diff options
Diffstat (limited to 'arch/x86/kernel')
61 files changed, 1838 insertions, 879 deletions
diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index fa88a1d7129..bfd10fd211c 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -97,6 +97,8 @@ static u64 acpi_lapic_addr __initdata = APIC_DEFAULT_PHYS_BASE; #warning ACPI uses CMPXCHG, i486 and later hardware #endif +static int acpi_mcfg_64bit_base_addr __initdata = FALSE; + /* -------------------------------------------------------------------------- Boot-time Configuration -------------------------------------------------------------------------- */ @@ -158,6 +160,14 @@ char *__init __acpi_map_table(unsigned long phys, unsigned long size) struct acpi_mcfg_allocation *pci_mmcfg_config; int pci_mmcfg_config_num; +static int __init acpi_mcfg_oem_check(struct acpi_table_mcfg *mcfg) +{ + if (!strcmp(mcfg->header.oem_id, "SGI")) + acpi_mcfg_64bit_base_addr = TRUE; + + return 0; +} + int __init acpi_parse_mcfg(struct acpi_table_header *header) { struct acpi_table_mcfg *mcfg; @@ -190,8 +200,12 @@ int __init acpi_parse_mcfg(struct acpi_table_header *header) } memcpy(pci_mmcfg_config, &mcfg[1], config_size); + + acpi_mcfg_oem_check(mcfg); + for (i = 0; i < pci_mmcfg_config_num; ++i) { - if (pci_mmcfg_config[i].address > 0xFFFFFFFF) { + if ((pci_mmcfg_config[i].address > 0xFFFFFFFF) && + !acpi_mcfg_64bit_base_addr) { printk(KERN_ERR PREFIX "MMCONFIG not in low 4GB of memory\n"); kfree(pci_mmcfg_config); diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index fa2161d5003..426e5d91b63 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -20,7 +20,7 @@ unsigned long acpi_realmode_flags; /* address in low memory of the wakeup routine. */ static unsigned long acpi_realmode; -#ifdef CONFIG_64BIT +#if defined(CONFIG_SMP) && defined(CONFIG_64BIT) static char temp_stack[10240]; #endif @@ -86,7 +86,7 @@ int acpi_save_state_mem(void) #endif /* !CONFIG_64BIT */ header->pmode_cr0 = read_cr0(); - header->pmode_cr4 = read_cr4(); + header->pmode_cr4 = read_cr4_safe(); header->realmode_flags = acpi_realmode_flags; header->real_magic = 0x12345678; diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c index 2763cb37b55..65a0c1b4869 100644 --- a/arch/x86/kernel/alternative.c +++ b/arch/x86/kernel/alternative.c @@ -145,35 +145,25 @@ static const unsigned char *const p6_nops[ASM_NOP_MAX+1] = { extern char __vsyscall_0; const unsigned char *const *find_nop_table(void) { - return boot_cpu_data.x86_vendor != X86_VENDOR_INTEL || - boot_cpu_data.x86 < 6 ? k8_nops : p6_nops; + if (boot_cpu_data.x86_vendor == X86_VENDOR_INTEL && + boot_cpu_has(X86_FEATURE_NOPL)) + return p6_nops; + else + return k8_nops; } #else /* CONFIG_X86_64 */ -static const struct nop { - int cpuid; - const unsigned char *const *noptable; -} noptypes[] = { - { X86_FEATURE_K8, k8_nops }, - { X86_FEATURE_K7, k7_nops }, - { X86_FEATURE_P4, p6_nops }, - { X86_FEATURE_P3, p6_nops }, - { -1, NULL } -}; - const unsigned char *const *find_nop_table(void) { - const unsigned char *const *noptable = intel_nops; - int i; - - for (i = 0; noptypes[i].cpuid >= 0; i++) { - if (boot_cpu_has(noptypes[i].cpuid)) { - noptable = noptypes[i].noptable; - break; - } - } - return noptable; + if (boot_cpu_has(X86_FEATURE_K8)) + return k8_nops; + else if (boot_cpu_has(X86_FEATURE_K7)) + return k7_nops; + else if (boot_cpu_has(X86_FEATURE_NOPL)) + return p6_nops; + else + return intel_nops; } #endif /* CONFIG_X86_64 */ diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 22d7d050905..042fdc27bc9 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -65,7 +65,7 @@ static int __iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd) u8 *target; tail = readl(iommu->mmio_base + MMIO_CMD_TAIL_OFFSET); - target = (iommu->cmd_buf + tail); + target = iommu->cmd_buf + tail; memcpy_toio(target, cmd, sizeof(*cmd)); tail = (tail + sizeof(*cmd)) % iommu->cmd_buf_size; head = readl(iommu->mmio_base + MMIO_CMD_HEAD_OFFSET); @@ -101,32 +101,39 @@ static int iommu_queue_command(struct amd_iommu *iommu, struct iommu_cmd *cmd) */ static int iommu_completion_wait(struct amd_iommu *iommu) { - int ret; + int ret = 0, ready = 0; + unsigned status = 0; struct iommu_cmd cmd; - volatile u64 ready = 0; - unsigned long ready_phys = virt_to_phys(&ready); - unsigned long i = 0; + unsigned long flags, i = 0; memset(&cmd, 0, sizeof(cmd)); - cmd.data[0] = LOW_U32(ready_phys) | CMD_COMPL_WAIT_STORE_MASK; - cmd.data[1] = upper_32_bits(ready_phys); - cmd.data[2] = 1; /* value written to 'ready' */ + cmd.data[0] = CMD_COMPL_WAIT_INT_MASK; CMD_SET_TYPE(&cmd, CMD_COMPL_WAIT); iommu->need_sync = 0; - ret = iommu_queue_command(iommu, &cmd); + spin_lock_irqsave(&iommu->lock, flags); + + ret = __iommu_queue_command(iommu, &cmd); if (ret) - return ret; + goto out; while (!ready && (i < EXIT_LOOP_COUNT)) { ++i; - cpu_relax(); + /* wait for the bit to become one */ + status = readl(iommu->mmio_base + MMIO_STATUS_OFFSET); + ready = status & MMIO_STATUS_COM_WAIT_INT_MASK; } + /* set bit back to zero */ + status &= ~MMIO_STATUS_COM_WAIT_INT_MASK; + writel(status, iommu->mmio_base + MMIO_STATUS_OFFSET); + if (unlikely((i == EXIT_LOOP_COUNT) && printk_ratelimit())) printk(KERN_WARNING "AMD IOMMU: Completion wait loop failed\n"); +out: + spin_unlock_irqrestore(&iommu->lock, flags); return 0; } @@ -137,6 +144,7 @@ static int iommu_completion_wait(struct amd_iommu *iommu) static int iommu_queue_inv_dev_entry(struct amd_iommu *iommu, u16 devid) { struct iommu_cmd cmd; + int ret; BUG_ON(iommu == NULL); @@ -144,9 +152,11 @@ static int iommu_queue_inv_dev_entry(struct amd_iommu *iommu, u16 devid) CMD_SET_TYPE(&cmd, CMD_INV_DEV_ENTRY); cmd.data[0] = devid; + ret = iommu_queue_command(iommu, &cmd); + iommu->need_sync = 1; - return iommu_queue_command(iommu, &cmd); + return ret; } /* @@ -156,21 +166,24 @@ static int iommu_queue_inv_iommu_pages(struct amd_iommu *iommu, u64 address, u16 domid, int pde, int s) { struct iommu_cmd cmd; + int ret; memset(&cmd, 0, sizeof(cmd)); address &= PAGE_MASK; CMD_SET_TYPE(&cmd, CMD_INV_IOMMU_PAGES); cmd.data[1] |= domid; - cmd.data[2] = LOW_U32(address); + cmd.data[2] = lower_32_bits(address); cmd.data[3] = upper_32_bits(address); if (s) /* size bit - we flush more than one 4kb page */ cmd.data[2] |= CMD_INV_IOMMU_PAGES_SIZE_MASK; if (pde) /* PDE bit - we wan't flush everything not only the PTEs */ cmd.data[2] |= CMD_INV_IOMMU_PAGES_PDE_MASK; + ret = iommu_queue_command(iommu, &cmd); + iommu->need_sync = 1; - return iommu_queue_command(iommu, &cmd); + return ret; } /* diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index d9a9da597e7..a69cc0f5204 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -801,6 +801,21 @@ static int __init init_memory_definitions(struct acpi_table_header *table) } /* + * Init the device table to not allow DMA access for devices and + * suppress all page faults + */ +static void init_device_table(void) +{ + u16 devid; + + for (devid = 0; devid <= amd_iommu_last_bdf; ++devid) { + set_dev_entry_bit(devid, DEV_ENTRY_VALID); + set_dev_entry_bit(devid, DEV_ENTRY_TRANSLATION); + set_dev_entry_bit(devid, DEV_ENTRY_NO_PAGE_FAULT); + } +} + +/* * This function finally enables all IOMMUs found in the system after * they have been initialized */ @@ -931,6 +946,9 @@ int __init amd_iommu_init(void) if (amd_iommu_pd_alloc_bitmap == NULL) goto free; + /* init the device table */ + init_device_table(); + /* * let all alias entries point to itself */ @@ -954,15 +972,15 @@ int __init amd_iommu_init(void) if (acpi_table_parse("IVRS", init_memory_definitions) != 0) goto free; - ret = amd_iommu_init_dma_ops(); + ret = sysdev_class_register(&amd_iommu_sysdev_class); if (ret) goto free; - ret = sysdev_class_register(&amd_iommu_sysdev_class); + ret = sysdev_register(&device_amd_iommu); if (ret) goto free; - ret = sysdev_register(&device_amd_iommu); + ret = amd_iommu_init_dma_ops(); if (ret) goto free; diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 039a8d4aaf6..f88bd0d982b 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1454,8 +1454,6 @@ void disconnect_bsp_APIC(int virt_wire_setup) } } -unsigned int __cpuinitdata maxcpus = NR_CPUS; - void __cpuinit generic_processor_info(int apicid, int version) { int cpu; @@ -1482,12 +1480,6 @@ void __cpuinit generic_processor_info(int apicid, int version) return; } - if (num_processors >= maxcpus) { - printk(KERN_WARNING "WARNING: maxcpus limit of %i reached." - " Processor ignored.\n", maxcpus); - return; - } - num_processors++; cpus_complement(tmp_map, cpu_present_map); cpu = first_cpu(tmp_map); diff --git a/arch/x86/kernel/apic_64.c b/arch/x86/kernel/apic_64.c index 7f1f030da7e..446c062e831 100644 --- a/arch/x86/kernel/apic_64.c +++ b/arch/x86/kernel/apic_64.c @@ -90,7 +90,6 @@ static unsigned long apic_phys; unsigned long mp_lapic_addr; -unsigned int __cpuinitdata maxcpus = NR_CPUS; /* * Get the LAPIC version */ @@ -1062,12 +1061,6 @@ void __cpuinit generic_processor_info(int apicid, int version) return; } - if (num_processors >= maxcpus) { - printk(KERN_WARNING "WARNING: maxcpus limit of %i reached." - " Processor ignored.\n", maxcpus); - return; - } - num_processors++; cpus_complement(tmp_map, cpu_present_map); cpu = first_cpu(tmp_map); diff --git a/arch/x86/kernel/apm_32.c b/arch/x86/kernel/apm_32.c index 9ee24e6bc4b..732d1f4e10e 100644 --- a/arch/x86/kernel/apm_32.c +++ b/arch/x86/kernel/apm_32.c @@ -234,6 +234,7 @@ #include <asm/uaccess.h> #include <asm/desc.h> #include <asm/i8253.h> +#include <asm/olpc.h> #include <asm/paravirt.h> #include <asm/reboot.h> @@ -2217,7 +2218,7 @@ static int __init apm_init(void) dmi_check_system(apm_dmi_table); - if (apm_info.bios.version == 0 || paravirt_enabled()) { + if (apm_info.bios.version == 0 || paravirt_enabled() || machine_is_olpc()) { printk(KERN_INFO "apm: BIOS not found.\n"); return -ENODEV; } diff --git a/arch/x86/kernel/asm-offsets_64.c b/arch/x86/kernel/asm-offsets_64.c index aa89387006f..505543a75a5 100644 --- a/arch/x86/kernel/asm-offsets_64.c +++ b/arch/x86/kernel/asm-offsets_64.c @@ -22,7 +22,7 @@ #define __NO_STUBS 1 #undef __SYSCALL -#undef _ASM_X86_64_UNISTD_H_ +#undef ASM_X86__UNISTD_64_H #define __SYSCALL(nr, sym) [nr] = 1, static char syscalls[] = { #include <asm/unistd.h> diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c index 84a8220a607..a6ef672adbb 100644 --- a/arch/x86/kernel/cpu/addon_cpuid_features.c +++ b/arch/x86/kernel/cpu/addon_cpuid_features.c @@ -56,9 +56,22 @@ void __cpuinit validate_pat_support(struct cpuinfo_x86 *c) switch (c->x86_vendor) { case X86_VENDOR_INTEL: - if (c->x86 == 0xF || (c->x86 == 6 && c->x86_model >= 15)) + /* + * There is a known erratum on Pentium III and Core Solo + * and Core Duo CPUs. + * " Page with PAT set to WC while associated MTRR is UC + * may consolidate to UC " + * Because of this erratum, it is better to stick with + * setting WC in MTRR rather than using PAT on these CPUs. + * + * Enable PAT WC only on P4, Core 2 or later CPUs. + */ + if (c->x86 > 0x6 || (c->x86 == 6 && c->x86_model >= 15)) return; - break; + + pat_disable("PAT WC disabled due to known CPU erratum."); + return; + case X86_VENDOR_AMD: case X86_VENDOR_CENTAUR: case X86_VENDOR_TRANSMETA: diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index cae9cabc303..18514ed2610 100644 --- a/arch/x86/kernel/cpu/amd.c +++ b/arch/x86/kernel/cpu/amd.c @@ -31,6 +31,11 @@ static void __cpuinit early_init_amd(struct cpuinfo_x86 *c) if (c->x86_power & (1<<8)) set_cpu_cap(c, X86_FEATURE_CONSTANT_TSC); } + + /* Set MTRR capability flag if appropriate */ + if (c->x86_model == 13 || c->x86_model == 9 || + (c->x86_model == 8 && c->x86_mask >= 8)) + set_cpu_cap(c, X86_FEATURE_K6_MTRR); } static void __cpuinit init_amd(struct cpuinfo_x86 *c) @@ -166,10 +171,6 @@ static void __cpuinit init_amd(struct cpuinfo_x86 *c) mbytes); } - /* Set MTRR capability flag if appropriate */ - if (c->x86_model == 13 || c->x86_model == 9 || - (c->x86_model == 8 && c->x86_mask >= 8)) - set_cpu_cap(c, X86_FEATURE_K6_MTRR); break; } diff --git a/arch/x86/kernel/cpu/centaur.c b/arch/x86/kernel/cpu/centaur.c index e0f45edd6a5..a0534c04d38 100644 --- a/arch/x86/kernel/cpu/centaur.c +++ b/arch/x86/kernel/cpu/centaur.c @@ -314,6 +314,16 @@ enum { EAMD3D = 1<<20, }; +static void __cpuinit early_init_centaur(struct cpuinfo_x86 *c) +{ + switch (c->x86) { + case 5: + /* Emulate MTRRs using Centaur's MCR. */ + set_cpu_cap(c, X86_FEATURE_CENTAUR_MCR); + break; + } +} + static void __cpuinit init_centaur(struct cpuinfo_x86 *c) { @@ -462,6 +472,7 @@ centaur_size_cache(struct cpuinfo_x86 *c, unsigned int size) static struct cpu_dev centaur_cpu_dev __cpuinitdata = { .c_vendor = "Centaur", .c_ident = { "CentaurHauls" }, + .c_early_init = early_init_centaur, .c_init = init_centaur, .c_size_cache = centaur_size_cache, }; diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 