered(void) return physid_isset(read_apic_id(), phys_cpu_present_map); } -static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask) +static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) { - return cpus_addr(cpumask)[0]; + return cpus_addr(*cpumask)[0]; } static inline u32 phys_pkg_id(u32 cpuid_apic, int index_msb) @@ -88,7 +88,7 @@ static inline int apicid_to_node(int logical_apicid) #endif } -static inline cpumask_t vector_allocation_domain(int cpu) +static inline void vector_allocation_domain(int cpu, cpumask_t *retmask) { /* Careful. Some cpus do not strictly honor the set of cpus * specified in the interrupt destination when using lowest @@ -98,8 +98,7 @@ static inline cpumask_t vector_allocation_domain(int cpu) * deliver interrupts to the wrong hyperthread when only one * hyperthread was specified in the interrupt desitination. */ - cpumask_t domain = { { [0] = APIC_ALL_CPUS, } }; - return domain; + *retmask = (cpumask_t) { { [0] = APIC_ALL_CPUS } }; } #endif @@ -131,7 +130,7 @@ static inline int cpu_to_logical_apicid(int cpu) static inline int cpu_present_to_apicid(int mps_cpu) { - if (mps_cpu < NR_CPUS && cpu_present(mps_cpu)) + if (mps_cpu < nr_cpu_ids && cpu_present(mps_cpu)) return (int)per_cpu(x86_bios_cpu_apicid, mps_cpu); else return BAD_APICID; diff --git a/arch/x86/include/asm/mach-default/mach_ipi.h b/arch/x86/include/asm/mach-default/mach_ipi.h index fabca01ebacf..9353ab854a10 100644 --- a/arch/x86/include/asm/mach-default/mach_ipi.h +++ b/arch/x86/include/asm/mach-default/mach_ipi.h @@ -4,7 +4,8 @@ /* Avoid include hell */ #define NMI_VECTOR 0x02 -void send_IPI_mask_bitmask(cpumask_t mask, int vector); +void send_IPI_mask_bitmask(const cpumask_t *mask, int vector); +void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); void __send_IPI_shortcut(unsigned int shortcut, int vector); extern int no_broadcast; @@ -12,28 +13,27 @@ extern int no_broadcast; #ifdef CONFIG_X86_64 #include #define send_IPI_mask (genapic->send_IPI_mask) +#define send_IPI_mask_allbutself (genapic->send_IPI_mask_allbutself) #else -static inline void send_IPI_mask(cpumask_t mask, int vector) +static inline void send_IPI_mask(const cpumask_t *mask, int vector) { send_IPI_mask_bitmask(mask, vector); } +void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); #endif static inline void __local_send_IPI_allbutself(int vector) { - if (no_broadcast || vector == NMI_VECTOR) { - cpumask_t mask = cpu_online_map; - - cpu_clear(smp_processor_id(), mask); - send_IPI_mask(mask, vector); - } else + if (no_broadcast || vector == NMI_VECTOR) + send_IPI_mask_allbutself(&cpu_online_map, vector); + else __send_IPI_shortcut(APIC_DEST_ALLBUT, vector); } static inline void __local_send_IPI_all(int vector) { if (no_broadcast || vector == NMI_VECTOR) - send_IPI_mask(cpu_online_map, vector); + send_IPI_mask(&cpu_online_map, vector); else __send_IPI_shortcut(APIC_DEST_ALLINC, vector); } diff --git a/arch/x86/include/asm/numaq/apic.h b/arch/x86/include/asm/numaq/apic.h index 0bf2a06b7a4e..1df7ebe738e5 100644 --- a/arch/x86/include/asm/numaq/apic.h +++ b/arch/x86/include/asm/numaq/apic.h @@ -7,9 +7,9 @@ #define APIC_DFR_VALUE (APIC_DFR_CLUSTER) -static inline cpumask_t target_cpus(void) +static inline const cpumask_t *target_cpus(void) { - return CPU_MASK_ALL; + return &CPU_MASK_ALL; } #define NO_BALANCE_IRQ (1) @@ -122,7 +122,7 @@ static inline void enable_apic_mode(void) * We use physical apicids here, not logical, so just return the default * physical broadcast to stop people from breaking us */ -static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask) +static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) { return (int) 0xF; } diff --git a/arch/x86/include/asm/numaq/ipi.h b/arch/x86/include/asm/numaq/ipi.h index 935588d286cf..c734d7acc430 100644 --- a/arch/x86/include/asm/numaq/ipi.h +++ b/arch/x86/include/asm/numaq/ipi.h @@ -1,9 +1,10 @@ #ifndef __ASM_NUMAQ_IPI_H #define __ASM_NUMAQ_IPI_H -void send_IPI_mask_sequence(cpumask_t, int vector); +void send_IPI_mask_sequence(const cpumask_t *mask, int vector); +void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); -static inline void send_IPI_mask(cpumask_t mask, int vector) +static inline void send_IPI_mask(const cpumask_t *mask, int vector) { send_IPI_mask_sequence(mask, vector); } @@ -14,12 +15,12 @@ static inline void send_IPI_allbutself(int vector) cpu_clear(smp_processor_id(), mask); if (!cpus_empty(mask)) - send_IPI_mask(mask, vector); + send_IPI_mask(&mask, vector); } static inline void send_IPI_all(int vector) { - send_IPI_mask(cpu_online_map, vector); + send_IPI_mask(&cpu_online_map, vector); } #endif /* __ASM_NUMAQ_IPI_H */ diff --git a/arch/x86/include/asm/smp.h b/arch/x86/include/asm/smp.h index d12811ce51d9..c4a9aa52df6e 100644 --- a/arch/x86/include/asm/smp.h +++ b/arch/x86/include/asm/smp.h @@ -60,7 +60,7 @@ struct smp_ops { void (*cpu_die)(unsigned int cpu); void (*play_dead)(void); - void (*send_call_func_ipi)(cpumask_t mask); + void (*send_call_func_ipi)(const cpumask_t *mask); void (*send_call_func_single_ipi)(int cpu); }; @@ -125,7 +125,7 @@ static inline void arch_send_call_function_single_ipi(int cpu) static inline void arch_send_call_function_ipi(cpumask_t mask) { - smp_ops.send_call_func_ipi(mask); + smp_ops.send_call_func_ipi(&mask); } void cpu_disable_common(void); @@ -138,7 +138,7 @@ void native_cpu_die(unsigned int cpu); void native_play_dead(void); void play_dead_common(void); -void native_send_call_func_ipi(cpumask_t mask); +void native_send_call_func_ipi(const cpumask_t *mask); void native_send_call_func_single_ipi(int cpu); extern void prefill_possible_map(void); diff --git a/arch/x86/include/asm/summit/apic.h b/arch/x86/include/asm/summit/apic.h index 9b3070f1c2ac..437dc83725ca 100644 --- a/arch/x86/include/asm/summit/apic.h +++ b/arch/x86/include/asm/summit/apic.h @@ -14,13 +14,13 @@ #define APIC_DFR_VALUE (APIC_DFR_CLUSTER) -static inline cpumask_t target_cpus(void) +static inline const cpumask_t *target_cpus(void) { /* CPU_MASK_ALL (0xff) has undefined behaviour with * dest_LowestPrio mode logical clustered apic interrupt routing * Just start on cpu 0. IRQ balancing will spread load */ - return cpumask_of_cpu(0); + return &cpumask_of_cpu(0); } #define INT_DELIVERY_MODE (dest_LowestPrio) @@ -137,14 +137,14 @@ static inline void enable_apic_mode(void) { } -static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask) +static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) { int num_bits_set; int cpus_found = 0; int cpu; int apicid; - num_bits_set = cpus_weight(cpumask); + num_bits_set = cpus_weight(*cpumask); /* Return id to all */ if (num_bits_set == NR_CPUS) return (int) 0xFF; @@ -152,10 +152,10 @@ static inline unsigned int cpu_mask_to_apicid(cpumask_t cpumask) * 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 = first_cpu(cpumask); + cpu = first_cpu(*cpumask); apicid = cpu_to_logical_apicid(cpu); while (cpus_found < num_bits_set) { - if (cpu_isset(cpu, cpumask)) { + if (cpu_isset(cpu, *cpumask)) { int new_apicid = cpu_to_logical_apicid(cpu); if (apicid_cluster(apicid) != apicid_cluster(new_apicid)){ diff --git a/arch/x86/include/asm/summit/ipi.h b/arch/x86/include/asm/summit/ipi.h index 53bd1e7bd7b4..a8a2c24f50cc 100644 --- a/arch/x86/include/asm/summit/ipi.h +++ b/arch/x86/include/asm/summit/ipi.h @@ -1,9 +1,10 @@ #ifndef __ASM_SUMMIT_IPI_H #define __ASM_SUMMIT_IPI_H -void send_IPI_mask_sequence(cpumask_t mask, int vector); +void send_IPI_mask_sequence(const cpumask_t *mask, int vector); +void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); -static inline void send_IPI_mask(cpumask_t mask, int vector) +static inline void send_IPI_mask(const cpumask_t *mask, int vector) { send_IPI_mask_sequence(mask, vector); } @@ -14,12 +15,12 @@ static inline void send_IPI_allbutself(int vector) cpu_clear(smp_processor_id(), mask); if (!cpus_empty(mask)) - send_IPI_mask(mask, vector); + send_IPI_mask(&mask, vector); } static inline void send_IPI_all(int vector) { - send_IPI_mask(cpu_online_map, vector); + send_IPI_mask(&cpu_online_map, vector); } #endif /* __ASM_SUMMIT_IPI_H */ diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index b2cef49f3085..a375791c08ca 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -141,7 +141,7 @@ 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 struct cpumask *mask); +static void lapic_timer_broadcast(const cpumask_t *mask); static void apic_pm_activate(void); /* @@ -453,10 +453,10 @@ static void lapic_timer_setup(enum clock_event_mode mode, /* * Local APIC timer broadcast function */ -static void lapic_timer_broadcast(const struct cpumask *mask) +static void lapic_timer_broadcast(const cpumask_t *mask) { #ifdef CONFIG_SMP - send_IPI_mask(*mask, LOCAL_TIMER_VECTOR); + send_IPI_mask(mask, LOCAL_TIMER_VECTOR); #endif } diff --git a/arch/x86/kernel/crash.c b/arch/x86/kernel/crash.c index 268553817909..81e01f7b1d12 100644 --- a/arch/x86/kernel/crash.c +++ b/arch/x86/kernel/crash.c @@ -77,10 +77,7 @@ static int crash_nmi_callback(struct notifier_block *self, static void smp_send_nmi_allbutself(void) { - cpumask_t mask = cpu_online_map; - cpu_clear(safe_smp_processor_id(), mask); - if (!cpus_empty(mask)) - send_IPI_mask(mask, NMI_VECTOR); + send_IPI_allbutself(NMI_VECTOR); } static struct notifier_block crash_nmi_nb = { diff --git a/arch/x86/kernel/genapic_flat_64.c b/arch/x86/kernel/genapic_flat_64.c index c0262791bda4..50eebd0328fe 100644 --- a/arch/x86/kernel/genapic_flat_64.c +++ b/arch/x86/kernel/genapic_flat_64.c @@ -30,12 +30,12 @@ static int flat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) return 1; } -static cpumask_t flat_target_cpus(void) +static const cpumask_t *flat_target_cpus(void) { - return cpu_online_map; + return &cpu_online_map; } -static cpumask_t flat_vector_allocation_domain(int cpu) +static void flat_vector_allocation_domain(int cpu, cpumask_t *retmask) { /* Careful. Some cpus do not strictly honor the set of cpus * specified in the interrupt destination when using lowest @@ -45,8 +45,7 @@ static cpumask_t flat_vector_allocation_domain(int cpu) * deliver interrupts to the wrong hyperthread when only one * hyperthread was specified in the interrupt desitination. */ - cpumask_t domain = { { [0] = APIC_ALL_CPUS, } }; - return domain; + *retmask = (cpumask_t) { {[0] = APIC_ALL_CPUS, } }; } /* @@ -69,9 +68,8 @@ static void flat_init_apic_ldr(void) apic_write(APIC_LDR, val); } -static void flat_send_IPI_mask(cpumask_t cpumask, int vector) +static inline void _flat_send_IPI_mask(unsigned long mask, int vector) { - unsigned long mask = cpus_addr(cpumask)[0]; unsigned long flags; local_irq_save(flags); @@ -79,20 +77,40 @@ static void flat_send_IPI_mask(cpumask_t cpumask, int vector) local_irq_restore(flags); } +static void flat_send_IPI_mask(const cpumask_t *cpumask, int vector) +{ + unsigned long mask = cpus_addr(*cpumask)[0]; + + _flat_send_IPI_mask(mask, vector); +} + +static void flat_send_IPI_mask_allbutself(const cpumask_t *cpumask, int vector) +{ + unsigned long mask = cpus_addr(*cpumask)[0]; + int cpu = smp_processor_id(); + + if (cpu < BITS_PER_LONG) + clear_bit(cpu, &mask); + _flat_send_IPI_mask(mask, vector); +} + static void flat_send_IPI_allbutself(int vector) { + int cpu = smp_processor_id(); #ifdef CONFIG_HOTPLUG_CPU int hotplug = 1; #else int hotplug = 0; #endif if (hotplug || vector == NMI_VECTOR) { - cpumask_t allbutme = cpu_online_map; + if (!cpus_equal(cpu_online_map, cpumask_of_cpu(cpu))) { + unsigned long mask = cpus_addr(cpu_online_map)[0]; - cpu_clear(smp_processor_id(), allbutme); + if (cpu < BITS_PER_LONG) + clear_bit(cpu, &mask); - if (!cpus_empty(allbutme)) - flat_send_IPI_mask(allbutme, vector); + _flat_send_IPI_mask(mask, vector); + } } else if (num_online_cpus() > 1) { __send_IPI_shortcut(APIC_DEST_ALLBUT, vector,APIC_DEST_LOGICAL); } @@ -101,7 +119,7 @@ static void flat_send_IPI_allbutself(int vector) static void flat_send_IPI_all(int vector) { if (vector == NMI_VECTOR) - flat_send_IPI_mask(cpu_online_map, vector); + flat_send_IPI_mask(&cpu_online_map, vector); else __send_IPI_shortcut(APIC_DEST_ALLINC, vector, APIC_DEST_LOGICAL); } @@ -135,9 +153,9 @@ static int flat_apic_id_registered(void) return physid_isset(read_xapic_id(), phys_cpu_present_map); } -static unsigned int flat_cpu_mask_to_apicid(cpumask_t cpumask) +static unsigned int flat_cpu_mask_to_apicid(const cpumask_t *cpumask) { - return cpus_addr(cpumask)[0] & APIC_ALL_CPUS; + return cpus_addr(*cpumask)[0] & APIC_ALL_CPUS; } static unsigned int phys_pkg_id(int index_msb) @@ -157,6 +175,7 @@ struct genapic apic_flat = { .send_IPI_all = flat_send_IPI_all, .send_IPI_allbutself = flat_send_IPI_allbutself, .send_IPI_mask = flat_send_IPI_mask, + .send_IPI_mask_allbutself = flat_send_IPI_mask_allbutself, .send_IPI_self = apic_send_IPI_self, .cpu_mask_to_apicid = flat_cpu_mask_to_apicid, .phys_pkg_id = phys_pkg_id, @@ -188,35 +207,39 @@ static int physflat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) return 0; } -static cpumask_t physflat_target_cpus(void) +static const cpumask_t *physflat_target_cpus(void) { - return cpu_online_map; + return &cpu_online_map; } -static cpumask_t physflat_vector_allocation_domain(int cpu) +static void physflat_vector_allocation_domain(int cpu, cpumask_t *retmask) { - return cpumask_of_cpu(cpu); + cpus_clear(*retmask); + cpu_set(cpu, *retmask); } -static void physflat_send_IPI_mask(cpumask_t cpumask, int vector) +static void physflat_send_IPI_mask(const cpumask_t *cpumask, int vector) { send_IPI_mask_sequence(cpumask, vector); } +static void physflat_send_IPI_mask_allbutself(const cpumask_t *cpumask, + int vector) +{ + send_IPI_mask_allbutself(cpumask, vector); +} + static void physflat_send_IPI_allbutself(int vector) { - cpumask_t allbutme = cpu_online_map; - - cpu_clear(smp_processor_id(), allbutme); - physflat_send_IPI_mask(allbutme, vector); + send_IPI_mask_allbutself(&cpu_online_map, vector); } static void physflat_send_IPI_all(int vector) { - physflat_send_IPI_mask(cpu_online_map, vector); + physflat_send_IPI_mask(&cpu_online_map, vector); } -static unsigned int physflat_cpu_mask_to_apicid(cpumask_t cpumask) +static unsigned int physflat_cpu_mask_to_apicid(const cpumask_t *cpumask) { int cpu; @@ -224,7 +247,7 @@ static unsigned int physflat_cpu_mask_to_apicid(cpumask_t cpumask) * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - cpu = first_cpu(cpumask); + cpu = first_cpu(*cpumask); if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_apicid, cpu); else @@ -243,6 +266,7 @@ struct genapic apic_physflat = { .send_IPI_all = physflat_send_IPI_all, .send_IPI_allbutself = physflat_send_IPI_allbutself, .send_IPI_mask = physflat_send_IPI_mask, + .send_IPI_mask_allbutself = physflat_send_IPI_mask_allbutself, .send_IPI_self = apic_send_IPI_self, .cpu_mask_to_apicid = physflat_cpu_mask_to_apicid, .phys_pkg_id = phys_pkg_id, diff --git a/arch/x86/kernel/genx2apic_cluster.c b/arch/x86/kernel/genx2apic_cluster.c index f6a2c8eb48a6..f5fa9a91ad38 100644 --- a/arch/x86/kernel/genx2apic_cluster.c +++ b/arch/x86/kernel/genx2apic_cluster.c @@ -22,19 +22,18 @@ static int x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id) /* Start with all IRQs pointing to boot CPU. IRQ balancing will shift them. */ -static cpumask_t x2apic_target_cpus(void) +static const cpumask_t *x2apic_target_cpus(void) { - return cpumask_of_cpu(0); + return &cpumask_of_cpu(0); } /* * for now each logical cpu is in its own vector allocation domain. */ -static cpumask_t x2apic_vector_allocation_domain(int cpu) +static void x2apic_vector_allocation_domain(int cpu, cpumask_t *retmask) { - cpumask_t domain = CPU_MASK_NONE; - cpu_set(cpu, domain); - return domain; + cpus_clear(*retmask); + cpu_set(cpu, *retmask); } static void __x2apic_send_IPI_dest(unsigned int apicid, int vector, @@ -56,32 +55,52 @@ static void __x2apic_send_IPI_dest(unsigned int apicid, int vector, * at once. We have 16 cpu's in a cluster. This will minimize IPI register * writes. */ -static void x2apic_send_IPI_mask(cpumask_t mask, int vector) +static void x2apic_send_IPI_mask(const cpumask_t *mask, int vector) { unsigned long flags; unsigned long query_cpu; local_irq_save(flags); - for_each_cpu_mask(query_cpu, mask) { - __x2apic_send_IPI_dest(per_cpu(x86_cpu_to_logical_apicid, query_cpu), - vector, APIC_DEST_LOGICAL); - } + for_each_cpu_mask_nr(query_cpu, *mask) + __x2apic_send_IPI_dest( + per_cpu(x86_cpu_to_logical_apicid, query_cpu), + vector, APIC_DEST_LOGICAL); + local_irq_restore(flags); +} + +static void x2apic_send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +{ + unsigned long flags; + unsigned long query_cpu; + unsigned long this_cpu = smp_processor_id(); + + local_irq_save(flags); + for_each_cpu_mask_nr(query_cpu, *mask) + if (query_cpu != this_cpu) + __x2apic_send_IPI_dest( + per_cpu(x86_cpu_to_logical_apicid, query_cpu), + vector, APIC_DEST_LOGICAL); local_irq_restore(flags); } static void x2apic_send_IPI_allbutself(int vector) { - cpumask_t mask = cpu_online_map; + unsigned long flags; + unsigned long query_cpu; + unsigned long this_cpu = smp_processor_id(); - cpu_clear(smp_processor_id(), mask); - - if (!cpus_empty(mask)) - x2apic_send_IPI_mask(mask, vector); + local_irq_save(flags); + for_each_online_cpu(query_cpu) + if (query_cpu != this_cpu) + __x2apic_send_IPI_dest( + per_cpu(x86_cpu_to_logical_apicid, query_cpu), + vector, APIC_DEST_LOGICAL); + local_irq_restore(flags); } static void x2apic_send_IPI_all(int vector) { - x2apic_send_IPI_mask(cpu_online_map, vector); + x2apic_send_IPI_mask(&cpu_online_map, vector); } static int x2apic_apic_id_registered(void) @@ -89,7 +108,7 @@ static int x2apic_apic_id_registered(void) return 1; } -static unsigned int x2apic_cpu_mask_to_apicid(cpumask_t cpumask) +static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) { int cpu; @@ -97,8 +116,8 @@ static unsigned int x2apic_cpu_mask_to_apicid(cpumask_t