From c10d38dda1774ed4540380333cabd229eff37094 Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Wed, 10 Sep 2008 13:37:17 +0200 Subject: x86: some lock annotations for user copy paths copy_to/from_user and all its variants (except the atomic ones) can take a page fault and perform non-trivial work like taking mmap_sem and entering the filesyste/pagecache. Unfortunately, this often escapes lockdep because a common pattern is to use it to read in some arguments just set up from userspace, or write data back to a hot buffer. In those cases, it will be unlikely for page reclaim to get a window in to cause copy_*_user to fault. With the new might_lock primitives, add some annotations to x86. I don't know if I caught all possible faulting points (it's a bit of a maze, and I didn't really look at 32-bit). But this is a starting point. Boots and runs OK so far. Signed-off-by: Nick Piggin Acked-by: Peter Zijlstra Signed-off-by: Ingo Molnar --- arch/x86/lib/usercopy_32.c | 7 ++++++- arch/x86/lib/usercopy_64.c | 4 ++++ 2 files changed, 10 insertions(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/lib/usercopy_32.c b/arch/x86/lib/usercopy_32.c index 24e60944971a..8eedde2a9cac 100644 --- a/arch/x86/lib/usercopy_32.c +++ b/arch/x86/lib/usercopy_32.c @@ -33,6 +33,8 @@ static inline int __movsl_is_ok(unsigned long a1, unsigned long a2, unsigned lon do { \ int __d0, __d1, __d2; \ might_sleep(); \ + if (current->mm) \ + might_lock_read(¤t->mm->mmap_sem); \ __asm__ __volatile__( \ " testl %1,%1\n" \ " jz 2f\n" \ @@ -120,6 +122,8 @@ EXPORT_SYMBOL(strncpy_from_user); do { \ int __d0; \ might_sleep(); \ + if (current->mm) \ + might_lock_read(¤t->mm->mmap_sem); \ __asm__ __volatile__( \ "0: rep; stosl\n" \ " movl %2,%0\n" \ @@ -148,7 +152,6 @@ do { \ unsigned long clear_user(void __user *to, unsigned long n) { - might_sleep(); if (access_ok(VERIFY_WRITE, to, n)) __do_clear_user(to, n); return n; @@ -191,6 +194,8 @@ long strnlen_user(const char __user *s, long n) unsigned long res, tmp; might_sleep(); + if (current->mm) + might_lock_read(¤t->mm->mmap_sem); __asm__ __volatile__( " testl %0, %0\n" diff --git a/arch/x86/lib/usercopy_64.c b/arch/x86/lib/usercopy_64.c index f4df6e7c718b..847d12945998 100644 --- a/arch/x86/lib/usercopy_64.c +++ b/arch/x86/lib/usercopy_64.c @@ -16,6 +16,8 @@ do { \ long __d0, __d1, __d2; \ might_sleep(); \ + if (current->mm) \ + might_lock_read(¤t->mm->mmap_sem); \ __asm__ __volatile__( \ " testq %1,%1\n" \ " jz 2f\n" \ @@ -65,6 +67,8 @@ unsigned long __clear_user(void __user *addr, unsigned long size) { long __d0; might_sleep(); + if (current->mm) + might_lock_read(¤t->mm->mmap_sem); /* no memory constraint because it doesn't change any memory gcc knows about */ asm volatile( -- cgit v1.2.3 From 3ee1afa308f2a38e5d1e2ad3752ad7abcf480da1 Mon Sep 17 00:00:00 2001 From: Nick Piggin Date: Wed, 10 Sep 2008 13:37:17 +0200 Subject: x86: some lock annotations for user copy paths, v2 - introduce might_fault() - handle the atomic user copy paths correctly [ mingo@elte.hu: move might_sleep() outside of in_atomic(). ] Signed-off-by: Nick Piggin Acked-by: Peter Zijlstra Signed-off-by: Ingo Molnar --- arch/x86/lib/usercopy_32.c | 12 +++--------- arch/x86/lib/usercopy_64.c | 8 ++------ include/asm-x86/uaccess.h | 18 ++++-------------- include/asm-x86/uaccess_32.h | 12 +++--------- include/asm-x86/uaccess_64.h | 12 +++--------- include/linux/kernel.h | 9 +++++++++ mm/memory.c | 15 +++++++++++++++ 7 files changed, 39 insertions(+), 47 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/lib/usercopy_32.c b/arch/x86/lib/usercopy_32.c index 8eedde2a9cac..7393152a252e 100644 --- a/arch/x86/lib/usercopy_32.c +++ b/arch/x86/lib/usercopy_32.c @@ -32,9 +32,7 @@ static inline int __movsl_is_ok(unsigned long a1, unsigned long a2, unsigned lon #define __do_strncpy_from_user(dst, src, count, res) \ do { \ int __d0, __d1, __d2; \ - might_sleep(); \ - if (current->mm) \ - might_lock_read(¤t->mm->mmap_sem); \ + might_fault(); \ __asm__ __volatile__( \ " testl %1,%1\n" \ " jz 2f\n" \ @@ -121,9 +119,7 @@ EXPORT_SYMBOL(strncpy_from_user); #define __do_clear_user(addr,size) \ do { \ int __d0; \ - might_sleep(); \ - if (current->mm) \ - might_lock_read(¤t->mm->mmap_sem); \ + might_fault(); \ __asm__ __volatile__( \ "0: rep; stosl\n" \ " movl %2,%0\n" \ @@ -193,9 +189,7 @@ long strnlen_user(const char __user *s, long n) unsigned long mask = -__addr_ok(s); unsigned long res, tmp; - might_sleep(); - if (current->mm) - might_lock_read(¤t->mm->mmap_sem); + might_fault(); __asm__ __volatile__( " testl %0, %0\n" diff --git a/arch/x86/lib/usercopy_64.c b/arch/x86/lib/usercopy_64.c index 847d12945998..64d6c84e6353 100644 --- a/arch/x86/lib/usercopy_64.c +++ b/arch/x86/lib/usercopy_64.c @@ -15,9 +15,7 @@ #define __do_strncpy_from_user(dst,src,count,res) \ do { \ long __d0, __d1, __d2; \ - might_sleep(); \ - if (current->mm) \ - might_lock_read(¤t->mm->mmap_sem); \ + might_fault(); \ __asm__ __volatile__( \ " testq %1,%1\n" \ " jz 2f\n" \ @@ -66,9 +64,7 @@ EXPORT_SYMBOL(strncpy_from_user); unsigned long __clear_user(void __user *addr, unsigned long size) { long __d0; - might_sleep(); - if (current->mm) - might_lock_read(¤t->mm->mmap_sem); + might_fault(); /* no memory constraint because it doesn't change any memory gcc knows about */ asm volatile( diff --git a/include/asm-x86/uaccess.h b/include/asm-x86/uaccess.h index ad29752a1713..39f8420c75d9 100644 --- a/include/asm-x86/uaccess.h +++ b/include/asm-x86/uaccess.h @@ -8,8 +8,6 @@ #include #include #include -#include -#include #include #include @@ -159,9 +157,7 @@ extern int __get_user_bad(void); int __ret_gu; \ unsigned long __val_gu; \ __chk_user_ptr(ptr); \ - might_sleep(); \ - if (current->mm) \ - might_lock_read(¤t->mm->mmap_sem); \ + might_fault(); \ switch (sizeof(*(ptr))) { \ case 1: \ __get_user_x(1, __ret_gu, __val_gu, ptr); \ @@ -246,9 +242,7 @@ extern void __put_user_8(void); int __ret_pu; \ __typeof__(*(ptr)) __pu_val; \ __chk_user_ptr(ptr); \ - might_sleep(); \ - if (current->mm) \ - might_lock_read(¤t->mm->mmap_sem); \ + might_fault(); \ __pu_val = x; \ switch (sizeof(*(ptr))) { \ case 1: \ @@ -273,9 +267,7 @@ extern void __put_user_8(void); #define __put_user_size(x, ptr, size, retval, errret) \ do { \ retval = 0; \ - might_sleep(); \ - if (current->mm) \ - might_lock_read(¤t->mm->mmap_sem); \ + might_fault(); \ __chk_user_ptr(ptr); \ switch (size) { \ case 1: \ @@ -328,9 +320,7 @@ do { \ #define __get_user_size(x, ptr, size, retval, errret) \ do { \ retval = 0; \ - might_sleep(); \ - if (current->mm) \ - might_lock_read(¤t->mm->mmap_sem); \ + might_fault(); \ __chk_user_ptr(ptr); \ switch (size) { \ case 1: \ diff --git a/include/asm-x86/uaccess_32.h b/include/asm-x86/uaccess_32.h index d725e2d703f7..d10e842ec3ee 100644 --- a/include/asm-x86/uaccess_32.h +++ b/include/asm-x86/uaccess_32.h @@ -82,9 +82,7 @@ __copy_to_user_inatomic(void __user *to, const void *from, unsigned long n) static __always_inline unsigned long __must_check __copy_to_user(void __user *to, const void *from, unsigned long n) { - might_sleep(); - if (current->mm) - might_lock_read(¤t->mm->mmap_sem); + might_fault(); return __copy_to_user_inatomic(to, from, n); } @@ -139,9 +137,7 @@ __copy_from_user_inatomic(void *to, const void __user *from, unsigned long n) static __always_inline unsigned long __copy_from_user(void *to, const void __user *from, unsigned long n) { - might_sleep(); - if (current->mm) - might_lock_read(¤t->mm->mmap_sem); + might_fault(); if (__builtin_constant_p(n)) { unsigned long ret; @@ -163,9 +159,7 @@ __copy_from_user(void *to, const void __user *from, unsigned long n) static __always_inline unsigned long __copy_from_user_nocache(void *to, const void __user *from, unsigned long n) { - might_sleep(); - if (current->mm) - might_lock_read(¤t->mm->mmap_sem); + might_fault(); if (__builtin_constant_p(n)) { unsigned long ret; diff --git a/include/asm-x86/uaccess_64.h b/include/asm-x86/uaccess_64.h index 40a7205fe576..13fd56fbc3ab 100644 --- a/include/asm-x86/uaccess_64.h +++ b/include/asm-x86/uaccess_64.h @@ -29,9 +29,7 @@ int __copy_from_user(void *dst, const void __user *src, unsigned size) { int ret = 0; - might_sleep(); - if (current->mm) - might_lock_read(¤t->mm->mmap_sem); + might_fault(); if (!__builtin_constant_p(size)) return copy_user_generic(dst, (__force void *)src, size); switch (size) { @@ -75,9 +73,7 @@ int __copy_to_user(void __user *dst, const void *src, unsigned size) { int ret = 0; - might_sleep(); - if (current->mm) - might_lock_read(¤t->mm->mmap_sem); + might_fault(); if (!__builtin_constant_p(size)) return copy_user_generic((__force void *)dst, src, size); switch (size) { @@ -121,9 +117,7 @@ int __copy_in_user(void __user *dst, const void __user *src, unsigned size) { int ret = 0; - might_sleep(); - if (current->mm) - might_lock_read(¤t->mm->mmap_sem); + might_fault(); if (!__builtin_constant_p(size)) return copy_user_generic((__force void *)dst, (__force void *)src, size); diff --git a/include/linux/kernel.h b/include/linux/kernel.h index 2651f805ba6d..e580ec095765 100644 --- a/include/linux/kernel.h +++ b/include/linux/kernel.h @@ -140,6 +140,15 @@ extern int _cond_resched(void); (__x < 0) ? -__x : __x; \ }) +#ifdef CONFIG_PROVE_LOCKING +void might_fault(void); +#else +static inline void might_fault(void) +{ + might_sleep(); +} +#endif + extern struct atomic_notifier_head panic_notifier_list; extern long (*panic_blink)(long time); NORET_TYPE void panic(const char * fmt, ...) diff --git a/mm/memory.c b/mm/memory.c index 1002f473f497..b8fdf4e5e65b 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -3016,3 +3016,18 @@ void print_vma_addr(char *prefix, unsigned long ip) } up_read(¤t->mm->mmap_sem); } + +#ifdef CONFIG_PROVE_LOCKING +void might_fault(void) +{ + might_sleep(); + /* + * it would be nicer only to annotate paths which are not under + * pagefault_disable, however that requires a larger audit and + * providing helpers like get_user_atomic. + */ + if (!in_atomic() && current->mm) + might_lock_read(¤t->mm->mmap_sem); +} +EXPORT_SYMBOL(might_fault); +#endif -- cgit v1.2.3 From 1d18ef489509314506328b9e464dd47c24c1d68f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 11 Sep 2008 20:53:21 +0200 Subject: x86: some lock annotations for user copy paths, v3 - add annotation back to clear_user() - change probe_kernel_address() to _inatomic*() method Signed-off-by: Ingo Molnar --- arch/x86/lib/usercopy_32.c | 1 + include/asm-x86/uaccess.h | 2 -- include/linux/uaccess.h | 2 +- 3 files changed, 2 insertions(+), 3 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/lib/usercopy_32.c b/arch/x86/lib/usercopy_32.c index 7393152a252e..fab5faba1d3e 100644 --- a/arch/x86/lib/usercopy_32.c +++ b/arch/x86/lib/usercopy_32.c @@ -148,6 +148,7 @@ do { \ unsigned long clear_user(void __user *to, unsigned long n) { + might_fault(); if (access_ok(VERIFY_WRITE, to, n)) __do_clear_user(to, n); return n; diff --git a/include/asm-x86/uaccess.h b/include/asm-x86/uaccess.h index 39f8420c75d9..dc8edb5c4659 100644 --- a/include/asm-x86/uaccess.h +++ b/include/asm-x86/uaccess.h @@ -267,7 +267,6 @@ extern void __put_user_8(void); #define __put_user_size(x, ptr, size, retval, errret) \ do { \ retval = 0; \ - might_fault(); \ __chk_user_ptr(ptr); \ switch (size) { \ case 1: \ @@ -320,7 +319,6 @@ do { \ #define __get_user_size(x, ptr, size, retval, errret) \ do { \ retval = 0; \ - might_fault(); \ __chk_user_ptr(ptr); \ switch (size) { \ case 1: \ diff --git a/include/linux/uaccess.h b/include/linux/uaccess.h index fec6decfb983..2062293e57e6 100644 --- a/include/linux/uaccess.h +++ b/include/linux/uaccess.h @@ -78,7 +78,7 @@ static inline unsigned long __copy_from_user_nocache(void *to, \ set_fs(KERNEL_DS); \ pagefault_disable(); \ - ret = __get_user(retval, (__force typeof(retval) __user *)(addr)); \ + ret = __copy_from_user_inatomic((__force typeof(retval) __user *)(addr), &(retval), sizeof(retval)); \ pagefault_enable(); \ set_fs(old_fs); \ ret; \ -- cgit v1.2.3 From e45f2c07742d447597df001c878bc4a8aafcde37 Mon Sep 17 00:00:00 2001 From: "Denis V. Lunev" Date: Mon, 24 Nov 2008 11:28:36 +0300 Subject: x86: correct link to HPET timer specification Impact: update documentation / help text Original link is dead. Signed-off-by: Denis V. Lunev Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 2 +- drivers/char/hpet.c | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index ac22bb7719f7..19f0d97829ee 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -482,7 +482,7 @@ config HPET_TIMER The HPET provides a stable time base on SMP systems, unlike the TSC, but it is more expensive to access, as it is off-chip. You can find the HPET spec at - . + . You can safely choose Y here. However, HPET will only be activated if the platform and the BIOS support this feature. diff --git a/drivers/char/hpet.c b/drivers/char/hpet.c index 53fdc7ff3870..32b8bbf5003e 100644 --- a/drivers/char/hpet.c +++ b/drivers/char/hpet.c @@ -46,7 +46,7 @@ /* * The High Precision Event Timer driver. * This driver is closely modelled after the rtc.c driver. - * http://www.intel.com/hardwaredesign/hpetspec.htm + * http://www.intel.com/hardwaredesign/hpetspec_1.pdf */ #define HPET_USER_FREQ (64) #define HPET_DRIFT (500) -- cgit v1.2.3 From 0b8f1efad30bd58f89961b82dfe68b9edf8fd2ac Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 18:58:31 -0800 Subject: sparse irq_desc[] array: core kernel and x86 changes Impact: new feature Problem on distro kernels: irq_desc[NR_IRQS] takes megabytes of RAM with NR_CPUS set to large values. The goal is to be able to scale up to much larger NR_IRQS value without impacting the (important) common case. To solve this, we generalize irq_desc[NR_IRQS] to an (optional) array of irq_desc pointers. When CONFIG_SPARSE_IRQ=y is used, we use kzalloc_node to get irq_desc, this also makes the IRQ descriptors NUMA-local (to the site that calls request_irq()). This gets rid of the irq_cfg[] static array on x86 as well: irq_cfg now uses desc->chip_data for x86 to store irq_cfg. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 10 ++ arch/x86/include/asm/irq_vectors.h | 9 ++ arch/x86/kernel/io_apic.c | 301 +++++++++++++++++++++++-------------- arch/x86/kernel/irq.c | 3 + arch/x86/kernel/irq_32.c | 2 + arch/x86/kernel/irq_64.c | 2 + arch/x86/kernel/irqinit_32.c | 1 - arch/x86/kernel/irqinit_64.c | 1 - drivers/char/random.c | 22 +-- drivers/pci/intr_remapping.c | 76 +++++++++- drivers/xen/events.c | 12 +- fs/proc/stat.c | 17 ++- include/linux/interrupt.h | 2 + include/linux/irq.h | 54 ++++++- include/linux/irqnr.h | 14 +- include/linux/kernel_stat.h | 14 +- include/linux/random.h | 51 +++++++ init/main.c | 11 ++ kernel/irq/autoprobe.c | 15 ++ kernel/irq/chip.c | 3 +- kernel/irq/handle.c | 181 +++++++++++++++++++++- kernel/irq/proc.c | 6 +- kernel/irq/spurious.c | 5 + 23 files changed, 649 insertions(+), 163 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index ac22bb7719f7..48ac688de3cd 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -238,6 +238,16 @@ config X86_HAS_BOOT_CPU_ID def_bool y depends on X86_VOYAGER +config SPARSE_IRQ + bool "Support sparse irq numbering" + depends on PCI_MSI || HT_IRQ + default y + help + This enables support for sparse irq, esp for msi/msi-x. You may need + if you have lots of cards supports msi-x installed. + + If you don't know what to do here, say Y. + config X86_FIND_SMP_CONFIG def_bool y depends on X86_MPPARSE || X86_VOYAGER diff --git a/arch/x86/include/asm/irq_vectors.h b/arch/x86/include/asm/irq_vectors.h index 0005adb0f941..bb6b69a6b125 100644 --- a/arch/x86/include/asm/irq_vectors.h +++ b/arch/x86/include/asm/irq_vectors.h @@ -102,11 +102,20 @@ #define invalid_vm86_irq(irq) ((irq) < 3 || (irq) > 15) #if defined(CONFIG_X86_IO_APIC) && !defined(CONFIG_X86_VOYAGER) + +#ifndef CONFIG_SPARSE_IRQ # if NR_CPUS < MAX_IO_APICS # define NR_IRQS (NR_VECTORS + (32 * NR_CPUS)) # else # define NR_IRQS (NR_VECTORS + (32 * MAX_IO_APICS)) # endif +#else +# if (8 * NR_CPUS) > (32 * MAX_IO_APICS) +# define NR_IRQS (NR_VECTORS + (8 * NR_CPUS)) +# else +# define NR_IRQS (NR_VECTORS + (32 * MAX_IO_APICS)) +# endif +#endif #elif defined(CONFIG_X86_VOYAGER) diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 9043251210fb..9de17f5c1125 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -108,8 +108,33 @@ static int __init parse_noapic(char *str) early_param("noapic", parse_noapic); struct irq_pin_list; + +/* + * This is performance-critical, we want to do it O(1) + * + * the indexing order of this array favors 1:1 mappings + * between pins and IRQs. + */ + +struct irq_pin_list { + int apic, pin; + struct irq_pin_list *next; +}; + +static struct irq_pin_list *get_one_free_irq_2_pin(int cpu) +{ + struct irq_pin_list *pin; + int node; + + node = cpu_to_node(cpu); + + pin = kzalloc_node(sizeof(*pin), GFP_ATOMIC, node); + printk(KERN_DEBUG " alloc irq_2_pin on cpu %d node %d\n", cpu, node); + + return pin; +} + struct irq_cfg { - unsigned int irq; struct irq_pin_list *irq_2_pin; cpumask_t domain; cpumask_t old_domain; @@ -119,83 +144,93 @@ struct irq_cfg { }; /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ +#ifdef CONFIG_SPARSE_IRQ +static struct irq_cfg irq_cfgx[] = { +#else static struct irq_cfg irq_cfgx[NR_IRQS] = { - [0] = { .irq = 0, .domain = CPU_MASK_ALL, .vector = IRQ0_VECTOR, }, - [1] = { .irq = 1, .domain = CPU_MASK_ALL, .vector = IRQ1_VECTOR, }, - [2] = { .irq = 2, .domain = CPU_MASK_ALL, .vector = IRQ2_VECTOR, }, - [3] = { .irq = 3, .domain = CPU_MASK_ALL, .vector = IRQ3_VECTOR, }, - [4] = { .irq = 4, .domain = CPU_MASK_ALL, .vector = IRQ4_VECTOR, }, - [5] = { .irq = 5, .domain = CPU_MASK_ALL, .vector = IRQ5_VECTOR, }, - [6] = { .irq = 6, .domain = CPU_MASK_ALL, .vector = IRQ6_VECTOR, }, - [7] = { .irq = 7, .domain = CPU_MASK_ALL, .vector = IRQ7_VECTOR, }, - [8] = { .irq = 8, .domain = CPU_MASK_ALL, .vector = IRQ8_VECTOR, }, - [9] = { .irq = 9, .domain = CPU_MASK_ALL, .vector = IRQ9_VECTOR, }, - [10] = { .irq = 10, .domain = CPU_MASK_ALL, .vector = IRQ10_VECTOR, }, - [11] = { .irq = 11, .domain = CPU_MASK_ALL, .vector = IRQ11_VECTOR, }, - [12] = { .irq = 12, .domain = CPU_MASK_ALL, .vector = IRQ12_VECTOR, }, - [13] = { .irq = 13, .domain = CPU_MASK_ALL, .vector = IRQ13_VECTOR, }, - [14] = { .irq = 14, .domain = CPU_MASK_ALL, .vector = IRQ14_VECTOR, }, - [15] = { .irq = 15, .domain = CPU_MASK_ALL, .vector = IRQ15_VECTOR, }, +#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, }, }; -#define for_each_irq_cfg(irq, cfg) \ - for (irq = 0, cfg = irq_cfgx; irq < nr_irqs; irq++, cfg++) - -static struct irq_cfg *irq_cfg(unsigned int irq) +void __init arch_early_irq_init(void) { - return irq < nr_irqs ? irq_cfgx + irq : NULL; -} + struct irq_cfg *cfg; + struct irq_desc *desc; + int count; + int i; -static struct irq_cfg *irq_cfg_alloc(unsigned int irq) -{ - return irq_cfg(irq); -} + cfg = irq_cfgx; + count = ARRAY_SIZE(irq_cfgx); -/* - * Rough estimation of how many shared IRQs there are, can be changed - * anytime. - */ -#define MAX_PLUS_SHARED_IRQS NR_IRQS -#define PIN_MAP_SIZE (MAX_PLUS_SHARED_IRQS + NR_IRQS) + for (i = 0; i < count; i++) { + desc = irq_to_desc(i); + desc->chip_data = &cfg[i]; + } +} -/* - * This is performance-critical, we want to do it O(1) - * - * the indexing order of this array favors 1:1 mappings - * between pins and IRQs. - */ +#ifdef CONFIG_SPARSE_IRQ +static struct irq_cfg *irq_cfg(unsigned int irq) +{ + struct irq_cfg *cfg = NULL; + struct irq_desc *desc; -struct irq_pin_list { - int apic, pin; - struct irq_pin_list *next; -}; + desc = irq_to_desc(irq); + if (desc) + cfg = desc->chip_data; -static struct irq_pin_list irq_2_pin_head[PIN_MAP_SIZE]; -static struct irq_pin_list *irq_2_pin_ptr; + return cfg; +} -static void __init irq_2_pin_init(void) +static struct irq_cfg *get_one_free_irq_cfg(int cpu) { - struct irq_pin_list *pin = irq_2_pin_head; - int i; + struct irq_cfg *cfg; + int node; + + node = cpu_to_node(cpu); - for (i = 1; i < PIN_MAP_SIZE; i++) - pin[i-1].next = &pin[i]; + cfg = kzalloc_node(sizeof(*cfg), GFP_ATOMIC, node); + printk(KERN_DEBUG " alloc irq_cfg on cpu %d node %d\n", cpu, node); - irq_2_pin_ptr = &pin[0]; + return cfg; } -static struct irq_pin_list *get_one_free_irq_2_pin(void) +void arch_init_chip_data(struct irq_desc *desc, int cpu) { - struct irq_pin_list *pin = irq_2_pin_ptr; + struct irq_cfg *cfg; - if (!pin) - panic("can not get more irq_2_pin\n"); + cfg = desc->chip_data; + if (!cfg) { + desc->chip_data = get_one_free_irq_cfg(cpu); + if (!desc->chip_data) { + printk(KERN_ERR "can not alloc irq_cfg\n"); + BUG_ON(1); + } + } +} - irq_2_pin_ptr = pin->next; - pin->next = NULL; - return pin; +#else +static struct irq_cfg *irq_cfg(unsigned int irq) +{ + return irq < nr_irqs ? irq_cfgx + irq : NULL; } +#endif + struct io_apic { unsigned int index; unsigned int unused[3]; @@ -397,16 +432,19 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) * shared ISA-space IRQs, so we have to support them. We are super * fast in the common case, and fast for shared ISA-space IRQs. */ -static void add_pin_to_irq(unsigned int irq, int apic, int pin) +static void add_pin_to_irq_cpu(unsigned int irq, int cpu, int apic, int pin) { - struct irq_cfg *cfg; struct irq_pin_list *entry; + struct irq_cfg *cfg = irq_cfg(irq); - /* first time to refer irq_cfg, so with new */ - cfg = irq_cfg_alloc(irq); entry = cfg->irq_2_pin; if (!entry) { - entry = get_one_free_irq_2_pin(); + entry = get_one_free_irq_2_pin(cpu); + if (!entry) { + printk(KERN_ERR "can not alloc irq_2_pin to add %d - %d\n", + apic, pin); + return; + } cfg->irq_2_pin = entry; entry->apic = apic; entry->pin = pin; @@ -421,7 +459,7 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) entry = entry->next; } - entry->next = get_one_free_irq_2_pin(); + entry->next = get_one_free_irq_2_pin(cpu); entry = entry->next; entry->apic = apic; entry->pin = pin; @@ -430,7 +468,7 @@ static void add_pin_to_irq(unsigned int irq, int apic, int pin) /* * Reroute an IRQ to a different pin. */ -static void __init replace_pin_at_irq(unsigned int irq, +static void __init replace_pin_at_irq(unsigned int irq, int cpu, int oldapic, int oldpin, int newapic, int newpin) { @@ -451,7 +489,7 @@ static void __init replace_pin_at_irq(unsigned int irq, /* why? call replace before add? */ if (!replaced) - add_pin_to_irq(irq, newapic, newpin); + add_pin_to_irq_cpu(irq, cpu, newapic, newpin); } static inline void io_apic_modify_irq(unsigned int irq, @@ -1162,9 +1200,13 @@ void __setup_vector_irq(int cpu) /* This function must be called with vector_lock held */ int irq, vector; struct irq_cfg *cfg; + struct irq_desc *desc; /* Mark the inuse vectors */ - for_each_irq_cfg(irq, cfg) { + for_each_irq_desc(irq, desc) { + if (!desc) + continue; + cfg = desc->chip_data; if (!cpu_isset(cpu, cfg->domain)) continue; vector = cfg->vector; @@ -1356,6 +1398,8 @@ static void __init setup_IO_APIC_irqs(void) { int apic, pin, idx, irq; int notcon = 0; + struct irq_desc *desc; + int cpu = boot_cpu_id; apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); @@ -1387,7 +1431,12 @@ static void __init setup_IO_APIC_irqs(void) if (multi_timer_check(apic, irq)) continue; #endif - add_pin_to_irq(irq, apic, pin); + desc = irq_to_desc_alloc_cpu(irq, cpu); + if (!desc) { + printk(KERN_INFO "can not get irq_desc for %d\n", irq); + continue; + } + add_pin_to_irq_cpu(irq, cpu, apic, pin); setup_IO_APIC_irq(apic, pin, irq, irq_trigger(idx), irq_polarity(idx)); @@ -1448,6 +1497,7 @@ __apicdebuginit(void) print_IO_APIC(void) union IO_APIC_reg_03 reg_03; unsigned long flags; struct irq_cfg *cfg; + struct irq_desc *desc; unsigned int irq; if (apic_verbosity == APIC_QUIET) @@ -1537,8 +1587,13 @@ __apicdebuginit(void) print_IO_APIC(void) } } printk(KERN_DEBUG "IRQ to pin mappings:\n"); - for_each_irq_cfg(irq, cfg) { - struct irq_pin_list *entry = cfg->irq_2_pin; + for_each_irq_desc(irq, desc) { + struct irq_pin_list *entry; + + if (!desc) + continue; + cfg = desc->chip_data; + entry = cfg->irq_2_pin; if (!entry) continue; printk(KERN_DEBUG "IRQ%d ", irq); @@ -2022,6 +2077,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) { int was_pending = 0; unsigned long flags; + struct irq_cfg *cfg; spin_lock_irqsave(&ioapic_lock, flags); if (irq < 16) { @@ -2029,6 +2085,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) if (i8259A_irq_pending(irq)) was_pending = 1; } + cfg = irq_cfg(irq); __unmask_IO_APIC_irq(irq); spin_unlock_irqrestore(&ioapic_lock, flags); @@ -2178,6 +2235,9 @@ static void ir_irq_migration(struct work_struct *work) struct irq_desc *desc; for_each_irq_desc(irq, desc) { + if (!desc) + continue; + if (desc->status & IRQ_MOVE_PENDING) { unsigned long flags; @@ -2229,6 +2289,9 @@ asmlinkage void smp_irq_move_cleanup_interrupt(void) struct irq_cfg *cfg; irq = __get_cpu_var(vector_irq)[vector]; + if (irq == -1) + continue; + desc = irq_to_desc(irq); if (!desc) continue; @@ -2430,8 +2493,12 @@ static inline void init_IO_APIC_traps(void) * Also, we've got to be careful not to trash gate * 0x80, because int 0x80 is hm, kind of importantish. ;) */ - for_each_irq_cfg(irq, cfg) { - if (IO_APIC_IRQ(irq) && !cfg->vector) { + for_each_irq_desc(irq, desc) { + if (!desc) + continue; + + cfg = desc->chip_data; + if (IO_APIC_IRQ(irq) && cfg && !cfg->vector) { /* * Hmm.. We don't have an entry for this, * so default to an old-fashioned 8259 @@ -2439,11 +2506,9 @@ static inline void init_IO_APIC_traps(void) */ if (irq < 16) make_8259A_irq(irq); - else { - desc = irq_to_desc(irq); + else /* Strange. Oh, well.. */ desc->chip = &no_irq_chip; - } } } } @@ -2654,7 +2719,7 @@ static inline void __init check_timer(void) * Ok, does IRQ0 through the IOAPIC work? */ if (no_pin1) { - add_pin_to_irq(0, apic1, pin1); + add_pin_to_irq_cpu(0, boot_cpu_id, apic1, pin1); setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); } unmask_IO_APIC_irq(0); @@ -2683,7 +2748,7 @@ static inline void __init check_timer(void) /* * legacy devices should be connected to IO APIC #0 */ - replace_pin_at_irq(0, apic1, pin1, apic2, pin2); + replace_pin_at_irq(0, boot_cpu_id, apic1, pin1, apic2, pin2); setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); unmask_IO_APIC_irq(0); enable_8259A_irq(0); @@ -2902,21 +2967,25 @@ unsigned int create_irq_nr(unsigned int irq_want) unsigned int irq; unsigned int new; unsigned long flags; - struct irq_cfg *cfg_new; - - irq_want = nr_irqs - 1; + struct irq_cfg *cfg_new = NULL; + int cpu = boot_cpu_id; + struct irq_desc *desc_new = NULL; irq = 0; spin_lock_irqsave(&vector_lock, flags); for (new = irq_want; new > 0; new--) { if (platform_legacy_irq(new)) continue; - cfg_new = irq_cfg(new); - if (cfg_new && cfg_new->vector != 0) + + desc_new = irq_to_desc_alloc_cpu(new, cpu); + if (!desc_new) { + printk(KERN_INFO "can not get irq_desc for %d\n", new); + continue; + } + cfg_new = desc_new->chip_data; + + if (cfg_new->vector != 0) continue; - /* check if need to create one */ - if (!cfg_new) - cfg_new = irq_cfg_alloc(new); if (__assign_irq_vector(new, TARGET_CPUS) == 0) irq = new; break; @@ -2925,6 +2994,9 @@ unsigned int create_irq_nr(unsigned int irq_want) if (irq > 0) { dynamic_irq_init(irq); + /* restore it, in case dynamic_irq_init clear it */ + if (desc_new) + desc_new->chip_data = cfg_new; } return irq; } @@ -2944,8 +3016,16 @@ int create_irq(void) void destroy_irq(unsigned int irq) { unsigned long flags; + struct irq_cfg *cfg; + struct irq_desc *desc; + /* store it, in case dynamic_irq_cleanup clear it */ + desc = irq_to_desc(irq); + cfg = desc->chip_data; dynamic_irq_cleanup(irq); + /* connect back irq_cfg */ + if (desc) + desc->chip_data = cfg; #ifdef CONFIG_INTR_REMAP free_irte(irq); @@ -3195,26 +3275,13 @@ static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) return 0; } -static unsigned int build_irq_for_pci_dev(struct pci_dev *dev) -{ - unsigned int irq; - - irq = dev->bus->number; - irq <<= 8; - irq |= dev->devfn; - irq <<= 12; - - return irq; -} - -int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) +int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *msidesc) { unsigned int irq; int ret; unsigned int irq_want; - irq_want = build_irq_for_pci_dev(dev) + 0x100; - + irq_want = nr_irqs - 1; irq = create_irq_nr(irq_want); if (irq == 0) return -1; @@ -3228,7 +3295,7 @@ int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc) goto error; no_ir: #endif - ret = setup_msi_irq(dev, desc, irq); + ret = setup_msi_irq(dev, msidesc, irq); if (ret < 0) { destroy_irq(irq); return ret; @@ -3246,7 +3313,7 @@ int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) { unsigned int irq; int ret, sub_handle; - struct msi_desc *desc; + struct msi_desc *msidesc; unsigned int irq_want; #ifdef CONFIG_INTR_REMAP @@ -3254,10 +3321,11 @@ int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) int index = 0; #endif - irq_want = build_irq_for_pci_dev(dev) + 0x100; + irq_want = nr_irqs - 1; sub_handle = 0; - list_for_each_entry(desc, &dev->msi_list, list) { - irq = create_irq_nr(irq_want--); + list_for_each_entry(msidesc, &dev->msi_list, list) { + irq = create_irq_nr(irq_want); + irq_want--; if (irq == 0) return -1; #ifdef CONFIG_INTR_REMAP @@ -3289,7 +3357,7 @@ int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) } no_ir: #endif - ret = setup_msi_irq(dev, desc, irq); + ret = setup_msi_irq(dev, msidesc, irq); if (ret < 0) goto error; sub_handle++; @@ -3707,17 +3775,29 @@ int __init io_apic_get_version(int ioapic) int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int polarity) { + struct irq_desc *desc; + struct irq_cfg *cfg; + int cpu = boot_cpu_id; + if (!IO_APIC_IRQ(irq)) { apic_printk(APIC_QUIET,KERN_ERR "IOAPIC[%d]: Invalid reference to IRQ 0\n", ioapic); return -EINVAL; } + desc = irq_to_desc_alloc_cpu(irq, cpu); + if (!desc) { + printk(KERN_INFO "can not get irq_desc %d\n", irq); + return 0; + } + /* * IRQs < 16 are already in the irq_2_pin[] map */ - if (irq >= 16) - add_pin_to_irq(irq, ioapic, pin); + if (irq >= 16) { + cfg = desc->chip_data; + add_pin_to_irq_cpu(irq, cpu, ioapic, pin); + } setup_IO_APIC_irq(ioapic, pin, irq, triggering, polarity); @@ -3773,7 +3853,8 @@ void __init setup_ioapic_dest(void) * when you have too many devices, because at that time only boot * cpu is online. */ - cfg = irq_cfg(irq); + desc = irq_to_desc(irq); + cfg = desc->chip_data; if (!cfg->vector) { setup_IO_APIC_irq(ioapic, pin, irq, irq_trigger(irq_entry), @@ -3785,7 +3866,6 @@ void __init setup_ioapic_dest(void) /* * Honour affinities which have been set in early boot */ - desc = irq_to_desc(irq); if (desc->status & (IRQ_NO_BALANCING | IRQ_AFFINITY_SET)) mask = desc->affinity; @@ -3846,7 +3926,6 @@ void __init ioapic_init_mappings(void) struct resource *ioapic_res; int i; - irq_2_pin_init(); ioapic_res = ioapic_setup_resources(); for (i = 0; i < nr_ioapics; i++) { if (smp_found_config) { diff --git a/arch/x86/kernel/irq.c b/arch/x86/kernel/irq.c index d1d4dc52f649..3f1d9d18df67 100644 --- a/arch/x86/kernel/irq.c +++ b/arch/x86/kernel/irq.c @@ -118,6 +118,9 @@ int show_interrupts(struct seq_file *p, void *v) } desc = irq_to_desc(i); + if (!desc) + return 0; + spin_lock_irqsave(&desc->lock, flags); #ifndef CONFIG_SMP any_count = kstat_irqs(i); diff --git a/arch/x86/kernel/irq_32.c b/arch/x86/kernel/irq_32.c index a51382672de0..119fc9c8ff7f 100644 --- a/arch/x86/kernel/irq_32.c +++ b/arch/x86/kernel/irq_32.c @@ -242,6 +242,8 @@ void fixup_irqs(cpumask_t map) for_each_irq_desc(irq, desc) { cpumask_t mask; + if (!desc) + continue; if (irq == 2) continue; diff --git a/arch/x86/kernel/irq_64.c b/arch/x86/kernel/irq_64.c index 60eb84eb77a0..900009c70591 100644 --- a/arch/x86/kernel/irq_64.c +++ b/arch/x86/kernel/irq_64.c @@ -94,6 +94,8 @@ void fixup_irqs(cpumask_t map) int break_affinity = 0; int set_affinity = 1; + if (!desc) + continue; if (irq == 2) continue; diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 845aa9803e80..5a5651b7f9e6 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -69,7 +69,6 @@ void __init init_ISA_irqs (void) * 16 old-style INTA-cycle interrupts: */ for (i = 0; i < 16; i++) { - /* first time call this irq_desc */ struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index ff0235391285..cd9f42d028d9 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -143,7 +143,6 @@ void __init init_ISA_irqs(void) init_8259A(0); for (i = 0; i < 16; i++) { - /* first time call this irq_desc */ struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; diff --git a/drivers/char/random.c b/drivers/char/random.c index 675076f5fca8..d26891bfcd41 100644 --- a/drivers/char/random.c +++ b/drivers/char/random.c @@ -558,23 +558,9 @@ struct timer_rand_state { unsigned dont_count_entropy:1; }; -static struct timer_rand_state *irq_timer_state[NR_IRQS]; - -static struct timer_rand_state *get_timer_rand_state(unsigned int irq) -{ - if (irq >= nr_irqs) - return NULL; - - return irq_timer_state[irq]; -} - -static void set_timer_rand_state(unsigned int irq, struct timer_rand_state *state) -{ - if (irq >= nr_irqs) - return; - - irq_timer_state[irq] = state; -} +#ifndef CONFIG_SPARSE_IRQ +struct timer_rand_state *irq_timer_state[NR_IRQS]; +#endif static struct timer_rand_state input_timer_state; @@ -933,8 +919,10 @@ void rand_initialize_irq(int irq) { struct timer_rand_state *state; +#ifndef CONFIG_SPARSE_IRQ if (irq >= nr_irqs) return; +#endif state = get_timer_rand_state(irq); diff --git a/drivers/pci/intr_remapping.c b/drivers/pci/intr_remapping.c index 2de5a3238c94..c9958ec5e25e 100644 --- a/drivers/pci/intr_remapping.c +++ b/drivers/pci/intr_remapping.c @@ -19,17 +19,75 @@ struct irq_2_iommu { u8 irte_mask; }; -static struct irq_2_iommu irq_2_iommuX[NR_IRQS]; +#ifdef CONFIG_SPARSE_IRQ +static struct irq_2_iommu *get_one_free_irq_2_iommu(int cpu) +{ + struct irq_2_iommu *iommu; + int node; + + node = cpu_to_node(cpu); + + iommu = kzalloc_node(sizeof(*iommu), GFP_ATOMIC, node); + printk(KERN_DEBUG "alloc irq_2_iommu on cpu %d node %d\n", cpu, node); + + return iommu; +} static struct irq_2_iommu *irq_2_iommu(unsigned int irq) { - return (irq < nr_irqs) ? irq_2_iommuX + irq : NULL; + struct irq_desc *desc; + + desc = irq_to_desc(irq); + + if (WARN_ON_ONCE(!desc)) + return NULL; + + return desc->irq_2_iommu; +} + +static struct irq_2_iommu *irq_2_iommu_alloc_cpu(unsigned int irq, int cpu) +{ + struct irq_desc *desc; + struct irq_2_iommu *irq_iommu; + + /* + * alloc irq desc if not allocated already. + */ + desc = irq_to_desc_alloc_cpu(irq, cpu); + if (!desc) { + printk(KERN_INFO "can not get irq_desc for %d\n", irq); + return NULL; + } + + irq_iommu = desc->irq_2_iommu; + + if (!irq_iommu) + desc->irq_2_iommu = get_one_free_irq_2_iommu(cpu); + + return desc->irq_2_iommu; } +static struct irq_2_iommu *irq_2_iommu_alloc(unsigned int irq) +{ + return irq_2_iommu_alloc_cpu(irq, boot_cpu_id); +} + +#else /* !CONFIG_SPARSE_IRQ */ + +static struct irq_2_iommu irq_2_iommuX[NR_IRQS]; + +static struct irq_2_iommu *irq_2_iommu(unsigned int irq) +{ + if (irq < nr_irqs) + return &irq_2_iommuX[irq]; + + return NULL; +} static struct irq_2_iommu *irq_2_iommu_alloc(unsigned int irq) { return irq_2_iommu(irq); } +#endif static DEFINE_SPINLOCK(irq_2_ir_lock); @@ -86,9 +144,11 @@ int alloc_irte(struct intel_iommu *iommu, int irq, u16 count) if (!count) return -1; +#ifndef CONFIG_SPARSE_IRQ /* protect irq_2_iommu_alloc later */ if (irq >= nr_irqs) return -1; +#endif /* * start the IRTE search from index 0. @@ -130,6 +190,12 @@ int alloc_irte(struct intel_iommu *iommu, int irq, u16 count) table->base[i].present = 1; irq_iommu = irq_2_iommu_alloc(irq); + if (!irq_iommu) { + spin_unlock(&irq_2_ir_lock); + printk(KERN_ERR "can't allocate irq_2_iommu\n"); + return -1; + } + irq_iommu->iommu = iommu; irq_iommu->irte_index = index; irq_iommu->sub_handle = 0; @@ -177,6 +243,12 @@ int set_irte_irq(int irq, struct intel_iommu *iommu, u16 index, u16 subhandle) irq_iommu = irq_2_iommu_alloc(irq); + if (!irq_iommu) { + spin_unlock(&irq_2_ir_lock); + printk(KERN_ERR "can't allocate irq_2_iommu\n"); + return -1; + } + irq_iommu->iommu = iommu; irq_iommu->irte_index = index; irq_iommu->sub_handle = subhandle; diff --git a/drivers/xen/events.c b/drivers/xen/events.c index 1e3b934a4cf7..2924faa7f6c4 100644 --- a/drivers/xen/events.c +++ b/drivers/xen/events.c @@ -141,8 +141,12 @@ static void init_evtchn_cpu_bindings(void) int i; /* By default all event channels notify CPU#0. */ - for_each_irq_desc(i, desc) + for_each_irq_desc(i, desc) { + if (!desc) + continue; + desc->affinity = cpumask_of_cpu(0); + } #endif memset(cpu_evtchn, 0, sizeof(cpu_evtchn)); @@ -231,7 +235,7 @@ static int find_unbound_irq(void) int irq; /* Only allocate from dynirq range */ - for_each_irq_nr(irq) + for (irq = 0; irq < nr_irqs; irq++) if (irq_bindcount[irq] == 0) break; @@ -792,7 +796,7 @@ void xen_irq_resume(void) mask_evtchn(evtchn); /* No IRQ <-> event-channel mappings. */ - for_each_irq_nr(irq) + for (irq = 0; irq < nr_irqs; irq++) irq_info[irq].evtchn = 0; /* zap event-channel binding */ for (evtchn = 0; evtchn < NR_EVENT_CHANNELS; evtchn++) @@ -824,7 +828,7 @@ void __init xen_init_IRQ(void) mask_evtchn(i); /* Dynamic IRQ space is currently unbound. Zero the refcnts. */ - for_each_irq_nr(i) + for (i = 0; i < nr_irqs; i++) irq_bindcount[i] = 0; irq_ctx_init(smp_processor_id()); diff --git a/fs/proc/stat.c b/fs/proc/stat.c index 81904f07679d..a13431ab7c65 100644 --- a/fs/proc/stat.c +++ b/fs/proc/stat.c @@ -27,6 +27,7 @@ static int show_stat(struct seq_file *p, void *v) u64 sum = 0; struct timespec boottime; unsigned int per_irq_sum; + struct irq_desc *desc; user = nice = system = idle = iowait = irq = softirq = steal = cputime64_zero; @@ -44,10 +45,11 @@ static int show_stat(struct seq_file *p, void *v) softirq = cputime64_add(softirq, kstat_cpu(i).cpustat.softirq); steal = cputime64_add(steal, kstat_cpu(i).cpustat.steal); guest = cputime64_add(guest, kstat_cpu(i).cpustat.guest); - - for_each_irq_nr(j) + for_each_irq_desc(j, desc) { + if (!desc) + continue; sum += kstat_irqs_cpu(j, i); - + } sum += arch_irq_stat_cpu(i); } sum += arch_irq_stat(); @@ -90,11 +92,14 @@ static int show_stat(struct seq_file *p, void *v) seq_printf(p, "intr %llu", (unsigned long long)sum); /* sum again ? it could be updated? */ - for_each_irq_nr(j) { + for (j = 0; j < NR_IRQS; j++) { + desc = irq_to_desc(j); per_irq_sum = 0; - for_each_possible_cpu(i) - per_irq_sum += kstat_irqs_cpu(j, i); + if (desc) { + for_each_possible_cpu(i) + per_irq_sum += kstat_irqs_cpu(j, i); + } seq_printf(p, " %u", per_irq_sum); } diff --git a/include/linux/interrupt.h b/include/linux/interrupt.h index f58a0cf8929a..79e915e7e8a5 100644 --- a/include/linux/interrupt.h +++ b/include/linux/interrupt.h @@ -18,6 +18,8 @@ #include #include +extern int nr_irqs; + /* * These correspond to the IORESOURCE_IRQ_* defines in * linux/ioport.h to select the interrupt line behaviour. When diff --git a/include/linux/irq.h b/include/linux/irq.h index 3dddfa703ebd..63b00439d4d2 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -129,6 +129,8 @@ struct irq_chip { const char *typename; }; +struct timer_rand_state; +struct irq_2_iommu; /** * struct irq_desc - interrupt descriptor * @irq: interrupt number for this descriptor @@ -154,6 +156,13 @@ struct irq_chip { */ struct irq_desc { unsigned int irq; +#ifdef CONFIG_SPARSE_IRQ + struct timer_rand_state *timer_rand_state; + unsigned int *kstat_irqs; +# ifdef CONFIG_INTR_REMAP + struct irq_2_iommu *irq_2_iommu; +# endif +#endif irq_flow_handler_t handle_irq; struct irq_chip *chip; struct msi_desc *msi_desc; @@ -181,14 +190,52 @@ struct irq_desc { const char *name; } ____cacheline_internodealigned_in_smp; +extern void early_irq_init(void); +extern void arch_early_irq_init(void); +extern void arch_init_chip_data(struct irq_desc *desc, int cpu); +extern void arch_init_copy_chip_data(struct irq_desc *old_desc, + struct irq_desc *desc, int cpu); +extern void arch_free_chip_data(struct irq_desc *old_desc, struct irq_desc *desc); + +#ifndef CONFIG_SPARSE_IRQ extern struct irq_desc irq_desc[NR_IRQS]; static inline struct irq_desc *irq_to_desc(unsigned int irq) { - return (irq < nr_irqs) ? irq_desc + irq : NULL; + return (irq < NR_IRQS) ? irq_desc + irq : NULL; +} +static inline struct irq_desc *irq_to_desc_alloc_cpu(unsigned int irq, int cpu) +{ + return irq_to_desc(irq); } +#ifdef CONFIG_GENERIC_HARDIRQS +# define for_each_irq_desc(irq, desc) \ + for (irq = 0, desc = irq_desc; irq < nr_irqs; irq++, desc++) +# define for_each_irq_desc_reverse(irq, desc) \ + for (irq = nr_irqs - 1, desc = irq_desc + (nr_irqs - 1); \ + irq >= 0; irq--, desc--) +#endif + +#else + +extern struct irq_desc *irq_to_desc(unsigned int irq); +extern struct irq_desc *irq_to_desc_alloc_cpu(unsigned int irq, int cpu); +extern struct irq_desc *move_irq_desc(struct irq_desc *old_desc, int cpu); + +# define for_each_irq_desc(irq, desc) \ + for (irq = 0, desc = irq_to_desc(irq); irq < nr_irqs; irq++, desc = irq_to_desc(irq)) +# define for_each_irq_desc_reverse(irq, desc) \ + for (irq = nr_irqs - 1, desc = irq_to_desc(irq); irq >= 0; irq--, desc = irq_to_desc(irq)) + +#define kstat_irqs_this_cpu(DESC) \ + ((DESC)->kstat_irqs[smp_processor_id()]) +#define kstat_incr_irqs_this_cpu(irqno, DESC) \ + ((DESC)->kstat_irqs[smp_processor_id()]++) + +#endif + /* * Migration helpers for obsolete names, they will go away: */ @@ -380,6 +427,11 @@ extern int set_irq_msi(unsigned int irq, struct msi_desc *entry); #define get_irq_data(irq) (irq_to_desc(irq)->handler_data) #define get_irq_msi(irq) (irq_to_desc(irq)->msi_desc) +#define get_irq_desc_chip(desc) ((desc)->chip) +#define get_irq_desc_chip_data(desc) ((desc)->chip_data) +#define get_irq_desc_data(desc) ((desc)->handler_data) +#define get_irq_desc_msi(desc) ((desc)->msi_desc) + #endif /* CONFIG_GENERIC_HARDIRQS */ #endif /* !CONFIG_S390 */ diff --git a/include/linux/irqnr.h b/include/linux/irqnr.h index 452c280c8115..7a299e989f8b 100644 --- a/include/linux/irqnr.h +++ b/include/linux/irqnr.h @@ -7,18 +7,10 @@ # define for_each_irq_desc(irq, desc) \ for (irq = 0; irq < nr_irqs; irq++) -#else -extern int nr_irqs; -# define for_each_irq_desc(irq, desc) \ - for (irq = 0, desc = irq_desc; irq < nr_irqs; irq++, desc++) - -# define for_each_irq_desc_reverse(irq, desc) \ - for (irq = nr_irqs - 1, desc = irq_desc + (nr_irqs - 1); \ - irq >= 0; irq--, desc--) +static inline early_sparse_irq_init(void) +{ +} #endif -#define for_each_irq_nr(irq) \ - for (irq = 0; irq < nr_irqs; irq++) - #endif diff --git a/include/linux/kernel_stat.h b/include/linux/kernel_stat.h index 4a145caeee07..4ee4b3d2316f 100644 --- a/include/linux/kernel_stat.h +++ b/include/linux/kernel_stat.h @@ -28,7 +28,9 @@ struct cpu_usage_stat { struct kernel_stat { struct cpu_usage_stat cpustat; - unsigned int irqs[NR_IRQS]; +#ifndef CONFIG_SPARSE_IRQ + unsigned int irqs[NR_IRQS]; +#endif }; DECLARE_PER_CPU(struct kernel_stat, kstat); @@ -39,6 +41,10 @@ DECLARE_PER_CPU(struct kernel_stat, kstat); extern unsigned long long nr_context_switches(void); +#ifndef CONFIG_SPARSE_IRQ +#define kstat_irqs_this_cpu(irq) \ + (kstat_this_cpu.irqs[irq]) + struct irq_desc; static inline void kstat_incr_irqs_this_cpu(unsigned int irq, @@ -46,11 +52,17 @@ static inline void kstat_incr_irqs_this_cpu(unsigned int irq, { kstat_this_cpu.irqs[irq]++; } +#endif + +#ifndef CONFIG_SPARSE_IRQ static inline unsigned int kstat_irqs_cpu(unsigned int irq, int cpu) { return kstat_cpu(cpu).irqs[irq]; } +#else +extern unsigned int kstat_irqs_cpu(unsigned int irq, int cpu); +#endif /* * Number of interrupts per specific IRQ source, since bootup diff --git a/include/linux/random.h b/include/linux/random.h index 36f125c0c603..ad9daa2374d5 100644 --- a/include/linux/random.h +++ b/include/linux/random.h @@ -44,6 +44,57 @@ struct rand_pool_info { extern void rand_initialize_irq(int irq); +struct timer_rand_state; +#ifndef CONFIG_SPARSE_IRQ + +extern struct timer_rand_state *irq_timer_state[]; + +extern int nr_irqs; +static inline struct timer_rand_state *get_timer_rand_state(unsigned int irq) +{ + if (irq >= nr_irqs) + return NULL; + + return irq_timer_state[irq]; +} + +static inline void set_timer_rand_state(unsigned int irq, struct timer_rand_state *state) +{ + if (irq >= nr_irqs) + return; + + irq_timer_state[irq] = state; +} + +#else + +#include +static inline struct timer_rand_state *get_timer_rand_state(unsigned int irq) +{ + struct irq_desc *desc; + + desc = irq_to_desc(irq); + + if (!desc) + return NULL; + + return desc->timer_rand_state; +} + +static inline void set_timer_rand_state(unsigned int irq, struct timer_rand_state *state) +{ + struct irq_desc *desc; + + desc = irq_to_desc(irq); + + if (!desc) + return; + + desc->timer_rand_state = state; +} +#endif + + extern void add_input_randomness(unsigned int type, unsigned int code, unsigned int value); extern void add_interrupt_randomness(int irq); diff --git a/init/main.c b/init/main.c index 7e117a231af1..c1f999a3cf31 100644 --- a/init/main.c +++ b/init/main.c @@ -539,6 +539,15 @@ void __init __weak thread_info_cache_init(void) { } +void __init __weak arch_early_irq_init(void) +{ +} + +void __init __weak early_irq_init(void) +{ + arch_early_irq_init(); +} + asmlinkage void __init start_kernel(void) { char * command_line; @@ -603,6 +612,8 @@ asmlinkage void __init start_kernel(void) sort_main_extable(); trap_init(); rcu_init(); + /* init some links before init_ISA_irqs() */ + early_irq_init(); init_IRQ(); pidhash_init(); init_timers(); diff --git a/kernel/irq/autoprobe.c b/kernel/irq/autoprobe.c index cc0f7321b8ce..650ce4102a63 100644 --- a/kernel/irq/autoprobe.c +++ b/kernel/irq/autoprobe.c @@ -40,6 +40,9 @@ unsigned long probe_irq_on(void) * flush such a longstanding irq before considering it as spurious. */ for_each_irq_desc_reverse(i, desc) { + if (!desc) + continue; + spin_lock_irq(&desc->lock); if (!desc->action && !(desc->status & IRQ_NOPROBE)) { /* @@ -68,6 +71,9 @@ unsigned long probe_irq_on(void) * happened in the previous stage, it may have masked itself) */ for_each_irq_desc_reverse(i, desc) { + if (!desc) + continue; + spin_lock_irq(&desc->lock); if (!desc->action && !(desc->status & IRQ_NOPROBE)) { desc->status |= IRQ_AUTODETECT | IRQ_WAITING; @@ -86,6 +92,9 @@ unsigned long probe_irq_on(void) * Now filter out any obviously spurious interrupts */ for_each_irq_desc(i, desc) { + if (!desc) + continue; + spin_lock_irq(&desc->lock); status = desc->status; @@ -124,6 +133,9 @@ unsigned int probe_irq_mask(unsigned long val) int i; for_each_irq_desc(i, desc) { + if (!desc) + continue; + spin_lock_irq(&desc->lock); status = desc->status; @@ -166,6 +178,9 @@ int probe_irq_off(unsigned long val) unsigned int status; for_each_irq_desc(i, desc) { + if (!desc) + continue; + spin_lock_irq(&desc->lock); status = desc->status; diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 10b5092e9bfe..8e4fce4a1b1f 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -24,9 +24,10 @@ */ void dynamic_irq_init(unsigned int irq) { - struct irq_desc *desc = irq_to_desc(irq); + struct irq_desc *desc; unsigned long flags; + desc = irq_to_desc(irq); if (!desc) { WARN(1, KERN_ERR "Trying to initialize invalid IRQ%d\n", irq); return; diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index c815b42d0f5b..96ca203eb51b 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -15,9 +15,16 @@ #include #include #include +#include +#include #include "internals.h" +/* + * lockdep: we want to handle all irq_desc locks as a single lock-class: + */ +static struct lock_class_key irq_desc_lock_class; + /** * handle_bad_irq - handle spurious and unhandled irqs * @irq: the interrupt number @@ -49,6 +56,155 @@ void handle_bad_irq(unsigned int irq, struct irq_desc *desc) int nr_irqs = NR_IRQS; EXPORT_SYMBOL_GPL(nr_irqs); +void __init __attribute__((weak)) arch_early_irq_init(void) +{ +} + +#ifdef CONFIG_SPARSE_IRQ +static struct irq_desc irq_desc_init = { + .irq = -1, + .status = IRQ_DISABLED, + .chip = &no_irq_chip, + .handle_irq = handle_bad_irq, + .depth = 1, + .lock = __SPIN_LOCK_UNLOCKED(irq_desc_init.lock), +#ifdef CONFIG_SMP + .affinity = CPU_MASK_ALL +#endif +}; + +static void init_kstat_irqs(struct irq_desc *desc, int cpu, int nr) +{ + unsigned long bytes; + char *ptr; + int node; + + /* Compute how many bytes we need per irq and allocate them */ + bytes = nr * sizeof(unsigned int); + + node = cpu_to_node(cpu); + ptr = kzalloc_node(bytes, GFP_ATOMIC, node); + printk(KERN_DEBUG " alloc kstat_irqs on cpu %d node %d\n", cpu, node); + + if (ptr) + desc->kstat_irqs = (unsigned int *)ptr; +} + +void __attribute__((weak)) arch_init_chip_data(struct irq_desc *desc, int cpu) +{ +} + +static void init_one_irq_desc(int irq, struct irq_desc *desc, int cpu) +{ + memcpy(desc, &irq_desc_init, sizeof(struct irq_desc)); + desc->irq = irq; +#ifdef CONFIG_SMP + desc->cpu = cpu; +#endif + lockdep_set_class(&desc->lock, &irq_desc_lock_class); + init_kstat_irqs(desc, cpu, nr_cpu_ids); + if (!desc->kstat_irqs) { + printk(KERN_ERR "can not alloc kstat_irqs\n"); + BUG_ON(1); + } + arch_init_chip_data(desc, cpu); +} + +/* + * Protect the sparse_irqs: + */ +static DEFINE_SPINLOCK(sparse_irq_lock); + +struct irq_desc *irq_desc_ptrs[NR_IRQS] __read_mostly; + +static struct irq_desc irq_desc_legacy[16] __cacheline_aligned_in_smp = { + [0 ... 15] = { + .irq = -1, + .status = IRQ_DISABLED, + .chip = &no_irq_chip, + .handle_irq = handle_bad_irq, + .depth = 1, + .lock = __SPIN_LOCK_UNLOCKED(irq_desc_init.lock), +#ifdef CONFIG_SMP + .affinity = CPU_MASK_ALL +#endif + } +}; + +/* FIXME: use bootmem alloc ...*/ +static unsigned int kstat_irqs_legacy[16][NR_CPUS]; + +void __init early_irq_init(void) +{ + struct irq_desc *desc; + int legacy_count; + int i; + + desc = irq_desc_legacy; + legacy_count = ARRAY_SIZE(irq_desc_legacy); + + for (i = 0; i < legacy_count; i++) { + desc[i].irq = i; + desc[i].kstat_irqs = kstat_irqs_legacy[i]; + + irq_desc_ptrs[i] = desc + i; + } + + for (i = legacy_count; i < NR_IRQS; i++) + irq_desc_ptrs[i] = NULL; + + arch_early_irq_init(); +} + +struct irq_desc *irq_to_desc(unsigned int irq) +{ + return (irq < NR_IRQS) ? irq_desc_ptrs[irq] : NULL; +} + +struct irq_desc *irq_to_desc_alloc_cpu(unsigned int irq, int cpu) +{ + struct irq_desc *desc; + unsigned long flags; + int node; + + if (irq >= NR_IRQS) { + printk(KERN_WARNING "irq >= NR_IRQS in irq_to_desc_alloc: %d %d\n", + irq, NR_IRQS); + WARN_ON(1); + return NULL; + } + + desc = irq_desc_ptrs[irq]; + if (desc) + return desc; + + spin_lock_irqsave(&sparse_irq_lock, flags); + + /* We have to check it to avoid races with another CPU */ + desc = irq_desc_ptrs[irq]; + if (desc) + goto out_unlock; + + node = cpu_to_node(cpu); + desc = kzalloc_node(sizeof(*desc), GFP_ATOMIC, node); + printk(KERN_DEBUG " alloc irq_desc for %d on cpu %d node %d\n", + irq, cpu, node); + if (!desc) { + printk(KERN_ERR "can not alloc irq_desc\n"); + BUG_ON(1); + } + init_one_irq_desc(irq, desc, cpu); + + irq_desc_ptrs[irq] = desc; + +out_unlock: + spin_unlock_irqrestore(&sparse_irq_lock, flags); + + return desc; +} + +#else + struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { [0 ... NR_IRQS-1] = { .status = IRQ_DISABLED, @@ -62,6 +218,8 @@ struct irq_desc irq_desc[NR_IRQS] __cacheline_aligned_in_smp = { } }; +#endif + /* * What should we do if we get a hw irq event on an illegal vector? * Each architecture has to answer this themself. @@ -261,17 +419,28 @@ out: #ifdef CONFIG_TRACE_IRQFLAGS -/* - * lockdep: we want to handle all irq_desc locks as a single lock-class: - */ -static struct lock_class_key irq_desc_lock_class; - void early_init_irq_lock_class(void) { +#ifndef CONFIG_SPARSE_IRQ struct irq_desc *desc; int i; - for_each_irq_desc(i, desc) + for_each_irq_desc(i, desc) { + if (!desc) + continue; + lockdep_set_class(&desc->lock, &irq_desc_lock_class); + } +#endif +} +#endif + +#ifdef CONFIG_SPARSE_IRQ +unsigned int kstat_irqs_cpu(unsigned int irq, int cpu) +{ + struct irq_desc *desc = irq_to_desc(irq); + return desc->kstat_irqs[cpu]; } #endif +EXPORT_SYMBOL(kstat_irqs_cpu); + diff --git a/kernel/irq/proc.c b/kernel/irq/proc.c index d257e7d6a8a4..f6b3440f05bc 100644 --- a/kernel/irq/proc.c +++ b/kernel/irq/proc.c @@ -243,7 +243,11 @@ void init_irq_proc(void) /* * Create entries for all existing IRQs. */ - for_each_irq_desc(irq, desc) + for_each_irq_desc(irq, desc) { + if (!desc) + continue; + register_irq_proc(irq, desc); + } } diff --git a/kernel/irq/spurious.c b/kernel/irq/spurious.c index dd364c11e56e..3738107531fd 100644 --- a/kernel/irq/spurious.c +++ b/kernel/irq/spurious.c @@ -91,6 +91,9 @@ static int misrouted_irq(int irq) int i, ok = 0; for_each_irq_desc(i, desc) { + if (!desc) + continue; + if (!i) continue; @@ -112,6 +115,8 @@ static void poll_spurious_irqs(unsigned long dummy) for_each_irq_desc(i, desc) { unsigned int status; + if (!desc) + continue; if (!i) continue; -- cgit v1.2.3 From 99d093d12897562a253540a902bbf65ec16042ac Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 18:58:32 -0800 Subject: x86: use NR_IRQS_LEGACY Impact: cleanup Introduce NR_IRQS_LEGACY instead of hard coded number. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/include/asm/irq_vectors.h | 2 ++ arch/x86/kernel/io_apic.c | 10 +++++----- arch/x86/kernel/irqinit_32.c | 2 +- arch/x86/kernel/irqinit_64.c | 2 +- kernel/irq/handle.c | 6 +++--- 5 files changed, 12 insertions(+), 10 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/irq_vectors.h b/arch/x86/include/asm/irq_vectors.h index bb6b69a6b125..f7ff65032b9d 100644 --- a/arch/x86/include/asm/irq_vectors.h +++ b/arch/x86/include/asm/irq_vectors.h @@ -101,6 +101,8 @@ #define LAST_VM86_IRQ 15 #define invalid_vm86_irq(irq) ((irq) < 3 || (irq) > 15) +#define NR_IRQS_LEGACY 16 + #if defined(CONFIG_X86_IO_APIC) && !defined(CONFIG_X86_VOYAGER) #ifndef CONFIG_SPARSE_IRQ diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 9de17f5c1125..9870461ae933 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -847,7 +847,7 @@ EXPORT_SYMBOL(IO_APIC_get_PCI_irq_vector); */ static int EISA_ELCR(unsigned int irq) { - if (irq < 16) { + if (irq < NR_IRQS_LEGACY) { unsigned int port = 0x4d0 + (irq >> 3); return (inb(port) >> (irq & 7)) & 1; } @@ -1388,7 +1388,7 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, } ioapic_register_intr(irq, trigger); - if (irq < 16) + if (irq < NR_IRQS_LEGACY) disable_8259A_irq(irq); ioapic_write_entry(apic, pin, entry); @@ -2080,7 +2080,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) struct irq_cfg *cfg; spin_lock_irqsave(&ioapic_lock, flags); - if (irq < 16) { + if (irq < NR_IRQS_LEGACY) { disable_8259A_irq(irq); if (i8259A_irq_pending(irq)) was_pending = 1; @@ -2504,7 +2504,7 @@ static inline void init_IO_APIC_traps(void) * so default to an old-fashioned 8259 * interrupt if we can.. */ - if (irq < 16) + if (irq < NR_IRQS_LEGACY) make_8259A_irq(irq); else /* Strange. Oh, well.. */ @@ -3794,7 +3794,7 @@ int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int p /* * IRQs < 16 are already in the irq_2_pin[] map */ - if (irq >= 16) { + if (irq >= NR_IRQS_LEGACY) { cfg = desc->chip_data; add_pin_to_irq_cpu(irq, cpu, ioapic, pin); } diff --git a/arch/x86/kernel/irqinit_32.c b/arch/x86/kernel/irqinit_32.c index 5a5651b7f9e6..6a92f47c52e7 100644 --- a/arch/x86/kernel/irqinit_32.c +++ b/arch/x86/kernel/irqinit_32.c @@ -68,7 +68,7 @@ void __init init_ISA_irqs (void) /* * 16 old-style INTA-cycle interrupts: */ - for (i = 0; i < 16; i++) { + for (i = 0; i < NR_IRQS_LEGACY; i++) { struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; diff --git a/arch/x86/kernel/irqinit_64.c b/arch/x86/kernel/irqinit_64.c index cd9f42d028d9..40c1e62ec785 100644 --- a/arch/x86/kernel/irqinit_64.c +++ b/arch/x86/kernel/irqinit_64.c @@ -142,7 +142,7 @@ void __init init_ISA_irqs(void) init_bsp_APIC(); init_8259A(0); - for (i = 0; i < 16; i++) { + for (i = 0; i < NR_IRQS_LEGACY; i++) { struct irq_desc *desc = irq_to_desc(i); desc->status = IRQ_DISABLED; diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 96ca203eb51b..8aa09547f5ef 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -117,8 +117,8 @@ static DEFINE_SPINLOCK(sparse_irq_lock); struct irq_desc *irq_desc_ptrs[NR_IRQS] __read_mostly; -static struct irq_desc irq_desc_legacy[16] __cacheline_aligned_in_smp = { - [0 ... 15] = { +static struct irq_desc irq_desc_legacy[NR_IRQS_LEGACY] __cacheline_aligned_in_smp = { + [0 ... NR_IRQS_LEGACY-1] = { .irq = -1, .status = IRQ_DISABLED, .chip = &no_irq_chip, @@ -132,7 +132,7 @@ static struct irq_desc irq_desc_legacy[16] __cacheline_aligned_in_smp = { }; /* FIXME: use bootmem alloc ...*/ -static unsigned int kstat_irqs_legacy[16][NR_CPUS]; +static unsigned int kstat_irqs_legacy[NR_IRQS_LEGACY][NR_CPUS]; void __init early_irq_init(void) { -- cgit v1.2.3 From be5d5350a937cd8513b258739f1099420129e96f Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 18:58:33 -0800 Subject: x86: MSI start irq numbering from nr_irqs_gsi Impact: sanitize MSI irq number ordering from top-down to bottom-up Increase new MSI IRQs starting from nr_irqs_gsi (which is somewhere below 256), instead of decreasing from NR_IRQS. (The latter method can result in confusingly high IRQ numbers - if NR_CPUS is set to a high value and NR_IRQS scales up to a high value.) Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/include/asm/io_apic.h | 2 +- arch/x86/kernel/io_apic.c | 24 +++++++++++++++++------- arch/x86/kernel/setup.c | 2 +- 3 files changed, 19 insertions(+), 9 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h index 6afd9933a7dd..b35e94c2d6df 100644 --- a/arch/x86/include/asm/io_apic.h +++ b/arch/x86/include/asm/io_apic.h @@ -188,7 +188,7 @@ extern void restore_IO_APIC_setup(void); extern void reinit_intr_remapped_IO_APIC(int); #endif -extern int probe_nr_irqs(void); +extern void probe_nr_irqs_gsi(void); #else /* !CONFIG_X86_IO_APIC */ #define io_apic_assign_pci_irqs 0 diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 9870461ae933..0dcde74abd1d 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -2973,7 +2973,7 @@ unsigned int create_irq_nr(unsigned int irq_want) irq = 0; spin_lock_irqsave(&vector_lock, flags); - for (new = irq_want; new > 0; new--) { + for (new = irq_want; new < NR_IRQS; new++) { if (platform_legacy_irq(new)) continue; @@ -3001,11 +3001,14 @@ unsigned int create_irq_nr(unsigned int irq_want) return irq; } +static int nr_irqs_gsi = NR_IRQS_LEGACY; int create_irq(void) { + unsigned int irq_want; int irq; - irq = create_irq_nr(nr_irqs - 1); + irq_want = nr_irqs_gsi; + irq = create_irq_nr(irq_want); if (irq == 0) irq = -1; @@ -3281,7 +3284,7 @@ int arch_setup_msi_irq(struct pci_dev *dev, struct msi_desc *msidesc) int ret; unsigned int irq_want; - irq_want = nr_irqs - 1; + irq_want = nr_irqs_gsi; irq = create_irq_nr(irq_want); if (irq == 0) return -1; @@ -3321,11 +3324,11 @@ int arch_setup_msi_irqs(struct pci_dev *dev, int nvec, int type) int index = 0; #endif - irq_want = nr_irqs - 1; + irq_want = nr_irqs_gsi; sub_handle = 0; list_for_each_entry(msidesc, &dev->msi_list, list) { irq = create_irq_nr(irq_want); - irq_want--; + irq_want++; if (irq == 0) return -1; #ifdef CONFIG_INTR_REMAP @@ -3674,9 +3677,16 @@ int __init io_apic_get_redir_entries (int ioapic) return reg_01.bits.entries; } -int __init probe_nr_irqs(void) +void __init probe_nr_irqs_gsi(void) { - return NR_IRQS; + int idx; + int nr = 0; + + for (idx = 0; idx < nr_ioapics; idx++) + nr += io_apic_get_redir_entries(idx) + 1; + + if (nr > nr_irqs_gsi) + nr_irqs_gsi = nr; } /* -------------------------------------------------------------------------- diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 9d5674f7b6cc..a3834f123206 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -1082,7 +1082,7 @@ void __init setup_arch(char **cmdline_p) ioapic_init_mappings(); /* need to wait for io_apic is mapped */ - nr_irqs = probe_nr_irqs(); + probe_nr_irqs_gsi(); kvm_guest_init(); -- cgit v1.2.3 From 3145e941fcfe2548fa2270afb1a05bab3a6bc418 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 5 Dec 2008 18:58:34 -0800 Subject: x86, MSI: pass irq_cfg and irq_desc Impact: simplify code Pass irq_desc and cfg around, instead of raw IRQ numbers - this way we dont have to look it up again and again. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/io_apic.c | 318 ++++++++++++++++++++++++++-------------------- drivers/pci/msi.c | 55 +++++--- include/linux/msi.h | 3 + 3 files changed, 222 insertions(+), 154 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index 0dcde74abd1d..a1a2e070f31a 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -231,6 +231,10 @@ static struct irq_cfg *irq_cfg(unsigned int irq) #endif +static inline void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask) +{ +} + struct io_apic { unsigned int index; unsigned int unused[3]; @@ -272,11 +276,10 @@ static inline void io_apic_modify(unsigned int apic, unsigned int reg, unsigned writel(value, &io_apic->data); } -static bool io_apic_level_ack_pending(unsigned int irq) +static bool io_apic_level_ack_pending(struct irq_cfg *cfg) { struct irq_pin_list *entry; unsigned long flags; - struct irq_cfg *cfg = irq_cfg(irq); spin_lock_irqsave(&ioapic_lock, flags); entry = cfg->irq_2_pin; @@ -358,13 +361,12 @@ static void ioapic_mask_entry(int apic, int pin) } #ifdef CONFIG_SMP -static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) +static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, struct irq_cfg *cfg) { int apic, pin; - struct irq_cfg *cfg; struct irq_pin_list *entry; + u8 vector = cfg->vector; - cfg = irq_cfg(irq); entry = cfg->irq_2_pin; for (;;) { unsigned int reg; @@ -394,24 +396,27 @@ static void __target_IO_APIC_irq(unsigned int irq, unsigned int dest, u8 vector) } } -static int assign_irq_vector(int irq, cpumask_t mask); +static int assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask); -static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +static void set_ioapic_affinity_irq_desc(struct irq_desc *desc, cpumask_t mask) { struct irq_cfg *cfg; unsigned long flags; unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; + unsigned int irq; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; - cfg = irq_cfg(irq); - if (assign_irq_vector(irq, mask)) + irq = desc->irq; + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); /* @@ -419,12 +424,20 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) */ dest = SET_APIC_LOGICAL_ID(dest); - desc = irq_to_desc(irq); spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg->vector); + __target_IO_APIC_irq(irq, dest, cfg); desc->affinity = mask; spin_unlock_irqrestore(&ioapic_lock, flags); } + +static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +{ + struct irq_desc *desc; + + desc = irq_to_desc(irq); + + set_ioapic_affinity_irq_desc(desc, mask); +} #endif /* CONFIG_SMP */ /* @@ -432,10 +445,9 @@ static void set_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) * shared ISA-space IRQs, so we have to support them. We are super * fast in the common case, and fast for shared ISA-space IRQs. */ -static void add_pin_to_irq_cpu(unsigned int irq, int cpu, int apic, int pin) +static void add_pin_to_irq_cpu(struct irq_cfg *cfg, int cpu, int apic, int pin) { struct irq_pin_list *entry; - struct irq_cfg *cfg = irq_cfg(irq); entry = cfg->irq_2_pin; if (!entry) { @@ -468,11 +480,10 @@ static void add_pin_to_irq_cpu(unsigned int irq, int cpu, int apic, int pin) /* * Reroute an IRQ to a different pin. */ -static void __init replace_pin_at_irq(unsigned int irq, int cpu, +static void __init replace_pin_at_irq_cpu(struct irq_cfg *cfg, int cpu, int oldapic, int oldpin, int newapic, int newpin) { - struct irq_cfg *cfg = irq_cfg(irq); struct irq_pin_list *entry = cfg->irq_2_pin; int replaced = 0; @@ -489,18 +500,16 @@ static void __init replace_pin_at_irq(unsigned int irq, int cpu, /* why? call replace before add? */ if (!replaced) - add_pin_to_irq_cpu(irq, cpu, newapic, newpin); + add_pin_to_irq_cpu(cfg, cpu, newapic, newpin); } -static inline void io_apic_modify_irq(unsigned int irq, +static inline void io_apic_modify_irq(struct irq_cfg *cfg, int mask_and, int mask_or, void (*final)(struct irq_pin_list *entry)) { int pin; - struct irq_cfg *cfg; struct irq_pin_list *entry; - cfg = irq_cfg(irq); for (entry = cfg->irq_2_pin; entry != NULL; entry = entry->next) { unsigned int reg; pin = entry->pin; @@ -513,9 +522,9 @@ static inline void io_apic_modify_irq(unsigned int irq, } } -static void __unmask_IO_APIC_irq(unsigned int irq) +static void __unmask_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~IO_APIC_REDIR_MASKED, 0, NULL); + io_apic_modify_irq(cfg, ~IO_APIC_REDIR_MASKED, 0, NULL); } #ifdef CONFIG_X86_64 @@ -530,47 +539,64 @@ void io_apic_sync(struct irq_pin_list *entry) readl(&io_apic->data); } -static void __mask_IO_APIC_irq(unsigned int irq) +static void __mask_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~0, IO_APIC_REDIR_MASKED, &io_apic_sync); + io_apic_modify_irq(cfg, ~0, IO_APIC_REDIR_MASKED, &io_apic_sync); } #else /* CONFIG_X86_32 */ -static void __mask_IO_APIC_irq(unsigned int irq) +static void __mask_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~0, IO_APIC_REDIR_MASKED, NULL); + io_apic_modify_irq(cfg, ~0, IO_APIC_REDIR_MASKED, NULL); } -static void __mask_and_edge_IO_APIC_irq(unsigned int irq) +static void __mask_and_edge_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~IO_APIC_REDIR_LEVEL_TRIGGER, + io_apic_modify_irq(cfg, ~IO_APIC_REDIR_LEVEL_TRIGGER, IO_APIC_REDIR_MASKED, NULL); } -static void __unmask_and_level_IO_APIC_irq(unsigned int irq) +static void __unmask_and_level_IO_APIC_irq(struct irq_cfg *cfg) { - io_apic_modify_irq(irq, ~IO_APIC_REDIR_MASKED, + io_apic_modify_irq(cfg, ~IO_APIC_REDIR_MASKED, IO_APIC_REDIR_LEVEL_TRIGGER, NULL); } #endif /* CONFIG_X86_32 */ -static void mask_IO_APIC_irq (unsigned int irq) +static void mask_IO_APIC_irq_desc(struct irq_desc *desc) { + struct irq_cfg *cfg = desc->chip_data; unsigned long flags; + BUG_ON(!cfg); + spin_lock_irqsave(&ioapic_lock, flags); - __mask_IO_APIC_irq(irq); + __mask_IO_APIC_irq(cfg); spin_unlock_irqrestore(&ioapic_lock, flags); } -static void unmask_IO_APIC_irq (unsigned int irq) +static void unmask_IO_APIC_irq_desc(struct irq_desc *desc) { + struct irq_cfg *cfg = desc->chip_data; unsigned long flags; spin_lock_irqsave(&ioapic_lock, flags); - __unmask_IO_APIC_irq(irq); + __unmask_IO_APIC_irq(cfg); spin_unlock_irqrestore(&ioapic_lock, flags); } +static void mask_IO_APIC_irq(unsigned int irq) +{ + struct irq_desc *desc = irq_to_desc(irq); + + mask_IO_APIC_irq_desc(desc); +} +static void unmask_IO_APIC_irq(unsigned int irq) +{ + struct irq_desc *desc = irq_to_desc(irq); + + unmask_IO_APIC_irq_desc(desc); +} + static void clear_IO_APIC_pin(unsigned int apic, unsigned int pin) { struct IO_APIC_route_entry entry; @@ -1072,7 +1098,7 @@ void unlock_vector_lock(void) spin_unlock(&vector_lock); } -static int __assign_irq_vector(int irq, cpumask_t mask) +static int __assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask) { /* * NOTE! The local APIC isn't very good at handling @@ -1088,16 +1114,13 @@ static int __assign_irq_vector(int irq, cpumask_t mask) static int current_vector = FIRST_DEVICE_VECTOR, current_offset = 0; unsigned int old_vector; int cpu; - struct irq_cfg *cfg; - cfg = irq_cfg(irq); + 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); - if ((cfg->move_in_progress) || cfg->move_cleanup_count) - return -EBUSY; - old_vector = cfg->vector; if (old_vector) { cpumask_t tmp; @@ -1151,24 +1174,22 @@ next: return -ENOSPC; } -static int assign_irq_vector(int irq, cpumask_t mask) +static int assign_irq_vector(int irq, struct irq_cfg *cfg, cpumask_t mask) { int err; unsigned long flags; spin_lock_irqsave(&vector_lock, flags); - err = __assign_irq_vector(irq, mask); + err = __assign_irq_vector(irq, cfg, mask); spin_unlock_irqrestore(&vector_lock, flags); return err; } -static void __clear_irq_vector(int irq) +static void __clear_irq_vector(int irq, struct irq_cfg *cfg) { - struct irq_cfg *cfg; cpumask_t mask; int cpu, vector; - cfg = irq_cfg(irq); BUG_ON(!cfg->vector); vector = cfg->vector; @@ -1257,11 +1278,8 @@ static inline int IO_APIC_irq_trigger(int irq) } #endif -static void ioapic_register_intr(int irq, unsigned long trigger) +static void ioapic_register_intr(int irq, struct irq_desc *desc, unsigned long trigger) { - struct irq_desc *desc; - - desc = irq_to_desc(irq); if ((trigger == IOAPIC_AUTO && IO_APIC_irq_trigger(irq)) || trigger == IOAPIC_LEVEL) @@ -1353,7 +1371,7 @@ static int setup_ioapic_entry(int apic, int irq, return 0; } -static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, +static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, struct irq_desc *desc, int trigger, int polarity) { struct irq_cfg *cfg; @@ -1363,10 +1381,10 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, if (!IO_APIC_IRQ(irq)) return; - cfg = irq_cfg(irq); + cfg = desc->chip_data; mask = TARGET_CPUS; - if (assign_irq_vector(irq, mask)) + if (assign_irq_vector(irq, cfg, mask)) return; cpus_and(mask, cfg->domain, mask); @@ -1383,11 +1401,11 @@ static void setup_IO_APIC_irq(int apic, int pin, unsigned int irq, cfg->vector)) { printk("Failed to setup ioapic entry for ioapic %d, pin %d\n", mp_ioapics[apic].mp_apicid, pin); - __clear_irq_vector(irq); + __clear_irq_vector(irq, cfg); return; } - ioapic_register_intr(irq, trigger); + ioapic_register_intr(irq, desc, trigger); if (irq < NR_IRQS_LEGACY) disable_8259A_irq(irq); @@ -1399,6 +1417,7 @@ static void __init setup_IO_APIC_irqs(void) int apic, pin, idx, irq; int notcon = 0; struct irq_desc *desc; + struct irq_cfg *cfg; int cpu = boot_cpu_id; apic_printk(APIC_VERBOSE, KERN_DEBUG "init IO_APIC IRQs\n"); @@ -1436,9 +1455,10 @@ static void __init setup_IO_APIC_irqs(void) printk(KERN_INFO "can not get irq_desc for %d\n", irq); continue; } - add_pin_to_irq_cpu(irq, cpu, apic, pin); + cfg = desc->chip_data; + add_pin_to_irq_cpu(cfg, cpu, apic, pin); - setup_IO_APIC_irq(apic, pin, irq, + setup_IO_APIC_irq(apic, pin, irq, desc, irq_trigger(idx), irq_polarity(idx)); } } @@ -2086,7 +2106,7 @@ static unsigned int startup_ioapic_irq(unsigned int irq) was_pending = 1; } cfg = irq_cfg(irq); - __unmask_IO_APIC_irq(irq); + __unmask_IO_APIC_irq(cfg); spin_unlock_irqrestore(&ioapic_lock, flags); return was_pending; @@ -2149,35 +2169,37 @@ 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(int irq, cpumask_t mask) +static void migrate_ioapic_irq_desc(struct irq_desc *desc, cpumask_t mask) { struct irq_cfg *cfg; - struct irq_desc *desc; cpumask_t tmp, cleanup_mask; 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)) return; + irq = desc->irq; if (get_irte(irq, &irte)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); - desc = irq_to_desc(irq); modify_ioapic_rte = desc->status & IRQ_LEVEL; if (modify_ioapic_rte) { spin_lock_irqsave(&ioapic_lock, flags); - __target_IO_APIC_irq(irq, dest, cfg->vector); + __target_IO_APIC_irq(irq, dest, cfg); spin_unlock_irqrestore(&ioapic_lock, flags); } @@ -2199,14 +2221,14 @@ static void migrate_ioapic_irq(int irq, cpumask_t mask) desc->affinity = mask; } -static int migrate_irq_remapped_level(int irq) +static int migrate_irq_remapped_level_desc(struct irq_desc *desc) { int ret = -1; - struct irq_desc *desc = irq_to_desc(irq); + struct irq_cfg *cfg = desc->chip_data; - mask_IO_APIC_irq(irq); + mask_IO_APIC_irq_desc(desc); - if (io_apic_level_ack_pending(irq)) { + if (io_apic_level_ack_pending(cfg)) { /* * Interrupt in progress. Migrating irq now will change the * vector information in the IO-APIC RTE and that will confuse @@ -2218,14 +2240,15 @@ static int migrate_irq_remapped_level(int irq) } /* everthing is clear. we have right of way */ - migrate_ioapic_irq(irq, desc->pending_mask); + migrate_ioapic_irq_desc(desc, desc->pending_mask); ret = 0; desc->status &= ~IRQ_MOVE_PENDING; cpus_clear(desc->pending_mask); unmask: - unmask_IO_APIC_irq(irq); + unmask_IO_APIC_irq_desc(desc); + return ret; } @@ -2258,18 +2281,22 @@ static void ir_irq_migration(struct work_struct *work) /* * Migrates the IRQ destination in the process context. */ -static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +static void set_ir_ioapic_affinity_irq_desc(struct irq_desc *desc, cpumask_t mask) { - struct irq_desc *desc = irq_to_desc(irq); - if (desc->status & IRQ_LEVEL) { desc->status |= IRQ_MOVE_PENDING; desc->pending_mask = mask; - migrate_irq_remapped_level(irq); + migrate_irq_remapped_level_desc(desc); return; } - migrate_ioapic_irq(irq, mask); + migrate_ioapic_irq_desc(desc, mask); +} +static void set_ir_ioapic_affinity_irq(unsigned int irq, cpumask_t mask) +{ + struct irq_desc *desc = irq_to_desc(irq); + + set_ir_ioapic_affinity_irq_desc(desc, mask); } #endif @@ -2313,9 +2340,10 @@ unlock: irq_exit(); } -static void irq_complete_move(unsigned int irq) +static void irq_complete_move(struct irq_desc **descp) { - struct irq_cfg *cfg = irq_cfg(irq); + struct irq_desc *desc = *descp; + struct irq_cfg *cfg = desc->chip_data; unsigned vector, me; if (likely(!cfg->move_in_progress)) @@ -2333,8 +2361,9 @@ static void irq_complete_move(unsigned int irq) } } #else -static inline void irq_complete_move(unsigned int irq) {} +static inline void irq_complete_move(struct irq_desc **descp) {} #endif + #ifdef CONFIG_INTR_REMAP static void ack_x2apic_level(unsigned int irq) { @@ -2345,11 +2374,14 @@ static void ack_x2apic_edge(unsigned int irq) { ack_x2APIC_irq(); } + #endif static void ack_apic_edge(unsigned int irq) { - irq_complete_move(irq); + struct irq_desc *desc = irq_to_desc(irq); + + irq_complete_move(&desc); move_native_irq(irq); ack_APIC_irq(); } @@ -2358,18 +2390,21 @@ atomic_t irq_mis_count; static void ack_apic_level(unsigned int irq) { + struct irq_desc *desc = irq_to_desc(irq); + #ifdef CONFIG_X86_32 unsigned long v; int i; #endif + struct irq_cfg *cfg; int do_unmask_irq = 0; - irq_complete_move(irq); + irq_complete_move(&desc); #ifdef CONFIG_GENERIC_PENDING_IRQ /* If we are moving the irq we need to mask it */ - if (unlikely(irq_to_desc(irq)->status & IRQ_MOVE_PENDING)) { + if (unlikely(desc->status & IRQ_MOVE_PENDING)) { do_unmask_irq = 1; - mask_IO_APIC_irq(irq); + mask_IO_APIC_irq_desc(desc); } #endif @@ -2393,7 +2428,8 @@ static void ack_apic_level(unsigned int irq) * operation to prevent an edge-triggered interrupt escaping meanwhile. * The idea is from Manfred Spraul. --macro */ - i = irq_cfg(irq)->vector; + cfg = desc->chip_data; + i = cfg->vector; v = apic_read(APIC_TMR + ((i & ~0x1f) >> 1)); #endif @@ -2432,17 +2468,18 @@ static void ack_apic_level(unsigned int irq) * accurate and is causing problems then it is a hardware bug * and you can go talk to the chipset vendor about it. */ - if (!io_apic_level_ack_pending(irq)) + cfg = desc->chip_data; + if (!io_apic_level_ack_pending(cfg)) move_masked_irq(irq); - unmask_IO_APIC_irq(irq); + unmask_IO_APIC_irq_desc(desc); } #ifdef CONFIG_X86_32 if (!(v & (1 << (i & 0x1f)))) { atomic_inc(&irq_mis_count); spin_lock(&ioapic_lock); - __mask_and_edge_IO_APIC_irq(irq); - __unmask_and_level_IO_APIC_irq(irq); + __mask_and_edge_IO_APIC_irq(cfg); + __unmask_and_level_IO_APIC_irq(cfg); spin_unlock(&ioapic_lock); } #endif @@ -2533,7 +2570,7 @@ static void unmask_lapic_irq(unsigned int irq) apic_write(APIC_LVT0, v & ~APIC_LVT_MASKED); } -static void ack_lapic_irq (unsigned int irq) +static void ack_lapic_irq(unsigned int irq) { ack_APIC_irq(); } @@ -2545,11 +2582,8 @@ static struct irq_chip lapic_chip __read_mostly = { .ack = ack_lapic_irq, }; -static void lapic_register_intr(int irq) +static void lapic_register_intr(int irq, struct irq_desc *desc) { - struct irq_desc *desc; - - desc = irq_to_desc(irq); desc->status &= ~IRQ_LEVEL; set_irq_chip_and_handler_name(irq, &lapic_chip, handle_edge_irq, "edge"); @@ -2653,7 +2687,9 @@ int timer_through_8259 __initdata; */ static inline void __init check_timer(void) { - struct irq_cfg *cfg = irq_cfg(0); + struct irq_desc *desc = irq_to_desc(0); + struct irq_cfg *cfg = desc->chip_data; + int cpu = boot_cpu_id; int apic1, pin1, apic2, pin2; unsigned long flags; unsigned int ver; @@ -2668,7 +2704,7 @@ static inline void __init check_timer(void) * get/set the timer IRQ vector: */ disable_8259A_irq(0); - assign_irq_vector(0, TARGET_CPUS); + assign_irq_vector(0, cfg, TARGET_CPUS); /* * As IRQ0 is to be enabled in the 8259A, the virtual @@ -2719,10 +2755,10 @@ static inline void __init check_timer(void) * Ok, does IRQ0 through the IOAPIC work? */ if (no_pin1) { - add_pin_to_irq_cpu(0, boot_cpu_id, apic1, pin1); + add_pin_to_irq_cpu(cfg, cpu, apic1, pin1); setup_timer_IRQ0_pin(apic1, pin1, cfg->vector); } - unmask_IO_APIC_irq(0); + unmask_IO_APIC_irq_desc(desc); if (timer_irq_works()) { if (nmi_watchdog == NMI_IO_APIC) { setup_nmi(); @@ -2748,9 +2784,9 @@ static inline void __init check_timer(void) /* * legacy devices should be connected to IO APIC #0 */ - replace_pin_at_irq(0, boot_cpu_id, apic1, pin1, apic2, pin2); + replace_pin_at_irq_cpu(cfg, cpu, apic1, pin1, apic2, pin2); setup_timer_IRQ0_pin(apic2, pin2, cfg->vector); - unmask_IO_APIC_irq(0); + unmask_IO_APIC_irq_desc(desc); enable_8259A_irq(0); if (timer_irq_works()) { apic_printk(APIC_QUIET, KERN_INFO "....... works.\n"); @@ -2782,7 +2818,7 @@ static inline void __init check_timer(void) apic_printk(APIC_QUIET, KERN_INFO "...trying to set up timer as Virtual Wire IRQ...\n"); - lapic_register_intr(0); + lapic_register_intr(0, desc); apic_write(APIC_LVT0, APIC_DM_FIXED | cfg->vector); /* Fixed mode */ enable_8259A_irq(0); @@ -2986,7 +3022,7 @@ unsigned int create_irq_nr(unsigned int irq_want) if (cfg_new->vector != 0) continue; - if (__assign_irq_vector(new, TARGET_CPUS) == 0) + if (__assign_irq_vector(new, cfg_new, TARGET_CPUS) == 0) irq = new; break; } @@ -3034,7 +3070,7 @@ void destroy_irq(unsigned int irq) free_irte(irq); #endif spin_lock_irqsave(&vector_lock, flags); - __clear_irq_vector(irq); + __clear_irq_vector(irq, cfg); spin_unlock_irqrestore(&vector_lock, flags); } @@ -3049,12 +3085,12 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms unsigned dest; cpumask_t tmp; + cfg = irq_cfg(irq); tmp = TARGET_CPUS; - err = assign_irq_vector(irq, tmp); + err = assign_irq_vector(irq, cfg, tmp); if (err) return err; - cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, tmp); dest = cpu_mask_to_apicid(tmp); @@ -3112,35 +3148,35 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq, struct msi_ms #ifdef CONFIG_SMP static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); - read_msi_msg(irq, &msg); + read_msi_msg_desc(desc, &msg); msg.data &= ~MSI_DATA_VECTOR_MASK; msg.data |= MSI_DATA_VECTOR(cfg->vector); msg.address_lo &= ~MSI_ADDR_DEST_ID_MASK; msg.address_lo |= MSI_ADDR_DEST_ID(dest); - write_msi_msg(irq, &msg); - desc = irq_to_desc(irq); + write_msi_msg_desc(desc, &msg); desc->affinity = mask; } - #ifdef CONFIG_INTR_REMAP /* * Migrate the MSI irq to another cpumask. This migration is @@ -3148,11 +3184,11 @@ static void set_msi_irq_affinity(unsigned int irq, cpumask_t mask) */ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; unsigned int dest; cpumask_t tmp, cleanup_mask; struct irte irte; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) @@ -3161,10 +3197,12 @@ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) if (get_irte(irq, &irte)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -3188,9 +3226,9 @@ static void ir_set_msi_irq_affinity(unsigned int irq, cpumask_t mask) cfg->move_in_progress = 0; } - desc = irq_to_desc(irq); desc->affinity = mask; } + #endif #endif /* CONFIG_SMP */ @@ -3249,7 +3287,7 @@ static int msi_alloc_irte(struct pci_dev *dev, int irq, int nvec) } #endif -static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) +static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *msidesc, int irq) { int ret; struct msi_msg msg; @@ -3258,7 +3296,7 @@ static int setup_msi_irq(struct pci_dev *dev, struct msi_desc *desc, int irq) if (ret < 0) return ret; - set_irq_msi(irq, desc); + set_irq_msi(irq, msidesc); write_msi_msg(irq, &msg); #ifdef CONFIG_INTR_REMAP @@ -3381,20 +3419,22 @@ void arch_teardown_msi_irq(unsigned int irq) #ifdef CONFIG_SMP static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; struct msi_msg msg; unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -3406,9 +3446,9 @@ static void dmar_msi_set_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); dmar_msi_write(irq, &msg); - desc = irq_to_desc(irq); desc->affinity = mask; } + #endif /* CONFIG_SMP */ struct irq_chip dmar_msi_type = { @@ -3442,8 +3482,8 @@ int arch_setup_dmar_msi(unsigned int irq) #ifdef CONFIG_SMP static void hpet_msi_set_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; - struct irq_desc *desc; struct msi_msg msg; unsigned int dest; cpumask_t tmp; @@ -3452,10 +3492,12 @@ static void hpet_msi_set_affinity(unsigned int irq, cpumask_t mask) if (cpus_empty(tmp)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); @@ -3467,9 +3509,9 @@ static void hpet_msi_set_affinity(unsigned int irq, cpumask_t mask) msg.address_lo |= MSI_ADDR_DEST_ID(dest); hpet_msi_write(irq, &msg); - desc = irq_to_desc(irq); desc->affinity = mask; } + #endif /* CONFIG_SMP */ struct irq_chip hpet_msi_type = { @@ -3524,26 +3566,28 @@ static void target_ht_irq(unsigned int irq, unsigned int dest, u8 vector) static void set_ht_irq_affinity(unsigned int irq, cpumask_t mask) { + struct irq_desc *desc = irq_to_desc(irq); struct irq_cfg *cfg; unsigned int dest; cpumask_t tmp; - struct irq_desc *desc; cpus_and(tmp, mask, cpu_online_map); if (cpus_empty(tmp)) return; - if (assign_irq_vector(irq, mask)) + cfg = desc->chip_data; + if (assign_irq_vector(irq, cfg, mask)) return; - cfg = irq_cfg(irq); + set_extra_move_desc(desc, mask); + cpus_and(tmp, cfg->domain, mask); dest = cpu_mask_to_apicid(tmp); target_ht_irq(irq, dest, cfg->vector); - desc = irq_to_desc(irq); desc->affinity = mask; } + #endif static struct irq_chip ht_irq_chip = { @@ -3563,13 +3607,13 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev) int err; cpumask_t tmp; + cfg = irq_cfg(irq); tmp = TARGET_CPUS; - err = assign_irq_vector(irq, tmp); + err = assign_irq_vector(irq, cfg, tmp); if (!err) { struct ht_irq_msg msg; unsigned dest; - cfg = irq_cfg(irq); cpus_and(tmp, cfg->domain, tmp); dest = cpu_mask_to_apicid(tmp); @@ -3615,7 +3659,9 @@ int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, unsigned long flags; int err; - err = assign_irq_vector(irq, *eligible_cpu); + cfg = irq_cfg(irq); + + err = assign_irq_vector(irq, cfg, *eligible_cpu); if (err != 0) return err; @@ -3624,8 +3670,6 @@ int arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade, irq_name); spin_unlock_irqrestore(&vector_lock, flags); - cfg = irq_cfg(irq); - mmr_value = 0; entry = (struct uv_IO_APIC_route_entry *)&mmr_value; BUG_ON(sizeof(struct uv_IO_APIC_route_entry) != sizeof(unsigned long)); @@ -3806,10 +3850,10 @@ int io_apic_set_pci_routing (int ioapic, int pin, int irq, int triggering, int p */ if (irq >= NR_IRQS_LEGACY) { cfg = desc->chip_data; - add_pin_to_irq_cpu(irq, cpu, ioapic, pin); + add_pin_to_irq_cpu(cfg, cpu, ioapic, pin); } - setup_IO_APIC_irq(ioapic, pin, irq, triggering, polarity); + setup_IO_APIC_irq(ioapic, pin, irq, desc, triggering, polarity); return 0; } @@ -3866,7 +3910,7 @@ void __init setup_ioapic_dest(void) desc = irq_to_desc(irq); cfg = desc->chip_data; if (!cfg->vector) { - setup_IO_APIC_irq(ioapic, pin, irq, + setup_IO_APIC_irq(ioapic, pin, irq, desc, irq_trigger(irq_entry), irq_polarity(irq_entry)); continue; @@ -3884,10 +3928,10 @@ void __init setup_ioapic_dest(void) #ifdef CONFIG_INTR_REMAP if (intr_remapping_enabled) - set_ir_ioapic_affinity_irq(irq, mask); + set_ir_ioapic_affinity_irq_desc(desc, mask); else #endif - set_ioapic_affinity_irq(irq, mask); + set_ioapic_affinity_irq_desc(desc, mask); } } diff --git a/drivers/pci/msi.c b/drivers/pci/msi.c index 74801f7df9c9..11a51f8ed3b3 100644 --- a/drivers/pci/msi.c +++ b/drivers/pci/msi.c @@ -103,11 +103,11 @@ static void msix_set_enable(struct pci_dev *dev, int enable) } } -static void msix_flush_writes(unsigned int irq) +static void msix_flush_writes(struct irq_desc *desc) { struct msi_desc *entry; - entry = get_irq_msi(irq); + entry = get_irq_desc_msi(desc); BUG_ON(!entry || !entry->dev); switch (entry->msi_attrib.type) { case PCI_CAP_ID_MSI: @@ -135,11 +135,11 @@ static void msix_flush_writes(unsigned int irq) * Returns 1 if it succeeded in masking the interrupt and 0 if the device * doesn't support MSI masking. */ -static int msi_set_mask_bits(unsigned int irq, u32 mask, u32 flag) +static int msi_set_mask_bits(struct irq_desc *desc, u32 mask, u32 flag) { struct msi_desc *entry; - entry = get_irq_msi(irq); + entry = get_irq_desc_msi(desc); BUG_ON(!entry || !entry->dev); switch (entry->msi_attrib.type) { case PCI_CAP_ID_MSI: @@ -172,9 +172,9 @@ static int msi_set_mask_bits(unsigned int irq, u32 mask, u32 flag) return 1; } -void read_msi_msg(unsigned int irq, struct msi_msg *msg) +void read_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg) { - struct msi_desc *entry = get_irq_msi(irq); + struct msi_desc *entry = get_irq_desc_msi(desc); switch(entry->msi_attrib.type) { case PCI_CAP_ID_MSI: { @@ -211,9 +211,16 @@ void read_msi_msg(unsigned int irq, struct msi_msg *msg) } } -void write_msi_msg(unsigned int irq, struct msi_msg *msg) +void read_msi_msg(unsigned int irq, struct msi_msg *msg) { - struct msi_desc *entry = get_irq_msi(irq); + struct irq_desc *desc = irq_to_desc(irq); + + read_msi_msg_desc(desc, msg); +} + +void write_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg) +{ + struct msi_desc *entry = get_irq_desc_msi(desc); switch (entry->msi_attrib.type) { case PCI_CAP_ID_MSI: { @@ -252,21 +259,31 @@ void write_msi_msg(unsigned int irq, struct msi_msg *msg) entry->msg = *msg; } +void write_msi_msg(unsigned int irq, struct msi_msg *msg) +{ + struct irq_desc *desc = irq_to_desc(irq); + + write_msi_msg_desc(desc, msg); +} + void mask_msi_irq(unsigned int irq) { - msi_set_mask_bits(irq, 1, 1); - msix_flush_writes(irq); + struct irq_desc *desc = irq_to_desc(irq); + + msi_set_mask_bits(desc, 1, 1); + msix_flush_writes(desc); } void unmask_msi_irq(unsigned int irq) { - msi_set_mask_bits(irq, 1, 0); - msix_flush_writes(irq); + struct irq_desc *desc = irq_to_desc(irq); + + msi_set_mask_bits(desc, 1, 0); + msix_flush_writes(desc); } static int msi_free_irqs(struct pci_dev* dev); - static struct msi_desc* alloc_msi_entry(void) { struct msi_desc *entry; @@ -303,9 +320,11 @@ static void __pci_restore_msi_state(struct pci_dev *dev) pci_intx_for_msi(dev, 0); msi_set_enable(dev, 0); write_msi_msg(dev->irq, &entry->msg); - if (entry->msi_attrib.maskbit) - msi_set_mask_bits(dev->irq, entry->msi_attrib.maskbits_mask, + if (entry->msi_attrib.maskbit) { + struct irq_desc *desc = irq_to_desc(dev->irq); + msi_set_mask_bits(desc, entry->msi_attrib.maskbits_mask, entry->msi_attrib.masked); + } pci_read_config_word(dev, pos + PCI_MSI_FLAGS, &control); control &= ~PCI_MSI_FLAGS_QSIZE; @@ -327,8 +346,9 @@ static void __pci_restore_msix_state(struct pci_dev *dev) msix_set_enable(dev, 0); list_for_each_entry(entry, &dev->msi_list, list) { + struct irq_desc *desc = irq_to_desc(entry->irq); write_msi_msg(entry->irq, &entry->msg); - msi_set_mask_bits(entry->irq, 1, entry->msi_attrib.masked); + msi_set_mask_bits(desc, 1, entry->msi_attrib.masked); } BUG_ON(list_empty(&dev->msi_list)); @@ -596,7 +616,8 @@ void pci_msi_shutdown(struct pci_dev* dev) /* Return the the pci reset with msi irqs unmasked */ if (entry->msi_attrib.maskbit) { u32 mask = entry->msi_attrib.maskbits_mask; - msi_set_mask_bits(dev->irq, mask, ~mask); + struct irq_desc *desc = irq_to_desc(dev->irq); + msi_set_mask_bits(desc, mask, ~mask); } if (!entry->dev || entry->msi_attrib.type != PCI_CAP_ID_MSI) return; diff --git a/include/linux/msi.h b/include/linux/msi.h index 8f2939227207..d2b8a1e8ca11 100644 --- a/include/linux/msi.h +++ b/include/linux/msi.h @@ -10,8 +10,11 @@ struct msi_msg { }; /* Helper functions */ +struct irq_desc; extern void mask_msi_irq(unsigned int irq); extern void unmask_msi_irq(unsigned int irq); +extern void read_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg); +extern void write_msi_msg_desc(struct irq_desc *desc, struct msi_msg *msg); extern void read_msi_msg(unsigned int irq, struct msi_msg *msg); extern void write_msi_msg(unsigned int irq, struct msi_msg *msg); -- cgit v1.2.3 From 50dd94e017ec39f85c26b6c10ed9fb2d7a7d8042 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 8 Dec 2008 18:47:51 +0100 Subject: sparseirq: fix typo in !CONFIG_IO_APIC case Impact: build fix Signed-off-by: Ingo Molnar --- arch/x86/include/asm/io_apic.h | 7 ++----- 1 file changed, 2 insertions(+), 5 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/io_apic.h b/arch/x86/include/asm/io_apic.h index b35e94c2d6df..25d527ca1362 100644 --- a/arch/x86/include/asm/io_apic.h +++ b/arch/x86/include/asm/io_apic.h @@ -193,12 +193,9 @@ extern void probe_nr_irqs_gsi(void); #else /* !CONFIG_X86_IO_APIC */ #define io_apic_assign_pci_irqs 0 static const int timer_through_8259 = 0; -static