80ab20d4fa3..4e456bd955b 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -13,6 +13,7 @@ #include <asm/mtrr.h> #include <asm/mce.h> #include <asm/pat.h> +#include <asm/asm.h> #ifdef CONFIG_X86_LOCAL_APIC #include <asm/mpspec.h> #include <asm/apic.h> @@ -334,11 +335,24 @@ static void __init early_cpu_detect(void) get_cpu_vendor(c, 1); + early_get_cap(c); + if (c->x86_vendor != X86_VENDOR_UNKNOWN && cpu_devs[c->x86_vendor]->c_early_init) cpu_devs[c->x86_vendor]->c_early_init(c); +} - early_get_cap(c); +/* + * The NOPL instruction is supposed to exist on all CPUs with + * family >= 6; unfortunately, that's not true in practice because + * of early VIA chips and (more importantly) broken virtualizers that + * are not easy to detect. In the latter case it doesn't even *fail* + * reliably, so probing for it doesn't even work. Disable it completely + * unless we can find a reliable way to detect all the broken cases. + */ +static void __cpuinit detect_nopl(struct cpuinfo_x86 *c) +{ + clear_cpu_cap(c, X86_FEATURE_NOPL); } static void __cpuinit generic_identify(struct cpuinfo_x86 *c) @@ -395,8 +409,8 @@ static void __cpuinit generic_identify(struct cpuinfo_x86 *c) } init_scattered_cpuid_features(c); + detect_nopl(c); } - } static void __cpuinit squash_the_stupid_serial_number(struct cpuinfo_x86 *c) diff --git a/arch/x86/kernel/cpu/common_64.c b/arch/x86/kernel/cpu/common_64.c index dd6e3f15017..a11f5d4477c 100644 --- a/arch/x86/kernel/cpu/common_64.c +++ b/arch/x86/kernel/cpu/common_64.c @@ -18,6 +18,7 @@ #include <asm/mtrr.h> #include <asm/mce.h> #include <asm/pat.h> +#include <asm/asm.h> #include <asm/numa.h> #ifdef CONFIG_X86_LOCAL_APIC #include <asm/mpspec.h> @@ -215,6 +216,39 @@ static void __init early_cpu_support_print(void) } } +/* + * The NOPL instruction is supposed to exist on all CPUs with + * family >= 6, unfortunately, that's not true in practice because + * of early VIA chips and (more importantly) broken virtualizers that + * are not easy to detect. Hence, probe for it based on first + * principles. + * + * Note: no 64-bit chip is known to lack these, but put the code here + * for consistency with 32 bits, and to make it utterly trivial to + * diagnose the problem should it ever surface. + */ +static void __cpuinit detect_nopl(struct cpuinfo_x86 *c) +{ + const u32 nopl_signature = 0x888c53b1; /* Random number */ + u32 has_nopl = nopl_signature; + + clear_cpu_cap(c, X86_FEATURE_NOPL); + if (c->x86 >= 6) { + asm volatile("\n" + "1: .byte 0x0f,0x1f,0xc0\n" /* nopl %eax */ + "2:\n" + " .section .fixup,\"ax\"\n" + "3: xor %0,%0\n" + " jmp 2b\n" + " .previous\n" + _ASM_EXTABLE(1b,3b) + : "+a" (has_nopl)); + + if (has_nopl == nopl_signature) + set_cpu_cap(c, X86_FEATURE_NOPL); + } +} + static void __cpuinit early_identify_cpu(struct cpuinfo_x86 *c); void __init early_cpu_init(void) @@ -313,6 +347,8 @@ static void __cpuinit early_identify_cpu(struct cpuinfo_x86 *c) c->x86_phys_bits = eax & 0xff; } + detect_nopl(c); + if (c->x86_vendor != X86_VENDOR_UNKNOWN && cpu_devs[c->x86_vendor]->c_early_init) cpu_devs[c->x86_vendor]->c_early_init(c); @@ -493,17 +529,20 @@ void pda_init(int cpu) /* others are initialized in smpboot.c */ pda->pcurrent = &init_task; pda->irqstackptr = boot_cpu_stack; + pda->irqstackptr += IRQSTACKSIZE - 64; } else { - pda->irqstackptr = (char *) - __get_free_pages(GFP_ATOMIC, IRQSTACK_ORDER); - if (!pda->irqstackptr) - panic("cannot allocate irqstack for cpu %d", cpu); + if (!pda->irqstackptr) { + pda->irqstackptr = (char *) + __get_free_pages(GFP_ATOMIC, IRQSTACK_ORDER); + if (!pda->irqstackptr) + panic("cannot allocate irqstack for cpu %d", + cpu); + pda->irqstackptr += IRQSTACKSIZE - 64; + } if (pda->nodenumber == 0 && cpu_to_node(cpu) != NUMA_NO_NODE) pda->nodenumber = cpu_to_node(cpu); } - - pda->irqstackptr += IRQSTACKSIZE-64; } char boot_exception_stacks[(N_EXCEPTION_STACKS - 1) * EXCEPTION_STKSZ + @@ -601,19 +640,22 @@ void __cpuinit cpu_init(void) /* * set up and load the per-CPU TSS */ - for (v = 0; v < N_EXCEPTION_STACKS; v++) { + if (!orig_ist->ist[0]) { static const unsigned int order[N_EXCEPTION_STACKS] = { - [0 ... N_EXCEPTION_STACKS - 1] = EXCEPTION_STACK_ORDER, - [DEBUG_STACK - 1] = DEBUG_STACK_ORDER + [0 ... N_EXCEPTION_STACKS - 1] = EXCEPTION_STACK_ORDER, + [DEBUG_STACK - 1] = DEBUG_STACK_ORDER }; - if (cpu) { - estacks = (char *)__get_free_pages(GFP_ATOMIC, order[v]); - if (!estacks) - panic("Cannot allocate exception stack %ld %d\n", - v, cpu); + for (v = 0; v < N_EXCEPTION_STACKS; v++) { + if (cpu) { + estacks = (char *)__get_free_pages(GFP_ATOMIC, order[v]); + if (!estacks) + panic("Cannot allocate exception " + "stack %ld %d\n", v, cpu); + } + estacks += PAGE_SIZE << order[v]; + orig_ist->ist[v] = t->x86_tss.ist[v] = + (unsigned long)estacks; } - estacks += PAGE_SIZE << order[v]; - orig_ist->ist[v] = t->x86_tss.ist[v] = (unsigned long)estacks; } t->x86_tss.io_bitmap_base = offsetof(struct tss_struct, io_bitmap); diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index 4e7271999a7..84bb395038d 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -737,63 +737,44 @@ static int find_psb_table(struct powernow_k8_data *data) #ifdef CONFIG_X86_POWERNOW_K8_ACPI static void powernow_k8_acpi_pst_values(struct powernow_k8_data *data, unsigned int index) { - if (!data->acpi_data->state_count || (cpu_family == CPU_HW_PSTATE)) + if (!data->acpi_data.state_count || (cpu_family == CPU_HW_PSTATE)) return; - data->irt = (data->acpi_data->states[index].control >> IRT_SHIFT) & IRT_MASK; - data->rvo = (data->acpi_data->states[index].control >> RVO_SHIFT) & RVO_MASK; - data->exttype = (data->acpi_data->states[index].control >> EXT_TYPE_SHIFT) & EXT_TYPE_MASK; - data->plllock = (data->acpi_data->states[index].control >> PLL_L_SHIFT) & PLL_L_MASK; - data->vidmvs = 1 << ((data->acpi_data->states[index].control >> MVS_SHIFT) & MVS_MASK); - data->vstable = (data->acpi_data->states[index].control >> VST_SHIFT) & VST_MASK; -} - - -static struct acpi_processor_performance *acpi_perf_data; -static int preregister_valid; - -static int powernow_k8_cpu_preinit_acpi(void) -{ - acpi_perf_data = alloc_percpu(struct acpi_processor_performance); - if (!acpi_perf_data) - return -ENODEV; - - if (acpi_processor_preregister_performance(acpi_perf_data)) - return -ENODEV; - else - preregister_valid = 1; - return 0; + data->irt = (data->acpi_data.states[index].control >> IRT_SHIFT) & IRT_MASK; + data->rvo = (data->acpi_data.states[index].control >> RVO_SHIFT) & RVO_MASK; + data->exttype = (data->acpi_data.states[index].control >> EXT_TYPE_SHIFT) & EXT_TYPE_MASK; + data->plllock = (data->acpi_data.states[index].control >> PLL_L_SHIFT) & PLL_L_MASK; + data->vidmvs = 1 << ((data->acpi_data.states[index].control >> MVS_SHIFT) & MVS_MASK); + data->vstable = (data->acpi_data.states[index].control >> VST_SHIFT) & VST_MASK; } static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) { struct cpufreq_frequency_table *powernow_table; int ret_val; - int cpu = 0; - data->acpi_data = percpu_ptr(acpi_perf_data, cpu); - if (acpi_processor_register_performance(data->acpi_data, data->cpu)) { + if (acpi_processor_register_performance(&data->acpi_data, data->cpu)) { dprintk("register performance failed: bad ACPI data\n"); return -EIO; } /* verify the data contained in the ACPI structures */ - if (data->acpi_data->state_count <= 1) { + if (data->acpi_data.state_count <= 1) { dprintk("No ACPI P-States\n"); goto err_out; } - if ((data->acpi_data->control_register.space_id != ACPI_ADR_SPACE_FIXED_HARDWARE) || - (data->acpi_data->status_register.space_id != ACPI_ADR_SPACE_FIXED_HARDWARE)) { + if ((data->acpi_data.control_register.space_id != ACPI_ADR_SPACE_FIXED_HARDWARE) || + (data->acpi_data.status_register.space_id != ACPI_ADR_SPACE_FIXED_HARDWARE)) { dprintk("Invalid control/status registers (%x - %x)\n", - data->acpi_data->control_register.space_id, - data->acpi_data->status_register.space_id); + data->acpi_data.control_register.space_id, + data->acpi_data.status_register.space_id); goto err_out; } /* fill in data->powernow_table */ powernow_table = kmalloc((sizeof(struct cpufreq_frequency_table) - * (data->acpi_data->state_count + 1)), GFP_KERNEL); + * (data->acpi_data.state_count + 1)), GFP_KERNEL); if (!powernow_table) { dprintk("powernow_table memory alloc failure\n"); goto err_out; @@ -806,12 +787,12 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) if (ret_val) goto err_out_mem; - powernow_table[data->acpi_data->state_count].frequency = CPUFREQ_TABLE_END; - powernow_table[data->acpi_data->state_count].index = 0; + powernow_table[data->acpi_data.state_count].frequency = CPUFREQ_TABLE_END; + powernow_table[data->acpi_data.state_count].index = 0; data->powernow_table = powernow_table; /* fill in data */ - data->numps = data->acpi_data->state_count; + data->numps = data->acpi_data.state_count; if (first_cpu(per_cpu(cpu_core_map, data->cpu)) == data->cpu) print_basics(data); powernow_k8_acpi_pst_values(data, 0); @@ -819,31 +800,16 @@ static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) /* notify BIOS that we exist */ acpi_processor_notify_smm(THIS_MODULE); - /* determine affinity, from ACPI if available */ - if (preregister_valid) { - if ((data->acpi_data->shared_type == CPUFREQ_SHARED_TYPE_ALL) || - (data->acpi_data->shared_type == CPUFREQ_SHARED_TYPE_ANY)) - data->starting_core_affinity = data->acpi_data->shared_cpu_map; - else - data->starting_core_affinity = cpumask_of_cpu(data->cpu); - } else { - /* best guess from family if not */ - if (cpu_family == CPU_HW_PSTATE) - data->starting_core_affinity = cpumask_of_cpu(data->cpu); - else - data->starting_core_affinity = per_cpu(cpu_core_map, data->cpu); - } - return 0; err_out_mem: kfree(powernow_table); err_out: - acpi_processor_unregister_performance(data->acpi_data, data->cpu); + acpi_processor_unregister_performance(&data->acpi_data, data->cpu); /* data->acpi_data.state_count informs us at ->exit() whether ACPI was used */ - data->acpi_data->state_count = 0; + data->acpi_data.state_count = 0; return -ENODEV; } @@ -855,10 +821,10 @@ static int fill_powernow_table_pstate(struct powernow_k8_data *data, struct cpuf rdmsr(MSR_PSTATE_CUR_LIMIT, hi, lo); data->max_hw_pstate = (hi & HW_PSTATE_MAX_MASK) >> HW_PSTATE_MAX_SHIFT; - for (i = 0; i < data->acpi_data->state_count; i++) { + for (i = 0; i < data->acpi_data.state_count; i++) { u32 index; - index = data->acpi_data->states[i].control & HW_PSTATE_MASK; + index = data->acpi_data.states[i].control & HW_PSTATE_MASK; if (index > data->max_hw_pstate) { printk(KERN_ERR PFX "invalid pstate %d - bad value %d.\n", i, index); printk(KERN_ERR PFX "Please report to BIOS manufacturer\n"); @@ -874,7 +840,7 @@ static int fill_powernow_table_pstate(struct powernow_k8_data *data, struct cpuf powernow_table[i].index = index; - powernow_table[i].frequency = data->acpi_data->states[i].core_frequency * 1000; + powernow_table[i].frequency = data->acpi_data.states[i].core_frequency * 1000; } return 0; } @@ -883,16 +849,16 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data, struct cpuf { int i; int cntlofreq = 0; - for (i = 0; i < data->acpi_data->state_count; i++) { + for (i = 0; i < data->acpi_data.state_count; i++) { u32 fid; u32 vid; if (data->exttype) { - fid = data->acpi_data->states[i].status & EXT_FID_MASK; - vid = (data->acpi_data->states[i].status >> VID_SHIFT) & EXT_VID_MASK; + fid = data->acpi_data.states[i].status & EXT_FID_MASK; + vid = (data->acpi_data.states[i].status >> VID_SHIFT) & EXT_VID_MASK; } else { - fid = data->acpi_data->states[i].control & FID_MASK; - vid = (data->acpi_data->states[i].control >> VID_SHIFT) & VID_MASK; + fid = data->acpi_data.states[i].control & FID_MASK; + vid = (data->acpi_data.states[i].control >> VID_SHIFT) & VID_MASK; } dprintk(" %d : fid 0x%x, vid 0x%x\n", i, fid, vid); @@ -933,10 +899,10 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data, struct cpuf cntlofreq = i; } - if (powernow_table[i].frequency != (data->acpi_data->states[i].core_frequency * 1000)) { + if (powernow_table[i].frequency != (data->acpi_data.states[i].core_frequency * 1000)) { printk(KERN_INFO PFX "invalid freq entries %u kHz vs. %u kHz\n", powernow_table[i].frequency, - (unsigned int) (data->acpi_data->states[i].core_frequency * 1000)); + (unsigned int) (data->acpi_data.states[i].core_frequency * 1000)); powernow_table[i].frequency = CPUFREQ_ENTRY_INVALID; continue; } @@ -946,12 +912,11 @@ static int fill_powernow_table_fidvid(struct powernow_k8_data *data, struct cpuf static void powernow_k8_cpu_exit_acpi(struct powernow_k8_data *data) { - if (data->acpi_data->state_count) - acpi_processor_unregister_performance(data->acpi_data, data->cpu); + if (data->acpi_data.state_count) + acpi_processor_unregister_performance(&data->acpi_data, data->cpu); } #else -static int powernow_k8_cpu_preinit_acpi(void) { return -ENODEV; } static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) { return -ENODEV; } static void powernow_k8_cpu_exit_acpi(struct powernow_k8_data *data) { return; } static void powernow_k8_acpi_pst_values(struct powernow_k8_data *data, unsigned int index) { return; } @@ -1136,7 +1101,7 @@ static int powernowk8_verify(struct cpufreq_policy *pol) static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) { struct powernow_k8_data *data; - cpumask_t oldmask = CPU_MASK_ALL; + cpumask_t oldmask; int rc; if (!cpu_online(pol->cpu)) @@ -1209,7 +1174,10 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) /* run on any CPU again */ set_cpus_allowed_ptr(current, &oldmask); - pol->cpus = data->starting_core_affinity; + if (cpu_family == CPU_HW_PSTATE) + pol->cpus = cpumask_of_cpu(pol->cpu); + else + pol->cpus = per_cpu(cpu_core_map, pol->cpu); data->available_cores = &(pol->cpus); /* Take a crude guess here. @@ -1332,7 +1300,6 @@ static int __cpuinit powernowk8_init(void) } if (supported_cpus == num_online_cpus()) { - powernow_k8_cpu_preinit_acpi(); printk(KERN_INFO PFX "Found %d %s " "processors (%d cpu cores) (" VERSION ")\n", num_online_nodes(), @@ -1349,10 +1316,6 @@ static void __exit powernowk8_exit(void) dprintk("exit\n"); cpufreq_unregister_driver(&cpufreq_amd64_driver); - -#ifdef CONFIG_X86_POWERNOW_K8_ACPI - free_percpu(acpi_perf_data); -#endif } MODULE_AUTHOR("Paul Devriendt <paul.devriendt@amd.com> and Mark Langsdorf <mark.langsdorf@amd.com>"); diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h index a62612cd4be..ab48cfed4d9 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h @@ -33,13 +33,12 @@ struct powernow_k8_data { #ifdef CONFIG_X86_POWERNOW_K8_ACPI /* the acpi table needs to be kept. it's only available if ACPI was * used to determine valid frequency/vid/fid states */ - struct acpi_processor_performance *acpi_data; + struct acpi_processor_performance acpi_data; #endif /* we need to keep track of associated cores, but let cpufreq * handle hotplug events - so just point at cpufreq pol->cpus * structure */ cpumask_t *available_cores; - cpumask_t starting_core_affinity; }; diff --git a/arch/x86/kernel/cpu/cyrix.c b/arch/x86/kernel/cpu/cyrix.c index 3fd7a67bb06..898a5a2002e 100644 --- a/arch/x86/kernel/cpu/cyrix.c +++ b/arch/x86/kernel/cpu/cyrix.c @@ -15,13 +15,11 @@ /* * Read NSC/Cyrix DEVID registers (DIR) to get more detailed info. about the CPU */ -static void __cpuinit do_cyrix_devid(unsigned char *dir0, unsigned char *dir1) +static void __cpuinit __do_cyrix_devid(unsigned char *dir0, unsigned char *dir1) { unsigned char ccr2, ccr3; - unsigned long flags; /* we test for DEVID by checking whether CCR3 is writable */ - local_irq_save(flags); ccr3 = getCx86(CX86_CCR3); setCx86(CX86_CCR3, ccr3 ^ 0x80); getCx86(0xc0); /* dummy to change bus */ @@ -44,9 +42,16 @@ static void __cpuinit do_cyrix_devid(unsigned char *dir0, unsigned char *dir1) *dir0 = getCx86(CX86_DIR0); *dir1 = getCx86(CX86_DIR1); } - local_irq_restore(flags); } +static void __cpuinit do_cyrix_devid(unsigned char *dir0, unsigned char *dir1) +{ + unsigned long flags; + + local_irq_save(flags); + __do_cyrix_devid(dir0, dir1); + local_irq_restore(flags); +} /* * Cx86_dir0_msb is a HACK needed by check_cx686_cpuid/slop in bugs.h in * order to identify the Cyrix CPU model after we're out of setup.c @@ -134,23 +139,6 @@ static void __cpuinit set_cx86_memwb(void) setCx86(CX86_CCR2, getCx86(CX86_CCR2) | 0x14); } -static void __cpuinit set_cx86_inc(void) -{ - unsigned char ccr3; - - printk(KERN_INFO "Enable Incrementor on Cyrix/NSC processor.\n"); - - ccr3 = getCx86(CX86_CCR3); - setCx86(CX86_CCR3, (ccr3 & 0x0f) | 0x10); /* enable MAPEN */ - /* PCR1 -- Performance Control */ - /* Incrementor on, whatever that is */ - setCx86(CX86_PCR1, getCx86(CX86_PCR1) | 0x02); - /* PCR0 -- Performance Control */ - /* Incrementor Margin 10 */ - setCx86(CX86_PCR0, getCx86(CX86_PCR0) | 0x04); - setCx86(CX86_CCR3, ccr3); /* disable MAPEN */ -} - /* * Configure later MediaGX and/or Geode processor. */ @@ -174,11 +162,28 @@ static void __cpuinit geode_configure(void) set_cx86_memwb(); set_cx86_reorder(); - set_cx86_inc(); local_irq_restore(flags); } +static void __cpuinit early_init_cyrix(struct cpuinfo_x86 *c) +{ + unsigned char dir0, dir0_msn, dir1 = 0; + + __do_cyrix_devid(&dir0, &dir1); + dir0_msn = dir0 >> 4; /* identifies CPU "family" */ + + switch (dir0_msn) { + case 3: /* 6x86/6x86L */ + /* Emulate MTRRs using Cyrix's ARRs. */ + set_cpu_cap(c, X86_FEATURE_CYRIX_ARR); + break; + case 5: /* 6x86MX/M II */ + /* Emulate MTRRs using Cyrix's ARRs. */ + set_cpu_cap(c, X86_FEATURE_CYRIX_ARR); + break; + } +} static void __cpuinit init_cyrix(struct cpuinfo_x86 *c) { @@ -434,6 +439,7 @@ static void __cpuinit cyrix_identify(struct cpuinfo_x86 *c) static struct cpu_dev cyrix_cpu_dev __cpuinitdata = { .c_vendor = "Cyrix", .c_ident = { "CyrixInstead" }, + .c_early_init = early_init_cyrix, .c_init = init_cyrix, .c_identify = cyrix_identify, }; diff --git a/arch/x86/kernel/cpu/feature_names.c b/arch/x86/kernel/cpu/feature_names.c index e43ad4ad4cb..c9017799497 100644 --- a/arch/x86/kernel/cpu/feature_names.c +++ b/arch/x86/kernel/cpu/feature_names.c @@ -39,7 +39,8 @@ const char * const x86_cap_flags[NCAPINTS*32] = { NULL, NULL, NULL, NULL, "constant_tsc", "up", NULL, "arch_perfmon", "pebs", "bts", NULL, NULL, - "rep_good", NULL, NULL, NULL, NULL, NULL, NULL, NULL, + "rep_good", NULL, NULL, NULL, + "nopl", NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL, /* Intel-defined (#2) */ diff --git a/arch/x86/kernel/cpu/intel.c b/arch/x86/kernel/cpu/intel.c index b75f2569b8f..f113ef4595f 100644 --- a/arch/x86/kernel/cpu/intel.c +++ b/arch/x86/kernel/cpu/intel.c @@ -222,10 +222,11 @@ static void __cpuinit init_intel(struct cpuinfo_x86 *c) set_cpu_cap(c, X86_FEATURE_BTS); if (!(l1 & (1<<12))) set_cpu_cap(c, X86_FEATURE_PEBS); + ds_init_intel(c); } if (cpu_has_bts) - ds_init_intel(c); + ptrace_bts_init_intel(c); /* * See if we have a good local APIC by checking for buggy Pentia, diff --git a/arch/x86/kernel/cpu/mcheck/mce_64.c b/arch/x86/kernel/cpu/mcheck/mce_64.c index 65a339678ec..726a5fcdf34 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_64.c @@ -759,6 +759,7 @@ static struct sysdev_class mce_sysclass = { }; DEFINE_PER_CPU(struct sys_device, device_mce); +void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu) __cpuinitdata; /* Why are there no generic functions for this? */ #define ACCESSOR(name, var, start) \ @@ -883,9 +884,13 @@ static int __cpuinit mce_cpu_callback(struct notifier_block *nfb, case CPU_ONLINE: case CPU_ONLINE_FROZEN: mce_create_device(cpu); + if (threshold_cpu_callback) + threshold_cpu_callback(action, cpu); break; case CPU_DEAD: case CPU_DEAD_FROZEN: + if (threshold_cpu_callback) + threshold_cpu_callback(action, cpu); mce_remove_device(cpu); break; } diff --git a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c index 88736cadbaa..5eb390a4b2e 100644 --- a/arch/x86/kernel/cpu/mcheck/mce_amd_64.c +++ b/arch/x86/kernel/cpu/mcheck/mce_amd_64.c @@ -628,6 +628,7 @@ static void threshold_remove_bank(unsigned int cpu, int bank) deallocate_threshold_block(cpu, bank); free_out: + kobject_del(b->kobj); kobject_put(b->kobj); kfree(b); per_cpu(threshold_banks, cpu)[bank] = NULL; @@ -645,14 +646,11 @@ static void threshold_remove_device(unsigned int cpu) } /* get notified when a cpu comes on/off */ -static int __cpuinit threshold_cpu_callback(struct notifier_block *nfb, - unsigned long action, void *hcpu) +static void __cpuinit amd_64_threshold_cpu_callback(unsigned long action, + unsigned int cpu) { - /* cpu was unsigned int to begin with */ - unsigned int cpu = (unsigned long)hcpu; - if (cpu >= NR_CPUS) - goto out; + return; switch (action) { case CPU_ONLINE: @@ -666,14 +664,8 @@ static int __cpuinit threshold_cpu_callback(struct notifier_block *nfb, default: break; } - out: - return NOTIFY_OK; } -static struct notifier_block threshold_cpu_notifier __cpuinitdata = { - .notifier_call = threshold_cpu_callback, -}; - static __init int threshold_init_device(void) { unsigned lcpu = 0; @@ -684,7 +676,7 @@ static __init int threshold_init_device(void) if (err) return err; } - register_hotcpu_notifier(&threshold_cpu_notifier); + threshold_cpu_callback = amd_64_threshold_cpu_callback; return 0; } diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c index 509bd3d9eac..cb7d3b6a80e 100644 --- a/arch/x86/kernel/cpu/mtrr/generic.c +++ b/arch/x86/kernel/cpu/mtrr/generic.c @@ -379,6 +379,7 @@ static void generic_get_mtrr(unsigned int reg, unsigned long *base, unsigned long *size, mtrr_type *type) { unsigned int mask_lo, mask_hi, base_lo, base_hi; + unsigned int tmp, hi; rdmsr(MTRRphysMask_MSR(reg), mask_lo, mask_hi); if ((mask_lo & 0x800) == 0) { @@ -392,8 +393,23 @@ static void generic_get_mtrr(unsigned int reg, unsigned long *base, rdmsr(MTRRphysBase_MSR(reg), base_lo, base_hi); /* Work out the shifted address mask. */ - mask_lo = size_or_mask | mask_hi << (32 - PAGE_SHIFT) - | mask_lo >> PAGE_SHIFT; + tmp = mask_hi << (32 - PAGE_SHIFT) | mask_lo >> PAGE_SHIFT; + mask_lo = size_or_mask | tmp; + /* Expand tmp with high bits to all 1s*/ + hi = fls(tmp); + if (hi > 0) { + tmp |= ~((1<<(hi - 1)) - 1); + + if (tmp != mask_lo) { + static int once = 1; + + if (once) { + printk(KERN_INFO "mtrr: your BIOS has set up an incorrect mask, fixing it up.\n"); + once = 0; + } + mask_lo = tmp; + } + } /* This works correctly if size is a power of two, i.e. a contiguous range. */ diff --git a/arch/x86/kernel/cpu/mtrr/main.c b/arch/x86/kernel/cpu/mtrr/main.c index b5ade28ca8f..5df16d81837 100644 --- a/arch/x86/kernel/cpu/mtrr/main.c +++ b/arch/x86/kernel/cpu/mtrr/main.c @@ -834,7 +834,7 @@ static int __init enable_mtrr_cleanup_setup(char *str) enable_mtrr_cleanup = 1; return 0; } -early_param("enble_mtrr_cleanup", enable_mtrr_cleanup_setup); +early_param("enable_mtrr_cleanup", enable_mtrr_cleanup_setup); struct var_mtrr_state { unsigned long range_startk; @@ -1496,11 +1496,8 @@ int __init mtrr_trim_uncached_memory(unsigned long end_pfn) /* kvm/qemu doesn't have mtrr set right, don't trim them all */ if (!highest_pfn) { - if (!kvm_para_available()) { - printk(KERN_WARNING + WARN(!kvm_para_available(), KERN_WARNING "WARNING: strange, CPU MTRRs all blank?\n"); - WARN_ON(1); - } return 0; } diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c index de7439f82b9..05cc22dbd4f 100644 --- a/arch/x86/kernel/cpu/perfctr-watchdog.c +++ b/arch/x86/kernel/cpu/perfctr-watchdog.c @@ -478,7 +478,13 @@ static int setup_p4_watchdog(unsigned nmi_hz) perfctr_msr = MSR_P4_IQ_PERFCTR1; evntsel_msr = MSR_P4_CRU_ESCR0; cccr_msr = MSR_P4_IQ_CCCR1; - cccr_val = P4_CCCR_OVF_PMI1 | P4_CCCR_ESCR_SELECT(4); + + /* Pentium 4 D processors don't support P4_CCCR_OVF_PMI1 */ + if (boot_cpu_data.x86_model == 4 && boot_cpu_data.x86_mask == 4) + cccr_val = P4_CCCR_OVF_PMI0; + else + cccr_val = P4_CCCR_OVF_PMI1; + cccr_val |= P4_CCCR_ESCR_SELECT(4); } evntsel = P4_ESCR_EVENT_SELECT(0x3F) diff --git a/arch/x86/kernel/cpuid.c b/arch/x86/kernel/cpuid.c index 14b11b3be31..8e9cd6a8ec1 100644 --- a/arch/x86/kernel/cpuid.c +++ b/arch/x86/kernel/cpuid.c @@ -89,6 +89,8 @@ static ssize_t cpuid_read(struct file *file, char __user *buf, struct cpuid_regs cmd; int cpu = iminor(file->f_path.dentry->d_inode); u64 pos = *ppos; + ssize_t bytes = 0; + int err = 0; if (count % 16) return -EINVAL; /* Invalid chunk size */ @@ -96,14 +98,19 @@ static ssize_t cpuid_read(struct file *file, char __user *buf, for (; count; count -= 16) { cmd.eax = pos; cmd.ecx = pos >> 32; - smp_call_function_single(cpu, cpuid_smp_cpuid, &cmd, 1); - if (copy_to_user(tmp, &cmd, 16)) - return -EFAULT; + err = smp_call_function_single(cpu, cpuid_smp_cpuid, &cmd, 1); + if (err) + break; + if (copy_to_user(tmp, &cmd, 16)) { + err = -EFAULT; + break; + } tmp += 16; + bytes += 16; *ppos = ++pos; } - return tmp - buf; + return bytes ? bytes : err; } static int cpuid_open(struct inode *inode, struct file *file) diff --git a/arch/x86/kernel/ds.c b/arch/x86/kernel/ds.c index 11c11b8ec48..2b69994fd3a 100644 --- a/arch/x86/kernel/ds.c +++ b/arch/x86/kernel/ds.c @@ -2,26 +2,49 @@ * Debug Store support * * This provides a low-level interface to the hardware's Debug Store - * feature that is used for last branch recording (LBR) and + * feature that is used for branch trace store (BTS) and * precise-event based sampling (PEBS). * - * Different architectures use a different DS layout/pointer size. - * The below functions therefore work on a void*. + * It manages: + * - per-thread and per-cpu allocation of BTS and PEBS + * - buffer memory allocation (optional) + * - buffer overflow handling + * - buffer access * + * It assumes: + * - get_task_struct on all parameter tasks + * - current is allowed to trace parameter tasks * - * Since there is no user for