cpumask) * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - cpu = first_cpu(cpumask); - if ((unsigned)cpu < NR_CPUS) + cpu = first_cpu(*cpumask); + if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_logical_apicid, cpu); else return BAD_APICID; @@ -150,6 +169,7 @@ struct genapic apic_x2apic_cluster = { .send_IPI_all = x2apic_send_IPI_all, .send_IPI_allbutself = x2apic_send_IPI_allbutself, .send_IPI_mask = x2apic_send_IPI_mask, + .send_IPI_mask_allbutself = x2apic_send_IPI_mask_allbutself, .send_IPI_self = x2apic_send_IPI_self, .cpu_mask_to_apicid = x2apic_cpu_mask_to_apicid, .phys_pkg_id = phys_pkg_id, diff --git a/arch/x86/kernel/genx2apic_phys.c b/arch/x86/kernel/genx2apic_phys.c index d042211768b7..41c27b2f3d01 100644 --- a/arch/x86/kernel/genx2apic_phys.c +++ b/arch/x86/kernel/genx2apic_phys.c @@ -29,16 +29,15 @@ static int x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id) /* Start with all IRQs pointing to boot CPU. IRQ balancing will shift them. */ -static cpumask_t x2apic_target_cpus(void) +static const cpumask_t *x2apic_target_cpus(void) { - return cpumask_of_cpu(0); + return &cpumask_of_cpu(0); } -static cpumask_t x2apic_vector_allocation_domain(int cpu) +static void x2apic_vector_allocation_domain(int cpu, cpumask_t *retmask) { - cpumask_t domain = CPU_MASK_NONE; - cpu_set(cpu, domain); - return domain; + cpus_clear(*retmask); + cpu_set(cpu, *retmask); } static void __x2apic_send_IPI_dest(unsigned int apicid, int vector, @@ -54,32 +53,53 @@ static void __x2apic_send_IPI_dest(unsigned int apicid, int vector, x2apic_icr_write(cfg, apicid); } -static void x2apic_send_IPI_mask(cpumask_t mask, int vector) +static void x2apic_send_IPI_mask(const cpumask_t *mask, int vector) { unsigned long flags; unsigned long query_cpu; local_irq_save(flags); - for_each_cpu_mask(query_cpu, mask) { + for_each_cpu_mask_nr(query_cpu, *mask) { __x2apic_send_IPI_dest(per_cpu(x86_cpu_to_apicid, query_cpu), vector, APIC_DEST_PHYSICAL); } local_irq_restore(flags); } +static void x2apic_send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +{ + unsigned long flags; + unsigned long query_cpu; + unsigned long this_cpu = smp_processor_id(); + + local_irq_save(flags); + for_each_cpu_mask_nr(query_cpu, *mask) { + if (query_cpu != this_cpu) + __x2apic_send_IPI_dest( + per_cpu(x86_cpu_to_apicid, query_cpu), + vector, APIC_DEST_PHYSICAL); + } + local_irq_restore(flags); +} + static void x2apic_send_IPI_allbutself(int vector) { - cpumask_t mask = cpu_online_map; + unsigned long flags; + unsigned long query_cpu; + unsigned long this_cpu = smp_processor_id(); - cpu_clear(smp_processor_id(), mask); - - if (!cpus_empty(mask)) - x2apic_send_IPI_mask(mask, vector); + local_irq_save(flags); + for_each_online_cpu(query_cpu) + if (query_cpu != this_cpu) + __x2apic_send_IPI_dest( + per_cpu(x86_cpu_to_apicid, query_cpu), + vector, APIC_DEST_PHYSICAL); + local_irq_restore(flags); } static void x2apic_send_IPI_all(int vector) { - x2apic_send_IPI_mask(cpu_online_map, vector); + x2apic_send_IPI_mask(&cpu_online_map, vector); } static int x2apic_apic_id_registered(void) @@ -87,7 +107,7 @@ static int x2apic_apic_id_registered(void) return 1; } -static unsigned int x2apic_cpu_mask_to_apicid(cpumask_t cpumask) +static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) { int cpu; @@ -95,8 +115,8 @@ static unsigned int x2apic_cpu_mask_to_apicid(cpumask_t cpumask) * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - cpu = first_cpu(cpumask); - if ((unsigned)cpu < NR_CPUS) + cpu = first_cpu(*cpumask); + if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_apicid, cpu); else return BAD_APICID; @@ -145,6 +165,7 @@ struct genapic apic_x2apic_phys = { .send_IPI_all = x2apic_send_IPI_all, .send_IPI_allbutself = x2apic_send_IPI_allbutself, .send_IPI_mask = x2apic_send_IPI_mask, + .send_IPI_mask_allbutself = x2apic_send_IPI_mask_allbutself, .send_IPI_self = x2apic_send_IPI_self, .cpu_mask_to_apicid = x2apic_cpu_mask_to_apicid, .phys_pkg_id = phys_pkg_id, diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 2c7dbdb98278..010659415ae4 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -75,16 +75,15 @@ EXPORT_SYMBOL(sn_rtc_cycles_per_second); /* Start with all IRQs pointing to boot CPU. IRQ balancing will shift them. */ -static cpumask_t uv_target_cpus(void) +static const cpumask_t *uv_target_cpus(void) { - return cpumask_of_cpu(0); + return &cpumask_of_cpu(0); } -static cpumask_t uv_vector_allocation_domain(int cpu) +static void uv_vector_allocation_domain(int cpu, cpumask_t *retmask) { - cpumask_t domain = CPU_MASK_NONE; - cpu_set(cpu, domain); - return domain; + cpus_clear(*retmask); + cpu_set(cpu, *retmask); } int uv_wakeup_secondary(int phys_apicid, unsigned int start_rip) @@ -123,28 +122,37 @@ static void uv_send_IPI_one(int cpu, int vector) uv_write_global_mmr64(pnode, UVH_IPI_INT, val); } -static void uv_send_IPI_mask(cpumask_t mask, int vector) +static void uv_send_IPI_mask(const cpumask_t *mask, int vector) { unsigned int cpu; - for_each_possible_cpu(cpu) - if (cpu_isset(cpu, mask)) + for_each_cpu_mask_nr(cpu, *mask) + uv_send_IPI_one(cpu, vector); +} + +static void uv_send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +{ + unsigned int cpu; + unsigned int this_cpu = smp_processor_id(); + + for_each_cpu_mask_nr(cpu, *mask) + if (cpu != this_cpu) uv_send_IPI_one(cpu, vector); } static void uv_send_IPI_allbutself(int vector) { - cpumask_t mask = cpu_online_map; + unsigned int cpu; + unsigned int this_cpu = smp_processor_id(); - cpu_clear(smp_processor_id(), mask); - - if (!cpus_empty(mask)) - uv_send_IPI_mask(mask, vector); + for_each_online_cpu(cpu) + if (cpu != this_cpu) + uv_send_IPI_one(cpu, vector); } static void uv_send_IPI_all(int vector) { - uv_send_IPI_mask(cpu_online_map, vector); + uv_send_IPI_mask(&cpu_online_map, vector); } static int uv_apic_id_registered(void) @@ -156,7 +164,7 @@ static void uv_init_apic_ldr(void) { } -static unsigned int uv_cpu_mask_to_apicid(cpumask_t cpumask) +static unsigned int uv_cpu_mask_to_apicid(const cpumask_t *cpumask) { int cpu; @@ -164,7 +172,7 @@ static unsigned int uv_cpu_mask_to_apicid(cpumask_t cpumask) * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - cpu = first_cpu(cpumask); + cpu = first_cpu(*cpumask); if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_apicid, cpu); else @@ -218,6 +226,7 @@ struct genapic apic_x2apic_uv_x = { .send_IPI_all = uv_send_IPI_all, .send_IPI_allbutself = uv_send_IPI_allbutself, .send_IPI_mask = uv_send_IPI_mask, + .send_IPI_mask_allbutself = uv_send_IPI_mask_allbutself, .send_IPI_self = uv_send_IPI_self, .cpu_mask_to_apicid = uv_cpu_mask_to_apicid, .phys_pkg_id = phys_pkg_id, diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 3d7d0d55253f..7f23ce7f5518 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -231,7 +231,8 @@ static struct irq_cfg *irq_cfg(unsigned int irq) #endif -static inline void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask) +static inline void +set_extra_move_desc(struct irq_desc *desc, const struct cpumask *mask) { } @@ -396,7 +397,8 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, struct irq } } -static int assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask); +static int +assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask); static void set_ioapic_affinity_irq_desc(struct irq_desc *desc, const struct cpumask *mask) @@ -412,13 +414,13 @@ static void set_ioapic_affinity_irq_desc(struct irq_desc *desc, irq = desc->irq; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, *mask)) + if (assign_irq_vector(irq, cfg, mask)) return; - set_extra_move_desc(desc, *mask); + set_extra_move_desc(desc, mask); cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); + dest = cpu_mask_to_apicid(&tmp); /* * Only the high 8 bits are valid. */ @@ -1099,7 +1101,8 @@ void unlock_vector_lock(void) spin_unlock(&vector_lock); } -static int __assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask) +static int +__assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask) { /* * NOTE! The local APIC isn't very good at handling @@ -1115,35 +1118,32 @@ static int __assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask) static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; unsigned int old_vector; int cpu; + cpumask_t tmp_mask; if ((cfg->move_in_progress) || cfg->move_cleanup_count) return -EBUSY; - /* Only try and allocate irqs on cpus that are present */ - cpus_and(mask, mask, cpu_online_map); - old_vector = cfg->vector; if (old_vector) { - cpumask_t tmp; - cpus_and(tmp, cfg->domain, mask); - if (!cpus_empty(tmp)) + cpus_and(tmp_mask, *mask, cpu_online_map); + cpus_and(tmp_mask, cfg->domain, tmp_mask); + if (!cpus_empty(tmp_mask)) return 0; } - for_each_cpu_mask_nr(cpu, mask) { - cpumask_t domain, new_mask; + /* Only try and allocate irqs on cpus that are present */ + for_each_cpu_and(cpu, mask, &cpu_online_map) { int new_cpu; int vector, offset; - domain = vector_allocation_domain(cpu); - cpus_and(new_mask, domain, cpu_online_map); + vector_allocation_domain(cpu, &tmp_mask); vector = current_vector; offset = current_offset; next: vector += 8; if (vector >= first_system_vector) { - /* If we run out of vectors on large boxen, must share them. */ + /* If out of vectors on large boxen, must share them. */ offset = (offset + 1) % 8; vector = FIRST_DEVICE_VECTOR + offset; } @@ -1156,7 +1156,7 @@ next: if (vector == SYSCALL_VECTOR) goto next; #endif - for_each_cpu_mask_nr(new_cpu, new_mask) + for_each_cpu_and(new_cpu, &tmp_mask, &cpu_online_map) if (per_cpu(vector_irq, new_cpu)[vector] != -1) goto next; /* Found one! */ @@ -1166,16 +1166,17 @@ next: cfg->move_in_progress = 1; cfg->old_domain = cfg->domain; } - for_each_cpu_mask_nr(new_cpu, new_mask) + for_each_cpu_and(new_cpu, &tmp_mask, &cpu_online_map) per_cpu(vector_irq, new_cpu)[vector] = irq; cfg->vector = vector; - cfg->domain = domain; + cfg->domain = tmp_mask; return 0; } return -ENOSPC; } -static int assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask) +static int +assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask) { int err; unsigned long flags; @@ -1384,8 +1385,8 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, struct irq_de cfg = desc->chip_data; - mask = TARGET_CPUS; - if (assign_irq_vector(irq, cfg, mask)) + mask = *TARGET_CPUS; + if (assign_irq_vector(irq, cfg, &mask)) return; cpus_and(mask, cfg->domain, mask); @@ -1398,7 +1399,7 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, struct irq_de if (setup_ioapic_entry(mp_ioapics[apic].mp_apicid, irq, &entry, - cpu_mask_to_apicid(mask), trigger, polarity, + cpu_mask_to_apicid(&mask), trigger, polarity, cfg->vector)) { printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", mp_ioapics[apic].mp_apicid, pin); @@ -2121,7 +2122,7 @@ static int ioapic_retrigger_irq(unsigned int irq) unsigned long flags; spin_lock_irqsave(&vector_lock, flags); - send_IPI_mask(cpumask_of_cpu(first_cpu(cfg->domain)), cfg->vector); + send_IPI_mask(&cpumask_of_cpu(first_cpu(cfg->domain)), cfg->vector); spin_unlock_irqrestore(&vector_lock, flags); return 1; @@ -2170,18 +2171,19 @@ static DECLARE_DELAYED_WORK(ir_migration_work, ir_irq_migration); * as simple as edge triggered migration and we can do the irq migration * with a simple atomic update to IO-APIC RTE. */ -static void migrate_ioapic_irq_desc(struct irq_desc *desc, cpumask_t mask) +static void +migrate_ioapic_irq_desc(struct irq_desc *desc, const struct cpumask *mask) { struct irq_cfg *cfg; - cpumask_t tmp, cleanup_mask; + cpumask_t tmpmask; struct irte irte; int modify_ioapic_rte; unsigned int dest; unsigned long flags; unsigned int irq; - cpus_and(tmp, mask, cpu_online_map); - if (cpus_empty(tmp)) + cpus_and(tmpmask, *mask, cpu_online_map); + if (cpus_empty(tmpmask)) return; irq = desc->irq; @@ -2194,8 +2196,8 @@ static void migrate_ioapic_irq_desc(struct irq_desc *desc, cpumask_t mask) set_extra_move_desc(desc, mask); - cpus_and(tmp, cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); + cpus_and(tmpmask, cfg->domain, *mask); + dest = cpu_mask_to_apicid(&tmpmask); modify_ioapic_rte = desc->status & IRQ_LEVEL; if (modify_ioapic_rte) { @@ -2213,13 +2215,13 @@ static void migrate_ioapic_irq_desc(struct irq_desc *desc, cpumask_t mask) modify_irte(irq, &irte); if (cfg->move_in_progress) { - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + cpus_and(tmpmask, cfg->old_domain, cpu_online_map); + cfg->move_cleanup_count = cpus_weight(tmpmask); + send_IPI_mask(&tmpmask, IRQ_MOVE_CLEANUP_VECTOR); cfg->move_in_progress = 0; } - desc->affinity = mask; + desc->affinity = *mask; } static int migrate_irq_remapped_level_desc(struct irq_desc *desc) @@ -2241,7 +2243,7 @@ static int migrate_irq_remapped_level_desc(struct irq_desc *desc) } /* everthing is clear. we have right of way */ - migrate_ioapic_irq_desc(desc, desc->pending_mask); + migrate_ioapic_irq_desc(desc, &desc->pending_mask); ret = 0; desc->status &= ~IRQ_MOVE_PENDING; @@ -2292,7 +2294,7 @@ static void set_ir_ioapic_affinity_irq_desc(struct irq_desc *desc, return; } - migrate_ioapic_irq_desc(desc, *mask); + migrate_ioapic_irq_desc(desc, mask); } static void set_ir_ioapic_affinity_irq(unsigned int irq, const struct cpumask *mask) @@ -2359,7 +2361,7 @@ static void irq_complete_move(struct irq_desc **descp) cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + send_IPI_mask(&cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); cfg->move_in_progress = 0; } } @@ -3089,13 +3091,13 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms cpumask_t tmp; cfg = irq_cfg(irq); - tmp = TARGET_CPUS; - err = assign_irq_vector(irq, cfg, tmp); + tmp = *TARGET_CPUS; + err = assign_irq_vector(irq, cfg, &tmp); if (err) return err; cpus_and(tmp, cfg->domain, tmp); - dest = cpu_mask_to_apicid(tmp); + dest = cpu_mask_to_apicid(&tmp); #ifdef CONFIG_INTR_REMAP if (irq_remapped(irq)) { @@ -3161,13 +3163,13 @@ static void set_msi_irq_affinity(unsigned int irq, const struct cpumask *mask) return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, *mask)) + if (assign_irq_vector(irq, cfg, mask)) return; - set_extra_move_desc(desc, *mask); + set_extra_move_desc(desc, mask); cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); + dest = cpu_mask_to_apicid(&tmp); read_msi_msg_desc(desc, &msg); @@ -3184,8 +3186,8 @@ static void set_msi_irq_affinity(unsigned int irq, const struct cpumask *mask) * Migrate the MSI irq to another cpumask. This migration is * done in the process context using interrupt-remapping hardware. */ -static void ir_set_msi_irq_affinity(unsigned int irq, - const struct cpumask *mask) +static void +ir_set_msi_irq_affinity(unsigned int irq, const struct cpumask *mask) { struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; @@ -3200,13 +3202,13 @@ static void ir_set_msi_irq_affinity(unsigned int irq, return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, *mask)) + if (assign_irq_vector(irq, cfg, mask)) return; - set_extra_move_desc(desc, *mask); + set_extra_move_desc(desc, mask); cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); + dest = cpu_mask_to_apicid(&tmp); irte.vector = cfg->vector; irte.dest_id = IRTE_DEST(dest); @@ -3224,7 +3226,7 @@ static void ir_set_msi_irq_affinity(unsigned int irq, if (cfg->move_in_progress) { cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + send_IPI_mask(&cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); cfg->move_in_progress = 0; } @@ -3419,7 +3421,7 @@ void arch_teardown_msi_irq(unsigned int irq) #ifdef CONFIG_DMAR #ifdef CONFIG_SMP -static void dmar_msi_set_affinity(unsigned int irq, const struct cpumask *mask) +static void dmar_msi_set_affinity(unsigned int irq, const cpumask_t *mask) { struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; @@ -3431,13 +3433,13 @@ static void dmar_msi_set_affinity(unsigned int irq, const struct cpumask *mask) return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, *mask)) + if (assign_irq_vector(irq, cfg, mask)) return; - set_extra_move_desc(desc, *mask); + set_extra_move_desc(desc, mask); cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); + dest = cpu_mask_to_apicid(&tmp); dmar_msi_read(irq, &msg); @@ -3481,7 +3483,7 @@ int arch_setup_dmar_msi(unsigned int irq) #ifdef CONFIG_HPET_TIMER #ifdef CONFIG_SMP -static void hpet_msi_set_affinity(unsigned int irq, const struct cpumask *mask) +static void hpet_msi_set_affinity(unsigned int irq, const cpumask_t *mask) { struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; @@ -3493,13 +3495,13 @@ static void hpet_msi_set_affinity(unsigned int irq, const struct cpumask *mask) return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, *mask)) + if (assign_irq_vector(irq, cfg, mask)) return; - set_extra_move_desc(desc, *mask); + set_extra_move_desc(desc, mask); cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); + dest = cpu_mask_to_apicid(&tmp); hpet_msi_read(irq, &msg); @@ -3564,7 +3566,7 @@ static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) write_ht_irq_msg(irq, &msg); } -static void set_ht_irq_affinity(unsigned int irq, const struct cpumask *mask) +static void set_ht_irq_affinity(unsigned int irq, const cpumask_t *mask) { struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; @@ -3575,13 +3577,13 @@ static void set_ht_irq_affinity(unsigned int irq, const struct cpumask *mask) return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, *mask)) + if (assign_irq_vector(irq, cfg, mask)) return; - set_extra_move_desc(desc, *mask); + set_extra_move_desc(desc, mask); cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(tmp); + dest = cpu_mask_to_apicid(&tmp); target_ht_irq(irq, dest, cfg->vector); cpumask_copy(&desc->affinity, mask); @@ -3607,14 +3609,13 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) cpumask_t tmp; cfg = irq_cfg(irq); - tmp = TARGET_CPUS; - err = assign_irq_vector(irq, cfg, tmp); + err = assign_irq_vector(irq, cfg, TARGET_CPUS); if (!err) { struct ht_irq_msg msg; unsigned dest; cpus_and(tmp, cfg->domain, tmp); - dest = cpu_mask_to_apicid(tmp); + dest = cpu_mask_to_apicid(&tmp); msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest); @@ -3650,7 +3651,7 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, unsigned long mmr_offset) { - const cpumask_t *eligible_cpu = get_cpu_mask(cpu); + const cpumask_t *eligible_cpu = &cpumask_of_cpu(cpu); struct irq_cfg *cfg; int mmr_pnode; unsigned long mmr_value; @@ -3660,7 +3661,7 @@ int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, cfg = irq_cfg(irq); - err = assign_irq_vector(irq, cfg, *eligible_cpu); + err = assign_irq_vector(irq, cfg, eligible_cpu); if (err != 0) return err; @@ -3679,7 +3680,7 @@ int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, entry->polarity = 0; entry->trigger = 0; entry->mask = 0; - entry->dest = cpu_mask_to_apicid(*eligible_cpu); + entry->dest = cpu_mask_to_apicid(eligible_cpu); mmr_pnode = uv_blade_to_pnode(mmr_blade); uv_write_global_mmr64(mmr_pnode, mmr_offset, mmr_value); @@ -3890,7 +3891,7 @@ void __init setup_ioapic_dest(void) int pin, ioapic, irq, irq_entry; struct irq_desc *desc; struct irq_cfg *cfg; - cpumask_t mask; + const cpumask_t *mask; if (skip_ioapic_setup == 1) return; @@ -3921,16 +3922,16 @@ void __init setup_ioapic_dest(void) */ if (desc->status & (IRQ_NO_BALANCING | IRQ_AFFINITY_SET)) - mask = desc->affinity; + mask = &desc->affinity; else mask = TARGET_CPUS; #ifdef CONFIG_INTR_REMAP if (intr_remapping_enabled) - set_ir_ioapic_affinity_irq_desc(desc, &mask); + set_ir_ioapic_affinity_irq_desc(desc, mask); else #endif - set_ioapic_affinity_irq_desc(desc, &mask); + set_ioapic_affinity_irq_desc(desc, mask); } } diff --git a/arch/x86/kernel/ipi.c b/arch/x86/kernel/ipi.c index f1c688e46f35..86aa50fc65a1 100644 --- a/arch/x86/kernel/ipi.c +++ b/arch/x86/kernel/ipi.c @@ -116,9 +116,9 @@ static inline void __send_IPI_dest_field(unsigned long mask, int vector) /* * This is only used on smaller machines. */ -void send_IPI_mask_bitmask(cpumask_t cpumask, int vector) +void send_IPI_mask_bitmask(const cpumask_t *cpumask, int vector) { - unsigned long mask = cpus_addr(cpumask)[0]; + unsigned long mask = cpus_addr(*cpumask)[0]; unsigned long flags; local_irq_save(flags); @@ -127,7 +127,7 @@ void send_IPI_mask_bitmask(cpumask_t cpumask, int vector) local_irq_restore(flags); } -void send_IPI_mask_sequence(cpumask_t mask, int vector) +void send_IPI_mask_sequence(const cpumask_t *mask, int vector) { unsigned long flags; unsigned int query_cpu; @@ -139,12 +139,24 @@ void send_IPI_mask_sequence(cpumask_t mask, int vector) */ local_irq_save(flags); - for_each_possible_cpu(query_cpu) { - if (cpu_isset(query_cpu, mask)) { + for_each_cpu_mask_nr(query_cpu, *mask) + __send_IPI_dest_field(cpu_to_logical_apicid(query_cpu), vector); + local_irq_restore(flags); +} + +void send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +{ + unsigned long flags; + unsigned int query_cpu; + unsigned int this_cpu = smp_processor_id(); + + /* See Hack comment above */ + + local_irq_save(flags); + for_each_cpu_mask_nr(query_cpu, *mask) + if (query_cpu != this_cpu) __send_IPI_dest_field(cpu_to_logical_apicid(query_cpu), vector); - } - } local_irq_restore(flags); } diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c index 3f92b134ab90..341df946f9a9 100644 --- a/arch/x86/kernel/smp.c +++ b/arch/x86/kernel/smp.c @@ -118,22 +118,22 @@ static void native_smp_send_reschedule(int cpu) WARN_ON(1); return; } - send_IPI_mask(cpumask_of_cpu(cpu), RESCHEDULE_VECTOR); + send_IPI_mask(&cpumask_of_cpu(cpu), RESCHEDULE_VECTOR); } void native_send_call_func_single_ipi(int cpu) { - send_IPI_mask(cpumask_of_cpu(cpu), CALL_FUNCTION_SINGLE_VECTOR); + send_IPI_mask(&cpumask_of_cpu(cpu), CALL_FUNCTION_SINGLE_VECTOR); } -void native_send_call_func_ipi(cpumask_t mask) +void native_send_call_func_ipi(const cpumask_t *mask) { cpumask_t allbutself; allbutself = cpu_online_map; cpu_clear(smp_processor_id(), allbutself); - if (cpus_equal(mask, allbutself) && + if (cpus_equal(*mask, allbutself) && cpus_equal(cpu_online_map, cpu_callout_map)) send_IPI_allbutself(CALL_FUNCTION_VECTOR); else diff --git a/arch/x86/kernel/tlb_32.c b/arch/x86/kernel/tlb_32.c index f4049f3513b6..174ea90d1cbd 100644 --- a/arch/x86/kernel/tlb_32.c +++ b/arch/x86/kernel/tlb_32.c @@ -164,7 +164,7 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, * We have to send the IPI only to * CPUs affected. */ - send_IPI_mask(cpumask, INVALIDATE_TLB_VECTOR); + send_IPI_mask(&cpumask, INVALIDATE_TLB_VECTOR); while (!cpus_empty(flush_cpumask)) /* nothing. lockup detection does not belong here */ diff --git a/arch/x86/kernel/tlb_64.c b/arch/x86/kernel/tlb_64.c index 8f919ca69494..de6f1bda0c50 100644 --- a/arch/x86/kernel/tlb_64.c +++ b/arch/x86/kernel/tlb_64.c @@ -191,7 +191,7 @@ void native_flush_tlb_others(const cpumask_t *cpumaskp, struct mm_struct *mm, * We have to send the IPI only to * CPUs affected. */ - send_IPI_mask(cpumask, INVALIDATE_TLB_VECTOR_START + sender); + send_IPI_mask(&cpumask, INVALIDATE_TLB_VECTOR_START + sender); while (!cpus_empty(f->flush_cpumask)) cpu_relax(); diff --git a/arch/x86/mach-generic/bigsmp.c b/arch/x86/mach-generic/bigsmp.c index 3624a364b7f3..bc4c7840b2a8 100644 --- a/arch/x86/mach-generic/bigsmp.c +++ b/arch/x86/mach-generic/bigsmp.c @@ -42,9 +42,10 @@ static const struct dmi_system_id bigsmp_dmi_table[] = { { } }; -static cpumask_t vector_allocation_domain(int cpu) +static void vector_allocation_domain(int cpu, cpumask_t *retmask) { - return cpumask_of_cpu(cpu); + cpus_clear(*retmask); + cpu_set(cpu, *retmask); } static int probe_bigsmp(void) diff --git a/arch/x86/mach-generic/es7000.c b/arch/x86/mach-generic/es7000.c index 7b4e6d0d1690..4ba5ccaa1584 100644 --- a/arch/x86/mach-generic/es7000.c +++ b/arch/x86/mach-generic/es7000.c @@ -87,7 +87,7 @@ static int __init acpi_madt_oem_check(char *oem_id, char *oem_table_id) } #endif -static cpumask_t vector_allocation_domain(int cpu) +static void vector_allocation_domain(int cpu, cpumask_t *retmask) { /* Careful. Some cpus do not strictly honor the set of cpus * specified in the interrupt destination when using lowest @@ -97,8 +97,7 @@ static cpumask_t vector_allocation_domain(int cpu) * deliver interrupts to the wrong hyperthread when only one * hyperthread was specified in the interrupt desitination. */ - cpumask_t domain = { { [0] = APIC_ALL_CPUS, } }; - return domain; + *retmask = (cpumask_t){ { [0] = APIC_ALL_CPUS, } }; } struct genapic __initdata_refok apic_es7000 = APIC_INIT("es7000", probe_es7000); diff --git a/arch/x86/mach-generic/numaq.c b/arch/x86/mach-generic/numaq.c index 71a309b122e6..511d7941364f 100644 --- a/arch/x86/mach-generic/numaq.c +++ b/arch/x86/mach-generic/numaq.c @@ -38,7 +38,7 @@ static int acpi_madt_oem_check(char *oem_id, char *oem_table_id) return 0; } -static cpumask_t vector_allocation_domain(int cpu) +static void vector_allocation_domain(int cpu, cpumask_t *retmask) { /* Careful. Some cpus do not strictly honor the set of cpus * specified in the interrupt destination when using lowest @@ -48,8 +48,7 @@ static cpumask_t vector_allocation_domain(int cpu) * deliver interrupts to the wrong hyperthread when only one * hyperthread was specified in the interrupt desitination. */ - cpumask_t domain = { { [0] = APIC_ALL_CPUS, } }; - return domain; + *retmask = (cpumask_t){ { [0] = APIC_ALL_CPUS, } }; } struct genapic apic_numaq = APIC_INIT("NUMAQ", probe_numaq); diff --git a/arch/x86/mach-generic/summit.c b/arch/x86/mach-generic/summit.c index 2c6d234e0009..2821ffc188b5 100644 --- a/arch/x86/mach-generic/summit.c +++ b/arch/x86/mach-generic/summit.c @@ -24,7 +24,7 @@ static int probe_summit(void) return 0; } -static cpumask_t vector_allocation_domain(int cpu) +static void vector_allocation_domain(int cpu, cpumask_t *retmask) { /* Careful. Some cpus do not strictly honor the set of cpus * specified in the interrupt destination when using lowest @@ -34,8 +34,7 @@ static cpumask_t vector_allocation_domain(int cpu) * deliver interrupts to the wrong hyperthread when only one * hyperthread was specified in the interrupt desitination. */ - cpumask_t domain = { { [0] = APIC_ALL_CPUS, } }; - return domain; + *retmask = (cpumask_t){ { [0] = APIC_ALL_CPUS, } }; } struct genapic apic_summit = APIC_INIT("summit", probe_summit); diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c index acd9b6705e02..2cce362c9874 100644 --- a/arch/x86/xen/smp.c +++ b/arch/x86/xen/smp.c @@ -158,7 +158,7 @@ static void __init xen_fill_possible_map(void) { int i, rc; - for (i = 0; i < NR_CPUS; i++) { + for (i = 0; i < nr_cpu_ids; i++) { rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL); if (rc >= 0) { num_processors++; @@ -196,7 +196,7 @@ static void __init xen_smp_prepare_cpus(unsigned int max_cpus) /* Restrict the possible_map according to max_cpus. */ while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) { - for (cpu = NR_CPUS - 1; !cpu_possible(cpu); cpu--) + for (cpu = nr_cpu_ids - 1; !cpu_possible(cpu); cpu--) continue; cpu_clear(cpu, cpu_possible_map); } @@ -408,24 +408,22 @@ static void xen_smp_send_reschedule(int cpu) xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR); } -static void xen_send_IPI_mask(cpumask_t mask, enum ipi_vector vector) +static void xen_send_IPI_mask(const cpumask_t *mask, enum ipi_vector vector) { unsigned cpu; - cpus_and(mask, mask, cpu_online_map); - - for_each_cpu_mask_nr(cpu, mask) + for_each_cpu_and(cpu, mask, &cpu_online_map) xen_send_IPI_one(cpu, vector); } -static void xen_smp_send_call_function_ipi(cpumask_t mask) +static void xen_smp_send_call_function_ipi(const cpumask_t *mask) { int cpu; xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR); /* Make sure other vcpus get a chance to run if they need to. */ - for_each_cpu_mask_nr(cpu, mask) { + for_each_cpu_mask_nr(cpu, *mask) { if (xen_vcpu_stolen(cpu)) { HYPERVISOR_sched_op(SCHEDOP_yield, 0); break; @@ -435,7 +433,8 @@ static void xen_smp_send_call_function_ipi(cpumask_t mask) static void xen_smp_send_call_function_single_ipi(int cpu) { - xen_send_IPI_mask(cpumask_of_cpu(cpu), XEN_CALL_FUNCTION_SINGLE_VECTOR); + xen_send_IPI_mask(&cpumask_of_cpu(cpu), + XEN_CALL_FUNCTION_SINGLE_VECTOR); } static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id) From a1681965011916c2f1f0f1f87e70784f5d5d5be5 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:53 -0800 Subject: [PATCH 03/16] x86: move and enhance debug printk for nr_cpu_ids etc. Impact: cleanup, better debugging This has proven useful in debugging, *before* we try to use for_each_possible_cpu(). It also now shows nr_cpumask_bits. Signed-off-by: Mike Travis Signed-off-by: Rusty Russell --- arch/x86/kernel/setup_percpu.c | 17 +++++++++++------ 1 file changed, 11 insertions(+), 6 deletions(-) diff --git a/arch/x86/kernel/setup_percpu.c b/arch/x86/kernel/setup_percpu.c index 1c2084291f97..0b63b08e7530 100644 --- a/arch/x86/kernel/setup_percpu.c +++ b/arch/x86/kernel/setup_percpu.c @@ -152,6 +152,11 @@ void __init setup_per_cpu_areas(void) old_size = PERCPU_ENOUGH_ROOM; 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", + NR_CPUS, nr_cpumask_bits, nr_cpu_ids, nr_node_ids); + printk(KERN_INFO "PERCPU: Allocating %zd bytes of per cpu data\n", size); @@ -168,24 +173,24 @@ void __init setup_per_cpu_areas(void) "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", + printk(KERN_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)); + printk(KERN_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; memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); } - printk(KERN_DEBUG "NR_CPUS: %d, nr_cpu_ids: %d, nr_node_ids %d\n", - NR_CPUS, nr_cpu_ids, nr_node_ids); - /* Setup percpu data maps */ setup_per_cpu_maps(); From 95d313cf1c1ecedc8bec5727b09bdacbf67dfc45 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:54 -0800 Subject: [PATCH 04/16] x86: Add cpu_mask_to_apicid_and Impact: new API Add a helper function that takes two cpumask's, and's them and then returns the apicid of the result. This removes a need in io_apic.c that uses a temporary cpumask to hold (mask & cfg->domain). Signed-off-by: Mike Travis Signed-off-by: Rusty Russell --- arch/x86/include/asm/bigsmp/apic.h | 16 +++++++ arch/x86/include/asm/es7000/apic.h | 47 +++++++++++++++++++ arch/x86/include/asm/genapic_32.h | 3 ++ arch/x86/include/asm/genapic_64.h | 2 + arch/x86/include/asm/mach-default/mach_apic.h | 10 ++++ arch/x86/include/asm/mach-generic/mach_apic.h | 1 + arch/x86/include/asm/numaq/apic.h | 6 +++ arch/x86/include/asm/summit/apic.h | 39 +++++++++++++++ arch/x86/kernel/genapic_flat_64.c | 26 ++++++++++ arch/x86/kernel/genx2apic_cluster.c | 16 +++++++ arch/x86/kernel/genx2apic_phys.c | 16 +++++++ arch/x86/kernel/genx2apic_uv_x.c | 16 +++++++ 12 files changed, 198 insertions(+) diff --git a/arch/x86/include/asm/bigsmp/apic.h b/arch/x86/include/asm/bigsmp/apic.h index dc6225ca48ad..99f9abacf6a2 100644 --- a/arch/x86/include/asm/bigsmp/apic.h +++ b/arch/x86/include/asm/bigsmp/apic.h @@ -129,6 +129,22 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return apicid; } +static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + int cpu; + + /* + * We're using fixed IRQ delivery, can only return one phys APIC ID. + * May as well be the first. + */ + while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) + if (cpu_isset(cpu, *andmask)) + return cpu_to_logical_apicid(cpu); + + return BAD_APICID; +} + static inline u32 phys_pkg_id(u32 cpuid_apic, int index_msb) { return cpuid_apic >> index_msb; diff --git a/arch/x86/include/asm/es7000/apic.h b/arch/x86/include/asm/es7000/apic.h index 4cac0837bb40..c2bed772af88 100644 --- a/arch/x86/include/asm/es7000/apic.h +++ b/arch/x86/include/asm/es7000/apic.h @@ -214,6 +214,53 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return apicid; } +static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + int num_bits_set; + int num_bits_set2; + int cpus_found = 0; + int cpu; + int apicid = 0; + + num_bits_set = cpus_weight(*cpumask); + num_bits_set2 = cpus_weight(*andmask); + num_bits_set = min_t(int, num_bits_set, num_bits_set2); + /* Return id to all */ + if (num_bits_set >= nr_cpu_ids) +#if defined CONFIG_ES7000_CLUSTERED_APIC + return 0xFF; +#else + return cpu_to_logical_apicid(0); +#endif + /* + * 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. + */ + while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) + if (cpu_isset(cpu, *andmask) + apicid = cpu_to_logical_apicid(cpu); + while (cpus_found < num_bits_set) { + if (cpu_isset(cpu, *cpumask) && cpu_isset(cpu, *andmask)) { + int new_apicid = cpu_to_logical_apicid(cpu); + if (apicid_cluster(apicid) != + apicid_cluster(new_apicid)) { + printk(KERN_WARNING + "%s: Not a valid mask!\n", __func__); +#if defined CONFIG_ES7000_CLUSTERED_APIC + return 0xFF; +#else + return cpu_to_logical_apicid(0); +#endif + } + apicid = new_apicid; + cpus_found++; + } + cpu++; + } + return apicid; +} + static inline u32 phys_pkg_id(u32 cpuid_apic, int index_msb) { return cpuid_apic >> index_msb; diff --git a/arch/x86/include/asm/genapic_32.h b/arch/x86/include/asm/genapic_32.h index b21ed21c574d..325298a82237 100644 --- a/arch/x86/include/asm/genapic_32.h +++ b/arch/x86/include/asm/genapic_32.h @@ -58,6 +58,8 @@ struct genapic { unsigned (*get_apic_id)(unsigned long x); unsigned long apic_id_mask; unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask); + unsigned int (*cpu_mask_to_apicid_and)(const cpumask_t *cpumask, + const cpumask_t *andmask); void (*vector_allocation_domain)(int cpu, cpumask_t *retmask); #ifdef CONFIG_SMP @@ -115,6 +117,7 @@ struct genapic { APICFUNC(get_apic_id) \ .apic_id_mask = APIC_ID_MASK, \ APICFUNC(cpu_mask_to_apicid) \ + APICFUNC(cpu_mask_to_apicid_and) \ APICFUNC(vector_allocation_domain) \ APICFUNC(acpi_madt_oem_check) \ IPIFUNC(send_IPI_mask) \ diff --git a/arch/x86/include/asm/genapic_64.h b/arch/x86/include/asm/genapic_64.h index a020e7d35a40..301c7f41125d 100644 --- a/arch/x86/include/asm/genapic_64.h +++ b/arch/x86/include/asm/genapic_64.h @@ -31,6 +31,8 @@ struct genapic { void (*send_IPI_self)(int vector); /* */ unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask); + unsigned int (*cpu_mask_to_apicid_and)(const cpumask_t *cpumask, + const cpumask_t *andmask); unsigned int (*phys_pkg_id)(int index_msb); unsigned int (*get_apic_id)(unsigned long x); unsigned long (*set_apic_id)(unsigned int id); diff --git a/arch/x86/include/asm/mach-default/mach_apic.h b/arch/x86/include/asm/mach-default/mach_apic.h index c18896b0508c..229b605d104e 100644 --- a/arch/x86/include/asm/mach-default/mach_apic.h +++ b/arch/x86/include/asm/mach-default/mach_apic.h @@ -28,6 +28,7 @@ static inline const cpumask_t *target_cpus(void) #define apic_id_registered (genapic->apic_id_registered) #define init_apic_ldr (genapic->init_apic_ldr) #define cpu_mask_to_apicid (genapic->cpu_mask_to_apicid) +#define cpu_mask_to_apicid_and (genapic->cpu_mask_to_apicid_and) #define phys_pkg_id (genapic->phys_pkg_id) #define vector_allocation_domain (genapic->vector_allocation_domain) #define read_apic_id() (GET_APIC_ID(apic_read(APIC_ID))) @@ -66,6 +67,15 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return cpus_addr(*cpumask)[0]; } +static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + unsigned long mask1 = cpus_addr(*cpumask)[0]; + unsigned long mask2 = cpus_addr(*andmask)[0]; + + return (unsigned int)(mask1 & mask2); +} + static inline u32 phys_pkg_id(u32 cpuid_apic, int index_msb) { return cpuid_apic >> index_msb; diff --git a/arch/x86/include/asm/mach-generic/mach_apic.h b/arch/x86/include/asm/mach-generic/mach_apic.h index e430f47df667..48553e958ad5 100644 --- a/arch/x86/include/asm/mach-generic/mach_apic.h +++ b/arch/x86/include/asm/mach-generic/mach_apic.h @@ -24,6 +24,7 @@ #define check_phys_apicid_present (genapic->check_phys_apicid_present) #define check_apicid_used (genapic->check_apicid_used) #define cpu_mask_to_apicid (genapic->cpu_mask_to_apicid) +#define cpu_mask_to_apicid_and (genapic->cpu_mask_to_apicid_and) #define vector_allocation_domain (genapic->vector_allocation_domain) #define enable_apic_mode (genapic->enable_apic_mode) #define phys_pkg_id (genapic->phys_pkg_id) diff --git a/arch/x86/include/asm/numaq/apic.h b/arch/x86/include/asm/numaq/apic.h index 1df7ebe738e5..abf668ced503 100644 --- a/arch/x86/include/asm/numaq/apic.h +++ b/arch/x86/include/asm/numaq/apic.h @@ -127,6 +127,12 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return (int) 0xF; } +static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + return (int) 0xF; +} + /* No NUMA-Q box has a HT CPU, but it can't hurt to use the default code. */ static inline u32 phys_pkg_id(u32 cpuid_apic, int index_msb) { diff --git a/arch/x86/include/asm/summit/apic.h b/arch/x86/include/asm/summit/apic.h index 437dc83725ca..cbcc2c7eb1d5 100644 --- a/arch/x86/include/asm/summit/apic.h +++ b/arch/x86/include/asm/summit/apic.h @@ -170,6 +170,45 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return apicid; } +static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + int num_bits_set; + int num_bits_set2; + int cpus_found = 0; + int cpu; + int apicid = 0; + + num_bits_set = cpus_weight(*cpumask); + num_bits_set2 = cpus_weight(*andmask); + num_bits_set = min_t(int, num_bits_set, num_bits_set2); + /* Return id to all */ + if (num_bits_set >= nr_cpu_ids) + return 0xFF; + /* + * 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. + */ + while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) + if (cpu_isset(cpu, *andmask) + apicid = cpu_to_logical_apicid(cpu); + while (cpus_found < num_bits_set) { + if (cpu_isset(cpu, *cpumask) && cpu_isset(cpu, *andmask)) { + int new_apicid = cpu_to_logical_apicid(cpu); + if (apicid_cluster(apicid) != + apicid_cluster(new_apicid)) { + printk(KERN_WARNING + "%s: Not a valid mask!