inline void ioapic_init_mappings(void) { } +static inline void ioapic_init_mappings(void) { } -static inline int probe_nr_irqs(void) -{ - return NR_IRQS; -} +static inline void probe_nr_irqs_gsi(void) { } #endif #endif /* _ASM_X86_IO_APIC_H */ -- cgit v1.2.3 From 8a4830f8891be6b4e04809693a24771a4694e0b0 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Tue, 9 Dec 2008 11:56:20 -0800 Subject: sparseirq: fix !SMP && !PCI_MSI && !HT_IRQ build Ingo Molnar wrote: >>> drivers/pci/intr_remapping.c: In function 'irq_2_iommu_alloc': >>> drivers/pci/intr_remapping.c:72: error: 'boot_cpu_id' undeclared (first use in this function) >>> drivers/pci/intr_remapping.c:72: error: (Each undeclared identifier is reported only once >>> drivers/pci/intr_remapping.c:72: error: for each function it appears in.) sparseirq should only be used with SMP for now. --- arch/x86/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 48ac688de3cd..8943c13502c6 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -240,7 +240,7 @@ config X86_HAS_BOOT_CPU_ID config SPARSE_IRQ bool "Support sparse irq numbering" - depends on PCI_MSI || HT_IRQ + depends on (PCI_MSI || HT_IRQ) && SMP default y help This enables support for sparse irq, esp for msi/msi-x. You may need -- cgit v1.2.3 From fd13f6c85144bb2026c534a35be1d7cb7628a64a Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Sun, 19 Oct 2008 21:00:09 +0200 Subject: oprofile: comment cleanup This fixes the coding style of some comments. Signed-off-by: Robert Richter --- arch/x86/oprofile/op_model_amd.c | 31 ++++++++++++++++--------------- drivers/oprofile/cpu_buffer.c | 25 ++++++++++++++++--------- 2 files changed, 32 insertions(+), 24 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c index 509513760a6e..fb67e1999d85 100644 --- a/arch/x86/oprofile/op_model_amd.c +++ b/arch/x86/oprofile/op_model_amd.c @@ -65,8 +65,10 @@ static unsigned long reset_value[NUM_COUNTERS]; #define IBS_FETCH_BEGIN 3 #define IBS_OP_BEGIN 4 -/* The function interface needs to be fixed, something like add - data. Should then be added to linux/oprofile.h. */ +/* + * The function interface needs to be fixed, something like add + * data. Should then be added to linux/oprofile.h. + */ extern void oprofile_add_ibs_sample(struct pt_regs *const regs, unsigned int *const ibs_sample, int ibs_code); @@ -106,7 +108,7 @@ struct ibs_op_sample { /* * unitialize the APIC for the IBS interrupts if needed on AMD Family10h+ -*/ + */ static void clear_ibs_nmi(void); static int ibs_allowed; /* AMD Family10h and later */ @@ -223,7 +225,7 @@ op_amd_handle_ibs(struct pt_regs * const regs, (unsigned int *)&ibs_fetch, IBS_FETCH_BEGIN); - /*reenable the IRQ */ + /* reenable the IRQ */ rdmsr(MSR_AMD64_IBSFETCHCTL, low, high); high &= ~IBS_FETCH_HIGH_VALID_BIT; high |= IBS_FETCH_HIGH_ENABLE; @@ -331,8 +333,10 @@ static void op_amd_stop(struct op_msrs const * const msrs) unsigned int low, high; int i; - /* Subtle: stop on all counters to avoid race with - * setting our pm callback */ + /* + * Subtle: stop on all counters to avoid race with setting our + * pm callback + */ for (i = 0 ; i < NUM_COUNTERS ; ++i) { if (!reset_value[i]) continue; @@ -343,13 +347,15 @@ static void op_amd_stop(struct op_msrs const * const msrs) #ifdef CONFIG_OPROFILE_IBS if (ibs_allowed && ibs_config.fetch_enabled) { - low = 0; /* clear max count and enable */ + /* clear max count and enable */ + low = 0; high = 0; wrmsr(MSR_AMD64_IBSFETCHCTL, low, high); } if (ibs_allowed && ibs_config.op_enabled) { - low = 0; /* clear max count and enable */ + /* clear max count and enable */ + low = 0; high = 0; wrmsr(MSR_AMD64_IBSOPCTL, low, high); } @@ -443,10 +449,7 @@ static int pfm_amd64_setup_eilvt(void) return 0; } -/* - * initialize the APIC for the IBS interrupts - * if available (AMD Family10h rev B0 and later) - */ +/* initialize the APIC for the IBS interrupts if available */ static void setup_ibs(void) { ibs_allowed = boot_cpu_has(X86_FEATURE_IBS); @@ -463,9 +466,7 @@ static void setup_ibs(void) } -/* - * unitialize the APIC for the IBS interrupts if needed on AMD Family10h - * rev B0 and later */ +/* uninitialize the APIC for the IBS interrupts if needed */ static void clear_ibs_nmi(void) { if (ibs_allowed) diff --git a/drivers/oprofile/cpu_buffer.c b/drivers/oprofile/cpu_buffer.c index 01d38e78cde1..3958107723fb 100644 --- a/drivers/oprofile/cpu_buffer.c +++ b/drivers/oprofile/cpu_buffer.c @@ -127,9 +127,10 @@ void end_cpu_work(void) /* Resets the cpu buffer to a sane state. */ void cpu_buffer_reset(struct oprofile_cpu_buffer *cpu_buf) { - /* reset these to invalid values; the next sample - * collected will populate the buffer with proper - * values to initialize the buffer + /* + * reset these to invalid values; the next sample collected + * will populate the buffer with proper values to initialize + * the buffer */ cpu_buf->last_is_kernel = -1; cpu_buf->last_task = NULL; @@ -151,8 +152,10 @@ static void increment_head(struct oprofile_cpu_buffer *b) { unsigned long new_head = b->head_pos + 1; - /* Ensure anything written to the slot before we - * increment is visible */ + /* + * Ensure anything written to the slot before we increment is + * visible + */ wmb(); if (new_head < b->buffer_size) @@ -253,8 +256,10 @@ void oprofile_add_ext_sample(unsigned long pc, struct pt_regs * const regs, if (!oprofile_begin_trace(cpu_buf)) return; - /* if log_sample() fail we can't backtrace since we lost the source - * of this event */ + /* + * if log_sample() fail we can't backtrace since we lost the + * source of this event + */ if (log_sample(cpu_buf, pc, is_kernel, event)) oprofile_ops.backtrace(regs, backtrace_depth); oprofile_end_trace(cpu_buf); @@ -338,8 +343,10 @@ void oprofile_add_trace(unsigned long pc) return; } - /* broken frame can give an eip with the same value as an escape code, - * abort the trace if we get it */ + /* + * broken frame can give an eip with the same value as an + * escape code, abort the trace if we get it + */ if (pc == ESCAPE_CODE) { cpu_buf->tracing = 0; cpu_buf->backtrace_aborted++; -- cgit v1.2.3 From cdc1834d1aa2e5b574a25e66f82625b44cdd0d8f Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Fri, 26 Sep 2008 22:18:44 -0400 Subject: oprofile: whitspace changes only Signed-off-by: Robert Richter --- arch/x86/oprofile/op_model_amd.c | 4 ++-- drivers/oprofile/cpu_buffer.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c index fb67e1999d85..f71bd218b488 100644 --- a/arch/x86/oprofile/op_model_amd.c +++ b/arch/x86/oprofile/op_model_amd.c @@ -70,8 +70,8 @@ static unsigned long reset_value[NUM_COUNTERS]; * data. Should then be added to linux/oprofile.h. */ extern void -oprofile_add_ibs_sample(struct pt_regs *const regs, - unsigned int *const ibs_sample, int ibs_code); +oprofile_add_ibs_sample(struct pt_regs * const regs, + unsigned int * const ibs_sample, int ibs_code); struct ibs_fetch_sample { /* MSRC001_1031 IBS Fetch Linear Address Register */ diff --git a/drivers/oprofile/cpu_buffer.c b/drivers/oprofile/cpu_buffer.c index 3958107723fb..2c4d54187b90 100644 --- a/drivers/oprofile/cpu_buffer.c +++ b/drivers/oprofile/cpu_buffer.c @@ -277,8 +277,8 @@ void oprofile_add_sample(struct pt_regs * const regs, unsigned long event) #define MAX_IBS_SAMPLE_SIZE 14 -void oprofile_add_ibs_sample(struct pt_regs *const regs, - unsigned int *const ibs_sample, int ibs_code) +void oprofile_add_ibs_sample(struct pt_regs * const regs, + unsigned int * const ibs_sample, int ibs_code) { int is_kernel = !user_mode(regs); struct oprofile_cpu_buffer *cpu_buf = &__get_cpu_var(cpu_buffer); -- cgit v1.2.3 From 9fa6812dbab9207f7af52c3d0417f1f9eb89c386 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Mon, 24 Nov 2008 14:21:03 +0100 Subject: x86/oprofile: reordering IBS code in op_model_amd.c This is part of the cpu buffer rework. Signed-off-by: Robert Richter --- arch/x86/oprofile/op_model_amd.c | 26 +++++++++++++------------- 1 file changed, 13 insertions(+), 13 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c index f71bd218b488..8ff657b3ff89 100644 --- a/arch/x86/oprofile/op_model_amd.c +++ b/arch/x86/oprofile/op_model_amd.c @@ -376,18 +376,7 @@ static void op_amd_shutdown(struct op_msrs const * const msrs) } } -#ifndef CONFIG_OPROFILE_IBS - -/* no IBS support */ - -static int op_amd_init(struct oprofile_operations *ops) -{ - return 0; -} - -static void op_amd_exit(void) {} - -#else +#ifdef CONFIG_OPROFILE_IBS static u8 ibs_eilvt_off; @@ -531,7 +520,18 @@ static void op_amd_exit(void) clear_ibs_nmi(); } -#endif +#else + +/* no IBS support */ + +static int op_amd_init(struct oprofile_operations *ops) +{ + return 0; +} + +static void op_amd_exit(void) {} + +#endif /* CONFIG_OPROFILE_IBS */ struct op_x86_model_spec const op_amd_spec = { .init = op_amd_init, -- cgit v1.2.3 From fe615cbf34fc6a1c53c359417da4696328a488ed Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Mon, 24 Nov 2008 14:58:03 +0100 Subject: x86/oprofile: cleanup IBS init/exit functions in op_model_amd.c Implementation of pairwise init/exit funcions for IBS and IBS NMI setup. There are also some function renames and the removal of forward function declarations. Signed-off-by: Robert Richter --- arch/x86/oprofile/op_model_amd.c | 32 +++++++++++++++++--------------- 1 file changed, 17 insertions(+), 15 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/oprofile/op_model_amd.c b/arch/x86/oprofile/op_model_amd.c index 8ff657b3ff89..98658f25f542 100644 --- a/arch/x86/oprofile/op_model_amd.c +++ b/arch/x86/oprofile/op_model_amd.c @@ -106,11 +106,6 @@ struct ibs_op_sample { unsigned int ibs_dc_phys_high; }; -/* - * unitialize the APIC for the IBS interrupts if needed on AMD Family10h+ - */ -static void clear_ibs_nmi(void); - static int ibs_allowed; /* AMD Family10h and later */ struct op_ibs_config { @@ -390,7 +385,7 @@ static inline void apic_clear_ibs_nmi_per_cpu(void *arg) setup_APIC_eilvt_ibs(0, APIC_EILVT_MSG_FIX, 1); } -static int pfm_amd64_setup_eilvt(void) +static int init_ibs_nmi(void) { #define IBSCTL_LVTOFFSETVAL (1 << 8) #define IBSCTL 0x1cc @@ -438,15 +433,22 @@ static int pfm_amd64_setup_eilvt(void) return 0; } +/* uninitialize the APIC for the IBS interrupts if needed */ +static void clear_ibs_nmi(void) +{ + if (ibs_allowed) + on_each_cpu(apic_clear_ibs_nmi_per_cpu, NULL, 1); +} + /* initialize the APIC for the IBS interrupts if available */ -static void setup_ibs(void) +static void ibs_init(void) { ibs_allowed = boot_cpu_has(X86_FEATURE_IBS); if (!ibs_allowed) return; - if (pfm_amd64_setup_eilvt()) { + if (init_ibs_nmi()) { ibs_allowed = 0; return; } @@ -454,12 +456,12 @@ static void setup_ibs(void) printk(KERN_INFO "oprofile: AMD IBS detected\n"); } - -/* uninitialize the APIC for the IBS interrupts if needed */ -static void clear_ibs_nmi(void) +static void ibs_exit(void) { - if (ibs_allowed) - on_each_cpu(apic_clear_ibs_nmi_per_cpu, NULL, 1); + if (!ibs_allowed) + return; + + clear_ibs_nmi(); } static int (*create_arch_files)(struct super_block *sb, struct dentry *root); @@ -509,7 +511,7 @@ static int setup_ibs_files(struct super_block *sb, struct dentry *root) static int op_amd_init(struct oprofile_operations *ops) { - setup_ibs(); + ibs_init(); create_arch_files = ops->create_files; ops->create_files = setup_ibs_files; return 0; @@ -517,7 +519,7 @@ static int op_amd_init(struct oprofile_operations *ops) static void op_amd_exit(void) { - clear_ibs_nmi(); + ibs_exit(); } #else -- cgit v1.2.3 From bacbe9994541c70aa3abd1a013ac738e58d4bfb2 Mon Sep 17 00:00:00 2001 From: Janne Kulmala Date: Tue, 16 Dec 2008 13:39:57 +0200 Subject: x86: enable HPET on Fujitsu u9200 Impact: auto-enable HPET on Fujitsu u9200 HPET timer is listed in the ACPI table, but needs a quirk entry in order to work. Unfortunately, the quirk code runs after first HPET hpet_enable() which has already determined that the timer doesn't work (reads 0xFFFFFFFF). This patch allows hpet_enable() to be called again after running the quirk code. Signed-off-by: Janne Kulmala Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 2 +- arch/x86/kernel/quirks.c | 2 ++ 2 files changed, 3 insertions(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 067d8de913f6..84089dc8fd1d 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -811,7 +811,7 @@ int __init hpet_enable(void) out_nohpet: hpet_clear_mapping(); - boot_hpet_disable = 1; + hpet_address = 0; return 0; } diff --git a/arch/x86/kernel/quirks.c b/arch/x86/kernel/quirks.c index 67465ed89310..309949e9e1c1 100644 --- a/arch/x86/kernel/quirks.c +++ b/arch/x86/kernel/quirks.c @@ -168,6 +168,8 @@ DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH7_31, ich_force_enable_hpet); DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH8_1, ich_force_enable_hpet); +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH8_4, + ich_force_enable_hpet); DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_INTEL, PCI_DEVICE_ID_INTEL_ICH9_7, ich_force_enable_hpet); -- cgit v1.2.3 From a79b7a2a758c39315344f0d86b5adb21d90d786e Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 16 Dec 2008 12:17:25 -0800 Subject: x86: remove unused iommu_nr_pages Impact: cleanup, remove dead code The last usage was removed by the patch set culminating in | commit e3c449f526cebb8d287241c7e82faafd9709668b | Author: Joerg Roedel | Date: Wed Oct 15 22:02:11 2008 -0700 | | x86, AMD IOMMU: convert driver to generic iommu_num_pages function Signed-off-by: Ian Campbell Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/include/asm/iommu.h | 2 -- arch/x86/kernel/pci-dma.c | 7 ------- 2 files changed, 9 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/iommu.h b/arch/x86/include/asm/iommu.h index 0b500c5b6446..35276ec5925b 100644 --- a/arch/x86/include/asm/iommu.h +++ b/arch/x86/include/asm/iommu.h @@ -7,8 +7,6 @@ extern struct dma_mapping_ops nommu_dma_ops; extern int force_iommu, no_iommu; extern int iommu_detected; -extern unsigned long iommu_nr_pages(unsigned long addr, unsigned long len); - /* 10 seconds */ #define DMAR_OPERATION_TIMEOUT ((cycles_t) tsc_khz*10*1000) diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 192624820217..e150ad4f0ccc 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -125,13 +125,6 @@ void __init pci_iommu_alloc(void) pci_swiotlb_init(); } -unsigned long iommu_nr_pages(unsigned long addr, unsigned long len) -{ - unsigned long size = roundup((addr & ~PAGE_MASK) + len, PAGE_SIZE); - - return size >> PAGE_SHIFT; -} -EXPORT_SYMBOL(iommu_nr_pages); #endif void *dma_generic_alloc_coherent(struct device *dev, size_t size, -- cgit v1.2.3 From 39c04b55240342d0742ac48538d3d8c71bfc0a94 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 16 Dec 2008 12:32:23 -0800 Subject: x86: make sure we really have an hpet mapping before using it Impact: prepare the hpet code for Xen dom0 booting When booting in Xen dom0, the hpet isn't really accessible, so make sure the mapping is non-NULL before use. Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 84089dc8fd1d..a1f6ed5e1a05 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -834,10 +834,11 @@ static __init int hpet_late_init(void) hpet_address = force_hpet_address; hpet_enable(); - if (!hpet_virt_address) - return -ENODEV; } + if (!hpet_virt_address) + return -ENODEV; + hpet_reserve_platform_timers(hpet_readl(HPET_ID)); for_each_online_cpu(cpu) { -- cgit v1.2.3 From 48a1b10aff588833b73994704c47bbd0deb73e9c Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 11 Dec 2008 00:15:01 -0800 Subject: x86, sparseirq: move irq_desc according to smp_affinity, v7 Impact: improve NUMA handling by migrating irq_desc on smp_affinity changes if CONFIG_NUMA_MIGRATE_IRQ_DESC is set: - make irq_desc to go with affinity aka irq_desc moving etc - call move_irq_desc in irq_complete_move() - legacy irq_desc is not moved, because they are allocated via static array for logical apic mode, need to add move_desc_in_progress_in_same_domain, otherwise it will not be moved ==> also could need two phases to get irq_desc moved. Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 9 +++ arch/x86/kernel/io_apic.c | 142 +++++++++++++++++++++++++++++++++++++++++++++- include/linux/irq.h | 10 ++++ kernel/irq/Makefile | 1 + kernel/irq/chip.c | 12 +++- kernel/irq/handle.c | 15 +++-- kernel/irq/internals.h | 5 ++ kernel/irq/numa_migrate.c | 127 +++++++++++++++++++++++++++++++++++++++++ 8 files changed, 313 insertions(+), 8 deletions(-) create mode 100644 kernel/irq/numa_migrate.c (limited to 'arch/x86') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 8943c13502c6..29073532f94c 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -248,6 +248,15 @@ config SPARSE_IRQ If you don't know what to do here, say Y. +config NUMA_MIGRATE_IRQ_DESC + bool "Move irq desc when changing irq smp_affinity" + depends on SPARSE_IRQ && SMP + default n + help + This enables moving irq_desc to cpu/node that irq will use handled. + + If you don't know what to do here, say N. + config X86_FIND_SMP_CONFIG def_bool y depends on X86_MPPARSE || X86_VOYAGER diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index a1a2e070f31a..bfe1245b1a3e 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -141,6 +141,9 @@ struct irq_cfg { unsigned move_cleanup_count; u8 vector; u8 move_in_progress : 1; +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + u8 move_desc_pending : 1; +#endif }; /* irq_cfg is indexed by the sum of all RTEs in all I/O APICs. */ @@ -223,6 +226,121 @@ void arch_init_chip_data(struct irq_desc *desc, int cpu) } } +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + +static void +init_copy_irq_2_pin(struct irq_cfg *old_cfg, struct irq_cfg *cfg, int cpu) +{ + struct irq_pin_list *old_entry, *head, *tail, *entry; + + cfg->irq_2_pin = NULL; + old_entry = old_cfg->irq_2_pin; + if (!old_entry) + return; + + entry = get_one_free_irq_2_pin(cpu); + if (!entry) + return; + + entry->apic = old_entry->apic; + entry->pin = old_entry->pin; + head = entry; + tail = entry; + old_entry = old_entry->next; + while (old_entry) { + entry = get_one_free_irq_2_pin(cpu); + if (!entry) { + entry = head; + while (entry) { + head = entry->next; + kfree(entry); + entry = head; + } + /* still use the old one */ + return; + } + entry->apic = old_entry->apic; + entry->pin = old_entry->pin; + tail->next = entry; + tail = entry; + old_entry = old_entry->next; + } + + tail->next = NULL; + cfg->irq_2_pin = head; +} + +static void free_irq_2_pin(struct irq_cfg *old_cfg, struct irq_cfg *cfg) +{ + struct irq_pin_list *entry, *next; + + if (old_cfg->irq_2_pin == cfg->irq_2_pin) + return; + + entry = old_cfg->irq_2_pin; + + while (entry) { + next = entry->next; + kfree(entry); + entry = next; + } + old_cfg->irq_2_pin = NULL; +} + +void arch_init_copy_chip_data(struct irq_desc *old_desc, + struct irq_desc *desc, int cpu) +{ + struct irq_cfg *cfg; + struct irq_cfg *old_cfg; + + cfg = get_one_free_irq_cfg(cpu); + + if (!cfg) + return; + + desc->chip_data = cfg; + + old_cfg = old_desc->chip_data; + + memcpy(cfg, old_cfg, sizeof(struct irq_cfg)); + + init_copy_irq_2_pin(old_cfg, cfg, cpu); +} + +static void free_irq_cfg(struct irq_cfg *old_cfg) +{ + kfree(old_cfg); +} + +void arch_free_chip_data(struct irq_desc *old_desc, struct irq_desc *desc) +{ + struct irq_cfg *old_cfg, *cfg; + + old_cfg = old_desc->chip_data; + cfg = desc->chip_data; + + if (old_cfg == cfg) + return; + + if (old_cfg) { + free_irq_2_pin(old_cfg, cfg); + free_irq_cfg(old_cfg); + old_desc->chip_data = NULL; + } +} + +static void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask) +{ + struct irq_cfg *cfg = desc->chip_data; + + if (!cfg->move_in_progress) { + /* it means that domain is not changed */ + if (!cpus_intersects(desc->affinity, mask)) + cfg->move_desc_pending = 1; + } +} +#endif + #else static struct irq_cfg *irq_cfg(unsigned int irq) { @@ -231,9 +349,11 @@ static struct irq_cfg *irq_cfg(unsigned int irq) #endif +#ifndef CONFIG_NUMA_MIGRATE_IRQ_DESC static inline void set_extra_move_desc(struct irq_desc *desc, cpumask_t mask) { } +#endif struct io_apic { unsigned int index; @@ -2346,14 +2466,34 @@ static void irq_complete_move(struct irq_desc **descp) struct irq_cfg *cfg = desc->chip_data; unsigned vector, me; - if (likely(!cfg->move_in_progress)) + if (likely(!cfg->move_in_progress)) { +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + if (likely(!cfg->move_desc_pending)) + return; + + /* domain is not change, but affinity is changed */ + me = smp_processor_id(); + if (cpu_isset(me, desc->affinity)) { + *descp = desc = move_irq_desc(desc, me); + /* get the new one */ + cfg = desc->chip_data; + cfg->move_desc_pending = 0; + } +#endif return; + } vector = ~get_irq_regs()->orig_ax; me = smp_processor_id(); if ((vector == cfg->vector) && cpu_isset(me, cfg->domain)) { cpumask_t cleanup_mask; +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + *descp = desc = move_irq_desc(desc, me); + /* get the new one */ + cfg = desc->chip_data; +#endif + 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); diff --git a/include/linux/irq.h b/include/linux/irq.h index b5749db3e5a1..36a015746788 100644 --- a/include/linux/irq.h +++ b/include/linux/irq.h @@ -227,6 +227,16 @@ extern struct irq_desc *move_irq_desc(struct irq_desc *old_desc, int cpu); #endif +static inline struct irq_desc * +irq_remap_to_desc(unsigned int irq, struct irq_desc *desc) +{ +#ifdef CONFIG_NUMA_MIGRATE_IRQ_DESC + return irq_to_desc(irq); +#else + return desc; +#endif +} + /* * Migration helpers for obsolete names, they will go away: */ diff --git a/kernel/irq/Makefile b/kernel/irq/Makefile index 681c52dbfe22..4dd5b1edac98 100644 --- a/kernel/irq/Makefile +++ b/kernel/irq/Makefile @@ -3,3 +3,4 @@ obj-y := handle.o manage.o spurious.o resend.o chip.o devres.o obj-$(CONFIG_GENERIC_IRQ_PROBE) += autoprobe.o obj-$(CONFIG_PROC_FS) += proc.o obj-$(CONFIG_GENERIC_PENDING_IRQ) += migration.o +obj-$(CONFIG_NUMA_MIGRATE_IRQ_DESC) += numa_migrate.o diff --git a/kernel/irq/chip.c b/kernel/irq/chip.c index 8e4fce4a1b1f..de210f4b7a92 100644 --- a/kernel/irq/chip.c +++ b/kernel/irq/chip.c @@ -353,6 +353,7 @@ handle_level_irq(unsigned int irq, struct irq_desc *desc) spin_lock(&desc->lock); mask_ack_irq(desc, irq); + desc = irq_remap_to_desc(irq, desc); if (unlikely(desc->status & IRQ_INPROGRESS)) goto out_unlock; @@ -430,6 +431,7 @@ handle_fasteoi_irq(unsigned int irq, struct irq_desc *desc) desc->status &= ~IRQ_INPROGRESS; out: desc->chip->eoi(irq); + desc = irq_remap_to_desc(irq, desc); spin_unlock(&desc->lock); } @@ -466,12 +468,14 @@ handle_edge_irq(unsigned int irq, struct irq_desc *desc) !desc->action)) { desc->status |= (IRQ_PENDING | IRQ_MASKED); mask_ack_irq(desc, irq); + desc = irq_remap_to_desc(irq, desc); goto out_unlock; } kstat_incr_irqs_this_cpu(irq, desc); /* Start handling the irq */ desc->chip->ack(irq); + desc = irq_remap_to_desc(irq, desc); /* Mark the IRQ currently in progress.*/ desc->status |= IRQ_INPROGRESS; @@ -532,8 +536,10 @@ handle_percpu_irq(unsigned int irq, struct irq_desc *desc) if (!noirqdebug) note_interrupt(irq, desc, action_ret); - if (desc->chip->eoi) + if (desc->chip->eoi) { desc->chip->eoi(irq); + desc = irq_remap_to_desc(irq, desc); + } } void @@ -568,8 +574,10 @@ __set_irq_handler(unsigned int irq, irq_flow_handler_t handle, int is_chained, /* Uninstall? */ if (handle == handle_bad_irq) { - if (desc->chip != &no_irq_chip) + if (desc->chip != &no_irq_chip) { mask_ack_irq(desc, irq); + desc = irq_remap_to_desc(irq, desc); + } desc->status |= IRQ_DISABLED; desc->depth = 1; } diff --git a/kernel/irq/handle.c b/kernel/irq/handle.c index 8aa09547f5ef..f1a23069c20a 100644 --- a/kernel/irq/handle.c +++ b/kernel/irq/handle.c @@ -23,7 +23,7 @@ /* * lockdep: we want to handle all irq_desc locks as a single lock-class: */ -static struct lock_class_key irq_desc_lock_class; +struct lock_class_key irq_desc_lock_class; /** * handle_bad_irq - handle spurious and unhandled irqs @@ -73,7 +73,7 @@ static struct irq_desc irq_desc_init = { #endif }; -static void init_kstat_irqs(struct irq_desc *desc, int cpu, int nr) +void init_kstat_irqs(struct irq_desc *desc, int cpu, int nr) { unsigned long bytes; char *ptr; @@ -113,7 +113,7 @@ static void init_one_irq_desc(int irq, struct irq_desc *desc, int cpu) /* * Protect the sparse_irqs: */ -static DEFINE_SPINLOCK(sparse_irq_lock); +DEFINE_SPINLOCK(sparse_irq_lock); struct irq_desc *irq_desc_ptrs[NR_IRQS] __read_mostly; @@ -337,8 +337,11 @@ unsigned int __do_IRQ(unsigned int irq) /* * No locking required for CPU-local interrupts: */ - if (desc->chip->ack) + if (desc->chip->ack) { desc->chip->ack(irq); + /* get new one */ + desc = irq_remap_to_desc(irq, desc); + } if (likely(!(desc->status & IRQ_DISABLED))) { action_ret = handle_IRQ_event(irq, desc->action); if (!noirqdebug) @@ -349,8 +352,10 @@ unsigned int __do_IRQ(unsigned int irq) } spin_lock(&desc->lock); - if (desc->chip->ack) + if (desc->chip->ack) { desc->chip->ack(irq); + desc = irq_remap_to_desc(irq, desc); + } /* * REPLAY is when Linux resends an IRQ that was dropped earlier * WAITING is used by probe to mark irqs that are being tested diff --git a/kernel/irq/internals.h b/kernel/irq/internals.h index 64c1c7253dae..e6d0a43cc125 100644 --- a/kernel/irq/internals.h +++ b/kernel/irq/internals.h @@ -13,6 +13,11 @@ extern void compat_irq_chip_set_default_handler(struct irq_desc *desc); extern int __irq_set_trigger(struct irq_desc *desc, unsigned int irq, unsigned long flags); +extern struct lock_class_key irq_desc_lock_class; +extern void init_kstat_irqs(struct irq_desc *desc, int cpu, int nr); +extern spinlock_t sparse_irq_lock; +extern struct irq_desc *irq_desc_ptrs[NR_IRQS]; + #ifdef CONFIG_PROC_FS extern void register_irq_proc(unsigned int irq, struct irq_desc *desc); extern void register_handler_proc(unsigned int irq, struct irqaction *action); diff --git a/kernel/irq/numa_migrate.c b/kernel/irq/numa_migrate.c new file mode 100644 index 000000000000..0178e2296990 --- /dev/null +++ b/kernel/irq/numa_migrate.c @@ -0,0 +1,127 @@ +/* + * linux/kernel/irq/handle.c + * + * Copyright (C) 1992, 1998-2006 Linus Torvalds, Ingo Molnar + * Copyright (C) 2005-2006, Thomas Gleixner, Russell King + * + * This file contains the core interrupt handling code. + * + * Detailed information is available in Documentation/DocBook/genericirq + * + */ + +#include +#include +#include +#include +#include + +#include "internals.h" + +static void init_copy_kstat_irqs(struct irq_desc *old_desc, + struct irq_desc *desc, + int cpu, int nr) +{ + unsigned long bytes; + + init_kstat_irqs(desc, cpu, nr); + + if (desc->kstat_irqs != old_desc->kstat_irqs) { + /* Compute how many bytes we need per irq and allocate them */ + bytes = nr * sizeof(unsigned int); + + memcpy(desc->kstat_irqs, old_desc->kstat_irqs, bytes); + } +} + +static void free_kstat_irqs(struct irq_desc *old_desc, struct irq_desc *desc) +{ + if (old_desc->kstat_irqs == desc->kstat_irqs) + return; + + kfree(old_desc->kstat_irqs); + old_desc->kstat_irqs = NULL; +} + +static void init_copy_one_irq_desc(int irq, struct irq_desc *old_desc, + struct irq_desc *desc, int cpu) +{ + memcpy(desc, old_desc, sizeof(struct irq_desc)); + desc->cpu = cpu; + lockdep_set_class(&desc->lock, &irq_desc_lock_class); + init_copy_kstat_irqs(old_desc, desc, cpu, nr_cpu_ids); + arch_init_copy_chip_data(old_desc, desc, cpu); +} + +static void free_one_irq_desc(struct irq_desc *old_desc, struct irq_desc *desc) +{ + free_kstat_irqs(old_desc, desc); + arch_free_chip_data(old_desc, desc); +} + +static struct irq_desc *__real_move_irq_desc(struct irq_desc *old_desc, + int cpu) +{ + struct irq_desc *desc; + unsigned int irq; + unsigned long flags; + int node; + + irq = old_desc->irq; + + spin_lock_irqsave(&sparse_irq_lock, flags); + + /* We have to check it to avoid races with another CPU */ + desc = irq_desc_ptrs[irq]; + + if (desc && old_desc != desc) + goto out_unlock; + + node = cpu_to_node(cpu); + desc = kzalloc_node(sizeof(*desc), GFP_ATOMIC, node); + printk(KERN_DEBUG " move irq_desc for %d to cpu %d node %d\n", + irq, cpu, node); + if (!desc) { + printk(KERN_ERR "can not get new irq_desc for moving\n"); + /* still use old one */ + desc = old_desc; + goto out_unlock; + } + init_copy_one_irq_desc(irq, old_desc, desc, cpu); + + irq_desc_ptrs[irq] = desc; + + /* free the old one */ + free_one_irq_desc(old_desc, desc); + kfree(old_desc); + +out_unlock: + spin_unlock_irqrestore(&sparse_irq_lock, flags); + + return desc; +} + +struct irq_desc *move_irq_desc(struct irq_desc *desc, int cpu) +{ + int old_cpu; + int node, old_node; + + /* those all static, do move them */ + if (desc->irq < NR_IRQS_LEGACY) + return desc; + + old_cpu = desc->cpu; + printk(KERN_DEBUG + "try to move irq_desc from cpu %d to %d\n", old_cpu, cpu); + if (old_cpu != cpu) { + node = cpu_to_node(cpu); + old_node = cpu_to_node(old_cpu); + if (old_node != node) + desc = __real_move_irq_desc(desc, cpu); + else + desc->cpu = cpu; + } + + return desc; +} + -- cgit v1.2.3 From 17483a1f34c970e6c2cb8c082d4441bfabbe88a9 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 12 Dec 2008 13:14:18 -0800 Subject: sparseirq: fix !SMP building, #2 Impact: build fix make intr_remapping.c to include smp.h, so could use boot_cpu_id there also remove old change that disabling sparseirq with !SMP Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 2 +- drivers/pci/intr_remapping.c | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 29073532f94c..60a008857a38 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -240,7 +240,7 @@ config X86_HAS_BOOT_CPU_ID config SPARSE_IRQ bool "Support sparse irq numbering" - depends on (PCI_MSI || HT_IRQ) && SMP + depends on PCI_MSI || HT_IRQ default y help This enables support for sparse irq, esp for msi/msi-x. You may need diff --git a/drivers/pci/intr_remapping.c b/drivers/pci/intr_remapping.c index c9958ec5e25e..f78371b22529 100644 --- a/drivers/pci/intr_remapping.c +++ b/drivers/pci/intr_remapping.c @@ -5,6 +5,7 @@ #include #include #include +#include #include #include "intr_remapping.h" -- cgit v1.2.3 From 8ce7996009bab7b2d23e7af7ad831fed7eb6faa1 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 16 Dec 2008 12:17:35 -0800 Subject: x86: add swiotlb allocation functions Add x86-specific swiotlb allocation functions. These are purely default for the moment. Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-swiotlb_64.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'arch/x86') diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index 3c539d111abb..f47a097a135b 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -3,6 +3,8 @@ #include #include #include +#include +#include #include #include @@ -11,6 +13,16 @@ int swiotlb __read_mostly; +void *swiotlb_alloc_boot(size_t size, unsigned long nslabs) +{ + return alloc_bootmem_low_pages(size); +} + +void *swiotlb_alloc(unsigned order, unsigned long nslabs) +{ + return (void *)__get_free_pages(GFP_DMA | __GFP_NOWARN, order); +} + static dma_addr_t swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size, int direction) -- cgit v1.2.3 From cfb80c9eae8c7ed8f2ee81090062d15ead51cbe8 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Tue, 16 Dec 2008 12:17:36 -0800 Subject: x86: unify pci iommu setup and allow swiotlb to compile for 32 bit swiotlb on 32 bit will be used by Xen domain 0 support. Signed-off-by: Ian Campbell Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/include/asm/dma-mapping.h | 2 +- arch/x86/include/asm/pci.h | 2 ++ arch/x86/include/asm/pci_64.h | 1 - arch/x86/kernel/Makefile | 3 ++- arch/x86/kernel/pci-dma.c | 6 ++++-- arch/x86/kernel/pci-swiotlb_64.c | 2 ++ arch/x86/mm/init_32.c | 3 +++ 7 files changed, 14 insertions(+), 5 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/dma-mapping.h b/arch/x86/include/asm/dma-mapping.h index 097794ff6b79..3b43a65894c4 100644 --- a/arch/x86/include/asm/dma-mapping.h +++ b/arch/x86/include/asm/dma-mapping.h @@ -65,7 +65,7 @@ static inline struct dma_mapping_ops *get_dma_ops(struct device *dev) return dma_ops; else return dev->archdata.dma_ops; -#endif /* _ASM_X86_DMA_MAPPING_H */ +#endif } /* Make sure we keep the same behaviour */ diff --git a/arch/x86/include/asm/pci.h b/arch/x86/include/asm/pci.h index 875b38edf193..50ac542c9382 100644 --- a/arch/x86/include/asm/pci.h +++ b/arch/x86/include/asm/pci.h @@ -82,6 +82,8 @@ static inline void pci_dma_burst_advice(struct pci_dev *pdev, static inline void early_quirks(void) { } #endif +extern void pci_iommu_alloc(void); + #endif /* __KERNEL__ */ #ifdef CONFIG_X86_32 diff --git a/arch/x86/include/asm/pci_64.h b/arch/x86/include/asm/pci_64.h index d02d936840a3..4da207982777 100644 --- a/arch/x86/include/asm/pci_64.h +++ b/arch/x86/include/asm/pci_64.h @@ -23,7 +23,6 @@ extern int (*pci_config_write)(int seg, int bus, int dev, int fn, int reg, int len, u32 value); extern void dma32_reserve_bootmem(void); -extern void pci_iommu_alloc(void); /* The PCI address space does equal the physical memory * address space. The networking and block device layers use diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile index b62a7667828e..a9c656f2d661 100644 --- a/arch/x86/kernel/Makefile +++ b/arch/x86/kernel/Makefile @@ -105,6 +105,8 @@ microcode-$(CONFIG_MICROCODE_INTEL) += microcode_intel.o microcode-$(CONFIG_MICROCODE_AMD) += microcode_amd.o obj-$(CONFIG_MICROCODE) += microcode.o +obj-$(CONFIG_SWIOTLB) += pci-swiotlb_64.o # NB rename without _64 + ### # 64 bit specific files ifeq ($(CONFIG_X86_64),y) @@ -118,7 +120,6 @@ ifeq ($(CONFIG_X86_64),y) obj-$(CONFIG_GART_IOMMU) += pci-gart_64.o aperture_64.o obj-$(CONFIG_CALGARY_IOMMU) += pci-calgary_64.o tce_64.o obj-$(CONFIG_AMD_IOMMU) += amd_iommu_init.o amd_iommu.o - obj-$(CONFIG_SWIOTLB) += pci-swiotlb_64.o obj-$(CONFIG_PCI_MMCONFIG) += mmconf-fam10h_64.o endif diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index e150ad4f0ccc..00e07447a5bd 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -105,11 +105,15 @@ static void __init dma32_free_bootmem(void) dma32_bootmem_ptr = NULL; dma32_bootmem_size = 0; } +#endif void __init pci_iommu_alloc(void) { +#ifdef CONFIG_X86_64 /* free the range so iommu could get some range less than 4G */ dma32_free_bootmem(); +#endif + /* * The order of these functions is important for * fall-back/fail-over reasons @@ -125,8 +129,6 @@ void __init pci_iommu_alloc(void) pci_swiotlb_init(); } -#endif - void *dma_generic_alloc_coherent(struct device *dev, size_t size, dma_addr_t *dma_addr, gfp_t flag) { diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index f47a097a135b..a991afea6700 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -62,8 +62,10 @@ struct dma_mapping_ops swiotlb_dma_ops = { void __init pci_swiotlb_init(void) { /* don't initialize swiotlb if iommu=off (no_iommu=1) */ +#ifdef CONFIG_X86_64 if (!iommu_detected && !no_iommu && max_pfn > MAX_DMA32_PFN) swiotlb = 1; +#endif if (swiotlb_force) swiotlb = 1; if (swiotlb) { diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c index c483f4242079..2b4b14fc0c04 100644 --- a/arch/x86/mm/init_32.c +++ b/arch/x86/mm/init_32.c @@ -21,6 +21,7 @@ #include #include #include +#include #include #include #include @@ -971,6 +972,8 @@ void __init mem_init(void) start_periodic_check_for_corruption(); + pci_iommu_alloc(); + #ifdef CONFIG_FLATMEM BUG_ON(!mem_map); #endif -- cgit v1.2.3 From 1d32251e846ccbcf9d2da041dffd1199f94b2a3b Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Tue, 16 Dec 2008 12:17:37 -0800 Subject: x86/swiotlb: add default phys<->bus conversion Xen will override these later on. Signed-off-by: Ian Campbell Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-swiotlb_64.