PEBS, yet, only LBR (or branch - * trace store, BTS) is supported. * - * - * Copyright (C) 2007 Intel Corporation. - * Markus Metzger <markus.t.metzger@intel.com>, Dec 2007 + * Copyright (C) 2007-2008 Intel Corporation. + * Markus Metzger <markus.t.metzger@intel.com>, 2007-2008 */ + +#ifdef CONFIG_X86_DS + #include <asm/ds.h> #include <linux/errno.h> #include <linux/string.h> #include <linux/slab.h> +#include <linux/sched.h> +#include <linux/mm.h> + + +/* + * The configuration for a particular DS hardware implementation. + */ +struct ds_configuration { + /* the size of the DS structure in bytes */ + unsigned char sizeof_ds; + /* the size of one pointer-typed field in the DS structure in bytes; + this covers the first 8 fields related to buffer management. */ + unsigned char sizeof_field; + /* the size of a BTS/PEBS record in bytes */ + unsigned char sizeof_rec[2]; +}; +static struct ds_configuration ds_cfg; /* @@ -44,378 +67,747 @@ * (interrupt occurs when write pointer passes interrupt pointer) * - value to which counter is reset following counter overflow * - * On later architectures, the last branch recording hardware uses - * 64bit pointers even in 32bit mode. - * - * - * Branch Trace Store (BTS) records store information about control - * flow changes. They at least provide the following information: - * - source linear address - * - destination linear address + * Later architectures use 64bit pointers throughout, whereas earlier + * architectures use 32bit pointers in 32bit mode. * - * Netburst supported a predicated bit that had been dropped in later - * architectures. We do not suppor it. * + * We compute the base address for the first 8 fields based on: + * - the field size stored in the DS configuration + * - the relative field position + * - an offset giving the start of the respective region * - * In order to abstract from the actual DS and BTS layout, we describe - * the access to the relevant fields. - * Thanks to Andi Kleen for proposing this design. + * This offset is further used to index various arrays holding + * information for BTS and PEBS at the respective index. * - * The implementation, however, is not as general as it might seem. In - * order to stay somewhat simple and efficient, we assume an - * underlying unsigned type (mostly a pointer type) and we expect the - * field to be at least as big as that type. + * On later 32bit processors, we only access the lower 32bit of the + * 64bit pointer fields. The upper halves will be zeroed out. */ -/* - * A special from_ip address to indicate that the BTS record is an - * info record that needs to be interpreted or skipped. - */ -#define BTS_ESCAPE_ADDRESS (-1) +enum ds_field { + ds_buffer_base = 0, + ds_index, + ds_absolute_maximum, + ds_interrupt_threshold, +}; -/* - * A field access descriptor - */ -struct access_desc { - unsigned char offset; - unsigned char size; +enum ds_qualifier { + ds_bts = 0, + ds_pebs }; +static inline unsigned long ds_get(const unsigned char *base, + enum ds_qualifier qual, enum ds_field field) +{ + base += (ds_cfg.sizeof_field * (field + (4 * qual))); + return *(unsigned long *)base; +} + +static inline void ds_set(unsigned char *base, enum ds_qualifier qual, + enum ds_field field, unsigned long value) +{ + base += (ds_cfg.sizeof_field * (field + (4 * qual))); + (*(unsigned long *)base) = value; +} + + /* - * The configuration for a particular DS/BTS hardware implementation. + * Locking is done only for allocating BTS or PEBS resources and for + * guarding context and buffer memory allocation. + * + * Most functions require the current task to own the ds context part + * they are going to access. All the locking is done when validating + * access to the context. */ -struct ds_configuration { - /* the DS configuration */ - unsigned char sizeof_ds; - struct access_desc bts_buffer_base; - struct access_desc bts_index; - struct access_desc bts_absolute_maximum; - struct access_desc bts_interrupt_threshold; - /* the BTS configuration */ - unsigned char sizeof_bts; - struct access_desc from_ip; - struct access_desc to_ip; - /* BTS variants used to store additional information like - timestamps */ - struct access_desc info_type; - struct access_desc info_data; - unsigned long debugctl_mask; -}; +static spinlock_t ds_lock = __SPIN_LOCK_UNLOCKED(ds_lock); /* - * The global configuration used by the below accessor functions + * Validate that the current task is allowed to access the BTS/PEBS + * buffer of the parameter task. + * + * Returns 0, if access is granted; -Eerrno, otherwise. */ -static struct ds_configuration ds_cfg; +static inline int ds_validate_access(struct ds_context *context, + enum ds_qualifier qual) +{ + if (!context) + return -EPERM; + + if (context->owner[qual] == current) + return 0; + + return -EPERM; +} + /* - * Accessor functions for some DS and BTS fields using the above - * global ptrace_bts_cfg. + * We either support (system-wide) per-cpu or per-thread allocation. + * We distinguish the two based on the task_struct pointer, where a + * NULL pointer indicates per-cpu allocation for the current cpu. + * + * Allocations are use-counted. As soon as resources are allocated, + * further allocations must be of the same type (per-cpu or + * per-thread). We model this by counting allocations (i.e. the number + * of tracers of a certain type) for one type negatively: + * =0 no tracers + * >0 number of per-thread tracers + * <0 number of per-cpu tracers + * + * The below functions to get and put tracers and to check the + * allocation type require the ds_lock to be held by the caller. + * + * Tracers essentially gives the number of ds contexts for a certain + * type of allocation. */ -static inline unsigned long get_bts_buffer_base(char *base) +static long tracers; + +static inline void get_tracer(struct task_struct *task) { - return *(unsigned long *)(base + ds_cfg.bts_buffer_base.offset); + tracers += (task ? 1 : -1); } -static inline void set_bts_buffer_base(char *base, unsigned long value) + +static inline void put_tracer(struct task_struct *task) { - (*(unsigned long *)(base + ds_cfg.bts_buffer_base.offset)) = value; + tracers -= (task ? 1 : -1); } -static inline unsigned long get_bts_index(char *base) + +static inline int check_tracer(struct task_struct *task) { - return *(unsigned long *)(base + ds_cfg.bts_index.offset); + return (task ? (tracers >= 0) : (tracers <= 0)); } -static inline void set_bts_index(char *base, unsigned long value) + + +/* + * The DS context is either attached to a thread or to a cpu: + * - in the former case, the thread_struct contains a pointer to the + * attached context. + * - in the latter case, we use a static array of per-cpu context + * pointers. + * + * Contexts are use-counted. They are allocated on first access and + * deallocated when the last user puts the context. + * + * We distinguish between an allocating and a non-allocating get of a + * context: + * - the allocating get is used for requesting BTS/PEBS resources. It + * requires the caller to hold the global ds_lock. + * - the non-allocating get is used for all other cases. A + * non-existing context indicates an error. It acquires and releases + * the ds_lock itself for obtaining the context. + * + * A context and its DS configuration are allocated and deallocated + * together. A context always has a DS configuration of the + * appropriate size. + */ +static DEFINE_PER_CPU(struct ds_context *, system_context); + +#define this_system_context per_cpu(system_context, smp_processor_id()) + +/* + * Returns the pointer to the parameter task's context or to the + * system-wide context, if task is NULL. + * + * Increases the use count of the returned context, if not NULL. + */ +static inline struct ds_context *ds_get_context(struct task_struct *task) { - (*(unsigned long *)(base + ds_cfg.bts_index.offset)) = value; + struct ds_context *context; + + spin_lock(&ds_lock); + + context = (task ? task->thread.ds_ctx : this_system_context); + if (context) + context->count++; + + spin_unlock(&ds_lock); + + return context; } -static inline unsigned long get_bts_absolute_maximum(char *base) + +/* + * Same as ds_get_context, but allocates the context and it's DS + * structure, if necessary; returns NULL; if out of memory. + * + * pre: requires ds_lock to be held + */ +static inline struct ds_context *ds_alloc_context(struct task_struct *task) { - return *(unsigned long *)(base + ds_cfg.bts_absolute_maximum.offset); + struct ds_context **p_context = + (task ? &task->thread.ds_ctx : &this_system_context); + struct ds_context *context = *p_context; + + if (!context) { + context = kzalloc(sizeof(*context), GFP_KERNEL); + + if (!context) + return NULL; + + context->ds = kzalloc(ds_cfg.sizeof_ds, GFP_KERNEL); + if (!context->ds) { + kfree(context); + return NULL; + } + + *p_context = context; + + context->this = p_context; + context->task = task; + + if (task) + set_tsk_thread_flag(task, TIF_DS_AREA_MSR); + + if (!task || (task == current)) + wrmsr(MSR_IA32_DS_AREA, (unsigned long)context->ds, 0); + + get_tracer(task); + } + + context->count++; + + return context; } -static inline void set_bts_absolute_maximum(char *base, unsigned long value) + +/* + * Decreases the use count of the parameter context, if not NULL. + * Deallocates the context, if the use count reaches zero. + */ +static inline void ds_put_context(struct ds_context *context) { - (*(unsigned long *)(base + ds_cfg.bts_absolute_maximum.offset)) = value; + if (!context) + return; + + spin_lock(&ds_lock); + + if (--context->count) + goto out; + + *(context->this) = NULL; + + if (context->task) + clear_tsk_thread_flag(context->task, TIF_DS_AREA_MSR); + + if (!context->task || (context->task == current)) + wrmsrl(MSR_IA32_DS_AREA, 0); + + put_tracer(context->task); + + /* free any leftover buffers from tracers that did not + * deallocate them properly. */ + kfree(context->buffer[ds_bts]); + kfree(context->buffer[ds_pebs]); + kfree(context->ds); + kfree(context); + out: + spin_unlock(&ds_lock); } -static inline unsigned long get_bts_interrupt_threshold(char *base) + + +/* + * Handle a buffer overflow + * + * task: the task whose buffers are overflowing; + * NULL for a buffer overflow on the current cpu + * context: the ds context + * qual: the buffer type + */ +static void ds_overflow(struct task_struct *task, struct ds_context *context, + enum ds_qualifier qual) { - return *(unsigned long *)(base + ds_cfg.bts_interrupt_threshold.offset); + if (!context) + return; + + if (context->callback[qual]) + (*context->callback[qual])(task); + + /* todo: do some more overflow handling */ } -static inline void set_bts_interrupt_threshold(char *base, unsigned long value) + + +/* + * Allocate a non-pageable buffer of the parameter size. + * Checks the memory and the locked memory rlimit. + * + * Returns the buffer, if successful; + * NULL, if out of memory or rlimit exceeded. + * + * size: the requested buffer size in bytes + * pages (out): if not NULL, contains the number of pages reserved + */ +static inline void *ds_allocate_buffer(size_t size, unsigned int *pages) { - (*(unsigned long *)(base + ds_cfg.bts_interrupt_threshold.offset)) = value; + unsigned long rlim, vm, pgsz; + void *buffer; + + pgsz = PAGE_ALIGN(size) >> PAGE_SHIFT; + + rlim = current->signal->rlim[RLIMIT_AS].rlim_cur >> PAGE_SHIFT; + vm = current->mm->total_vm + pgsz; + if (rlim < vm) + return NULL; + + rlim = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >> PAGE_SHIFT; + vm = current->mm->locked_vm + pgsz; + if (rlim < vm) + return NULL; + + buffer = kzalloc(size, GFP_KERNEL); + if (!buffer) + return NULL; + + current->mm->total_vm += pgsz; + current->mm->locked_vm += pgsz; + + if (pages) + *pages = pgsz; + + return buffer; } -static inline unsigned long get_from_ip(char *base) + +static int ds_request(struct task_struct *task, void *base, size_t size, + ds_ovfl_callback_t ovfl, enum ds_qualifier qual) { - return *(unsigned long *)(base + ds_cfg.from_ip.offset); + struct ds_context *context; + unsigned long buffer, adj; + const unsigned long alignment = (1 << 3); + int error = 0; + + if (!ds_cfg.sizeof_ds) + return -EOPNOTSUPP; + + /* we require some space to do alignment adjustments below */ + if (size < (alignment + ds_cfg.sizeof_rec[qual])) + return -EINVAL; + + /* buffer overflow notification is not yet implemented */ + if (ovfl) + return -EOPNOTSUPP; + + + spin_lock(&ds_lock); + + if (!check_tracer(task)) + return -EPERM; + + error = -ENOMEM; + context = ds_alloc_context(task); + if (!context) + goto out_unlock; + + error = -EALREADY; + if (context->owner[qual] == current) + goto out_unlock; + error = -EPERM; + if (context->owner[qual] != NULL) + goto out_unlock; + context->owner[qual] = current; + + spin_unlock(&ds_lock); + + + error = -ENOMEM; + if (!base) { + base = ds_allocate_buffer(size, &context->pages[qual]); + if (!base) + goto out_release; + + context->buffer[qual] = base; + } + error = 0; + + context->callback[qual] = ovfl; + + /* adjust the buffer address and size to meet alignment + * constraints: + * - buffer is double-word aligned + * - size is multiple of record size + * + * We checked the size at the very beginning; we have enough + * space to do the adjustment. + */ + buffer = (unsigned long)base; + + adj = ALIGN(buffer, alignment) - buffer; + buffer += adj; + size -= adj; + + size /= ds_cfg.sizeof_rec[qual]; + size *= ds_cfg.sizeof_rec[qual]; + + ds_set(context->ds, qual, ds_buffer_base, buffer); + ds_set(context->ds, qual, ds_index, buffer); + ds_set(context->ds, qual, ds_absolute_maximum, buffer + size); + + if (ovfl) { + /* todo: select a suitable interrupt threshold */ + } else + ds_set(context->ds, qual, + ds_interrupt_threshold, buffer + size + 1); + + /* we keep the context until ds_release */ + return error; + + out_release: + context->owner[qual] = NULL; + ds_put_context(context); + return error; + + out_unlock: + spin_unlock(&ds_lock); + ds_put_context(context); + return error; } -static inline void set_from_ip(char *base, unsigned long value) + +int ds_request_bts(struct task_struct *task, void *base, size_t size, + ds_ovfl_callback_t ovfl) { - (*(unsigned long *)(base + ds_cfg.from_ip.offset)) = value; + return ds_request(task, base, size, ovfl, ds_bts); } -static inline unsigned long get_to_ip(char *base) + +int ds_request_pebs(struct task_struct *task, void *base, size_t size, + ds_ovfl_callback_t ovfl) { - return *(unsigned long *)(base + ds_cfg.to_ip.offset); + return ds_request(task, base, size, ovfl, ds_pebs); } -static inline void set_to_ip(char *base, unsigned long value) + +static int ds_release(struct task_struct *task, enum ds_qualifier qual) { - (*(unsigned long *)(base + ds_cfg.to_ip.offset)) = value; + struct ds_context *context; + int error; + + context = ds_get_context(task); + error = ds_validate_access(context, qual); + if (error < 0) + goto out; + + kfree(context->buffer[qual]); + context->buffer[qual] = NULL; + + current->mm->total_vm -= context->pages[qual]; + current->mm->locked_vm -= context->pages[qual]; + context->pages[qual] = 0; + context->owner[qual] = NULL; + + /* + * we put the context twice: + * once for the ds_get_context + * once for the corresponding ds_request + */ + ds_put_context(context); + out: + ds_put_context(context); + return error; } -static inline unsigned char get_info_type(char *base) + +int ds_release_bts(struct task_struct *task) { - return *(unsigned char *)(base + ds_cfg.info_type.offset); + return ds_release(task, ds_bts); } -static inline void set_info_type(char *base, unsigned char value) + +int ds_release_pebs(struct task_struct *task) { - (*(unsigned char *)(base + ds_cfg.info_type.offset)) = value; + return ds_release(task, ds_pebs); } -static inline unsigned long get_info_data(char *base) + +static int ds_get_index(struct task_struct *task, size_t *pos, + enum ds_qualifier qual) { - return *(unsigned long *)(base + ds_cfg.info_data.offset); + struct ds_context *context; + unsigned long base, index; + int error; + + context = ds_get_context(task); + error = ds_validate_access(context, qual); + if (error < 0) + goto out; + + base = ds_get(context->ds, qual, ds_buffer_base); + index = ds_get(context->ds, qual, ds_index); + + error = ((index - base) / ds_cfg.sizeof_rec[qual]); + if (pos) + *pos = error; + out: + ds_put_context(context); + return error; } -static inline void set_info_data(char *base, unsigned long value) + +int ds_get_bts_index(struct task_struct *task, size_t *pos) { - (*(unsigned long *)(base + ds_cfg.info_data.offset)) = value; + return ds_get_index(task, pos, ds_bts); } +int ds_get_pebs_index(struct task_struct *task, size_t *pos) +{ + return ds_get_index(task, pos, ds_pebs); +} -int ds_allocate(void **dsp, size_t bts_size_in_bytes) +static int ds_get_end(struct task_struct *task, size_t *pos, + enum ds_qualifier qual) { - size_t bts_size_in_records; - unsigned long bts; - void *ds; + struct ds_context *context; + unsigned long base, end; + int error; + + context = ds_get_context(task); + error = ds_validate_access(context, qual); + if (error < 0) + goto out; + + base = ds_get(context->ds, qual, ds_buffer_base); + end = ds_get(context->ds, qual, ds_absolute_maximum); + + error = ((end - base) / ds_cfg.sizeof_rec[qual]); + if (pos) + *pos = error; + out: + ds_put_context(context); + return error; +} - if (!ds_cfg.sizeof_ds || !ds_cfg.sizeof_bts) - return -EOPNOTSUPP; +int ds_get_bts_end(struct task_struct *task, size_t *pos) +{ + return ds_get_end(task, pos, ds_bts); +} - if (bts_size_in_bytes < 0) - return -EINVAL; +int ds_get_pebs_end(struct task_struct *task, size_t *pos) +{ + return ds_get_end(task, pos, ds_pebs); +} - bts_size_in_records = - bts_size_in_bytes / ds_cfg.sizeof_bts; - bts_size_in_bytes = - bts_size_in_records * ds_cfg.sizeof_bts; +static int ds_access(struct task_struct *task, size_t index, + const void **record, enum ds_qualifier qual) +{ + struct ds_context *context; + unsigned long base, idx; + int error; - if (bts_size_in_bytes <= 0) + if (!record) return -EINVAL; - bts = (unsigned long)kzalloc(bts_size_in_bytes, GFP_KERNEL); - - if (!bts) - return -ENOMEM; + context = ds_get_context(task); + error = ds_validate_access(context, qual); + if (error < 0) + goto out; - ds = kzalloc(ds_cfg.sizeof_ds, GFP_KERNEL); + base = ds_get(context->ds, qual, ds_buffer_base); + idx = base + (index * ds_cfg.sizeof_rec[qual]); - if (!ds) { - kfree((void *)bts); - return -ENOMEM; - } - - set_bts_buffer_base(ds, bts); - set_bts_index(ds, bts); - set_bts_absolute_maximum(ds, bts + bts_size_in_bytes); - set_bts_interrupt_threshold(ds, bts + bts_size_in_bytes + 1); + error = -EINVAL; + if (idx > ds_get(context->ds, qual, ds_absolute_maximum)) + goto out; - *dsp = ds; - return 0; + *record = (const void *)idx; + error = ds_cfg.sizeof_rec[qual]; + out: + ds_put_context(context); + return error; } -int ds_free(void **dsp) +int ds_access_bts(struct task_struct *task, size_t index, const void **record) { - if (*dsp) { - kfree((void *)get_bts_buffer_base(*dsp)); - kfree(*dsp); - *dsp = NULL; - } - return 0; + return ds_access(task, index, record, ds_bts); } -int ds_get_bts_size(void *ds) +int ds_access_pebs(struct task_struct *task, size_t index, const void **record) { - int size_in_bytes; - - if (!ds_cfg.sizeof_ds || !ds_cfg.sizeof_bts) - return -EOPNOTSUPP; - - if (!ds) - return 0; - - size_in_bytes = - get_bts_absolute_maximum(ds) - - get_bts_buffer_base(ds); - return size_in_bytes; + return ds_access(task, index, record, ds_pebs); } -int ds_get_bts_end(void *ds) +static int ds_write(struct task_struct *task, const void *record, size_t size, + enum ds_qualifier qual, int force) { - int size_in_bytes = ds_get_bts_size(ds); - - if (size_in_bytes <= 0) - return size_in_bytes; + struct ds_context *context; + int error; - return size_in_bytes / ds_cfg.sizeof_bts; -} + if (!record) + return -EINVAL; -int ds_get_bts_index(void *ds) -{ - int index_offset_in_bytes; + error = -EPERM; + context = ds_get_context(task); + if (!context) + goto out; - if (!ds_cfg.sizeof_ds || !ds_cfg.sizeof_bts) - return -EOPNOTSUPP; + if (!force) { + error = ds_validate_access(context, qual); + if (error < 0) + goto out; + } - index_offset_in_bytes = - get_bts_index(ds) - - get_bts_buffer_base(ds); + error = 0; + while (size) { + unsigned long base, index, end, write_end, int_th; + unsigned long write_size, adj_write_size; + + /* + * write as much as possible without producing an + * overflow interrupt. + * + * interrupt_threshold must either be + * - bigger than absolute_maximum or + * - point to a record between buffer_base and absolute_maximum + * + * index points to a valid record. + */ + base = ds_get(context->ds, qual, ds_buffer_base); + index = ds_get(context->ds, qual, ds_index); + end = ds_get(context->ds, qual, ds_absolute_maximum); + int_th = ds_get(context->ds, qual, ds_interrupt_threshold); + + write_end = min(end, int_th); + + /* if we are already beyond the interrupt threshold, + * we fill the entire buffer */ + if (write_end <= index) + write_end = end; + + if (write_end <= index) + goto out; + + write_size = min((unsigned long) size, write_end - index); + memcpy((void *)index, record, write_size); + + record = (const char *)record + write_size; + size -= write_size; + error += write_size; + + adj_write_size = write_size / ds_cfg.sizeof_rec[qual]; + adj_write_size *= ds_cfg.sizeof_rec[qual]; + + /* zero out trailing bytes */ + memset((char *)index + write_size, 0, + adj_write_size - write_size); + index += adj_write_size; + + if (index >= end) + index = base; + ds_set(context->ds, qual, ds_index, index); + + if (index >= int_th) + ds_overflow(task, context, qual); + } - return index_offset_in_bytes / ds_cfg.sizeof_bts; + out: + ds_put_context(context); + return error; } -int ds_set_overflow(void *ds, int method) +int ds_write_bts(struct task_struct *task, const void *record, size_t size) { - switch (method) { - case DS_O_SIGNAL: - return -EOPNOTSUPP; - case DS_O_WRAP: - return 0; - default: - return -EINVAL; - } + return ds_write(task, record, size, ds_bts, /* force = */ 0); } -int ds_get_overflow(void *ds) +int ds_write_pebs(struct task_struct *task, const void *record, size_t size) { - return DS_O_WRAP; + return ds_write(task, record, size, ds_pebs, /* force = */ 0); } -int ds_clear(void *ds) +int ds_unchecked_write_bts(struct task_struct *task, + const void *record, size_t size) { - int bts_size = ds_get_bts_size(ds); - unsigned long bts_base; - - if (bts_size <= 0) - return bts_size; - - bts_base = get_bts_buffer_base(ds); - memset((void *)bts_base, 0, bts_size); - - set_bts_index(ds, bts_base); - return 0; + return ds_write(task, record, size, ds_bts, /* force = */ 1); } -int ds_read_bts(void *ds, int index, struct bts_struct *out) +int ds_unchecked_write_pebs(struct task_struct *task, + const void *record, size_t size) { - void *bts; + return ds_write(task, record, size, ds_pebs, /* force = */ 1); +} - if (!ds_cfg.sizeof_ds || !ds_cfg.sizeof_bts) - return -EOPNOTSUPP; +static int ds_reset_or_clear(struct task_struct *task, + enum ds_qualifier qual, int clear) +{ + struct ds_context *context; + unsigned long base, end; + int error; - if (index < 0) - return -EINVAL; + context = ds_get_context(task); + error = ds_validate_access(context, qual); + if (error < 0) + goto out; - if (index >= ds_get_bts_size(ds)) - return -EINVAL; + base = ds_get(context->ds, qual, ds_buffer_base); + end = ds_get(context->ds, qual, ds_absolute_maximum); - bts = (void *)(get_bts_buffer_base(ds) + (index * ds_cfg.sizeof_bts)); + if (clear) + memset((void *)base, 0, end - base); - memset(out, 0, sizeof(*out)); - if (get_from_ip(bts) == BTS_ESCAPE_ADDRESS) { - out->qualifier = get_info_type(bts); - out->variant.jiffies = get_info_data(bts); - } else { - out->qualifier = BTS_BRANCH; - out->variant.lbr.from_ip = get_from_ip(bts); - out->variant.lbr.to_ip = get_to_ip(bts); - } + ds_set(context->ds, qual, ds_index, base); - return sizeof(*out);; + error = 0; + out: + ds_put_context(context); + return error; } -int ds_write_bts(void *ds, const struct bts_struct *in) +int ds_reset_bts(struct task_struct *task) { - unsigned long bts; - - if (!ds_cfg.sizeof_ds || !ds_cfg.sizeof_bts) - return -EOPNOTSUPP; - - if (ds_get_bts_size(ds) <= 0) - return -ENXIO; + return ds_reset_or_clear(task, ds_bts, /* clear = */ 0); +} - bts = get_bts_index(ds); +int ds_reset_pebs(struct task_struct *task) +{ + return ds_reset_or_clear(task, ds_pebs, /* clear = */ 0); +} - memset((void *)bts, 0, ds_cfg.sizeof_bts); - switch (in->qualifier) { - case BTS_INVALID: - break; +int ds_clear_bts(struct task_struct *task) +{ + return ds_reset_or_clear(task, ds_bts, /* clear = */ 1); +} - case BTS_BRANCH: - set_from_ip((void *)bts, in->variant.lbr.from_ip); - set_to_ip((void *)bts, in->variant.lbr.to_ip); - break; +int ds_clear_pebs(struct task_struct *task) +{ + return ds_reset_or_clear(task, ds_pebs, /* clear = */ 1); +} - case BTS_TASK_ARRIVES: - case BTS_TASK_DEPARTS: - set_from_ip((void *)bts, BTS_ESCAPE_ADDRESS); - set_info_type((void *)bts, in->qualifier); - set_info_data((void *)bts, in->variant.jiffies); - break; +int ds_get_pebs_reset(struct task_struct *task, u64 *value) +{ + struct ds_context *context; + int error; - default: + if (!value) return -EINVAL; - } - bts = bts + ds_cfg.sizeof_bts; - if (bts >= get_bts_absolute_maximum(ds)) - bts = get_bts_buffer_base(ds); - set_bts_index(ds, bts); + context = ds_get_context(task); + error = ds_validate_access(context, ds_pebs); + if (error < 0) + goto out; - return ds_cfg.sizeof_bts; + *value = *(u64 *)(context->ds + (ds_cfg.sizeof_field * 8)); + + error = 0; + out: + ds_put_context(context); + return error; } -unsigned long