\n", __func__); + return 0xFF; + } + apicid = apicid | new_apicid; + cpus_found++; + } + cpu++; + } + return apicid; +} + /* cpuid returns the value latched in the HW at reset, not the APIC ID * register's value. For any box whose BIOS changes APIC IDs, like * clustered APIC systems, we must use hard_smp_processor_id. diff --git a/arch/x86/kernel/genapic_flat_64.c b/arch/x86/kernel/genapic_flat_64.c index 50eebd0328fe..1efecd206a74 100644 --- a/arch/x86/kernel/genapic_flat_64.c +++ b/arch/x86/kernel/genapic_flat_64.c @@ -158,6 +158,15 @@ static unsigned int flat_cpu_mask_to_apicid(const cpumask_t *cpumask) return cpus_addr(*cpumask)[0] & APIC_ALL_CPUS; } +static unsigned int flat_cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + unsigned long mask1 = cpus_addr(*cpumask)[0] & APIC_ALL_CPUS; + unsigned long mask2 = cpus_addr(*andmask)[0] & APIC_ALL_CPUS; + + return (int)(mask1 & mask2); +} + static unsigned int phys_pkg_id(int index_msb) { return hard_smp_processor_id() >> index_msb; @@ -178,6 +187,7 @@ struct genapic apic_flat = { .send_IPI_mask_allbutself = flat_send_IPI_mask_allbutself, .send_IPI_self = apic_send_IPI_self, .cpu_mask_to_apicid = flat_cpu_mask_to_apicid, + .cpu_mask_to_apicid_and = flat_cpu_mask_to_apicid_and, .phys_pkg_id = phys_pkg_id, .get_apic_id = get_apic_id, .set_apic_id = set_apic_id, @@ -254,6 +264,21 @@ static unsigned int physflat_cpu_mask_to_apicid(const cpumask_t *cpumask) return BAD_APICID; } +static unsigned int physflat_cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + int cpu; + + /* + * We're using fixed IRQ delivery, can only return one phys APIC ID. + * May as well be the first. + */ + while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) + if (cpu_isset(cpu, *andmask)) + return per_cpu(x86_cpu_to_apicid, cpu); + return BAD_APICID; +} + struct genapic apic_physflat = { .name = "physical flat", .acpi_madt_oem_check = physflat_acpi_madt_oem_check, @@ -269,6 +294,7 @@ struct genapic apic_physflat = { .send_IPI_mask_allbutself = physflat_send_IPI_mask_allbutself, .send_IPI_self = apic_send_IPI_self, .cpu_mask_to_apicid = physflat_cpu_mask_to_apicid, + .cpu_mask_to_apicid_and = physflat_cpu_mask_to_apicid_and, .phys_pkg_id = phys_pkg_id, .get_apic_id = get_apic_id, .set_apic_id = set_apic_id, diff --git a/arch/x86/kernel/genx2apic_cluster.c b/arch/x86/kernel/genx2apic_cluster.c index f5fa9a91ad38..fd8047f4e455 100644 --- a/arch/x86/kernel/genx2apic_cluster.c +++ b/arch/x86/kernel/genx2apic_cluster.c @@ -123,6 +123,21 @@ static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) return BAD_APICID; } +static unsigned int x2apic_cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + int cpu; + + /* + * We're using fixed IRQ delivery, can only return one phys APIC ID. + * May as well be the first. + */ + while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) + if (cpu_isset(cpu, *andmask)) + return per_cpu(x86_cpu_to_apicid, cpu); + return BAD_APICID; +} + static unsigned int get_apic_id(unsigned long x) { unsigned int id; @@ -172,6 +187,7 @@ struct genapic apic_x2apic_cluster = { .send_IPI_mask_allbutself = x2apic_send_IPI_mask_allbutself, .send_IPI_self = x2apic_send_IPI_self, .cpu_mask_to_apicid = x2apic_cpu_mask_to_apicid, + .cpu_mask_to_apicid_and = x2apic_cpu_mask_to_apicid_and, .phys_pkg_id = phys_pkg_id, .get_apic_id = get_apic_id, .set_apic_id = set_apic_id, diff --git a/arch/x86/kernel/genx2apic_phys.c b/arch/x86/kernel/genx2apic_phys.c index 41c27b2f3d01..d5578bb8f165 100644 --- a/arch/x86/kernel/genx2apic_phys.c +++ b/arch/x86/kernel/genx2apic_phys.c @@ -122,6 +122,21 @@ static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) return BAD_APICID; } +static unsigned int x2apic_cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + int cpu; + + /* + * We're using fixed IRQ delivery, can only return one phys APIC ID. + * May as well be the first. + */ + while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) + if (cpu_isset(cpu, *andmask)) + return per_cpu(x86_cpu_to_apicid, cpu); + return BAD_APICID; +} + static unsigned int get_apic_id(unsigned long x) { unsigned int id; @@ -168,6 +183,7 @@ struct genapic apic_x2apic_phys = { .send_IPI_mask_allbutself = x2apic_send_IPI_mask_allbutself, .send_IPI_self = x2apic_send_IPI_self, .cpu_mask_to_apicid = x2apic_cpu_mask_to_apicid, + .cpu_mask_to_apicid_and = x2apic_cpu_mask_to_apicid_and, .phys_pkg_id = phys_pkg_id, .get_apic_id = get_apic_id, .set_apic_id = set_apic_id, diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 010659415ae4..53bd2570272c 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -179,6 +179,21 @@ static unsigned int uv_cpu_mask_to_apicid(const cpumask_t *cpumask) return BAD_APICID; } +static unsigned int uv_cpu_mask_to_apicid_and(const cpumask_t *cpumask, + const cpumask_t *andmask) +{ + int cpu; + + /* + * We're using fixed IRQ delivery, can only return one phys APIC ID. + * May as well be the first. + */ + while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) + if (cpu_isset(cpu, *andmask)) + return per_cpu(x86_cpu_to_apicid, cpu); + return BAD_APICID; +} + static unsigned int get_apic_id(unsigned long x) { unsigned int id; @@ -229,6 +244,7 @@ struct genapic apic_x2apic_uv_x = { .send_IPI_mask_allbutself = uv_send_IPI_mask_allbutself, .send_IPI_self = uv_send_IPI_self, .cpu_mask_to_apicid = uv_cpu_mask_to_apicid, + .cpu_mask_to_apicid_and = uv_cpu_mask_to_apicid_and, .phys_pkg_id = phys_pkg_id, .get_apic_id = get_apic_id, .set_apic_id = set_apic_id, From 6eeb7c5a99434596c5953a95baa17d2f085664e3 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:55 -0800 Subject: [PATCH 05/16] x86: update add-cpu_mask_to_apicid_and to use struct cpumask* Impact: use updated APIs Various API updates for x86:add-cpu_mask_to_apicid_and (Note: separate because previous patch has been "backported" to 2.6.27.) Signed-off-by: Rusty Russell Signed-off-by: Mike Travis --- arch/x86/include/asm/bigsmp/apic.h | 10 ++++----- arch/x86/include/asm/es7000/apic.h | 19 +++++++++-------- arch/x86/include/asm/genapic_32.h | 4 ++-- arch/x86/include/asm/genapic_64.h | 4 ++-- arch/x86/include/asm/mach-default/mach_apic.h | 8 +++---- arch/x86/include/asm/numaq/apic.h | 4 ++-- arch/x86/include/asm/summit/apic.h | 18 ++++++++-------- arch/x86/kernel/genapic_flat_64.c | 21 ++++++++++--------- arch/x86/kernel/genx2apic_cluster.c | 10 ++++----- arch/x86/kernel/genx2apic_phys.c | 10 ++++----- arch/x86/kernel/genx2apic_uv_x.c | 10 ++++----- 11 files changed, 60 insertions(+), 58 deletions(-) diff --git a/arch/x86/include/asm/bigsmp/apic.h b/arch/x86/include/asm/bigsmp/apic.h index 99f9abacf6a2..976399debb3f 100644 --- a/arch/x86/include/asm/bigsmp/apic.h +++ b/arch/x86/include/asm/bigsmp/apic.h @@ -129,8 +129,8 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return apicid; } -static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { int cpu; @@ -138,9 +138,9 @@ static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) - if (cpu_isset(cpu, *andmask)) - return cpu_to_logical_apicid(cpu); + cpu = cpumask_any_and(cpumask, andmask); + if (cpu < nr_cpu_ids) + return cpu_to_logical_apicid(cpu); return BAD_APICID; } diff --git a/arch/x86/include/asm/es7000/apic.h b/arch/x86/include/asm/es7000/apic.h index c2bed772af88..ba8423c5363f 100644 --- a/arch/x86/include/asm/es7000/apic.h +++ b/arch/x86/include/asm/es7000/apic.h @@ -214,8 +214,8 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return apicid; } -static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { int num_bits_set; int num_bits_set2; @@ -223,9 +223,9 @@ static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, int cpu; int apicid = 0; - num_bits_set = cpus_weight(*cpumask); - num_bits_set2 = cpus_weight(*andmask); - num_bits_set = min_t(int, num_bits_set, num_bits_set2); + num_bits_set = cpumask_weight(cpumask); + num_bits_set2 = cpumask_weight(andmask); + num_bits_set = min(num_bits_set, num_bits_set2); /* Return id to all */ if (num_bits_set >= nr_cpu_ids) #if defined CONFIG_ES7000_CLUSTERED_APIC @@ -237,11 +237,12 @@ static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, * 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. */ - while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) - if (cpu_isset(cpu, *andmask) - apicid = cpu_to_logical_apicid(cpu); + cpu = cpumask_first_and(cpumask, andmask); + apicid = cpu_to_logical_apicid(cpu); + while (cpus_found < num_bits_set) { - if (cpu_isset(cpu, *cpumask) && cpu_isset(cpu, *andmask)) { + if (cpumask_test_cpu(cpu, cpumask) && + cpumask_test_cpu(cpu, andmask)) { int new_apicid = cpu_to_logical_apicid(cpu); if (apicid_cluster(apicid) != apicid_cluster(new_apicid)) { diff --git a/arch/x86/include/asm/genapic_32.h b/arch/x86/include/asm/genapic_32.h index 325298a82237..eed6e305291f 100644 --- a/arch/x86/include/asm/genapic_32.h +++ b/arch/x86/include/asm/genapic_32.h @@ -58,8 +58,8 @@ struct genapic { unsigned (*get_apic_id)(unsigned long x); unsigned long apic_id_mask; unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask); - unsigned int (*cpu_mask_to_apicid_and)(const cpumask_t *cpumask, - const cpumask_t *andmask); + unsigned int (*cpu_mask_to_apicid_and)(const struct cpumask *cpumask, + const struct cpumask *andmask); void (*vector_allocation_domain)(int cpu, cpumask_t *retmask); #ifdef CONFIG_SMP diff --git a/arch/x86/include/asm/genapic_64.h b/arch/x86/include/asm/genapic_64.h index 301c7f41125d..244b71729ecb 100644 --- a/arch/x86/include/asm/genapic_64.h +++ b/arch/x86/include/asm/genapic_64.h @@ -31,8 +31,8 @@ struct genapic { void (*send_IPI_self)(int vector); /* */ unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask); - unsigned int (*cpu_mask_to_apicid_and)(const cpumask_t *cpumask, - const cpumask_t *andmask); + unsigned int (*cpu_mask_to_apicid_and)(const struct cpumask *cpumask, + const struct cpumask *andmask); unsigned int (*phys_pkg_id)(int index_msb); unsigned int (*get_apic_id)(unsigned long x); unsigned long (*set_apic_id)(unsigned int id); diff --git a/arch/x86/include/asm/mach-default/mach_apic.h b/arch/x86/include/asm/mach-default/mach_apic.h index 229b605d104e..df8e024c43c5 100644 --- a/arch/x86/include/asm/mach-default/mach_apic.h +++ b/arch/x86/include/asm/mach-default/mach_apic.h @@ -67,11 +67,11 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return cpus_addr(*cpumask)[0]; } -static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask, - const cpumask_t *andmask) +static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { - unsigned long mask1 = cpus_addr(*cpumask)[0]; - unsigned long mask2 = cpus_addr(*andmask)[0]; + unsigned long mask1 = cpumask_bits(cpumask)[0]; + unsigned long mask2 = cpumask_bits(andmask)[0]; return (unsigned int)(mask1 & mask2); } diff --git a/arch/x86/include/asm/numaq/apic.h b/arch/x86/include/asm/numaq/apic.h index abf668ced503..c80f00d29965 100644 --- a/arch/x86/include/asm/numaq/apic.h +++ b/arch/x86/include/asm/numaq/apic.h @@ -127,8 +127,8 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return (int) 0xF; } -static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { return (int) 0xF; } diff --git a/arch/x86/include/asm/summit/apic.h b/arch/x86/include/asm/summit/apic.h index cbcc2c7eb1d5..651a93849341 100644 --- a/arch/x86/include/asm/summit/apic.h +++ b/arch/x86/include/asm/summit/apic.h @@ -170,8 +170,8 @@ static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) return apicid; } -static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { int num_bits_set; int num_bits_set2; @@ -179,9 +179,9 @@ static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, int cpu; int apicid = 0; - num_bits_set = cpus_weight(*cpumask); - num_bits_set2 = cpus_weight(*andmask); - num_bits_set = min_t(int, num_bits_set, num_bits_set2); + num_bits_set = cpumask_weight(cpumask); + num_bits_set2 = cpumask_weight(andmask); + num_bits_set = min(num_bits_set, num_bits_set2); /* Return id to all */ if (num_bits_set >= nr_cpu_ids) return 0xFF; @@ -189,11 +189,11 @@ static inline unsigned int cpu_mask_to_apicid_and(const cpumask_t *cpumask, * 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. */ - while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) - if (cpu_isset(cpu, *andmask) - apicid = cpu_to_logical_apicid(cpu); + cpu = cpumask_first_and(cpumask, andmask); + apicid = cpu_to_logical_apicid(cpu); while (cpus_found < num_bits_set) { - if (cpu_isset(cpu, *cpumask) && cpu_isset(cpu, *andmask)) { + if (cpumask_test_cpu(cpu, cpumask) + && cpumask_test_cpu(cpu, andmask)) { int new_apicid = cpu_to_logical_apicid(cpu); if (apicid_cluster(apicid) != apicid_cluster(new_apicid)) { diff --git a/arch/x86/kernel/genapic_flat_64.c b/arch/x86/kernel/genapic_flat_64.c index 1efecd206a74..c772bb10b173 100644 --- a/arch/x86/kernel/genapic_flat_64.c +++ b/arch/x86/kernel/genapic_flat_64.c @@ -158,13 +158,13 @@ static unsigned int flat_cpu_mask_to_apicid(const cpumask_t *cpumask) return cpus_addr(*cpumask)[0] & APIC_ALL_CPUS; } -static unsigned int flat_cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static unsigned int flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { - unsigned long mask1 = cpus_addr(*cpumask)[0] & APIC_ALL_CPUS; - unsigned long mask2 = cpus_addr(*andmask)[0] & APIC_ALL_CPUS; + unsigned long mask1 = cpumask_bits(cpumask)[0] & APIC_ALL_CPUS; + unsigned long mask2 = cpumask_bits(andmask)[0] & APIC_ALL_CPUS; - return (int)(mask1 & mask2); + return mask1 & mask2; } static unsigned int phys_pkg_id(int index_msb) @@ -264,8 +264,9 @@ static unsigned int physflat_cpu_mask_to_apicid(const cpumask_t *cpumask) return BAD_APICID; } -static unsigned int physflat_cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static unsigned int +physflat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { int cpu; @@ -273,9 +274,9 @@ static unsigned int physflat_cpu_mask_to_apicid_and(const cpumask_t *cpumask, * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) - if (cpu_isset(cpu, *andmask)) - return per_cpu(x86_cpu_to_apicid, cpu); + cpu = cpumask_any_and(cpumask, andmask); + if (cpu < nr_cpu_ids) + return per_cpu(x86_cpu_to_apicid, cpu); return BAD_APICID; } diff --git a/arch/x86/kernel/genx2apic_cluster.c b/arch/x86/kernel/genx2apic_cluster.c index fd8047f4e455..e7d16f53b9cd 100644 --- a/arch/x86/kernel/genx2apic_cluster.c +++ b/arch/x86/kernel/genx2apic_cluster.c @@ -123,8 +123,8 @@ static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) return BAD_APICID; } -static unsigned int x2apic_cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static unsigned int x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { int cpu; @@ -132,9 +132,9 @@ static unsigned int x2apic_cpu_mask_to_apicid_and(const cpumask_t *cpumask, * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) - if (cpu_isset(cpu, *andmask)) - return per_cpu(x86_cpu_to_apicid, cpu); + cpu = cpumask_any_and(cpumask, andmask); + if (cpu < nr_cpu_ids) + return per_cpu(x86_cpu_to_apicid, cpu); return BAD_APICID; } diff --git a/arch/x86/kernel/genx2apic_phys.c b/arch/x86/kernel/genx2apic_phys.c index d5578bb8f165..9d0386c7e798 100644 --- a/arch/x86/kernel/genx2apic_phys.c +++ b/arch/x86/kernel/genx2apic_phys.c @@ -122,8 +122,8 @@ static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) return BAD_APICID; } -static unsigned int x2apic_cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static unsigned int x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { int cpu; @@ -131,9 +131,9 @@ static unsigned int x2apic_cpu_mask_to_apicid_and(const cpumask_t *cpumask, * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) - if (cpu_isset(cpu, *andmask)) - return per_cpu(x86_cpu_to_apicid, cpu); + cpu = cpumask_any_and(cpumask, andmask); + if (cpu < nr_cpu_ids) + return per_cpu(x86_cpu_to_apicid, cpu); return BAD_APICID; } diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 53bd2570272c..22596ec94c82 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -179,8 +179,8 @@ static unsigned int uv_cpu_mask_to_apicid(const cpumask_t *cpumask) return BAD_APICID; } -static unsigned int uv_cpu_mask_to_apicid_and(const cpumask_t *cpumask, - const cpumask_t *andmask) +static unsigned int uv_cpu_mask_to_apicid_and(const struct cpumask *cpumask, + const struct cpumask *andmask) { int cpu; @@ -188,9 +188,9 @@ static unsigned int uv_cpu_mask_to_apicid_and(const cpumask_t *cpumask, * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - while ((cpu = next_cpu(-1, *cpumask)) < nr_cpu_ids) - if (cpu_isset(cpu, *andmask)) - return per_cpu(x86_cpu_to_apicid, cpu); + cpu = cpumask_any_and(cpumask, andmask); + if (cpu < nr_cpu_ids) + return per_cpu(x86_cpu_to_apicid, cpu); return BAD_APICID; } From 22f65d31b25a320a5246592160bcb102d2791c45 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:56 -0800 Subject: [PATCH 06/16] x86: Update io_apic.c to use new cpumask API Impact: cleanup, consolidate patches, use new API Consolidate the following into a single patch to adapt to new sparseirq code in arch/x86/kernel/io_apic.c, add allocation of cpumask_var_t's in domain and old_domain, and reduce further merge conflicts. Only one file (arch/x86/kernel/io_apic.c) is changed in all of these patches. 