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'arch/x86') diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index a991afea6700..93a8371f2c22 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -23,6 +23,16 @@ void *swiotlb_alloc(unsigned order, unsigned long nslabs) return (void *)__get_free_pages(GFP_DMA | __GFP_NOWARN, order); } +dma_addr_t swiotlb_phys_to_bus(phys_addr_t paddr) +{ + return paddr; +} + +phys_addr_t swiotlb_bus_to_phys(dma_addr_t baddr) +{ + return baddr; +} + static dma_addr_t swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size, int direction) -- cgit v1.2.3 From a08636690d06b2e36cfb4c2b3ee133a81c47e1e0 Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Tue, 16 Dec 2008 12:17:38 -0800 Subject: x86/swiotlb: add default swiotlb_arch_range_needs_mapping Xen will override these later on. Signed-off-by: Ian Campbell Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-swiotlb_64.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'arch/x86') diff --git a/arch/x86/kernel/pci-swiotlb_64.c b/arch/x86/kernel/pci-swiotlb_64.c index 93a8371f2c22..242c3440687f 100644 --- a/arch/x86/kernel/pci-swiotlb_64.c +++ b/arch/x86/kernel/pci-swiotlb_64.c @@ -33,6 +33,11 @@ phys_addr_t swiotlb_bus_to_phys(dma_addr_t baddr) return baddr; } +int __weak swiotlb_arch_range_needs_mapping(void *ptr, size_t size) +{ + return 0; +} + static dma_addr_t swiotlb_map_single_phys(struct device *hwdev, phys_addr_t paddr, size_t size, int direction) -- cgit v1.2.3 From b909895739427874c089bc0e03dc119f99cab2dd Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 19 Dec 2008 13:48:34 -0800 Subject: sparseirq: fix numa_migrate_irq_desc dependency and comments Impact: reduce kconfig variable scope and clean up Bartlomiej pointed out that the config dependencies and comments are not right. update it depend to NUMA, and fix some comments Reported-by: Bartlomiej Zolnierkiewicz Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 2 +- arch/x86/kernel/io_apic.c | 2 +- kernel/irq/numa_migrate.c | 11 +++-------- 3 files changed, 5 insertions(+), 10 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 60a008857a38..5c243826334a 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -250,7 +250,7 @@ config SPARSE_IRQ config NUMA_MIGRATE_IRQ_DESC bool "Move irq desc when changing irq smp_affinity" - depends on SPARSE_IRQ && SMP + depends on SPARSE_IRQ && NUMA default n help This enables moving irq_desc to cpu/node that irq will use handled. diff --git a/arch/x86/kernel/io_apic.c b/arch/x86/kernel/io_apic.c index bfe1245b1a3e..a74887b416cc 100644 --- a/arch/x86/kernel/io_apic.c +++ b/arch/x86/kernel/io_apic.c @@ -2471,7 +2471,7 @@ static void irq_complete_move(struct irq_desc **descp) if (likely(!cfg->move_desc_pending)) return; - /* domain is not change, but affinity is changed */ + /* domain has not changed, but affinity did */ me = smp_processor_id(); if (cpu_isset(me, desc->affinity)) { *descp = desc = move_irq_desc(desc, me); diff --git a/kernel/irq/numa_migrate.c b/kernel/irq/numa_migrate.c index 0178e2296990..089c3746358a 100644 --- a/kernel/irq/numa_migrate.c +++ b/kernel/irq/numa_migrate.c @@ -1,13 +1,8 @@ /* - * linux/kernel/irq/handle.c - * - * Copyright (C) 1992, 1998-2006 Linus Torvalds, Ingo Molnar - * Copyright (C) 2005-2006, Thomas Gleixner, Russell King - * - * This file contains the core interrupt handling code. - * - * Detailed information is available in Documentation/DocBook/genericirq + * NUMA irq-desc migration code * + * Migrate IRQ data structures (irq_desc, chip_data, etc.) over to + * the new "home node" of the IRQ. */ #include -- cgit v1.2.3 From 4e17fee24a39448f3a20e9cf98887b7665825848 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 25 Dec 2008 12:04:17 +0100 Subject: x86: turn CONFIG_SPARSE_IRQ off by default New feature - lets disable it by default first - can flip it around later. Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 1 - 1 file changed, 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 5c243826334a..d14a8806227d 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -241,7 +241,6 @@ config X86_HAS_BOOT_CPU_ID config SPARSE_IRQ bool "Support sparse irq numbering" depends on PCI_MSI || HT_IRQ - default y help This enables support for sparse irq, esp for msi/msi-x. You may need if you have lots of cards supports msi-x installed. -- cgit v1.2.3 From 973656fe1afb4adf95d7b9ab75d4660cd3821ea1 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 25 Dec 2008 16:26:47 +0100 Subject: x86, sparseirq: clean up Kconfig entry Impact: improve help text Signed-off-by: Ingo Molnar --- arch/x86/Kconfig | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index d14a8806227d..e4c038abb71c 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -242,10 +242,14 @@ config SPARSE_IRQ bool "Support sparse irq numbering" depends on PCI_MSI || HT_IRQ help - This enables support for sparse irq, esp for msi/msi-x. You may need - if you have lots of cards supports msi-x installed. + This enables support for sparse irqs. This is useful for distro + kernels that want to define a high CONFIG_NR_CPUS value but still + want to have low kernel memory footprint on smaller machines. - If you don't know what to do here, say Y. + ( Sparse IRQs can also be beneficial on NUMA boxes, as they spread + out the irq_desc[] array in a more NUMA-friendly way. ) + + If you don't know what to do here, say N. config NUMA_MIGRATE_IRQ_DESC bool "Move irq desc when changing irq smp_affinity" -- cgit v1.2.3 From 58a24566449892dda409b9ad92c2e56c76c5670c Mon Sep 17 00:00:00 2001 From: Matias Zabaljauregui Date: Mon, 29 Sep 2008 01:40:07 -0300 Subject: lguest: move the initial guest page table creation code to the host This patch moves the initial guest page table creation code to the host, so the launcher keeps working with PAE enabled configs. Signed-off-by: Matias Zabaljauregui Signed-off-by: Rusty Russell --- Documentation/lguest/lguest.c | 60 ++++------------------------------ arch/x86/lguest/i386_head.S | 15 --------- drivers/lguest/lg.h | 2 +- drivers/lguest/lguest_user.c | 13 +++----- drivers/lguest/page_tables.c | 72 +++++++++++++++++++++++++++++++++++++++-- include/linux/lguest_launcher.h | 2 +- 6 files changed, 83 insertions(+), 81 deletions(-) (limited to 'arch/x86') diff --git a/Documentation/lguest/lguest.c b/Documentation/lguest/lguest.c index aa2574ca94c7..f2dbbf3bdeab 100644 --- a/Documentation/lguest/lguest.c +++ b/Documentation/lguest/lguest.c @@ -481,51 +481,6 @@ static unsigned long load_initrd(const char *name, unsigned long mem) /* We return the initrd size. */ return len; } - -/* Once we know how much memory we have we can construct simple linear page - * tables which set virtual == physical which will get the Guest far enough - * into the boot to create its own. - * - * We lay them out of the way, just below the initrd (which is why we need to - * know its size here). */ -static unsigned long setup_pagetables(unsigned long mem, - unsigned long initrd_size) -{ - unsigned long *pgdir, *linear; - unsigned int mapped_pages, i, linear_pages; - unsigned int ptes_per_page = getpagesize()/sizeof(void *); - - mapped_pages = mem/getpagesize(); - - /* Each PTE page can map ptes_per_page pages: how many do we need? */ - linear_pages = (mapped_pages + ptes_per_page-1)/ptes_per_page; - - /* We put the toplevel page directory page at the top of memory. */ - pgdir = from_guest_phys(mem) - initrd_size - getpagesize(); - - /* Now we use the next linear_pages pages as pte pages */ - linear = (void *)pgdir - linear_pages*getpagesize(); - - /* Linear mapping is easy: put every page's address into the mapping in - * order. PAGE_PRESENT contains the flags Present, Writable and - * Executable. */ - for (i = 0; i < mapped_pages; i++) - linear[i] = ((i * getpagesize()) | PAGE_PRESENT); - - /* The top level points to the linear page table pages above. */ - for (i = 0; i < mapped_pages; i += ptes_per_page) { - pgdir[i/ptes_per_page] - = ((to_guest_phys(linear) + i*sizeof(void *)) - | PAGE_PRESENT); - } - - verbose("Linear mapping of %u pages in %u pte pages at %#lx\n", - mapped_pages, linear_pages, to_guest_phys(linear)); - - /* We return the top level (guest-physical) address: the kernel needs - * to know where it is. */ - return to_guest_phys(pgdir); -} /*:*/ /* Simple routine to roll all the commandline arguments together with spaces @@ -548,13 +503,13 @@ static void concat(char *dst, char *args[]) /*L:185 This is where we actually tell the kernel to initialize the Guest. We * saw the arguments it expects when we looked at initialize() in lguest_user.c: - * the base of Guest "physical" memory, the top physical page to allow, the - * top level pagetable and the entry point for the Guest. */ -static int tell_kernel(unsigned long pgdir, unsigned long start) + * the base of Guest "physical" memory, the top physical page to allow and the + * entry point for the Guest. */ +static int tell_kernel(unsigned long start) { unsigned long args[] = { LHREQ_INITIALIZE, (unsigned long)guest_base, - guest_limit / getpagesize(), pgdir, start }; + guest_limit / getpagesize(), start }; int fd; verbose("Guest: %p - %p (%#lx)\n", @@ -1941,7 +1896,7 @@ int main(int argc, char *argv[]) { /* Memory, top-level pagetable, code startpoint and size of the * (optional) initrd. */ - unsigned long mem = 0, pgdir, start, initrd_size = 0; + unsigned long mem = 0, start, initrd_size = 0; /* Two temporaries and the /dev/lguest file descriptor. */ int i, c, lguest_fd; /* The boot information for the Guest. */ @@ -2040,9 +1995,6 @@ int main(int argc, char *argv[]) boot->hdr.type_of_loader = 0xFF; } - /* Set up the initial linear pagetables, starting below the initrd. */ - pgdir = setup_pagetables(mem, initrd_size); - /* The Linux boot header contains an "E820" memory map: ours is a * simple, single region. */ boot->e820_entries = 1; @@ -2064,7 +2016,7 @@ int main(int argc, char *argv[]) /* We tell the kernel to initialize the Guest: this returns the open * /dev/lguest file descriptor. */ - lguest_fd = tell_kernel(pgdir, start); + lguest_fd = tell_kernel(start); /* We clone off a thread, which wakes the Launcher whenever one of the * input file descriptors needs attention. We call this the Waker, and diff --git a/arch/x86/lguest/i386_head.S b/arch/x86/lguest/i386_head.S index 5c7cef34c9e7..10b9bd35a8ff 100644 --- a/arch/x86/lguest/i386_head.S +++ b/arch/x86/lguest/i386_head.S @@ -30,21 +30,6 @@ ENTRY(lguest_entry) movl $lguest_data - __PAGE_OFFSET, %edx int $LGUEST_TRAP_ENTRY - /* The Host put the toplevel pagetable in lguest_data.pgdir. The movsl - * instruction uses %esi implicitly as the source for the copy we're - * about to do. */ - movl lguest_data - __PAGE_OFFSET + LGUEST_DATA_pgdir, %esi - - /* Copy first 32 entries of page directory to __PAGE_OFFSET entries. - * This means the first 128M of kernel memory will be mapped at - * PAGE_OFFSET where the kernel expects to run. This will get it far - * enough through boot to switch to its own pagetables. */ - movl $32, %ecx - movl %esi, %edi - addl $((__PAGE_OFFSET >> 22) * 4), %edi - rep - movsl - /* Set up the initial stack so we can run C code. */ movl $(init_thread_union+THREAD_SIZE),%esp diff --git a/drivers/lguest/lg.h b/drivers/lguest/lg.h index 5faefeaf6790..f2c641e0bdde 100644 --- a/drivers/lguest/lg.h +++ b/drivers/lguest/lg.h @@ -164,7 +164,7 @@ void copy_gdt(const struct lg_cpu *cpu, struct desc_struct *gdt); void copy_gdt_tls(const struct lg_cpu *cpu, struct desc_struct *gdt); /* page_tables.c: */ -int init_guest_pagetable(struct lguest *lg, unsigned long pgtable); +int init_guest_pagetable(struct lguest *lg); void free_guest_pagetable(struct lguest *lg); void guest_new_pagetable(struct lg_cpu *cpu, unsigned long pgtable); void guest_set_pmd(struct lguest *lg, unsigned long gpgdir, u32 i); diff --git a/drivers/lguest/lguest_user.c b/drivers/lguest/lguest_user.c index e73a000473cc..34bc017b8b3c 100644 --- a/drivers/lguest/lguest_user.c +++ b/drivers/lguest/lguest_user.c @@ -146,7 +146,7 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) return 0; } -/*L:020 The initialization write supplies 4 pointer sized (32 or 64 bit) +/*L:020 The initialization write supplies 3 pointer sized (32 or 64 bit) * values (in addition to the LHREQ_INITIALIZE value). These are: * * base: The start of the Guest-physical memory inside the Launcher memory. @@ -155,9 +155,6 @@ static int lg_cpu_start(struct lg_cpu *cpu, unsigned id, unsigned long start_ip) * allowed to access. The Guest memory lives inside the Launcher, so it sets * this to ensure the Guest can only reach its own memory. * - * pgdir: The (Guest-physical) address of the top of the initial Guest - * pagetables (which are set up by the Launcher). - * * start: The first instruction to execute ("eip" in x86-speak). */ static int initialize(struct file *file, const unsigned long __user *input) @@ -166,7 +163,7 @@ static int initialize(struct file *file, const unsigned long __user *input) * Guest. */ struct lguest *lg; int err; - unsigned long args[4]; + unsigned long args[3]; /* We grab the Big Lguest lock, which protects against multiple * simultaneous initializations. */ @@ -192,14 +189,14 @@ static int initialize(struct file *file, const unsigned long __user *input) lg->mem_base = (void __user *)args[0]; lg->pfn_limit = args[1]; - /* This is the first cpu (cpu 0) and it will start booting at args[3] */ - err = lg_cpu_start(&lg->cpus[0], 0, args[3]); + /* This is the first cpu (cpu 0) and it will start booting at args[2] */ + err = lg_cpu_start(&lg->cpus[0], 0, args[2]); if (err) goto release_guest; /* Initialize the Guest's shadow page tables, using the toplevel * address the Launcher gave us. This allocates memory, so can fail. */ - err = init_guest_pagetable(lg, args[2]); + err = init_guest_pagetable(lg); if (err) goto free_regs; diff --git a/drivers/lguest/page_tables.c b/drivers/lguest/page_tables.c index 81d0c6053447..576a8318221c 100644 --- a/drivers/lguest/page_tables.c +++ b/drivers/lguest/page_tables.c @@ -14,6 +14,7 @@ #include #include #include +#include #include "lg.h" /*M:008 We hold reference to pages, which prevents them from being swapped. @@ -581,15 +582,82 @@ void guest_set_pmd(struct lguest *lg, unsigned long gpgdir, u32 idx) release_pgd(lg, lg->pgdirs[pgdir].pgdir + idx); } +/* Once we know how much memory we have we can construct simple identity + * (which set virtual == physical) and linear mappings + * which will get the Guest far enough into the boot to create its own. + * + * We lay them out of the way, just below the initrd (which is why we need to + * know its size here). */ +static unsigned long setup_pagetables(struct lguest *lg, + unsigned long mem, + unsigned long initrd_size) +{ + pgd_t __user *pgdir; + pte_t __user *linear; + unsigned int mapped_pages, i, linear_pages, phys_linear; + unsigned long mem_base = (unsigned long)lg->mem_base; + + /* We have mapped_pages frames to map, so we need + * linear_pages page tables to map them. */ + mapped_pages = mem / PAGE_SIZE; + linear_pages = (mapped_pages + PTRS_PER_PTE - 1) / PTRS_PER_PTE; + + /* We put the toplevel page directory page at the top of memory. */ + pgdir = (pgd_t *)(mem + mem_base - initrd_size - PAGE_SIZE); + + /* Now we use the next linear_pages pages as pte pages */ + linear = (void *)pgdir - linear_pages * PAGE_SIZE; + + /* Linear mapping is easy: put every page's address into the + * mapping in order. */ + for (i = 0; i < mapped_pages; i++) { + pte_t pte; + pte = pfn_pte(i, __pgprot(_PAGE_PRESENT|_PAGE_RW|_PAGE_USER)); + if (copy_to_user(&linear[i], &pte, sizeof(pte)) != 0) + return -EFAULT; + } + + /* The top level points to the linear page table pages above. + * We setup the identity and linear mappings here. */ + phys_linear = (unsigned long)linear - mem_base; + for (i = 0; i < mapped_pages; i += PTRS_PER_PTE) { + pgd_t pgd; + pgd = __pgd((phys_linear + i * sizeof(pte_t)) | + (_PAGE_PRESENT | _PAGE_RW | _PAGE_USER)); + + if (copy_to_user(&pgdir[i / PTRS_PER_PTE], &pgd, sizeof(pgd)) + || copy_to_user(&pgdir[pgd_index(PAGE_OFFSET) + + i / PTRS_PER_PTE], + &pgd, sizeof(pgd))) + return -EFAULT; + } + + /* We return the top level (guest-physical) address: remember where + * this is. */ + return (unsigned long)pgdir - mem_base; +} + /*H:500 (vii) Setting up the page tables initially. * * When a Guest is first created, the Launcher tells us where the toplevel of * its first page table is. We set some things up here: */ -int init_guest_pagetable(struct lguest *lg, unsigned long pgtable) +int init_guest_pagetable(struct lguest *lg) { + u64 mem; + u32 initrd_size; + struct boot_params __user *boot = (struct boot_params *)lg->mem_base; + + /* Get the Guest memory size and the ramdisk size from the boot header + * located at lg->mem_base (Guest address 0). */ + if (copy_from_user(&mem, &boot->e820_map[0].size, sizeof(mem)) + || get_user(initrd_size, &boot->hdr.ramdisk_size)) + return -EFAULT; + /* We start on the first shadow page table, and give it a blank PGD * page. */ - lg->pgdirs[0].gpgdir = pgtable; + lg->pgdirs[0].gpgdir = setup_pagetables(lg, mem, initrd_size); + if (IS_ERR_VALUE(lg->pgdirs[0].gpgdir)) + return lg->pgdirs[0].gpgdir; lg->pgdirs[0].pgdir = (pgd_t *)get_zeroed_page(GFP_KERNEL); if (!lg->pgdirs[0].pgdir) return -ENOMEM; diff --git a/include/linux/lguest_launcher.h b/include/linux/lguest_launcher.h index bd0eba760522..a53407a4165c 100644 --- a/include/linux/lguest_launcher.h +++ b/include/linux/lguest_launcher.h @@ -54,7 +54,7 @@ struct lguest_vqconfig { /* Write command first word is a request. */ enum lguest_req { - LHREQ_INITIALIZE, /* + base, pfnlimit, pgdir, start */ + LHREQ_INITIALIZE, /* + base, pfnlimit, start */ LHREQ_GETDMA, /* No longer used */ LHREQ_IRQ, /* + irq */ LHREQ_BREAK, /* + on/off flag (on blocks until someone does off) */ -- cgit v1.2.3