ds_debugctl_mask(void) +int ds_set_pebs_reset(struct task_struct *task, u64 value) { - return ds_cfg.debugctl_mask; -} + struct ds_context *context; + int error; -#ifdef __i386__ -static const struct ds_configuration ds_cfg_netburst = { - .sizeof_ds = 9 * 4, - .bts_buffer_base = { 0, 4 }, - .bts_index = { 4, 4 }, - .bts_absolute_maximum = { 8, 4 }, - .bts_interrupt_threshold = { 12, 4 }, - .sizeof_bts = 3 * 4, - .from_ip = { 0, 4 }, - .to_ip = { 4, 4 }, - .info_type = { 4, 1 }, - .info_data = { 8, 4 }, - .debugctl_mask = (1<<2)|(1<<3) -}; + context = ds_get_context(task); + error = ds_validate_access(context, ds_pebs); + if (error < 0) + goto out; -static const struct ds_configuration ds_cfg_pentium_m = { - .sizeof_ds = 9 * 4, - .bts_buffer_base = { 0, 4 }, - .bts_index = { 4, 4 }, - .bts_absolute_maximum = { 8, 4 }, - .bts_interrupt_threshold = { 12, 4 }, - .sizeof_bts = 3 * 4, - .from_ip = { 0, 4 }, - .to_ip = { 4, 4 }, - .info_type = { 4, 1 }, - .info_data = { 8, 4 }, - .debugctl_mask = (1<<6)|(1<<7) + *(u64 *)(context->ds + (ds_cfg.sizeof_field * 8)) = value; + + error = 0; + out: + ds_put_context(context); + return error; +} + +static const struct ds_configuration ds_cfg_var = { + .sizeof_ds = sizeof(long) * 12, + .sizeof_field = sizeof(long), + .sizeof_rec[ds_bts] = sizeof(long) * 3, + .sizeof_rec[ds_pebs] = sizeof(long) * 10 }; -#endif /* _i386_ */ - -static const struct ds_configuration ds_cfg_core2 = { - .sizeof_ds = 9 * 8, - .bts_buffer_base = { 0, 8 }, - .bts_index = { 8, 8 }, - .bts_absolute_maximum = { 16, 8 }, - .bts_interrupt_threshold = { 24, 8 }, - .sizeof_bts = 3 * 8, - .from_ip = { 0, 8 }, - .to_ip = { 8, 8 }, - .info_type = { 8, 1 }, - .info_data = { 16, 8 }, - .debugctl_mask = (1<<6)|(1<<7)|(1<<9) +static const struct ds_configuration ds_cfg_64 = { + .sizeof_ds = 8 * 12, + .sizeof_field = 8, + .sizeof_rec[ds_bts] = 8 * 3, + .sizeof_rec[ds_pebs] = 8 * 10 }; static inline void @@ -429,14 +821,13 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) switch (c->x86) { case 0x6: switch (c->x86_model) { -#ifdef __i386__ case 0xD: case 0xE: /* Pentium M */ - ds_configure(&ds_cfg_pentium_m); + ds_configure(&ds_cfg_var); break; -#endif /* _i386_ */ case 0xF: /* Core2 */ - ds_configure(&ds_cfg_core2); + case 0x1C: /* Atom */ + ds_configure(&ds_cfg_64); break; default: /* sorry, don't know about them */ @@ -445,13 +836,11 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) break; case 0xF: switch (c->x86_model) { -#ifdef __i386__ case 0x0: case 0x1: case 0x2: /* Netburst */ - ds_configure(&ds_cfg_netburst); + ds_configure(&ds_cfg_var); break; -#endif /* _i386_ */ default: /* sorry, don't know about them */ break; @@ -462,3 +851,14 @@ void __cpuinit ds_init_intel(struct cpuinfo_x86 *c) break; } } + +void ds_free(struct ds_context *context) +{ + /* This is called when the task owning the parameter context + * is dying. There should not be any user of that context left + * to disturb us, anymore. */ + unsigned long leftovers = context->count; + while (leftovers--) + ds_put_context(context); +} +#endif /* CONFIG_X86_DS */ diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index 9af89078f7b..66e48aa2dd1 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c @@ -1203,7 +1203,7 @@ static int __init parse_memmap_opt(char *p) if (!p) return -EINVAL; - if (!strcmp(p, "exactmap")) { + if (!strncmp(p, "exactmap", 8)) { #ifdef CONFIG_CRASH_DUMP /* * If we are doing a crash dump, we still need to know diff --git a/arch/x86/kernel/efi_32.c b/arch/x86/kernel/efi_32.c index 4b63c8e1f13..5cab48ee61a 100644 --- a/arch/x86/kernel/efi_32.c +++ b/arch/x86/kernel/efi_32.c @@ -53,7 +53,7 @@ void efi_call_phys_prelog(void) * directory. If I have PAE, I just need to duplicate one entry in * page directory. */ - cr4 = read_cr4(); + cr4 = read_cr4_safe(); if (cr4 & X86_CR4_PAE) { efi_bak_pg_dir_pointer[0].pgd = @@ -91,7 +91,7 @@ void efi_call_phys_epilog(void) gdt_descr.size = GDT_SIZE - 1; load_gdt(&gdt_descr); - cr4 = read_cr4(); + cr4 = read_cr4_safe(); if (cr4 & X86_CR4_PAE) { swapper_pg_dir[pgd_index(0)].pgd = diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 2cfcbded888..bfa837cb16b 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -222,7 +222,7 @@ static __init void map_low_mmrs(void) enum map_type {map_wb, map_uc}; -static void map_high(char *id, unsigned long base, int shift, enum map_type map_type) +static __init void map_high(char *id, unsigned long base, int shift, enum map_type map_type) { unsigned long bytes, paddr; @@ -293,7 +293,9 @@ static __init void uv_rtc_init(void) sn_rtc_cycles_per_second = ticks_per_sec; } -static __init void uv_system_init(void) +static bool uv_system_inited; + +void __init uv_system_init(void) { union uvh_si_addr_map_config_u m_n_config; union uvh_node_id_u node_id; @@ -383,6 +385,7 @@ static __init void uv_system_init(void) map_mmr_high(max_pnode); map_config_high(max_pnode); map_mmioh_high(max_pnode); + uv_system_inited = true; } /* @@ -391,8 +394,7 @@ static __init void uv_system_init(void) */ void __cpuinit uv_cpu_init(void) { - if (!uv_node_to_blade) - uv_system_init(); + BUG_ON(!uv_system_inited); uv_blade_info[uv_numa_blade_id()].nr_online_cpus++; diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c index 1b318e903bf..9bfc4d72fb2 100644 --- a/arch/x86/kernel/head64.c +++ b/arch/x86/kernel/head64.c @@ -88,6 +88,7 @@ void __init x86_64_start_kernel(char * real_mode_data) BUILD_BUG_ON(!(MODULES_VADDR > __START_KERNEL)); BUILD_BUG_ON(!(((MODULES_END - 1) & PGDIR_MASK) == (__START_KERNEL & PGDIR_MASK))); + BUILD_BUG_ON(__fix_to_virt(__end_of_fixed_addresses) <= MODULES_END); /* clear bss before set_intr_gate with early_idt_handler */ clear_bss(); diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index ad2b15a1334..73deaffadd0 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -210,8 +210,8 @@ static void hpet_legacy_clockevent_register(void) /* Calculate the min / max delta */ hpet_clockevent.max_delta_ns = clockevent_delta2ns(0x7FFFFFFF, &hpet_clockevent); - hpet_clockevent.min_delta_ns = clockevent_delta2ns(0x30, - &hpet_clockevent); + /* 5 usec minimum reprogramming delta. */ + hpet_clockevent.min_delta_ns = 5000; /* * Start hpet with the boot cpu mask and make it @@ -270,15 +270,22 @@ static void hpet_legacy_set_mode(enum clock_event_mode mode, } static int hpet_legacy_next_event(unsigned long delta, - struct clock_event_device *evt) + struct clock_event_device *evt) { - unsigned long cnt; + u32 cnt; cnt = hpet_readl(HPET_COUNTER); - cnt += delta; + cnt += (u32) delta; hpet_writel(cnt, HPET_T0_CMP); - return ((long)(hpet_readl(HPET_COUNTER) - cnt ) > 0) ? -ETIME : 0; + /* + * We need to read back the CMP register to make sure that + * what we wrote hit the chip before we compare it to the + * counter. + */ + WARN_ON((u32)hpet_readl(HPET_T0_CMP) != cnt); + + return (s32)((u32)hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; } /* @@ -359,6 +366,7 @@ static int hpet_clocksource_register(void) int __init hpet_enable(void) { unsigned long id; + int i; if (!is_hpet_capable()) return 0; @@ -369,6 +377,29 @@ int __init hpet_enable(void) * Read the period and check for a sane value: */ hpet_period = hpet_readl(HPET_PERIOD); + + /* + * AMD SB700 based systems with spread spectrum enabled use a + * SMM based HPET emulation to provide proper frequency + * setting. The SMM code is initialized with the first HPET + * register access and takes some time to complete. During + * this time the config register reads 0xffffffff. We check + * for max. 1000 loops whether the config register reads a non + * 0xffffffff value to make sure that HPET is up and running + * before we go further. A counting loop is safe, as the HPET + * access takes thousands of CPU cycles. On non SB700 based + * machines this check is only done once and has no side + * effects. + */ + for (i = 0; hpet_readl(HPET_CFG) == 0xFFFFFFFF; i++) { + if (i == 1000) { + printk(KERN_WARNING + "HPET config register value = 0xFFFFFFFF. " + "Disabling HPET\n"); + goto out_nohpet; + } + } + if (hpet_period < HPET_MIN_PERIOD || hpet_period > HPET_MAX_PERIOD) goto out_nohpet; diff --git a/arch/x86/kernel/io_delay.c b/arch/x86/kernel/io_delay.c index 1c3a66a67f8..720d2607aac 100644 --- a/arch/x86/kernel/io_delay.c +++ b/arch/x86/kernel/io_delay.c @@ -92,6 +92,14 @@ static struct dmi_system_id __initdata io_delay_0xed_port_dmi_table[] = { DMI_MATCH(DMI_BOARD_NAME, "30BF") } }, + { + .callback = dmi_io_delay_0xed_port, + .ident = "Presario F700", + .matches = { + DMI_MATCH(DMI_BOARD_VENDOR, "Quanta"), + DMI_MATCH(DMI_BOARD_NAME, "30D3") + } + }, { } }; diff --git a/arch/x86/kernel/kdebugfs.c b/arch/x86/kernel/kdebugfs.c index f2d43bc7551..ff7d3b0124f 100644 --- a/arch/x86/kernel/kdebugfs.c +++ b/arch/x86/kernel/kdebugfs.c @@ -139,6 +139,7 @@ static int __init create_setup_data_nodes(struct dentry *parent) if (PageHighMem(pg)) { data = ioremap_cache(pa_data, sizeof(*data)); if (!data) { + kfree(node); error = -ENXIO; goto err_dir; } diff --git a/arch/x86/kernel/kgdb.c b/arch/x86/kernel/kgdb.c index f47f0eb886b..8282a213968 100644 --- a/arch/x86/kernel/kgdb.c +++ b/arch/x86/kernel/kgdb.c @@ -69,6 +69,9 @@ static int gdb_x86vector = -1; */ void pt_regs_to_gdb_regs(unsigned long *gdb_regs, struct pt_regs *regs) { +#ifndef CONFIG_X86_32 + u32 *gdb_regs32 = (u32 *)gdb_regs; +#endif gdb_regs[GDB_AX] = regs->ax; gdb_regs[GDB_BX] = regs->bx; gdb_regs[GDB_CX] = regs->cx; @@ -76,9 +79,9 @@ void pt_regs_to_gdb_regs(unsigned long *gdb_regs, struct pt_regs *regs) gdb_regs[GDB_SI] = regs->si; gdb_regs[GDB_DI] = regs->di; gdb_regs[GDB_BP] = regs->bp; - gdb_regs[GDB_PS] = regs->flags; gdb_regs[GDB_PC] = regs->ip; #ifdef CONFIG_X86_32 + gdb_regs[GDB_PS] = regs->flags; gdb_regs[GDB_DS] = regs->ds; gdb_regs[GDB_ES] = regs->es; gdb_regs[GDB_CS] = regs->cs; @@ -94,6 +97,9 @@ void pt_regs_to_gdb_regs(unsigned long *gdb_regs, struct pt_regs *regs) gdb_regs[GDB_R13] = regs->r13; gdb_regs[GDB_R14] = regs->r14; gdb_regs[GDB_R15] = regs->r15; + gdb_regs32[GDB_PS] = regs->flags; + gdb_regs32[GDB_CS] = regs->cs; + gdb_regs32[GDB_SS] = regs->ss; #endif gdb_regs[GDB_SP] = regs->sp; } @@ -112,6 +118,9 @@ void pt_regs_to_gdb_regs(unsigned long *gdb_regs, struct pt_regs *regs) */ void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs, struct task_struct *p) { +#ifndef CONFIG_X86_32 + u32 *gdb_regs32 = (u32 *)gdb_regs; +#endif gdb_regs[GDB_AX] = 0; gdb_regs[GDB_BX] = 0; gdb_regs[GDB_CX] = 0; @@ -129,8 +138,10 @@ void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs, struct task_struct *p) gdb_regs[GDB_FS] = 0xFFFF; gdb_regs[GDB_GS] = 0xFFFF; #else - gdb_regs[GDB_PS] = *(unsigned long *)(p->thread.sp + 8); - gdb_regs[GDB_PC] = 0; + gdb_regs32[GDB_PS] = *(unsigned long *)(p->thread.sp + 8); + gdb_regs32[GDB_CS] = __KERNEL_CS; + gdb_regs32[GDB_SS] = __KERNEL_DS; + gdb_regs[GDB_PC] = p->thread.ip; gdb_regs[GDB_R8] = 0; gdb_regs[GDB_R9] = 0; gdb_regs[GDB_R10] = 0; @@ -153,6 +164,9 @@ void sleeping_thread_to_gdb_regs(unsigned long *gdb_regs, struct task_struct *p) */ void gdb_regs_to_pt_regs(unsigned long *gdb_regs, struct pt_regs *regs) { +#ifndef CONFIG_X86_32 + u32 *gdb_regs32 = (u32 *)gdb_regs; +#endif regs->ax = gdb_regs[GDB_AX]; regs->bx = gdb_regs[GDB_BX]; regs->cx = gdb_regs[GDB_CX]; @@ -160,9 +174,9 @@ void gdb_regs_to_pt_regs(unsigned long *gdb_regs, struct pt_regs *regs) regs->si = gdb_regs[GDB_SI]; regs->di = gdb_regs[GDB_DI]; regs->bp = gdb_regs[GDB_BP]; - regs->flags = gdb_regs[GDB_PS]; regs->ip = gdb_regs[GDB_PC]; #ifdef CONFIG_X86_32 + regs->flags = gdb_regs[GDB_PS]; regs->ds = gdb_regs[GDB_DS]; regs->es = gdb_regs[GDB_ES]; regs->cs = gdb_regs[GDB_CS]; @@ -175,6 +189,9 @@ void gdb_regs_to_pt_regs(unsigned long *gdb_regs, struct pt_regs *regs) regs->r13 = gdb_regs[GDB_R13]; regs->r14 = gdb_regs[GDB_R14]; regs->r15 = gdb_regs[GDB_R15]; + regs->flags = gdb_regs32[GDB_PS]; + regs->cs = gdb_regs32[GDB_CS]; + regs->ss = gdb_regs32[GDB_SS]; #endif } @@ -378,10 +395,8 @@ int kgdb_arch_handle_exception(int e_vector, int signo, int err_code, if (remcomInBuffer[0] == 's') { linux_regs->flags |= X86_EFLAGS_TF; kgdb_single_step = 1; - if (kgdb_contthread) { - atomic_set(&kgdb_cpu_doing_single_step, - raw_smp_processor_id()); - } + atomic_set(&kgdb_cpu_doing_single_step, + raw_smp_processor_id()); } get_debugreg(dr6, 6); @@ -466,9 +481,15 @@ static int __kgdb_notify(struct die_args *args, unsigned long cmd) case DIE_DEBUG: if (atomic_read(&kgdb_cpu_doing_single_step) == - raw_smp_processor_id() && - user_mode(regs)) - return single_step_cont(regs, args); + raw_smp_processor_id()) { + if (user_mode(regs)) + return single_step_cont(regs, args); + break; + } else if (test_thread_flag(TIF_SINGLESTEP)) + /* This means a user thread is single stepping + * a system call which should be ignored + */ + return NOTIFY_DONE; /* fall through */ default: if (user_mode(regs)) diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c index 9fe478d9840..0732adba05c 100644 --- a/arch/x86/kernel/machine_kexec_32.c +++ b/arch/x86/kernel/machine_kexec_32.c @@ -12,6 +12,7 @@ #include <linux/init.h> #include <linux/numa.h> #include <linux/ftrace.h> +#include <linux/suspend.h> #include <asm/pgtable.h> #include <asm/pgalloc.h> @@ -78,7 +79,7 @@ static void load_segments(void) /* * A architecture hook called to validate the * proposed image and prepare the control pages - * as needed. The pages for KEXEC_CONTROL_CODE_SIZE + * as needed. The pages for KEXEC_CONTROL_PAGE_SIZE * have been allocated, but the segments have yet * been copied into the kernel. * @@ -113,6 +114,7 @@ void machine_kexec(struct kimage *image) { unsigned long page_list[PAGES_NR]; void *control_page; + int save_ftrace_enabled; asmlinkage unsigned long (*relocate_kernel_ptr)(unsigned long indirection_page, unsigned long control_page, @@ -120,7 +122,12 @@ void machine_kexec(struct kimage *image) unsigned int has_pae, unsigned int preserve_context); - tracer_disable(); +#ifdef CONFIG_KEXEC_JUMP + if (kexec_image->preserve_context) + save_processor_state(); +#endif + + save_ftrace_enabled = __ftrace_enabled_save(); /* Interrupts aren't acceptable while we reboot */ local_irq_disable(); @@ -138,7 +145,7 @@ void machine_kexec(struct kimage *image) } control_page = page_address(image->control_code_page); - memcpy(control_page, relocate_kernel, PAGE_SIZE/2); + memcpy(control_page, relocate_kernel, KEXEC_CONTROL_CODE_MAX_SIZE); relocate_kernel_ptr = control_page; page_list[PA_CONTROL_PAGE] = __pa(control_page); @@ -178,6 +185,13 @@ void machine_kexec(struct kimage *image) (unsigned long)page_list, image->start, cpu_has_pae, image->preserve_context); + +#ifdef CONFIG_KEXEC_JUMP + if (kexec_image->preserve_context) + restore_processor_state(); +#endif + + __ftrace_enabled_restore(save_ftrace_enabled); } void arch_crash_save_vmcoreinfo(void) diff --git a/arch/x86/kernel/mfgpt_32.c b/arch/x86/kernel/mfgpt_32.c index 07c0f828f48..3b599518c32 100644 --- a/arch/x86/kernel/mfgpt_32.c +++ b/arch/x86/kernel/mfgpt_32.c @@ -33,6 +33,8 @@ #include <linux/module.h> #include <asm/geode.h> +#define MFGPT_DEFAULT_IRQ 7 + static struct mfgpt_timer_t { unsigned int avail:1; } mfgpt_timers[MFGPT_MAX_TIMERS]; @@ -157,29 +159,48 @@ int geode_mfgpt_toggle_event(int timer, int cmp, int event, int enable) } EXPORT_SYMBOL_GPL(geode_mfgpt_toggle_event); -int geode_mfgpt_set_irq(int timer, int cmp, int irq, int enable) +int geode_mfgpt_set_irq(int timer, int cmp, int *irq, int enable) { - u32 val, dummy; - int offset; + u32 zsel, lpc, dummy; + int shift; if (timer < 0 || timer >= MFGPT_MAX_TIMERS) return -EIO; - if (geode_mfgpt_toggle_event(timer, cmp, MFGPT_EVENT_IRQ, enable)) + /* + * Unfortunately, MFGPTs come in pairs sharing their IRQ lines. If VSA + * is using the same CMP of the timer's Siamese twin, the IRQ is set to + * 2, and we mustn't use nor change it. + * XXX: Likewise, 2 Linux drivers might clash if the 2nd overwrites the + * IRQ of the 1st. This can only happen if forcing an IRQ, calling this + * with *irq==0 is safe. Currently there _are_ no 2 drivers. + */ + rdmsr(MSR_PIC_ZSEL_LOW, zsel, dummy); + shift = ((cmp == MFGPT_CMP1 ? 0 : 4) + timer % 4) * 4; + if (((zsel >> shift) & 0xF) == 2) return -EIO; - rdmsr(MSR_PIC_ZSEL_LOW, val, dummy); + /* Choose IRQ: if none supplied, keep IRQ already set or use default */ + if (!*irq) + *irq = (zsel >> shift) & 0xF; + if (!*irq) + *irq = MFGPT_DEFAULT_IRQ; - offset = (timer % 4) * 4; - - val &= ~((0xF << offset) | (0xF << (offset + 16))); + /* Can't use IRQ if it's 0 (=disabled), 2, or routed to LPC */ + if (*irq < 1 || *irq == 2 || *irq > 15) + return -EIO; + rdmsr(MSR_PIC_IRQM_LPC, lpc, dummy); + if (lpc & (1 << *irq)) + return -EIO; + /* All chosen and checked - go for it */ + if (geode_mfgpt_toggle_event(timer, cmp, MFGPT_EVENT_IRQ, enable)) + return -EIO; if (enable) { - val |= (irq & 0x0F) << (offset); - val |= (irq & 0x0F) << (offset + 16); + zsel = (zsel & ~(0xF << shift)) | (*irq << shift); + wrmsr(MSR_PIC_ZSEL_LOW, zsel, dummy); } - wrmsr(MSR_PIC_ZSEL_LOW, val, dummy); return 0; } @@ -242,7 +263,7 @@ EXPORT_SYMBOL_GPL(geode_mfgpt_alloc_timer); static unsigned int mfgpt_tick_mode = CLOCK_EVT_MODE_SHUTDOWN; static u16 mfgpt_event_clock; -static int irq = 7; +static int irq; static int __init mfgpt_setup(char *str) { get_option(&str, &irq); @@ -346,7 +367,7 @@ int __init mfgpt_timer_setup(void) mfgpt_event_clock = timer; /* Set up the IRQ on the MFGPT side */ - if (geode_mfgpt_setup_irq(mfgpt_event_clock, MFGPT_CMP2, irq)) { + if (geode_mfgpt_setup_irq(mfgpt_event_clock, MFGPT_CMP2, &irq)) { printk(KERN_ERR "mfgpt-timer: Could not set up IRQ %d\n", irq); return -EIO; } @@ -374,13 +395,14 @@ int __init mfgpt_timer_setup(void) &mfgpt_clockevent); printk(KERN_INFO - "mfgpt-timer: registering the MFGPT timer as a clock event.\n"); + "mfgpt-timer: Registering MFGPT timer %d as a clock event, using IRQ %d\n", + timer, irq); clockevents_register_device(&mfgpt_clockevent); return 0; err: - geode_mfgpt_release_irq(mfgpt_event_clock, MFGPT_CMP2, irq); + geode_mfgpt_release_irq(mfgpt_event_clock, MFGPT_CMP2, &irq); printk(KERN_ERR "mfgpt-timer: Unable to set up the MFGPT clock source\n"); return -EIO; diff --git a/arch/x86/kernel/mmconf-fam10h_64.c b/arch/x86/kernel/mmconf-fam10h_64.c index fdfdc550b36..efc2f361fe8 100644 --- a/arch/x86/kernel/mmconf-fam10h_64.c +++ b/arch/x86/kernel/mmconf-fam10h_64.c @@ -238,7 +238,7 @@ static struct dmi_system_id __devinitdata mmconf_dmi_table[] = { {} }; -void __init check_enable_amd_mmconf_dmi(void) +void __cpuinit check_enable_amd_mmconf_dmi(void) { dmi_check_system(mmconf_dmi_table); } diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index 678090508a6..b3fb430725c 100644 --- a/arch/x86/kernel/mpparse.c +++ b/arch/x86/kernel/mpparse.c @@ -49,7 +49,7 @@ static int __init mpf_checksum(unsigned char *mp, int len) return sum & 0xFF; } -static void __cpuinit MP_processor_info(struct mpc_config_processor *m) +static void __init MP_processor_info(struct mpc_config_processor *m) { int apicid; char *bootup_cpu = ""; @@ -484,7 +484,7 @@ static void __init construct_default_ioirq_mptable(int mpc_default_type) } -static void construct_ioapic_table(int mpc_default_type) +static void __init construct_ioapic_table(int mpc_default_type) { struct mpc_config_ioapic ioapic; struct mpc_config_bus bus; @@ -529,7 +529,7 @@ static void construct_ioapic_table(int mpc_default_type) construct_default_ioirq_mptable(mpc_default_type); } #else -static inline void construct_ioapic_table(int mpc_default_type) { } +static inline void __init construct_ioapic_table(int mpc_default_type) { } #endif static inline void __init construct_default_ISA_mptable(int mpc_default_type) diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c index 9fd80955244..2e2af5d1819 100644 --- a/arch/x86/kernel/msr.c +++ b/arch/x86/kernel/msr.c @@ -72,21 +72,28 @@ static ssize_t msr_read(struct file *file, char __user *buf, u32 data[2]; u32 reg = *ppos; int cpu = iminor(file->f_path.dentry->d_inode); - int err; + int err = 0; + ssize_t bytes = 0; if (count % 8) return -EINVAL; /* Invalid chunk size */ for (; count; count -= 8) { err = rdmsr_safe_on_cpu(cpu, reg, &data[0], &data[1]); - if (err) - return -EIO; - if (copy_to_user(tmp, &data, 8)) - return -EFAULT; + if (err) { + if (err == -EFAULT) /* Fix idiotic error code */ + err = -EIO; + break; + } + if (copy_to_user(tmp, &data, 8)) { + err = -EFAULT; + break; + } tmp += 2; + bytes += 8; } - return ((char __user *)tmp) - buf; + return bytes ? bytes : err; } static ssize_t msr_write(struct file *file, const char __user *buf, @@ -96,21 +103,28 @@ static ssize_t msr_write(struct file *file, const char __user *buf, u32 data[2]; u32 reg = *ppos; int cpu = iminor(file->f_path.dentry->d_inode); - int err; + int err = 0; + ssize_t bytes = 0; if (count % 8) return -EINVAL; /* Invalid chunk size */ for (; count; count -= 8) { - if (copy_from_user(&data, tmp, 8)) - return -EFAULT; + if (copy_from_user(&data, tmp, 8)) { + err = -EFAULT; + break; + } err = wrmsr_safe_on_cpu(cpu, reg, data[0], data[1]); - if (err) - return -EIO; + if (err) { + if (err == -EFAULT) /* Fix idiotic error code */ + err = -EIO; + break; + } tmp += 2; + bytes += 8; } - return ((char __user *)tmp) - buf; + return bytes ? bytes : err; } static int msr_open(struct inode *inode, struct file *file) @@ -131,7 +145,7 @@ static int msr_open(struct inode *inode, struct file *file) ret = -EIO; /* MSR not supported */ out: unlock_kernel(); - return 0; + return ret; } /* diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c index ac6d51222e7..abb78a2cc4a 100644 --- a/arch/x86/kernel/nmi.c +++ b/arch/x86/kernel/nmi.c @@ -114,6 +114,23 @@ static __init void nmi_cpu_busy(void *data) } #endif +static void report_broken_nmi(int cpu, int *prev_nmi_count) +{ + printk(KERN_CONT "\n"); + + printk(KERN_WARNING + "WARNING: CPU#%d: NMI appears to be stuck (%d->%d)!\n", + cpu, prev_nmi_count[cpu], get_nmi_count(cpu)); + + printk(KERN_WARNING + "Please report this to bugzilla.kernel.org,\n"); + printk(KERN_WARNING + "and attach the output of the 'dmesg' command.\n"); + + per_cpu(wd_enabled, cpu) = 0; + atomic_dec(&nmi_active); +} + int __init check_nmi_watchdog(void) { unsigned int *prev_nmi_count; @@ -141,15 +158,8 @@ int __init check_nmi_watchdog(void) for_each_online_cpu(cpu) { if (!per_cpu(wd_enabled, cpu)) continue; - if (get_nmi_count(cpu) - prev_nmi_count[cpu] <= 5) { - printk(KERN_WARNING "WARNING: CPU#%d: NMI " - "appears to be stuck (%d->%d)!\n", - cpu, - prev_nmi_count[cpu], - get_nmi_count(cpu)); - per_cpu(wd_enabled, cpu) = 0; - atomic_dec(&nmi_active); - } + if (get_nmi_count(cpu) - prev_nmi_count[cpu] <= 5) + report_broken_nmi(cpu, prev_nmi_count); } endflag = 1; if (!atomic_read(&nmi_active)) { diff --git a/arch/x86/kernel/numaq_32.c b/arch/x86/kernel/numaq_32.c index b8c45610b20..eecc8c18f01 100644 --- a/arch/x86/kernel/numaq_32.c +++ b/arch/x86/kernel/numaq_32.c @@ -73,7 +73,7 @@ static void __init smp_dump_qct(void) } -void __init numaq_tsc_disable(void) +void __cpuinit numaq_tsc_disable(void) { if (!found_numaq) return; diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c index 94da4d52d79..300da17e61c 100644 --- a/arch/x86/kernel/paravirt.c +++ b/arch/x86/kernel/paravirt.c @@ -471,7 +471,7 @@ struct pv_lock_ops pv_lock_ops = { .spin_unlock = __ticket_spin_unlock, #endif }; -EXPORT_SYMBOL_GPL(pv_lock_ops); +EXPORT_SYMBOL(pv_lock_ops); EXPORT_SYMBOL_GPL(pv_time_ops); EXPORT_SYMBOL (pv_cpu_ops); diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index 02d19328525..dcdac6c826e 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -343,9 +343,8 @@ static void iommu_free(struct iommu_table *tbl, dma_addr_t dma_addr, /* were we called with bad_dma_address? */ badend = bad_dma_address + (EMERGENCY_PAGES * PAGE_SIZE); if (unlikely((dma_addr >= bad_dma_address) && (dma_addr < badend))) { - printk(KERN_ERR "Calgary: driver tried unmapping bad DMA " + WARN(1, KERN_ERR "Calgary: driver tried unmapping bad DMA " "address 0x%Lx\n", dma_addr); - WARN_ON(1); return; } @@ -1269,13 +1268,15 @@ static inline int __init determine_tce_table_size(u64 ram) static int __init build_detail_arrays(void) { unsigned long ptr; - int i, scal_detail_size, rio_detail_size; + unsigned numnodes, i; + int scal_detail_size, rio_detail_size; - if (rio_table_hdr->num_scal_dev > MAX_NUMNODES){ + numnodes = rio_table_hdr->num_scal_dev; + if (numnodes > MAX_NUMNODES){ printk(KERN_WARNING "Calgary: MAX_NUMNODES too low! Defined as %d, " "but system has %d nodes.