0006-x86-io_apic-change-irq_cfg-domain-old_domain-to.patch 0007-x86-io_apic-set_desc_affinity.patch 0008-x86-io_apic-send_cleanup_vector.patch 0009-x86-io_apic-eliminate-remaining-cpumask_ts-from-st.patch 0021-x86-final-cleanups-in-io_apic-to-use-new-cpumask-AP.patch Signed-off-by: Rusty Russell Signed-off-by: Mike Travis --- arch/x86/kernel/io_apic.c | 302 ++++++++++++++++++-------------------- 1 file changed, 145 insertions(+), 157 deletions(-) diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 7f23ce7f5518..60bb8b19f4cd 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -136,8 +136,8 @@ static struct irq_pin_list *get_one_free_irq_2_pin(int cpu) struct irq_cfg { struct irq_pin_list *irq_2_pin; - cpumask_t domain; - cpumask_t old_domain; + cpumask_var_t domain; + cpumask_var_t old_domain; unsigned move_cleanup_count; u8 vector; u8 move_in_progress : 1; @@ -149,22 +149,22 @@ static struct irq_cfg irq_cfgx[] = { #else static struct irq_cfg irq_cfgx[NR_IRQS] = { #endif - [0] = { .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, - [1] = { .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, - [2] = { .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, - [3] = { .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, - [4] = { .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, - [5] = { .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, - [6] = { .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, - [7] = { .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, - [8] = { .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, - [9] = { .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, - [10] = { .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, - [11] = { .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, - [12] = { .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, - [13] = { .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, - [14] = { .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, - [15] = { .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, + [0] = { .vector = IRQ0_VECTOR, }, + [1] = { .vector = IRQ1_VECTOR, }, + [2] = { .vector = IRQ2_VECTOR, }, + [3] = { .vector = IRQ3_VECTOR, }, + [4] = { .vector = IRQ4_VECTOR, }, + [5] = { .vector = IRQ5_VECTOR, }, + [6] = { .vector = IRQ6_VECTOR, }, + [7] = { .vector = IRQ7_VECTOR, }, + [8] = { .vector = IRQ8_VECTOR, }, + [9] = { .vector = IRQ9_VECTOR, }, + [10] = { .vector = IRQ10_VECTOR, }, + [11] = { .vector = IRQ11_VECTOR, }, + [12] = { .vector = IRQ12_VECTOR, }, + [13] = { .vector = IRQ13_VECTOR, }, + [14] = { .vector = IRQ14_VECTOR, }, + [15] = { .vector = IRQ15_VECTOR, }, }; void __init arch_early_irq_init(void) @@ -180,6 +180,10 @@ void __init arch_early_irq_init(void) for (i = 0; i < count; i++) { desc = irq_to_desc(i); desc->chip_data = &cfg[i]; + alloc_bootmem_cpumask_var(&cfg[i].domain); + alloc_bootmem_cpumask_var(&cfg[i].old_domain); + if (i < NR_IRQS_LEGACY) + cpumask_setall(cfg[i].domain); } } @@ -204,6 +208,20 @@ static struct irq_cfg *get_one_free_irq_cfg(int cpu) node = cpu_to_node(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)) { + kfree(cfg); + cfg = NULL; + } else if (!alloc_cpumask_var(&cfg->old_domain, GFP_ATOMIC)) { + free_cpumask_var(cfg->domain); + kfree(cfg); + cfg = NULL; + } else { + cpumask_clear(cfg->domain); + cpumask_clear(cfg->old_domain); + } + } printk(KERN_DEBUG " alloc irq_cfg on cpu %d node %d\n", cpu, node); return cfg; @@ -362,6 +380,26 @@ static void ioapic_mask_entry(int apic, int pin) } #ifdef CONFIG_SMP +static void send_cleanup_vector(struct irq_cfg *cfg) +{ + cpumask_var_t cleanup_mask; + + if (unlikely(!alloc_cpumask_var(&cleanup_mask, GFP_ATOMIC))) { + unsigned int i; + cfg->move_cleanup_count = 0; + for_each_cpu_and(i, cfg->old_domain, cpu_online_mask) + cfg->move_cleanup_count++; + for_each_cpu_and(i, cfg->old_domain, cpu_online_mask) + send_IPI_mask(cpumask_of(i), IRQ_MOVE_CLEANUP_VECTOR); + } else { + cpumask_and(cleanup_mask, cfg->old_domain, cpu_online_mask); + cfg->move_cleanup_count = cpumask_weight(cleanup_mask); + send_IPI_mask(cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); + free_cpumask_var(cleanup_mask); + } + cfg->move_in_progress = 0; +} + static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, struct irq_cfg *cfg) { int apic, pin; @@ -400,40 +438,52 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, struct irq static int assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask); -static void set_ioapic_affinity_irq_desc(struct irq_desc *desc, - const struct cpumask *mask) +/* + * Either sets desc->affinity to a valid value, and returns cpu_mask_to_apicid + * of that, or returns BAD_APICID and leaves desc->affinity untouched. + */ +static unsigned int +set_desc_affinity(struct irq_desc *desc, const struct cpumask *mask) { struct irq_cfg *cfg; - unsigned long flags; - unsigned int dest; - cpumask_t tmp; unsigned int irq; if (!cpumask_intersects(mask, cpu_online_mask)) - return; + return BAD_APICID; irq = desc->irq; cfg = desc->chip_data; if (assign_irq_vector(irq, cfg, mask)) - return; + return BAD_APICID; + cpumask_and(&desc->affinity, cfg->domain, mask); set_extra_move_desc(desc, mask); + return cpu_mask_to_apicid_and(&desc->affinity, cpu_online_mask); +} - cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(&tmp); - /* - * Only the high 8 bits are valid. - */ - dest = SET_APIC_LOGICAL_ID(dest); +static void +set_ioapic_affinity_irq_desc(struct irq_desc *desc, const struct cpumask *mask) +{ + struct irq_cfg *cfg; + unsigned long flags; + unsigned int dest; + unsigned int irq; + + irq = desc->irq; + cfg = desc->chip_data; spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg); - cpumask_copy(&desc->affinity, mask); + dest = set_desc_affinity(desc, mask); + if (dest != BAD_APICID) { + /* Only the high 8 bits are valid. */ + dest = SET_APIC_LOGICAL_ID(dest); + __target_IO_APIC_irq(irq, dest, cfg); + } spin_unlock_irqrestore(&ioapic_lock, flags); } -static void set_ioapic_affinity_irq(unsigned int irq, - const struct cpumask *mask) +static void +set_ioapic_affinity_irq(unsigned int irq, const struct cpumask *mask) { struct irq_desc *desc; @@ -1117,26 +1167,32 @@ __assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask) */ static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; unsigned int old_vector; - int cpu; - cpumask_t tmp_mask; + int cpu, err; + cpumask_var_t tmp_mask; if ((cfg->move_in_progress) || cfg->move_cleanup_count) return -EBUSY; + if (!alloc_cpumask_var(&tmp_mask, GFP_ATOMIC)) + return -ENOMEM; + old_vector = cfg->vector; if (old_vector) { - cpus_and(tmp_mask, *mask, cpu_online_map); - cpus_and(tmp_mask, cfg->domain, tmp_mask); - if (!cpus_empty(tmp_mask)) + cpumask_and(tmp_mask, mask, cpu_online_mask); + cpumask_and(tmp_mask, cfg->domain, tmp_mask); + if (!cpumask_empty(tmp_mask)) { + free_cpumask_var(tmp_mask); return 0; + } } /* Only try and allocate irqs on cpus that are present */ - for_each_cpu_and(cpu, mask, &cpu_online_map) { + err = -ENOSPC; + for_each_cpu_and(cpu, mask, cpu_online_mask) { int new_cpu; int vector, offset; - vector_allocation_domain(cpu, &tmp_mask); + vector_allocation_domain(cpu, tmp_mask); vector = current_vector; offset = current_offset; @@ -1156,7 +1212,7 @@ next: if (vector == SYSCALL_VECTOR) goto next; #endif - for_each_cpu_and(new_cpu, &tmp_mask, &cpu_online_map) + for_each_cpu_and(new_cpu, tmp_mask, cpu_online_mask) if (per_cpu(vector_irq, new_cpu)[vector] != -1) goto next; /* Found one! */ @@ -1164,15 +1220,17 @@ next: current_offset = offset; if (old_vector) { cfg->move_in_progress = 1; - cfg->old_domain = cfg->domain; + cpumask_copy(cfg->old_domain, cfg->domain); } - for_each_cpu_and(new_cpu, &tmp_mask, &cpu_online_map) + for_each_cpu_and(new_cpu, tmp_mask, cpu_online_mask) per_cpu(vector_irq, new_cpu)[vector] = irq; cfg->vector = vector; - cfg->domain = tmp_mask; - return 0; + cpumask_copy(cfg->domain, tmp_mask); + err = 0; + break; } - return -ENOSPC; + free_cpumask_var(tmp_mask); + return err; } static int @@ -1189,23 +1247,20 @@ assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask) static void __clear_irq_vector(int irq, struct irq_cfg *cfg) { - cpumask_t mask; int cpu, vector; BUG_ON(!cfg->vector); vector = cfg->vector; - cpus_and(mask, cfg->domain, cpu_online_map); - for_each_cpu_mask_nr(cpu, mask) + for_each_cpu_and(cpu, cfg->domain, cpu_online_mask) per_cpu(vector_irq, cpu)[vector] = -1; cfg->vector = 0; - cpus_clear(cfg->domain); + cpumask_clear(cfg->domain); if (likely(!cfg->move_in_progress)) return; - cpus_and(mask, cfg->old_domain, cpu_online_map); - for_each_cpu_mask_nr(cpu, mask) { + for_each_cpu_and(cpu, cfg->old_domain, cpu_online_mask) { for (vector = FIRST_EXTERNAL_VECTOR; vector < NR_VECTORS; vector++) { if (per_cpu(vector_irq, cpu)[vector] != irq) @@ -1230,7 +1285,7 @@ void __setup_vector_irq(int cpu) if (!desc) continue; cfg = desc->chip_data; - if (!cpu_isset(cpu, cfg->domain)) + if (!cpumask_test_cpu(cpu, cfg->domain)) continue; vector = cfg->vector; per_cpu(vector_irq, cpu)[vector] = irq; @@ -1242,7 +1297,7 @@ void __setup_vector_irq(int cpu) continue; cfg = irq_cfg(irq); - if (!cpu_isset(cpu, cfg->domain)) + if (!cpumask_test_cpu(cpu, cfg->domain)) per_cpu(vector_irq, cpu)[vector] = -1; } } @@ -1378,18 +1433,17 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, struct irq_de { struct irq_cfg *cfg; struct IO_APIC_route_entry entry; - cpumask_t mask; + unsigned int dest; if (!IO_APIC_IRQ(irq)) return; cfg = desc->chip_data; - mask = *TARGET_CPUS; - if (assign_irq_vector(irq, cfg, &mask)) + if (assign_irq_vector(irq, cfg, TARGET_CPUS)) return; - cpus_and(mask, cfg->domain, mask); + dest = cpu_mask_to_apicid_and(cfg->domain, TARGET_CPUS); apic_printk(APIC_VERBOSE,KERN_DEBUG "IOAPIC[%d]: Set routing entry (%d-%d -> 0x%x -> " @@ -1399,8 +1453,7 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, struct irq_de if (setup_ioapic_entry(mp_ioapics[apic].mp_apicid, irq, &entry, - cpu_mask_to_apicid(&mask), trigger, polarity, - cfg->vector)) { + dest, trigger, polarity, cfg->vector)) { printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", mp_ioapics[apic].mp_apicid, pin); __clear_irq_vector(irq, cfg); @@ -2122,7 +2175,7 @@ static int ioapic_retrigger_irq(unsigned int irq) unsigned long flags; spin_lock_irqsave(&vector_lock, flags); - send_IPI_mask(&cpumask_of_cpu(first_cpu(cfg->domain)), cfg->vector); + send_IPI_mask(cpumask_of(cpumask_first(cfg->domain)), cfg->vector); spin_unlock_irqrestore(&vector_lock, flags); return 1; @@ -2175,15 +2228,13 @@ static void migrate_ioapic_irq_desc(struct irq_desc *desc, const struct cpumask *mask) { struct irq_cfg *cfg; - cpumask_t tmpmask; struct irte irte; int modify_ioapic_rte; unsigned int dest; unsigned long flags; unsigned int irq; - cpus_and(tmpmask, *mask, cpu_online_map); - if (cpus_empty(tmpmask)) + if (!cpumask_intersects(mask, cpu_online_mask)) return; irq = desc->irq; @@ -2196,8 +2247,7 @@ migrate_ioapic_irq_desc(struct irq_desc *desc, const struct cpumask *mask) set_extra_move_desc(desc, mask); - cpus_and(tmpmask, cfg->domain, *mask); - dest = cpu_mask_to_apicid(&tmpmask); + dest = cpu_mask_to_apicid_and(cfg->domain, mask); modify_ioapic_rte = desc->status & IRQ_LEVEL; if (modify_ioapic_rte) { @@ -2214,14 +2264,10 @@ migrate_ioapic_irq_desc(struct irq_desc *desc, const struct cpumask *mask) */ modify_irte(irq, &irte); - if (cfg->move_in_progress) { - cpus_and(tmpmask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(tmpmask); - send_IPI_mask(&tmpmask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } + if (cfg->move_in_progress) + send_cleanup_vector(cfg); - desc->affinity = *mask; + cpumask_copy(&desc->affinity, mask); } static int migrate_irq_remapped_level_desc(struct irq_desc *desc) @@ -2247,7 +2293,7 @@ static int migrate_irq_remapped_level_desc(struct irq_desc *desc) ret = 0; desc->status &= ~IRQ_MOVE_PENDING; - cpus_clear(desc->pending_mask); + cpumask_clear(&desc->pending_mask); unmask: unmask_IO_APIC_irq_desc(desc); @@ -2333,7 +2379,7 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) if (!cfg->move_cleanup_count) goto unlock; - if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) + if (vector == cfg->vector && cpumask_test_cpu(me, cfg->domain)) goto unlock; __get_cpu_var(vector_irq)[vector] = -1; @@ -2356,14 +2402,8 @@ static void irq_complete_move(struct irq_desc **descp) vector = ~get_irq_regs()->orig_ax; me = smp_processor_id(); - if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { - cpumask_t cleanup_mask; - - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(&cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } + if (vector == cfg->vector && cpumask_test_cpu(me, cfg->domain)) + send_cleanup_vector(cfg); } #else static inline void irq_complete_move(struct irq_desc **descp) {} @@ -3088,16 +3128,13 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms struct irq_cfg *cfg; int err; unsigned dest; - cpumask_t tmp; cfg = irq_cfg(irq); - tmp = *TARGET_CPUS; - err = assign_irq_vector(irq, cfg, &tmp); + err = assign_irq_vector(irq, cfg, TARGET_CPUS); if (err) return err; - cpus_and(tmp, cfg->domain, tmp); - dest = cpu_mask_to_apicid(&tmp); + dest = cpu_mask_to_apicid_and(cfg->domain, TARGET_CPUS); #ifdef CONFIG_INTR_REMAP if (irq_remapped(irq)) { @@ -3157,19 +3194,12 @@ static void set_msi_irq_affinity(unsigned int irq, const struct cpumask *mask) struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; - cpumask_t tmp; - if (!cpumask_intersects(mask, cpu_online_mask)) + dest = set_desc_affinity(desc, mask); + if (dest == BAD_APICID) return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, mask)) - return; - - set_extra_move_desc(desc, mask); - - cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(&tmp); read_msi_msg_desc(desc, &msg); @@ -3179,7 +3209,6 @@ static void set_msi_irq_affinity(unsigned int irq, const struct cpumask *mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); write_msi_msg_desc(desc, &msg); - cpumask_copy(&desc->affinity, mask); } #ifdef CONFIG_INTR_REMAP /* @@ -3192,24 +3221,15 @@ ir_set_msi_irq_affinity(unsigned int irq, const struct cpumask *mask) struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; unsigned int dest; - cpumask_t tmp, cleanup_mask; struct irte irte; - if (!cpumask_intersects(mask, cpu_online_mask)) - return; - if (get_irte(irq, &irte)) return; - cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, mask)) + dest = set_desc_affinity(desc, mask); + if (dest == BAD_APICID) return; - set_extra_move_desc(desc, mask); - - cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(&tmp); - irte.vector = cfg->vector; irte.dest_id = IRTE_DEST(dest); @@ -3223,14 +3243,8 @@ ir_set_msi_irq_affinity(unsigned int irq, const struct cpumask *mask) * at the new destination. So, time to cleanup the previous * vector allocation. */ - if (cfg->move_in_progress) { - cpus_and(cleanup_mask, cfg->old_domain, cpu_online_map); - cfg->move_cleanup_count = cpus_weight(cleanup_mask); - send_IPI_mask(&cleanup_mask, IRQ_MOVE_CLEANUP_VECTOR); - cfg->move_in_progress = 0; - } - - cpumask_copy(&desc->affinity, mask); + if (cfg->move_in_progress) + send_cleanup_vector(cfg); } #endif @@ -3421,25 +3435,18 @@ void arch_teardown_msi_irq(unsigned int irq) #ifdef CONFIG_DMAR #ifdef CONFIG_SMP -static void dmar_msi_set_affinity(unsigned int irq, const cpumask_t *mask) +static void dmar_msi_set_affinity(unsigned int irq, const struct cpumask *mask) { struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; - cpumask_t tmp; - if (!cpumask_intersects(mask, cpu_online_mask)) + dest = set_desc_affinity(desc, mask); + if (dest == BAD_APICID) return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, mask)) - return; - - set_extra_move_desc(desc, mask); - - cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(&tmp); dmar_msi_read(irq, &msg); @@ -3449,7 +3456,6 @@ static void dmar_msi_set_affinity(unsigned int irq, const cpumask_t *mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); dmar_msi_write(irq, &msg); - cpumask_copy(&desc->affinity, mask); } #endif /* CONFIG_SMP */ @@ -3483,25 +3489,18 @@ int arch_setup_dmar_msi(unsigned int irq) #ifdef CONFIG_HPET_TIMER #ifdef CONFIG_SMP -static void hpet_msi_set_affinity(unsigned int irq, const cpumask_t *mask) +static void hpet_msi_set_affinity(unsigned int irq, const struct cpumask *mask) { struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; - cpumask_t tmp; - if (!cpumask_intersects(mask, cpu_online_mask)) + dest = set_desc_affinity(desc, mask); + if (dest == BAD_APICID) return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, mask)) - return; - - set_extra_move_desc(desc, mask); - - cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(&tmp); hpet_msi_read(irq, &msg); @@ -3511,7 +3510,6 @@ static void hpet_msi_set_affinity(unsigned int irq, const cpumask_t *mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); hpet_msi_write(irq, &msg); - cpumask_copy(&desc->affinity, mask); } #endif /* CONFIG_SMP */ @@ -3566,27 +3564,19 @@ static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) write_ht_irq_msg(irq, &msg); } -static void set_ht_irq_affinity(unsigned int irq, const cpumask_t *mask) +static void set_ht_irq_affinity(unsigned int irq, const struct cpumask *mask) { struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; unsigned int dest; - cpumask_t tmp; - if (!cpumask_intersects(mask, cpu_online_mask)) + dest = set_desc_affinity(desc, mask); + if (dest == BAD_APICID) return; cfg = desc->chip_data; - if (assign_irq_vector(irq, cfg, mask)) - return; - - set_extra_move_desc(desc, mask); - - cpumask_and(&tmp, &cfg->domain, mask); - dest = cpu_mask_to_apicid(&tmp); target_ht_irq(irq, dest, cfg->vector); - cpumask_copy(&desc->affinity, mask); } #endif @@ -3606,7 +3596,6 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) { struct irq_cfg *cfg; int err; - cpumask_t tmp; cfg = irq_cfg(irq); err = assign_irq_vector(irq, cfg, TARGET_CPUS); @@ -3614,8 +3603,7 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) struct ht_irq_msg msg; unsigned dest; - cpus_and(tmp, cfg->domain, tmp); - dest = cpu_mask_to_apicid(&tmp); + dest = cpu_mask_to_apicid_and(cfg->domain, TARGET_CPUS); msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest); @@ -3651,7 +3639,7 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, unsigned long mmr_offset) { - const cpumask_t *eligible_cpu = &cpumask_of_cpu(cpu); + const struct cpumask *eligible_cpu = cpumask_of(cpu); struct irq_cfg *cfg; int mmr_pnode; unsigned long mmr_value; @@ -3891,7 +3879,7 @@ void __init setup_ioapic_dest(void) int pin, ioapic, irq, irq_entry; struct irq_desc *desc; struct irq_cfg *cfg; - const cpumask_t *mask; + const struct cpumask *mask; if (skip_ioapic_setup == 1) return; From b78936e14ee47b6b2d628501a0eab5270db80132 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:57 -0800 Subject: [PATCH 07/16] xen: convert to cpumask_var_t and new cpumask primitives. Simple change, and eventual space saving when NR_CPUS >> nr_cpu_ids. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis Cc: Jeremy Fitzhardinge --- arch/x86/xen/smp.c | 9 ++++++--- arch/x86/xen/suspend.c | 3 ++- arch/x86/xen/xen-ops.h | 2 +- 3 files changed, 9 insertions(+), 5 deletions(-) diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c index 2cce362c9874..b3a95868839b 100644 --- a/arch/x86/xen/smp.c +++ b/arch/x86/xen/smp.c @@ -33,7 +33,7 @@ #include "xen-ops.h" #include "mmu.h" -cpumask_t xen_cpu_initialized_map; +cpumask_var_t xen_cpu_initialized_map; static DEFINE_PER_CPU(int, resched_irq); static DEFINE_PER_CPU(int, callfunc_irq); @@ -192,7 +192,10 @@ static void __init xen_smp_prepare_cpus(unsigned int max_cpus) if (xen_smp_intr_init(0)) BUG(); - xen_cpu_initialized_map = cpumask_of_cpu(0); + if (!alloc_cpumask_var(&xen_cpu_initialized_map, GFP_KERNEL)) + panic("could not allocate xen_cpu_initialized_map\n"); + + cpumask_copy(xen_cpu_initialized_map, cpumask_of(0)); /* Restrict the possible_map according to max_cpus. */ while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) { @@ -221,7 +224,7 @@ cpu_initialize_context(unsigned int cpu, struct task_struct *idle) struct vcpu_guest_context *ctxt; struct desc_struct *gdt; - if (cpu_test_and_set(cpu, xen_cpu_initialized_map)) + if (cpumask_test_and_set_cpu(cpu, xen_cpu_initialized_map)) return 0; ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL); diff --git a/arch/x86/xen/suspend.c b/arch/x86/xen/suspend.c index 2a234db5949b..212ffe012b76 100644 --- a/arch/x86/xen/suspend.c +++ b/arch/x86/xen/suspend.c @@ -35,7 +35,8 @@ void xen_post_suspend(int suspend_cancelled) pfn_to_mfn(xen_start_info->console.domU.mfn); } else { #ifdef CONFIG_SMP - xen_cpu_initialized_map = cpu_online_map; + BUG_ON(xen_cpu_initialized_map == NULL); + cpumask_copy(xen_cpu_initialized_map, cpu_online_mask); #endif xen_vcpu_restore(); } diff --git a/arch/x86/xen/xen-ops.h b/arch/x86/xen/xen-ops.h index 9e1afae8461f..c1f8faf0a2c5 100644 --- a/arch/x86/xen/xen-ops.h +++ b/arch/x86/xen/xen-ops.h @@ -58,7 +58,7 @@ void __init xen_init_spinlocks(void); __cpuinit void xen_init_lock_cpu(int cpu); void xen_uninit_lock_cpu(int cpu); -extern cpumask_t xen_cpu_initialized_map; +extern cpumask_var_t xen_cpu_initialized_map; #else static inline void xen_smp_init(void) {} #endif From d7b381bb7b1ad69ff008ea063d26e988b686c8de Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:58 -0800 Subject: [PATCH 08/16] x86: fixup_irqs() doesnt need an argument. Impact: cleanup, remove on-stack cpumask. The "map" arg is always cpu_online_mask. Importantly, set_affinity always ands the argument with cpu_online_mask anyway, so we don't need to do it in fixup_irqs(), avoiding a temporary. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis --- arch/x86/include/asm/irq.h | 2 +- arch/x86/kernel/irq_32.c | 13 +++++++------ arch/x86/kernel/irq_64.c | 15 ++++++++------- arch/x86/kernel/smpboot.c | 2 +- 4 files changed, 17 insertions(+), 15 deletions(-) diff --git a/arch/x86/include/asm/irq.h b/arch/x86/include/asm/irq.h index bae0eda95486..8766d30fb746 100644 --- a/arch/x86/include/asm/irq.h +++ b/arch/x86/include/asm/irq.h @@ -37,7 +37,7 @@ extern int irqbalance_disable(char *str); #ifdef CONFIG_HOTPLUG_CPU #include -extern void fixup_irqs(cpumask_t map); +extern void fixup_irqs(void); #endif extern unsigned int do_IRQ(struct pt_regs *regs); diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index 9cf9cbbf7a02..9dc5588f336a 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -233,27 +233,28 @@ unsigned int do_IRQ(struct pt_regs *regs) #ifdef CONFIG_HOTPLUG_CPU #include -void fixup_irqs(cpumask_t map) +/* A cpu has been removed from cpu_online_mask. Reset irq affinities. */ +void fixup_irqs(void) { unsigned int irq; static int warned; struct irq_desc *desc; for_each_irq_desc(irq, desc) { - cpumask_t mask; + const struct cpumask *affinity; if (!desc) continue; if (irq == 2) continue; - cpus_and(mask, desc->affinity, map); - if (any_online_cpu(mask) == NR_CPUS) { + affinity = &desc->affinity; + if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { printk("Breaking affinity for irq %i\n", irq); - mask = map; + affinity = cpu_all_mask; } if (desc->chip->set_affinity) - desc->chip->set_affinity(irq, &mask); + desc->chip->set_affinity(irq, affinity); else if (desc->action && !(warned++)) printk("Cannot set affinity for irq %i\n", irq); } diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 27f2307b0a34..fca2991443f5 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -83,16 +83,17 @@ asmlinkage unsigned int do_IRQ(struct pt_regs *regs) } #ifdef CONFIG_HOTPLUG_CPU -void fixup_irqs(cpumask_t map) +/* A cpu has been removed from cpu_online_mask. Reset irq affinities. */ +void fixup_irqs(void) { unsigned int irq; static int warned; struct irq_desc *desc; for_each_irq_desc(irq, desc) { - cpumask_t mask; int break_affinity = 0; int set_affinity = 1; + const struct cpumask *affinity; if (!desc) continue; @@ -102,23 +103,23 @@ void fixup_irqs(cpumask_t map) /* interrupt's are disabled at this point */ spin_lock(&desc->lock); + affinity = &desc->affinity; if (!irq_has_action(irq) || - cpus_equal(desc->affinity, map)) { + cpumask_equal(affinity, cpu_online_mask)) { spin_unlock(&desc->lock); continue; } - cpus_and(mask, desc->affinity, map); - if (cpus_empty(mask)) { + if (cpumask_any_and(affinity, cpu_online_mask) >= nr_cpu_ids) { break_affinity = 1; - mask = map; + affinity = cpu_all_mask; } if (desc->chip->mask) desc->chip->mask(irq); if (desc->chip->set_affinity) - desc->chip->set_affinity(irq, &mask); + desc->chip->set_affinity(irq, affinity); else if (!(warned++)) set_affinity = 0; diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 9d58134e0231..8b6f675b363b 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -1346,7 +1346,7 @@ void cpu_disable_common(void) lock_vector_lock(); remove_cpu_from_maps(cpu); unlock_vector_lock(); - fixup_irqs(cpu_online_map); + fixup_irqs(); } int native_cpu_disable(void) From bcda016eddd7a8b374bb371473c821a91ff1d8cc Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:33:59 -0800 Subject: [PATCH 09/16] x86: cosmetic changes apic-related files. This patch simply changes cpumask_t to struct cpumask and similar trivial modernizations. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis --- arch/x86/include/asm/bigsmp/ipi.h | 14 ++---- arch/x86/include/asm/es7000/ipi.h | 13 ++--- arch/x86/include/asm/genapic_32.h | 11 ++-- arch/x86/include/asm/genapic_64.h | 11 ++-- arch/x86/include/asm/ipi.h | 10 ++-- arch/x86/include/asm/mach-default/mach_apic.h | 12 ++--- arch/x86/include/asm/mach-default/mach_ipi.h | 12 ++--- arch/x86/include/asm/numaq/ipi.h | 14 ++---- arch/x86/include/asm/smp.h | 4 +- arch/x86/kernel/genapic_flat_64.c | 50 ++++++++++--------- arch/x86/kernel/genx2apic_cluster.c | 25 +++++----- arch/x86/kernel/genx2apic_phys.c | 25 +++++----- arch/x86/kernel/genx2apic_uv_x.c | 24 ++++----- arch/x86/kernel/ipi.c | 14 +++--- arch/x86/kernel/smp.c | 6 +-- arch/x86/xen/smp.c | 11 ++-- 16 files changed, 127 insertions(+), 129 deletions(-) diff --git a/arch/x86/include/asm/bigsmp/ipi.h b/arch/x86/include/asm/bigsmp/ipi.h index 63553e9f22b2..27fcd01b3ae6 100644 --- a/arch/x86/include/asm/bigsmp/ipi.h +++ b/arch/x86/include/asm/bigsmp/ipi.h @@ -1,26 +1,22 @@ #ifndef __ASM_MACH_IPI_H #define __ASM_MACH_IPI_H -void send_IPI_mask_sequence(const cpumask_t *mask, int vector); -void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); +void send_IPI_mask_sequence(const struct cpumask *mask, int vector); +void send_IPI_mask_allbutself(const struct cpumask *mask, int vector); -static inline void send_IPI_mask(const cpumask_t *mask, int vector) +static inline void send_IPI_mask(const struct cpumask *mask, int vector) { send_IPI_mask_sequence(mask, vector); } static inline void send_IPI_allbutself(int vector) { - cpumask_t mask = cpu_online_map; - cpu_clear(smp_processor_id(), mask); - - if (!cpus_empty(mask)) - send_IPI_mask(&mask, vector); + send_IPI_mask_allbutself(cpu_online_mask, vector); } static inline void send_IPI_all(int vector) { - send_IPI_mask(&cpu_online_map, vector); + send_IPI_mask(cpu_online_mask, vector); } #endif /* __ASM_MACH_IPI_H */ diff --git a/arch/x86/include/asm/es7000/ipi.h b/arch/x86/include/asm/es7000/ipi.h index 1a8507265f91..7e8ed24d4b8a 100644 --- a/arch/x86/include/asm/es7000/ipi.h +++ b/arch/x86/include/asm/es7000/ipi.h @@ -1,25 +1,22 @@ #ifndef __ASM_ES7000_IPI_H #define __ASM_ES7000_IPI_H -void send_IPI_mask_sequence(const cpumask_t *mask, int vector); -void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); +void send_IPI_mask_sequence(const struct cpumask *mask, int vector); +void send_IPI_mask_allbutself(const struct cpumask *mask, int vector); -static inline void send_IPI_mask(const cpumask_t *mask, int vector) +static inline void send_IPI_mask(const struct cpumask *mask, int vector) { send_IPI_mask_sequence(mask, vector); } static inline void send_IPI_allbutself(int vector) { - cpumask_t mask = cpu_online_map; - cpu_clear(smp_processor_id(), mask); - if (!cpus_empty(mask)) - send_IPI_mask(&mask, vector); + send_IPI_mask_allbutself(cpu_online_mask, vector); } static inline void send_IPI_all(int vector) { - send_IPI_mask(&cpu_online_map, vector); + send_IPI_mask(cpu_online_mask, vector); } #endif /* __ASM_ES7000_IPI_H */ diff --git a/arch/x86/include/asm/genapic_32.h b/arch/x86/include/asm/genapic_32.h index eed6e305291f..746f37a7963a 100644 --- a/arch/x86/include/asm/genapic_32.h +++ b/arch/x86/include/asm/genapic_32.h @@ -24,7 +24,7 @@ struct genapic { int (*probe)(void); int (*apic_id_registered)(void); - const cpumask_t *(*target_cpus)(void); + const struct cpumask *(*target_cpus)(void); int int_delivery_mode; int int_dest_mode; int ESR_DISABLE; @@ -57,15 +57,16 @@ struct genapic { unsigned (*get_apic_id)(unsigned long x); unsigned long apic_id_mask; - unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask); + unsigned int (*cpu_mask_to_apicid)(const struct cpumask *cpumask); unsigned int (*cpu_mask_to_apicid_and)(const struct cpumask *cpumask, const struct cpumask *andmask); - void (*vector_allocation_domain)(int cpu, cpumask_t *retmask); + void (*vector_allocation_domain)(int cpu, struct cpumask *retmask); #ifdef CONFIG_SMP /* ipi */ - void (*send_IPI_mask)(const cpumask_t *mask, int vector); - void (*send_IPI_mask_allbutself)(const cpumask_t *mask, int vector); + void (*send_IPI_mask)(const struct cpumask *mask, int vector); + void (*send_IPI_mask_allbutself)(const struct cpumask *mask, + int vector); void (*send_IPI_allbutself)(int vector); void (*send_IPI_all)(int vector); #endif diff --git a/arch/x86/include/asm/genapic_64.h b/arch/x86/include/asm/genapic_64.h index 244b71729ecb..adf32fb56aa6 100644 --- a/arch/x86/include/asm/genapic_64.h +++ b/arch/x86/include/asm/genapic_64.h @@ -20,17 +20,18 @@ struct genapic { u32 int_delivery_mode; u32 int_dest_mode; int (*apic_id_registered)(void); - const cpumask_t *(*target_cpus)(void); - void (*vector_allocation_domain)(int cpu, cpumask_t *retmask); + const struct cpumask *(*target_cpus)(void); + void (*vector_allocation_domain)(int cpu, struct cpumask *retmask); void (*init_apic_ldr)(void); /* ipi */ - void (*send_IPI_mask)(const cpumask_t *mask, int vector); - void (*send_IPI_mask_allbutself)(const cpumask_t *mask, int vector); + void (*send_IPI_mask)(const struct cpumask *mask, int vector); + void (*send_IPI_mask_allbutself)(const struct cpumask *mask, + int vector); void (*send_IPI_allbutself)(int vector); void (*send_IPI_all)(int vector); void (*send_IPI_self)(int vector); /* */ - unsigned int (*cpu_mask_to_apicid)(const cpumask_t *cpumask); + unsigned int (*cpu_mask_to_apicid)(const struct cpumask *cpumask); unsigned int (*cpu_mask_to_apicid_and)(const struct cpumask *cpumask, const struct cpumask *andmask); unsigned int (*phys_pkg_id)(int index_msb); diff --git a/arch/x86/include/asm/ipi.h b/arch/x86/include/asm/ipi.h index 24b6e613edfa..c745a306f7d3 100644 --- a/arch/x86/include/asm/ipi.h +++ b/arch/x86/include/asm/ipi.h @@ -117,7 +117,8 @@ static inline void __send_IPI_dest_field(unsigned int mask, int vector, native_apic_mem_write(APIC_ICR, cfg); } -static inline void send_IPI_mask_sequence(const cpumask_t *mask, int vector) +static inline void send_IPI_mask_sequence(const struct cpumask *mask, + int vector) { unsigned long flags; unsigned long query_cpu; @@ -128,14 +129,15 @@ static inline void send_IPI_mask_sequence(const cpumask_t *mask, int vector) * - mbligh */ local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, *mask) { + for_each_cpu(query_cpu, mask) { __send_IPI_dest_field(per_cpu(x86_cpu_to_apicid, query_cpu), vector, APIC_DEST_PHYSICAL); } local_irq_restore(flags); } -static inline void send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +static inline void send_IPI_mask_allbutself(const struct cpumask *mask, + int vector) { unsigned long flags; unsigned int query_cpu; @@ -144,7 +146,7 @@ static inline void send_IPI_mask_allbutself(const cpumask_t *mask, int vector) /* See Hack comment above */ local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, *mask) + for_each_cpu(query_cpu, mask) if (query_cpu != this_cpu) __send_IPI_dest_field( per_cpu(x86_cpu_to_apicid, query_cpu), diff --git a/arch/x86/include/asm/mach-default/mach_apic.h b/arch/x86/include/asm/mach-default/mach_apic.h index df8e024c43c5..8863d978cb96 100644 --- a/arch/x86/include/asm/mach-default/mach_apic.h +++ b/arch/x86/include/asm/mach-default/mach_apic.h @@ -8,12 +8,12 @@ #define APIC_DFR_VALUE (APIC_DFR_FLAT) -static inline const cpumask_t *target_cpus(void) +static inline const struct cpumask *target_cpus(void) { #ifdef CONFIG_SMP - return &cpu_online_map; + return cpu_online_mask; #else - return &cpumask_of_cpu(0); + return cpumask_of(0); #endif } @@ -62,9 +62,9 @@ static inline int apic_id_registered(void) return physid_isset(read_apic_id(), phys_cpu_present_map); } -static inline unsigned int cpu_mask_to_apicid(const cpumask_t *cpumask) +static inline unsigned int cpu_mask_to_apicid(const struct cpumask *cpumask) { - return cpus_addr(*cpumask)[0]; + return cpumask_bits(cpumask)[0]; } static inline unsigned int cpu_mask_to_apicid_and(const struct cpumask *cpumask, @@ -98,7 +98,7 @@ static inline int apicid_to_node(int logical_apicid) #endif } -static inline void vector_allocation_domain(int cpu, cpumask_t *retmask) +static inline void vector_allocation_domain(int cpu, struct cpumask *retmask) { /* Careful. Some cpus do not strictly honor the set of cpus * specified in the interrupt destination when using lowest diff --git a/arch/x86/include/asm/mach-default/mach_ipi.h b/arch/x86/include/asm/mach-default/mach_ipi.h index 9353ab854a10..191312d155da 100644 --- a/arch/x86/include/asm/mach-default/mach_ipi.h +++ b/arch/x86/include/asm/mach-default/mach_ipi.h @@ -4,8 +4,8 @@ /* Avoid include hell */ #define NMI_VECTOR 0x02 -void send_IPI_mask_bitmask(const cpumask_t *mask, int vector); -void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); +void send_IPI_mask_bitmask(const struct cpumask *mask, int vector); +void send_IPI_mask_allbutself(const struct cpumask *mask, int vector); void __send_IPI_shortcut(unsigned int shortcut, int vector); extern int no_broadcast; @@ -15,17 +15,17 @@ extern int no_broadcast; #define send_IPI_mask (genapic->send_IPI_mask) #define send_IPI_mask_allbutself (genapic->send_IPI_mask_allbutself) #else -static inline void send_IPI_mask(const cpumask_t *mask, int vector) +static inline void send_IPI_mask(const struct cpumask *mask, int vector) { send_IPI_mask_bitmask(mask, vector); } -void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); +void send_IPI_mask_allbutself(const struct cpumask *mask, int vector); #endif static inline void __local_send_IPI_allbutself(int vector) { if (no_broadcast || vector == NMI_VECTOR) - send_IPI_mask_allbutself(&cpu_online_map, vector); + send_IPI_mask_allbutself(cpu_online_mask, vector); else __send_IPI_shortcut(APIC_DEST_ALLBUT, vector); } @@ -33,7 +33,7 @@ static inline void __local_send_IPI_allbutself(int vector) static inline void __local_send_IPI_all(int vector) { if (no_broadcast || vector == NMI_VECTOR) - send_IPI_mask(&cpu_online_map, vector); + send_IPI_mask(cpu_online_mask, vector); else __send_IPI_shortcut(APIC_DEST_ALLINC, vector); } diff --git a/arch/x86/include/asm/numaq/ipi.h b/arch/x86/include/asm/numaq/ipi.h index c734d7acc430..a8374c652778 100644 --- a/arch/x86/include/asm/numaq/ipi.h +++ b/arch/x86/include/asm/numaq/ipi.h @@ -1,26 +1,22 @@ #ifndef __ASM_NUMAQ_IPI_H #define __ASM_NUMAQ_IPI_H -void send_IPI_mask_sequence(const cpumask_t *mask, int vector); -void send_IPI_mask_allbutself(const cpumask_t *mask, int vector); +void send_IPI_mask_sequence(const struct cpumask *mask, int vector); +void send_IPI_mask_allbutself(const struct cpumask *mask, int vector); -static inline void send_IPI_mask(const cpumask_t *mask, int vector) +static inline void send_IPI_mask(const struct cpumask *mask, int vector) { send_IPI_mask_sequence(mask, vector); } static inline void send_IPI_allbutself(int vector) { - cpumask_t mask = cpu_online_map; - cpu_clear(smp_processor_id(), mask); - - if (!cpus_empty(mask)) - send_IPI_mask(&mask, vector); + send_IPI_mask_allbutself(cpu_online_mask, vector); } static inline void send_IPI_all(int vector) { - send_IPI_mask(&cpu_online_map, vector); + send_IPI_mask(cpu_online_mask, vector); } #endif /* __ASM_NUMAQ_IPI_H */ diff --git a/arch/x86/include/asm/smp.h b/arch/x86/include/asm/smp.h index c4a9aa52df6e..830b9fcb6427 100644 --- a/arch/x86/include/asm/smp.h +++ b/arch/x86/include/asm/smp.h @@ -60,7 +60,7 @@ struct smp_ops { void (*cpu_die)(unsigned int cpu); void (*play_dead)(void); - void (*send_call_func_ipi)(const cpumask_t *mask); + void (*send_call_func_ipi)(const struct cpumask *mask); void (*send_call_func_single_ipi)(int cpu); }; @@ -138,7 +138,7 @@ void native_cpu_die(unsigned int cpu); void native_play_dead(void); void play_dead_common(void); -void native_send_call_func_ipi(const cpumask_t *mask); +void native_send_call_func_ipi(const struct cpumask *mask); void native_send_call_func_single_ipi(int cpu); extern void prefill_possible_map(void); diff --git a/arch/x86/kernel/genapic_flat_64.c b/arch/x86/kernel/genapic_flat_64.c index c772bb10b173..7fa5f49c2dda 100644 --- a/arch/x86/kernel/genapic_flat_64.c +++ b/arch/x86/kernel/genapic_flat_64.c @@ -30,12 +30,12 @@ static int flat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) return 1; } -static const cpumask_t *flat_target_cpus(void) +static const struct cpumask *flat_target_cpus(void) { - return &cpu_online_map; + return cpu_online_mask; } -static void flat_vector_allocation_domain(int cpu, cpumask_t *retmask) +static void flat_vector_allocation_domain(int cpu, struct cpumask *retmask) { /* Careful. Some cpus do not strictly honor the set of cpus * specified in the interrupt destination when using lowest @@ -45,7 +45,8 @@ static void flat_vector_allocation_domain(int cpu, cpumask_t *retmask) * deliver interrupts to the wrong hyperthread when only one * hyperthread was specified in the interrupt desitination. */ - *retmask = (cpumask_t) { {[0] = APIC_ALL_CPUS, } }; + cpumask_clear(retmask); + cpumask_bits(retmask)[0] = APIC_ALL_CPUS; } /* @@ -77,16 +78,17 @@ static inline void _flat_send_IPI_mask(unsigned long mask, int vector) local_irq_restore(flags); } -static void flat_send_IPI_mask(const cpumask_t *cpumask, int vector) +static void flat_send_IPI_mask(const struct cpumask *cpumask, int vector) { - unsigned long mask = cpus_addr(*cpumask)[0]; + unsigned long mask = cpumask_bits(cpumask)[0]; _flat_send_IPI_mask(mask, vector); } -static void flat_send_IPI_mask_allbutself(const cpumask_t *cpumask, int vector) +static void flat_send_IPI_mask_allbutself(const struct cpumask *cpumask, + int vector) { - unsigned long mask = cpus_addr(*cpumask)[0]; + unsigned long mask = cpumask_bits(cpumask)[0]; int cpu = smp_processor_id(); if (cpu < BITS_PER_LONG) @@ -103,8 +105,8 @@ static void flat_send_IPI_allbutself(int vector) int hotplug = 0; #endif if (hotplug || vector == NMI_VECTOR) { - if (!cpus_equal(cpu_online_map, cpumask_of_cpu(cpu))) { - unsigned long mask = cpus_addr(cpu_online_map)[0]; + if (!cpumask_equal(cpu_online_mask, cpumask_of(cpu))) { + unsigned long mask = cpumask_bits(cpu_online_mask)[0]; if (cpu < BITS_PER_LONG) clear_bit(cpu, &mask); @@ -119,7 +121,7 @@ static void flat_send_IPI_allbutself(int vector) static void flat_send_IPI_all(int vector) { if (vector == NMI_VECTOR) - flat_send_IPI_mask(&cpu_online_map, vector); + flat_send_IPI_mask(cpu_online_mask, vector); else __send_IPI_shortcut(APIC_DEST_ALLINC, vector, APIC_DEST_LOGICAL); } @@ -153,9 +155,9 @@ static int flat_apic_id_registered(void) return physid_isset(read_xapic_id(), phys_cpu_present_map); } -static unsigned int flat_cpu_mask_to_apicid(const cpumask_t *cpumask) +static unsigned int flat_cpu_mask_to_apicid(const struct cpumask *cpumask) { - return cpus_addr(*cpumask)[0] & APIC_ALL_CPUS; + return cpumask_bits(cpumask)[0] & APIC_ALL_CPUS; } static unsigned int flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask, @@ -217,23 +219,23 @@ static int physflat_acpi_madt_oem_check(char *oem_id, char *oem_table_id) return 0; } -static const cpumask_t *physflat_target_cpus(void) +static const struct cpumask *physflat_target_cpus(void) { - return &cpu_online_map; + return cpu_online_mask; } -static void physflat_vector_allocation_domain(int cpu, cpumask_t *retmask) +static void physflat_vector_allocation_domain(int cpu, struct cpumask *retmask) { - cpus_clear(*retmask); - cpu_set(cpu, *retmask); + cpumask_clear(retmask); + cpumask_set_cpu(cpu, retmask); } -static void physflat_send_IPI_mask(const cpumask_t *cpumask, int vector) +static void physflat_send_IPI_mask(const struct cpumask *cpumask, int vector) { send_IPI_mask_sequence(cpumask, vector); } -static void physflat_send_IPI_mask_allbutself(const cpumask_t *cpumask, +static void physflat_send_IPI_mask_allbutself(const struct cpumask *cpumask, int vector) { send_IPI_mask_allbutself(cpumask, vector); @@ -241,15 +243,15 @@ static void physflat_send_IPI_mask_allbutself(const cpumask_t *cpumask, static void physflat_send_IPI_allbutself(int vector) { - send_IPI_mask_allbutself(&cpu_online_map, vector); + send_IPI_mask_allbutself(cpu_online_mask, vector); } static void physflat_send_IPI_all(int vector) { - physflat_send_IPI_mask(&cpu_online_map, vector); + physflat_send_IPI_mask(cpu_online_mask, vector); } -static unsigned int physflat_cpu_mask_to_apicid(const cpumask_t *cpumask) +static unsigned int physflat_cpu_mask_to_apicid(const struct cpumask *cpumask) { int cpu; @@ -257,7 +259,7 @@ static unsigned int physflat_cpu_mask_to_apicid(const cpumask_t *cpumask) * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - cpu = first_cpu(*cpumask); + cpu = cpumask_first(cpumask); if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_apicid, cpu); else diff --git a/arch/x86/kernel/genx2apic_cluster.c b/arch/x86/kernel/genx2apic_cluster.c index e7d16f53b9cd..4716a0c9f936 100644 --- a/arch/x86/kernel/genx2apic_cluster.c +++ b/arch/x86/kernel/genx2apic_cluster.c @@ -22,18 +22,18 @@ static int x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id) /* Start with all IRQs pointing to boot CPU. IRQ balancing will shift them. */ -static const cpumask_t *x2apic_target_cpus(void) +static const struct cpumask *x2apic_target_cpus(void) { - return &cpumask_of_cpu(0); + return cpumask_of(0); } /* * for now each logical cpu is in its own vector allocation domain. */ -static void x2apic_vector_allocation_domain(int cpu, cpumask_t *retmask) +static void x2apic_vector_allocation_domain(int cpu, struct cpumask *retmask) { - cpus_clear(*retmask); - cpu_set(cpu, *retmask); + cpumask_clear(retmask); + cpumask_set_cpu(cpu, retmask); } static void __x2apic_send_IPI_dest(unsigned int apicid, int vector, @@ -55,27 +55,28 @@ static void __x2apic_send_IPI_dest(unsigned int apicid, int vector, * at once. We have 16 cpu's in a cluster. This will minimize IPI register * writes. */ -static void x2apic_send_IPI_mask(const cpumask_t *mask, int vector) +static void x2apic_send_IPI_mask(const struct cpumask *mask, int vector) { unsigned long flags; unsigned long query_cpu; local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, *mask) + for_each_cpu(query_cpu, mask) __x2apic_send_IPI_dest( per_cpu(x86_cpu_to_logical_apicid, query_cpu), vector, APIC_DEST_LOGICAL); local_irq_restore(flags); } -static void x2apic_send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +static void x2apic_send_IPI_mask_allbutself(const struct cpumask *mask, + int vector) { unsigned long flags; unsigned long query_cpu; unsigned long this_cpu = smp_processor_id(); local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, *mask) + for_each_cpu(query_cpu, mask) if (query_cpu != this_cpu) __x2apic_send_IPI_dest( per_cpu(x86_cpu_to_logical_apicid, query_cpu), @@ -100,7 +101,7 @@ static void x2apic_send_IPI_allbutself(int vector) static void x2apic_send_IPI_all(int vector) { - x2apic_send_IPI_mask(&cpu_online_map, vector); + x2apic_send_IPI_mask(cpu_online_mask, vector); } static int x2apic_apic_id_registered(void) @@ -108,7 +109,7 @@ static int x2apic_apic_id_registered(void) return 1; } -static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) +static unsigned int x2apic_cpu_mask_to_apicid(const struct cpumask *cpumask) { int cpu; @@ -116,7 +117,7 @@ static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - cpu = first_cpu(*cpumask); + cpu = cpumask_first(cpumask); if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_logical_apicid, cpu); else diff --git a/arch/x86/kernel/genx2apic_phys.c b/arch/x86/kernel/genx2apic_phys.c index 9d0386c7e798..b255507884f2 100644 --- a/arch/x86/kernel/genx2apic_phys.c +++ b/arch/x86/kernel/genx2apic_phys.c @@ -29,15 +29,15 @@ static int x2apic_acpi_madt_oem_check(char *oem_id, char *oem_table_id) /* Start with all IRQs pointing to boot CPU. IRQ balancing will shift them. */ -static const cpumask_t *x2apic_target_cpus(void) +static const struct cpumask *x2apic_target_cpus(void) { - return &cpumask_of_cpu(0); + return cpumask_of(0); } -static void x2apic_vector_allocation_domain(int cpu, cpumask_t *retmask) +static void x2apic_vector_allocation_domain(int cpu, struct cpumask *retmask) { - cpus_clear(*retmask); - cpu_set(cpu, *retmask); + cpumask_clear(retmask); + cpumask_set_cpu(cpu, retmask); } static void __x2apic_send_IPI_dest(unsigned int apicid, int vector, @@ -53,27 +53,28 @@ static void __x2apic_send_IPI_dest(unsigned int apicid, int vector, x2apic_icr_write(cfg, apicid); } -static void x2apic_send_IPI_mask(const cpumask_t *mask, int vector) +static void x2apic_send_IPI_mask(const struct cpumask *mask, int vector) { unsigned long flags; unsigned long query_cpu; local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, *mask) { + for_each_cpu(query_cpu, mask) { __x2apic_send_IPI_dest(per_cpu(x86_cpu_to_apicid, query_cpu), vector, APIC_DEST_PHYSICAL); } local_irq_restore(flags); } -static void x2apic_send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +static void x2apic_send_IPI_mask_allbutself(const struct cpumask *mask, + int vector) { unsigned long flags; unsigned long query_cpu; unsigned long this_cpu = smp_processor_id(); local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, *mask) { + for_each_cpu(query_cpu, mask) { if (query_cpu != this_cpu) __x2apic_send_IPI_dest( per_cpu(x86_cpu_to_apicid, query_cpu), @@ -99,7 +100,7 @@ static void x2apic_send_IPI_allbutself(int vector) static void x2apic_send_IPI_all(int vector) { - x2apic_send_IPI_mask(&cpu_online_map, vector); + x2apic_send_IPI_mask(cpu_online_mask, vector); } static int x2apic_apic_id_registered(void) @@ -107,7 +108,7 @@ static int x2apic_apic_id_registered(void) return 1; } -static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) +static unsigned int x2apic_cpu_mask_to_apicid(const struct cpumask *cpumask) { int cpu; @@ -115,7 +116,7 @@ static unsigned int x2apic_cpu_mask_to_apicid(const cpumask_t *cpumask) * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - cpu = first_cpu(*cpumask); + cpu = cpumask_first(cpumask); if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_apicid, cpu); else diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 22596ec94c82..3984682cd849 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -75,15 +75,15 @@ EXPORT_SYMBOL(sn_rtc_cycles_per_second); /* Start with all IRQs pointing to boot CPU. IRQ balancing will shift them. */ -static const cpumask_t *uv_target_cpus(void) +static const struct cpumask *uv_target_cpus(void) { - return &cpumask_of_cpu(0); + return cpumask_of(0); } -static void uv_vector_allocation_domain(int cpu, cpumask_t *retmask) +static void uv_vector_allocation_domain(int cpu, struct cpumask *retmask) { - cpus_clear(*retmask); - cpu_set(cpu, *retmask); + cpumask_clear(retmask); + cpumask_set_cpu(cpu, retmask); } int uv_wakeup_secondary(int phys_apicid, unsigned int start_rip) @@ -122,20 +122,20 @@ static void uv_send_IPI_one(int cpu, int vector) uv_write_global_mmr64(pnode, UVH_IPI_INT, val); } -static void uv_send_IPI_mask(const cpumask_t *mask, int vector) +static void uv_send_IPI_mask(const struct cpumask *mask, int vector) { unsigned int cpu; - for_each_cpu_mask_nr(cpu, *mask) + for_each_cpu(cpu, mask) uv_send_IPI_one(cpu, vector); } -static void uv_send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +static void uv_send_IPI_mask_allbutself(const struct cpumask *mask, int vector) { unsigned int cpu; unsigned int this_cpu = smp_processor_id(); - for_each_cpu_mask_nr(cpu, *mask) + for_each_cpu(cpu, mask) if (cpu != this_cpu) uv_send_IPI_one(cpu, vector); } @@ -152,7 +152,7 @@ static void uv_send_IPI_allbutself(int vector) static void uv_send_IPI_all(int vector) { - uv_send_IPI_mask(&cpu_online_map, vector); + uv_send_IPI_mask(cpu_online_mask, vector); } static int uv_apic_id_registered(void) @@ -164,7 +164,7 @@ static void uv_init_apic_ldr(void) { } -static unsigned int uv_cpu_mask_to_apicid(const cpumask_t *cpumask) +static unsigned int uv_cpu_mask_to_apicid(const struct cpumask *cpumask) { int cpu; @@ -172,7 +172,7 @@ static unsigned int uv_cpu_mask_to_apicid(const cpumask_t *cpumask) * We're using fixed IRQ delivery, can only return one phys APIC ID. * May as well be the first. */ - cpu = first_cpu(*cpumask); + cpu = cpumask_first(cpumask); if ((unsigned)cpu < nr_cpu_ids) return per_cpu(x86_cpu_to_apicid, cpu); else diff --git a/arch/x86/kernel/ipi.c b/arch/x86/kernel/ipi.c index 86aa50fc65a1..285bbf8831fa 100644 --- a/arch/x86/kernel/ipi.c +++ b/arch/x86/kernel/ipi.c @@ -116,18 +116,18 @@ static inline void __send_IPI_dest_field(unsigned long mask, int vector) /* * This is only used on smaller machines. */ -void send_IPI_mask_bitmask(const cpumask_t *cpumask, int vector) +void send_IPI_mask_bitmask(const struct cpumask *cpumask, int vector) { - unsigned long mask = cpus_addr(*cpumask)[0]; + unsigned long mask = cpumask_bits(cpumask)[0]; unsigned long flags; local_irq_save(flags); - WARN_ON(mask & ~cpus_addr(cpu_online_map)[0]); + WARN_ON(mask & ~cpumask_bits(cpu_online_mask)[0]); __send_IPI_dest_field(mask, vector); local_irq_restore(flags); } -void send_IPI_mask_sequence(const cpumask_t *mask, int vector) +void send_IPI_mask_sequence(const struct cpumask *mask, int vector) { unsigned long flags; unsigned int query_cpu; @@ -139,12 +139,12 @@ void send_IPI_mask_sequence(const cpumask_t *mask, int vector) */ local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, *mask) + for_each_cpu(query_cpu, mask) __send_IPI_dest_field(cpu_to_logical_apicid(query_cpu), vector); local_irq_restore(flags); } -void send_IPI_mask_allbutself(const cpumask_t *mask, int vector) +void send_IPI_mask_allbutself(const struct cpumask *mask, int vector) { unsigned long flags; unsigned int query_cpu; @@ -153,7 +153,7 @@ void send_IPI_mask_allbutself(const cpumask_t *mask, int vector) /* See Hack comment above */ local_irq_save(flags); - for_each_cpu_mask_nr(query_cpu, *mask) + for_each_cpu(query_cpu, mask) if (query_cpu != this_cpu) __send_IPI_dest_field(cpu_to_logical_apicid(query_cpu), vector); diff --git a/arch/x86/kernel/smp.c b/arch/x86/kernel/smp.c index 341df946f9a9..49ed667b06f3 100644 --- a/arch/x86/kernel/smp.c +++ b/arch/x86/kernel/smp.c @@ -118,15 +118,15 @@ static void native_smp_send_reschedule(int cpu) WARN_ON(1); return; } - send_IPI_mask(&cpumask_of_cpu(cpu), RESCHEDULE_VECTOR); + send_IPI_mask(cpumask_of(cpu), RESCHEDULE_VECTOR); } void native_send_call_func_single_ipi(int cpu) { - send_IPI_mask(&cpumask_of_cpu(cpu), CALL_FUNCTION_SINGLE_VECTOR); + send_IPI_mask(cpumask_of(cpu), CALL_FUNCTION_SINGLE_VECTOR); } -void native_send_call_func_ipi(const cpumask_t *mask) +void native_send_call_func_ipi(const struct cpumask *mask) { cpumask_t allbutself; diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c index b3a95868839b..c44e2069c7c7 100644 --- a/arch/x86/xen/smp.c +++ b/arch/x86/xen/smp.c @@ -411,22 +411,23 @@ static void xen_smp_send_reschedule(int cpu) xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR); } -static void xen_send_IPI_mask(const cpumask_t *mask, enum ipi_vector vector) +static void xen_send_IPI_mask(const struct cpumask *mask, + enum ipi_vector vector) { unsigned cpu; - for_each_cpu_and(cpu, mask, &cpu_online_map) + for_each_cpu_and(cpu, mask, cpu_online_mask) xen_send_IPI_one(cpu, vector); } -static void xen_smp_send_call_function_ipi(const cpumask_t *mask) +static void xen_smp_send_call_function_ipi(const struct cpumask *mask) { int cpu; xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR); /* Make sure other vcpus get a chance to run if they need to. */ - for_each_cpu_mask_nr(cpu, *mask) { + for_each_cpu(cpu, mask) { if (xen_vcpu_stolen(cpu)) { HYPERVISOR_sched_op(SCHEDOP_yield, 0); break; @@ -436,7 +437,7 @@ static void xen_smp_send_call_function_ipi(const cpumask_t *mask) static void xen_smp_send_call_function_single_ipi(int cpu) { - xen_send_IPI_mask(&cpumask_of_cpu(cpu), + xen_send_IPI_mask(cpumask_of(cpu), XEN_CALL_FUNCTION_SINGLE_VECTOR); } From 78637a97b7fe1df51f40a460448df0b93d511176 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:34:00 -0800 Subject: [PATCH 10/16] x86: Set CONFIG_NR_CPUS even on UP Impact: cleanup Signed-off-by: Rusty Russell Signed-off-by: Mike Travis --- arch/x86/Kconfig | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 1fd44352f27c..4a3f5851ec64 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -599,12 +599,12 @@ config MAXSMP If unsure, say N. config NR_CPUS - depends on SMP int "Maximum number of CPUs" if SMP && !MAXSMP range 2 512 if SMP && !MAXSMP + default "1" if !SMP default "4096" if MAXSMP - default "32" if X86_NUMAQ || X86_SUMMIT || X86_BIGSMP || X86_ES7000 - default "8" + default "32" if SMP && (X86_NUMAQ || X86_SUMMIT || X86_BIGSMP || X86_ES7000) + default "8" if SMP help This allows you to specify the maximum number of CPUs which this kernel will support. The maximum supported value is 512 and the From 168ef543a43678146e06b3911e987ac021d575b8 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:34:01 -0800 Subject: [PATCH 11/16] x86: prepare for cpumask iterators to only go to nr_cpu_ids Impact: cleanup, futureproof In fact, all cpumask ops will only be valid (in general) for bit numbers < nr_cpu_ids. So use that instead of NR_CPUS in various places. This is always safe: no cpu number can be >= nr_cpu_ids, and nr_cpu_ids is initialized to NR_CPUS at boot. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis Acked-by: Ingo Molnar --- arch/x86/kernel/apic.c | 2 +- arch/x86/mach-voyager/voyager_smp.c | 2 +- arch/x86/mm/numa_64.c | 4 ++-- arch/x86/mm/srat_64.c | 2 +- 4 files changed, 5 insertions(+), 5 deletions(-) diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index a375791c08ca..3b630ec24935 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -2106,7 +2106,7 @@ __cpuinit int apic_is_clustered_box(void) bios_cpu_apicid = early_per_cpu_ptr(x86_bios_cpu_apicid); bitmap_zero(clustermap, NUM_APIC_CLUSTERS); - for (i = 0; i < NR_CPUS; i++) { + for (i = 0; i < nr_cpu_ids; i++) { /* are we being called early in kernel startup? */ if (bios_cpu_apicid) { id = bios_cpu_apicid[i]; diff --git a/arch/x86/mach-voyager/voyager_smp.c b/arch/x86/mach-voyager/voyager_smp.c index 9c990185e9f2..a5bc05492b1e 100644 --- a/arch/x86/mach-voyager/voyager_smp.c +++ b/arch/x86/mach-voyager/voyager_smp.c @@ -672,7 +672,7 @@ void __init smp_boot_cpus(void) /* loop over all the extended VIC CPUs and boot them. The * Quad CPUs must be bootstrapped by their extended VIC cpu */ - for (i = 0; i < NR_CPUS; i++) { + for (i = 0; i < nr_cpu_ids; i++) { if (i == boot_cpu_id || !cpu_isset(i, phys_cpu_present_map)) continue; do_boot_cpu(i); diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c index cebcbf152d46..71a14f89f89e 100644 --- a/arch/x86/mm/numa_64.c +++ b/arch/x86/mm/numa_64.c @@ -278,7 +278,7 @@ void __init numa_init_array(void) int rr, i; rr = first_node(node_online_map); - for (i = 0; i < NR_CPUS; i++) { + for (i = 0; i < nr_cpu_ids; i++) { if (early_cpu_to_node(i) != NUMA_NO_NODE) continue; numa_set_node(i, rr); @@ -549,7 +549,7 @@ void __init initmem_init(unsigned long start_pfn, unsigned long last_pfn) memnodemap[0] = 0; node_set_online(0); node_set(0, node_possible_map); - for (i = 0; i < NR_CPUS; i++) + for (i = 0; i < nr_cpu_ids; i++) numa_set_node(i, 0); e820_register_active_regions(0, start_pfn, last_pfn); setup_node_bootmem(0, start_pfn << PAGE_SHIFT, last_pfn << PAGE_SHIFT); diff --git a/arch/x86/mm/srat_64.c b/arch/x86/mm/srat_64.c index 51c0a2fc14fe..09737c8af074 100644 --- a/arch/x86/mm/srat_64.c +++ b/arch/x86/mm/srat_64.c @@ -382,7 +382,7 @@ int __init acpi_scan_nodes(unsigned long start, unsigned long end) if (!node_online(i)) setup_node_bootmem(i, nodes[i].start, nodes[i].end); - for (i = 0; i < NR_CPUS; i++) { + for (i = 0; i < nr_cpu_ids; i++) { int node = early_cpu_to_node(i); if (node == NUMA_NO_NODE) From 1de88cd4a33fcc2fcf70cbce01688723f728675d Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:34:02 -0800 Subject: [PATCH 12/16] x86: Use cpumask accessors code for possible/present maps. Impact: use new API Use the accessors rather than frobbing bits directly. Most of this is in arch code I haven't even compiled, but is straightforward. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis --- arch/x86/kernel/apic.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/x86/kernel/apic.c b/arch/x86/kernel/apic.c index 3b630ec24935..edda4c00e3d2 100644 --- a/arch/x86/kernel/apic.c +++ b/arch/x86/kernel/apic.c @@ -1903,8 +1903,8 @@ void __cpuinit generic_processor_info(int apicid, int version) } #endif - cpu_set(cpu, cpu_possible_map); - cpu_set(cpu, cpu_present_map); + set_cpu_possible(cpu, true); + set_cpu_present(cpu, true); } #ifdef CONFIG_X86_64 From b2bb85549134c005e997e5a7ed303bda6a1ae738 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:34:03 -0800 Subject: [PATCH 13/16] x86: Remove cpumask games in x86/kernel/cpu/intel_cacheinfo.c Impact: remove cpumask_t from stack. We should not try to save and restore cpus_allowed on current. We can't use work_on_cpu() here, since it's in the hotplug cpu path (if anyone else tries to get the hotplug lock from a workqueue we could deadlock against them). Fortunately, we can just use smp_call_function_single() since the function can run from an interrupt. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis Cc: Ingo Molnar Cc: Oleg Nesterov --- arch/x86/kernel/cpu/intel_cacheinfo.c | 45 +++++++++++++-------------- 1 file changed, 21 insertions(+), 24 deletions(-) diff --git a/arch/x86/kernel/cpu/intel_cacheinfo.c b/arch/x86/kernel/cpu/intel_cacheinfo.c index 43ea612d3e9d..fb7f946cb65e 100644 --- a/arch/x86/kernel/cpu/intel_cacheinfo.c +++ b/arch/x86/kernel/cpu/intel_cacheinfo.c @@ -534,12 +534,29 @@ static void __cpuinit free_cache_attributes(unsigned int cpu) per_cpu(cpuid4_info, cpu) = NULL; } +static void get_cpu_leaves(void *_retval) +{ + int j, *retval = _retval, cpu = smp_processor_id(); + + /* Do cpuid and store the results */ + for (j = 0; j < num_cache_leaves; j++) { + struct _cpuid4_info *this_leaf; + this_leaf = CPUID4_INFO_IDX(cpu, j); + *retval = cpuid4_cache_lookup(j, this_leaf); + if (unlikely(*retval < 0)) { + int i; + + for (i = 0; i < j; i++) + cache_remove_shared_cpu_map(cpu, i); + break; + } + cache_shared_cpu_map_setup(cpu, j); + } +} + static int __cpuinit detect_cache_attributes(unsigned int cpu) { - struct _cpuid4_info *this_leaf; - unsigned long j; int retval; - cpumask_t oldmask; if (num_cache_leaves == 0) return -ENOENT; @@ -549,27 +566,7 @@ static int __cpuinit detect_cache_attributes(unsigned int cpu) if (per_cpu(cpuid4_info, cpu) == NULL) return -ENOMEM; - oldmask = current->cpus_allowed; - retval = set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu)); - if (retval) - goto out; - - /* Do cpuid and store the results */ - for (j = 0; j < num_cache_leaves; j++) { - this_leaf = CPUID4_INFO_IDX(cpu, j); - retval = cpuid4_cache_lookup(j, this_leaf); - if (unlikely(retval < 0)) { - int i; - - for (i = 0; i < j; i++) - cache_remove_shared_cpu_map(cpu, i); - break; - } - cache_shared_cpu_map_setup(cpu, j); - } - set_cpus_allowed_ptr(current, &oldmask); - -out: + smp_call_function_single(cpu, get_cpu_leaves, &retval, true); if (retval) { kfree(per_cpu(cpuid4_info, cpu)); per_cpu(cpuid4_info, cpu) = NULL; From 4cd4601d592d07b26e4b7d2bb8fcd55bbfd6cf6e Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:34:04 -0800 Subject: [PATCH 14/16] x86: use work_on_cpu in x86/kernel/cpu/mcheck/mce_amd_64.c Impact: Remove cpumask_t's from stack. Simple transition to work_on_cpu(), rather than cpumask games. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis Cc: Ingo Molnar Cc: Robert Richter Cc: jacob.shin@amd.com --- arch/x86/kernel/cpu/mcheck/mce_amd_64.c | 108 ++++++++++++------------ 1 file changed, 55 insertions(+), 53 deletions(-) diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c index 5eb390a4b2e9..a1de80f368f1 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c @@ -83,34 +83,41 @@ static DEFINE_PER_CPU(unsigned char, bank_map); /* see which banks are on */ * CPU Initialization */ +struct thresh_restart { + struct threshold_block *b; + int reset; + u16 old_limit; +}; + /* must be called with correct cpu affinity */ -static void threshold_restart_bank(struct threshold_block *b, - int reset, u16 old_limit) +static long threshold_restart_bank(void *_tr) { + struct thresh_restart *tr = _tr; u32 mci_misc_hi, mci_misc_lo; - rdmsr(b->address, mci_misc_lo, mci_misc_hi); + rdmsr(tr->b->address, mci_misc_lo, mci_misc_hi); - if (b->threshold_limit < (mci_misc_hi & THRESHOLD_MAX)) - reset = 1; /* limit cannot be lower than err count */ + if (tr->b->threshold_limit < (mci_misc_hi & THRESHOLD_MAX)) + tr->reset = 1; /* limit cannot be lower than err count */ - if (reset) { /* reset err count and overflow bit */ + if (tr->reset) { /* reset err count and overflow bit */ mci_misc_hi = (mci_misc_hi & ~(MASK_ERR_COUNT_HI | MASK_OVERFLOW_HI)) | - (THRESHOLD_MAX - b->threshold_limit); - } else if (old_limit) { /* change limit w/o reset */ + (THRESHOLD_MAX - tr->b->threshold_limit); + } else if (tr->old_limit) { /* change limit w/o reset */ int new_count = (mci_misc_hi & THRESHOLD_MAX) + - (old_limit - b->threshold_limit); + (tr->old_limit - tr->b->threshold_limit); mci_misc_hi = (mci_misc_hi & ~MASK_ERR_COUNT_HI) | (new_count & THRESHOLD_MAX); } - b->interrupt_enable ? + tr->b->interrupt_enable ? (mci_misc_hi = (mci_misc_hi & ~MASK_INT_TYPE_HI) | INT_TYPE_APIC) : (mci_misc_hi &= ~MASK_INT_TYPE_HI); mci_misc_hi |= MASK_COUNT_EN_HI; - wrmsr(b->address, mci_misc_lo, mci_misc_hi); + wrmsr(tr->b->address, mci_misc_lo, mci_misc_hi); + return 0; } /* cpu init entry point, called from mce.c with preempt off */ @@ -120,6 +127,7 @@ void __cpuinit mce_amd_feature_init(struct cpuinfo_x86 *c) unsigned int cpu = smp_processor_id(); u8 lvt_off; u32 low = 0, high = 0, address = 0; + struct thresh_restart tr; for (bank = 0; bank < NR_BANKS; ++bank) { for (block = 0; block < NR_BLOCKS; ++block) { @@ -162,7 +170,10 @@ void __cpuinit mce_amd_feature_init(struct cpuinfo_x86 *c) wrmsr(address, low, high); threshold_defaults.address = address; - threshold_restart_bank(&threshold_defaults, 0, 0); + tr.b = &threshold_defaults; + tr.reset = 0; + tr.old_limit = 0; + threshold_restart_bank(&tr); } } } @@ -251,20 +262,6 @@ struct threshold_attr { ssize_t(*store) (struct threshold_block *, const char *, size_t count); }; -static void affinity_set(unsigned int cpu, cpumask_t *oldmask, - cpumask_t *newmask) -{ - *oldmask = current->cpus_allowed; - cpus_clear(*newmask); - cpu_set(cpu, *newmask); - set_cpus_allowed_ptr(current, newmask); -} - -static void affinity_restore(const cpumask_t *oldmask) -{ - set_cpus_allowed_ptr(current, oldmask); -} - #define SHOW_FIELDS(name) \ static ssize_t show_ ## name(struct threshold_block * b, char *buf) \ { \ @@ -277,15 +274,16 @@ static ssize_t store_interrupt_enable(struct threshold_block *b, const char *buf, size_t count) { char *end; - cpumask_t oldmask, newmask; + struct thresh_restart tr; unsigned long new = simple_strtoul(buf, &end, 0); if (end == buf) return -EINVAL; b->interrupt_enable = !!new; - affinity_set(b->cpu, &oldmask, &newmask); - threshold_restart_bank(b, 0, 0); - affinity_restore(&oldmask); + tr.b = b; + tr.reset = 0; + tr.old_limit = 0; + work_on_cpu(b->cpu, threshold_restart_bank, &tr); return end - buf; } @@ -294,8 +292,7 @@ static ssize_t store_threshold_limit(struct threshold_block *b, const char *buf, size_t count) { char *end; - cpumask_t oldmask, newmask; - u16 old; + struct thresh_restart tr; unsigned long new = simple_strtoul(buf, &end, 0); if (end == buf) return -EINVAL; @@ -303,34 +300,36 @@ static ssize_t store_threshold_limit(struct threshold_block *b, new = THRESHOLD_MAX; if (new < 1) new = 1; - old = b->threshold_limit; + tr.old_limit = b->threshold_limit; b->threshold_limit = new; + tr.b = b; + tr.reset = 0; - affinity_set(b->cpu, &oldmask, &newmask); - threshold_restart_bank(b, 0, old); - affinity_restore(&oldmask); + work_on_cpu(b->cpu, threshold_restart_bank, &tr); return end - buf; } +static long local_error_count(void *_b) +{ + struct threshold_block *b = _b; + u32 low, high; + + rdmsr(b->address, low, high); + return (high & 0xFFF) - (THRESHOLD_MAX - b->threshold_limit); +} + static ssize_t show_error_count(struct threshold_block *b, char *buf) { - u32 high, low; - cpumask_t oldmask, newmask; - affinity_set(b->cpu, &oldmask, &newmask); - rdmsr(b->address, low, high); - affinity_restore(&oldmask); - return sprintf(buf, "%x\n", - (high & 0xFFF) - (THRESHOLD_MAX - b->threshold_limit)); + return sprintf(buf, "%lx\n", work_on_cpu(b->cpu, local_error_count, b)); } static ssize_t store_error_count(struct threshold_block *b, const char *buf, size_t count) { - cpumask_t oldmask, newmask; - affinity_set(b->cpu, &oldmask, &newmask); - threshold_restart_bank(b, 1, 0); - affinity_restore(&oldmask); + struct thresh_restart tr = { .b = b, .reset = 1, .old_limit = 0 }; + + work_on_cpu(b->cpu, threshold_restart_bank, &tr); return 1; } @@ -463,12 +462,19 @@ out_free: return err; } +static long local_allocate_threshold_blocks(void *_bank) +{ + unsigned int *bank = _bank; + + return allocate_threshold_blocks(smp_processor_id(), *bank, 0, + MSR_IA32_MC0_MISC + *bank * 4); +} + /* symlinks sibling shared banks to first core. first core owns dir/files. */ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) { int i, err = 0; struct threshold_bank *b = NULL; - cpumask_t oldmask, newmask; char name[32]; sprintf(name, "threshold_bank%i", bank); @@ -519,11 +525,7 @@ static __cpuinit int threshold_create_bank(unsigned int cpu, unsigned int bank) per_cpu(threshold_banks, cpu)[bank] = b; - affinity_set(cpu, &oldmask, &newmask); - err = allocate_threshold_blocks(cpu, bank, 0, - MSR_IA32_MC0_MISC + bank * 4); - affinity_restore(&oldmask); - + err = work_on_cpu(cpu, local_allocate_threshold_blocks, &bank); if (err) goto out_free; From e4d98207ea3f3d15eb664282df16d18c4ac86f80 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:34:05 -0800 Subject: [PATCH 15/16] x86: xen: use smp_call_function_many() Impact: use new API, remove cpumask from stack. Change smp_call_function_mask() callers to smp_call_function_many(). This removes a cpumask from the stack, and falls back should allocating the cpumask var fail (only possible with CONFIG_CPUMASKS_OFFSTACK). Signed-off-by: Rusty Russell Signed-off-by: Mike Travis Cc: jeremy@xensource.com --- arch/x86/xen/mmu.c | 20 +++++++++++++++----- 1 file changed, 15 insertions(+), 5 deletions(-) diff --git a/arch/x86/xen/mmu.c b/arch/x86/xen/mmu.c index 636ef4caa52d..e59e53b11e2b 100644 --- a/arch/x86/xen/mmu.c +++ b/arch/x86/xen/mmu.c @@ -1079,7 +1079,7 @@ static void drop_other_mm_ref(void *info) static void xen_drop_mm_ref(struct mm_struct *mm) { - cpumask_t mask; + cpumask_var_t mask; unsigned cpu; if (current->active_mm == mm) { @@ -1091,7 +1091,16 @@ static void xen_drop_mm_ref(struct mm_struct *mm) } /* Get the "official" set of cpus referring to our pagetable. */ - mask = mm->cpu_vm_mask; + if (!alloc_cpumask_var(&mask, GFP_ATOMIC)) { + for_each_online_cpu(cpu) { + if (!cpumask_test_cpu(cpu, &mm->cpu_vm_mask) + && per_cpu(xen_current_cr3, cpu) != __pa(mm->pgd)) + continue; + smp_call_function_single(cpu, drop_other_mm_ref, mm, 1); + } + return; + } + cpumask_copy(mask, &mm->cpu_vm_mask); /* It's possible that a vcpu may have a stale reference to our cr3, because its in lazy mode, and it hasn't yet flushed @@ -1100,11 +1109,12 @@ static void xen_drop_mm_ref(struct mm_struct *mm) if needed. */ for_each_online_cpu(cpu) { if (per_cpu(xen_current_cr3, cpu) == __pa(mm->pgd)) - cpu_set(cpu, mask); + cpumask_set_cpu(cpu, mask); } - if (!cpus_empty(mask)) - smp_call_function_mask(mask, drop_other_mm_ref, mm, 1); + if (!cpumask_empty(mask)) + smp_call_function_many(mask, drop_other_mm_ref, mm, 1); + free_cpumask_var(mask); } #else static void xen_drop_mm_ref(struct mm_struct *mm) From 83b19597f793fd5f91533bda0dc2eb3d21936798 Mon Sep 17 00:00:00 2001 From: Mike Travis Date: Tue, 16 Dec 2008 17:34:06 -0800 Subject: [PATCH 16/16] x86: Introduce topology_core_cpumask()/topology_thread_cpumask() Impact: new API The old topology_core_siblings() and topology_thread_siblings() return a cpumask_t; these new ones return a (const) struct cpumask *. Signed-off-by: Rusty Russell Signed-off-by: Mike Travis --- arch/x86/include/asm/topology.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/arch/x86/include/asm/topology.h b/arch/x86/include/asm/topology.h index ff386ff50ed7..79e31e9dcdda 100644 --- a/arch/x86/include/asm/topology.h +++ b/arch/x86/include/asm/topology.h @@ -226,6 +226,8 @@ extern cpumask_t cpu_coregroup_map(int cpu); #define topology_core_id(cpu) (cpu_data(cpu).cpu_core_id) #define topology_core_siblings(cpu) (per_cpu(cpu_core_map, cpu)) #define topology_thread_siblings(cpu) (per_cpu(cpu_sibling_map, cpu)) +#define topology_core_cpumask(cpu) (&per_cpu(cpu_core_map, cpu)) +#define topology_thread_cpumask(cpu) (&per_cpu(cpu_sibling_map, cpu)) /* indicates that pointers to the topology cpumask_t maps are valid */ #define arch_provides_topology_pointers yes