\n", - MAX_NUMNODES, rio_table_hdr->num_scal_dev); + MAX_NUMNODES, numnodes); return -ENODEV; } @@ -1296,8 +1297,7 @@ static int __init build_detail_arrays(void) } ptr = ((unsigned long)rio_table_hdr) + 3; - for (i = 0; i < rio_table_hdr->num_scal_dev; - i++, ptr += scal_detail_size) + for (i = 0; i < numnodes; i++, ptr += scal_detail_size) scal_devs[i] = (struct scal_detail *)ptr; for (i = 0; i < rio_table_hdr->num_rio_dev; @@ -1350,7 +1350,7 @@ static void calgary_init_bitmap_from_tce_table(struct iommu_table *tbl) * Function for kdump case. Get the tce tables from first kernel * by reading the contents of the base adress register of calgary iommu */ -static void get_tce_space_from_tar(void) +static void __init get_tce_space_from_tar(void) { int bus; void __iomem *target; diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index 7fc4d5b0a6a..876e9189077 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -246,6 +246,14 @@ static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c) return 1; } +static cpumask_t c1e_mask = CPU_MASK_NONE; +static int c1e_detected; + +void c1e_remove_cpu(int cpu) +{ + cpu_clear(cpu, c1e_mask); +} + /* * C1E aware idle routine. We check for C1E active in the interrupt * pending message MSR. If we detect C1E, then we handle it the same @@ -253,9 +261,6 @@ static int __cpuinit check_c1e_idle(const struct cpuinfo_x86 *c) */ static void c1e_idle(void) { - static cpumask_t c1e_mask = CPU_MASK_NONE; - static int c1e_detected; - if (need_resched()) return; @@ -265,8 +270,10 @@ static void c1e_idle(void) rdmsr(MSR_K8_INT_PENDING_MSG, lo, hi); if (lo & K8_INTP_C1E_ACTIVE_MASK) { c1e_detected = 1; - mark_tsc_unstable("TSC halt in C1E"); - printk(KERN_INFO "System has C1E enabled\n"); + if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC)) + mark_tsc_unstable("TSC halt in AMD C1E"); + printk(KERN_INFO "System has AMD C1E enabled\n"); + set_cpu_cap(&boot_cpu_data, X86_FEATURE_AMDC1E); } } diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index 7b6e44a7c62..c8609dea443 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -55,6 +55,7 @@ #include <asm/tlbflush.h> #include <asm/cpu.h> #include <asm/kdebug.h> +#include <asm/idle.h> #include <asm/syscalls.h> #include <asm/smp.h> @@ -90,6 +91,7 @@ static void cpu_exit_clear(void) cpu_clear(cpu, cpu_callin_map); numa_remove_cpu(cpu); + c1e_remove_cpu(cpu); } /* We don't actually take CPU down, just spin without interrupts. */ @@ -97,7 +99,6 @@ static inline void play_dead(void) { /* This must be done before dead CPU ack */ cpu_exit_clear(); - wbinvd(); mb(); /* Ack it */ __get_cpu_var(cpu_state) = CPU_DEAD; @@ -106,8 +107,8 @@ static inline void play_dead(void) * With physical CPU hotplug, we should halt the cpu */ local_irq_disable(); - while (1) - halt(); + /* mask all interrupts, flush any and all caches, and halt */ + wbinvd_halt(); } #else static inline void play_dead(void) @@ -278,6 +279,14 @@ void exit_thread(void) tss->x86_tss.io_bitmap_base = INVALID_IO_BITMAP_OFFSET; put_cpu(); } +#ifdef CONFIG_X86_DS + /* Free any DS contexts that have not been properly released. */ + if (unlikely(current->thread.ds_ctx)) { + /* we clear debugctl to make sure DS is not used. */ + update_debugctlmsr(0); + ds_free(current->thread.ds_ctx); + } +#endif /* CONFIG_X86_DS */ } void flush_thread(void) @@ -439,6 +448,35 @@ int set_tsc_mode(unsigned int val) return 0; } +#ifdef CONFIG_X86_DS +static int update_debugctl(struct thread_struct *prev, + struct thread_struct *next, unsigned long debugctl) +{ + unsigned long ds_prev = 0; + unsigned long ds_next = 0; + + if (prev->ds_ctx) + ds_prev = (unsigned long)prev->ds_ctx->ds; + if (next->ds_ctx) + ds_next = (unsigned long)next->ds_ctx->ds; + + if (ds_next != ds_prev) { + /* we clear debugctl to make sure DS + * is not in use when we change it */ + debugctl = 0; + update_debugctlmsr(0); + wrmsr(MSR_IA32_DS_AREA, ds_next, 0); + } + return debugctl; +} +#else +static int update_debugctl(struct thread_struct *prev, + struct thread_struct *next, unsigned long debugctl) +{ + return debugctl; +} +#endif /* CONFIG_X86_DS */ + static noinline void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, struct tss_struct *tss) @@ -449,14 +487,7 @@ __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, prev = &prev_p->thread; next = &next_p->thread; - debugctl = prev->debugctlmsr; - if (next->ds_area_msr != prev->ds_area_msr) { - /* we clear debugctl to make sure DS - * is not in use when we change it */ - debugctl = 0; - update_debugctlmsr(0); - wrmsr(MSR_IA32_DS_AREA, next->ds_area_msr, 0); - } + debugctl = update_debugctl(prev, next, prev->debugctlmsr); if (next->debugctlmsr != debugctl) update_debugctlmsr(next->debugctlmsr); @@ -480,13 +511,13 @@ __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p, hard_enable_TSC(); } -#ifdef X86_BTS +#ifdef CONFIG_X86_PTRACE_BTS if (test_tsk_thread_flag(prev_p, TIF_BTS_TRACE_TS)) ptrace_bts_take_timestamp(prev_p, BTS_TASK_DEPARTS); if (test_tsk_thread_flag(next_p, TIF_BTS_TRACE_TS)) ptrace_bts_take_timestamp(next_p, BTS_TASK_ARRIVES); -#endif +#endif /* CONFIG_X86_PTRACE_BTS */ if (!test_tsk_thread_flag(next_p, TIF_IO_BITMAP)) { diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index 87d7dfdcf46..79e3e173ab4 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -94,14 +94,15 @@ DECLARE_PER_CPU(int, cpu_state); static inline void play_dead(void) { idle_task_exit(); - wbinvd(); + c1e_remove_cpu(raw_smp_processor_id()); + mb(); /* Ack it */ __get_cpu_var(cpu_state) = CPU_DEAD; local_irq_disable(); - while (1) - halt(); + /* mask all interrupts, flush any and all caches, and halt */ + wbinvd_halt(); } #else static inline void play_dead(void) @@ -240,6 +241,14 @@ void exit_thread(void) t->io_bitmap_max = 0; put_cpu(); } +#ifdef CONFIG_X86_DS + /* Free any DS contexts that have not been properly released. */ + if (unlikely(t->ds_ctx)) { + /* we clear debugctl to make sure DS is not used. */ + update_debugctlmsr(0); + ds_free(t->ds_ctx); + } +#endif /* CONFIG_X86_DS */ } void flush_thread(void) @@ -473,13 +482,27 @@ static inline void __switch_to_xtra(struct task_struct *prev_p, next = &next_p->thread; debugctl = prev->debugctlmsr; - if (next->ds_area_msr != prev->ds_area_msr) { - /* we clear debugctl to make sure DS - * is not in use when we change it */ - debugctl = 0; - update_debugctlmsr(0); - wrmsrl(MSR_IA32_DS_AREA, next->ds_area_msr); + +#ifdef CONFIG_X86_DS + { + unsigned long ds_prev = 0, ds_next = 0; + + if (prev->ds_ctx) + ds_prev = (unsigned long)prev->ds_ctx->ds; + if (next->ds_ctx) + ds_next = (unsigned long)next->ds_ctx->ds; + + if (ds_next != ds_prev) { + /* + * We clear debugctl to make sure DS + * is not in use when we change it: + */ + debugctl = 0; + update_debugctlmsr(0); + wrmsrl(MSR_IA32_DS_AREA, ds_next); + } } +#endif /* CONFIG_X86_DS */ if (next->debugctlmsr != debugctl) update_debugctlmsr(next->debugctlmsr); @@ -517,13 +540,13 @@ static inline void __switch_to_xtra(struct task_struct *prev_p, memset(tss->io_bitmap, 0xff, prev->io_bitmap_max); } -#ifdef X86_BTS +#ifdef CONFIG_X86_PTRACE_BTS if (test_tsk_thread_flag(prev_p, TIF_BTS_TRACE_TS)) ptrace_bts_take_timestamp(prev_p, BTS_TASK_DEPARTS); if (test_tsk_thread_flag(next_p, TIF_BTS_TRACE_TS)) ptrace_bts_take_timestamp(next_p, BTS_TASK_ARRIVES); -#endif +#endif /* CONFIG_X86_PTRACE_BTS */ } /* diff --git a/arch/x86/kernel/ptrace.c b/arch/x86/kernel/ptrace.c index fc3e8dcd9da..d00e756c05a 100644 --- a/arch/x86/kernel/ptrace.c +++ b/arch/x86/kernel/ptrace.c @@ -554,45 +554,115 @@ static int ptrace_set_debugreg(struct task_struct *child, return 0; } -#ifdef X86_BTS +#ifdef CONFIG_X86_PTRACE_BTS +/* + * The configuration for a particular BTS hardware implementation. + */ +struct bts_configuration { + /* the size of a BTS record in bytes; at most BTS_MAX_RECORD_SIZE */ + unsigned char sizeof_bts; + /* the size of a field in the BTS record in bytes */ + unsigned char sizeof_field; + /* a bitmask to enable/disable BTS in DEBUGCTL MSR */ + unsigned long debugctl_mask; +}; +static struct bts_configuration bts_cfg; + +#define BTS_MAX_RECORD_SIZE (8 * 3) + + +/* + * Branch Trace Store (BTS) uses the following format. Different + * architectures vary in the size of those fields. + * - source linear address + * - destination linear address + * - flags + * + * Later architectures use 64bit pointers throughout, whereas earlier + * architectures use 32bit pointers in 32bit mode. + * + * We compute the base address for the first 8 fields based on: + * - the field size stored in the DS configuration + * - the relative field position + * + * In order to store additional information in the BTS buffer, we use + * a special source address to indicate that the record requires + * special interpretation. + * + * Netburst indicated via a bit in the flags field whether the branch + * was predicted; this is ignored. + */ + +enum bts_field { + bts_from = 0, + bts_to, + bts_flags, + + bts_escape = (unsigned long)-1, + bts_qual = bts_to, + bts_jiffies = bts_flags +}; -static int ptrace_bts_get_size(struct task_struct *child) +static inline unsigned long bts_get(const char *base, enum bts_field field) { - if (!child->thread.ds_area_msr) - return -ENXIO; + base += (bts_cfg.sizeof_field * field); + return *(unsigned long *)base; +} + +static inline void bts_set(char *base, enum bts_field field, unsigned long val) +{ + base += (bts_cfg.sizeof_field * field);; + (*(unsigned long *)base) = val; +} - return ds_get_bts_index((void *)child->thread.ds_area_msr); +/* + * Translate a BTS record from the raw format into the bts_struct format + * + * out (out): bts_struct interpretation + * raw: raw BTS record + */ +static void ptrace_bts_translate_record(struct bts_struct *out, const void *raw) +{ + memset(out, 0, sizeof(*out)); + if (bts_get(raw, bts_from) == bts_escape) { + out->qualifier = bts_get(raw, bts_qual); + out->variant.jiffies = bts_get(raw, bts_jiffies); + } else { + out->qualifier = BTS_BRANCH; + out->variant.lbr.from_ip = bts_get(raw, bts_from); + out->variant.lbr.to_ip = bts_get(raw, bts_to); + } } -static int ptrace_bts_read_record(struct task_struct *child, - long index, +static int ptrace_bts_read_record(struct task_struct *child, size_t index, struct bts_struct __user *out) { struct bts_struct ret; - int retval; - int bts_end; - int bts_index; + const void *bts_record; + size_t bts_index, bts_end; + int error; - if (!child->thread.ds_area_msr) - return -ENXIO; + error = ds_get_bts_end(child, &bts_end); + if (error < 0) + return error; - if (index < 0) - return -EINVAL; - - bts_end = ds_get_bts_end((void *)child->thread.ds_area_msr); if (bts_end <= index) return -EINVAL; + error = ds_get_bts_index(child, &bts_index); + if (error < 0) + return error; + /* translate the ptrace bts index into the ds bts index */ - bts_index = ds_get_bts_index((void *)child->thread.ds_area_msr); - bts_index -= (index + 1); - if (bts_index < 0) - bts_index += bts_end; + bts_index += bts_end - (index + 1); + if (bts_end <= bts_index) + bts_index -= bts_end; + + error = ds_access_bts(child, bts_index, &bts_record); + if (error < 0) + return error; - retval = ds_read_bts((void *)child->thread.ds_area_msr, - bts_index, &ret); - if (retval < 0) - return retval; + ptrace_bts_translate_record(&ret, bts_record); if (copy_to_user(out, &ret, sizeof(ret))) return -EFAULT; @@ -600,101 +670,106 @@ static int ptrace_bts_read_record(struct task_struct *child, return sizeof(ret); } -static int ptrace_bts_clear(struct task_struct *child) -{ - if (!child->thread.ds_area_msr) - return -ENXIO; - - return ds_clear((void *)child->thread.ds_area_msr); -} - static int ptrace_bts_drain(struct task_struct *child, long size, struct bts_struct __user *out) { - int end, i; - void *ds = (void *)child->thread.ds_area_msr; - - if (!ds) - return -ENXIO; + struct bts_struct ret; + const unsigned char *raw; + size_t end, i; + int error; - end = ds_get_bts_index(ds); - if (end <= 0) - return end; + error = ds_get_bts_index(child, &end); + if (error < 0) + return error; if (size < (end * sizeof(struct bts_struct))) return -EIO; - for (i = 0; i < end; i++, out++) { - struct bts_struct ret; - int retval; + error = ds_access_bts(child, 0, (const void **)&raw); + if (error < 0) + return error; - retval = ds_read_bts(ds, i, &ret); - if (retval < 0) - return retval; + for (i = 0; i < end; i++, out++, raw += bts_cfg.sizeof_bts) { + ptrace_bts_translate_record(&ret, raw); if (copy_to_user(out, &ret, sizeof(ret))) return -EFAULT; } - ds_clear(ds); + error = ds_clear_bts(child); + if (error < 0) + return error; return end; } +static void ptrace_bts_ovfl(struct task_struct *child) +{ + send_sig(child->thread.bts_ovfl_signal, child, 0); +} + static int ptrace_bts_config(struct task_struct *child, long cfg_size, const struct ptrace_bts_config __user *ucfg) { struct ptrace_bts_config cfg; - int bts_size, ret = 0; - void *ds; + int error = 0; + + error = -EOPNOTSUPP; + if (!bts_cfg.sizeof_bts) + goto errout; + error = -EIO; if (cfg_size < sizeof(cfg)) - return -EIO; + goto errout; + error = -EFAULT; if (copy_from_user(&cfg, ucfg, sizeof(cfg))) - return -EFAULT; + goto errout; - if ((int)cfg.size < 0) - return -EINVAL; + error = -EINVAL; + if ((cfg.flags & PTRACE_BTS_O_SIGNAL) && + !(cfg.flags & PTRACE_BTS_O_ALLOC)) + goto errout; - bts_size = 0; - ds = (void *)child->thread.ds_area_msr; - if (ds) { - bts_size = ds_get_bts_size(ds); - if (bts_size < 0) - return bts_size; - } - cfg.size = PAGE_ALIGN(cfg.size); + if (cfg.flags & PTRACE_BTS_O_ALLOC) { + ds_ovfl_callback_t ovfl = NULL; + unsigned int sig = 0; + + /* we ignore the error in case we were not tracing child */ + (void)ds_release_bts(child); + + if (cfg.flags & PTRACE_BTS_O_SIGNAL) { + if (!cfg.signal) + goto errout; + + sig = cfg.signal; + ovfl = ptrace_bts_ovfl; + } - if (bts_size != cfg.size) { - ret = ptrace_bts_realloc(child, cfg.size, - cfg.flags & PTRACE_BTS_O_CUT_SIZE); - if (ret < 0) + error = ds_request_bts(child, /* base = */ NULL, cfg.size, ovfl); + if (error < 0) goto errout; - ds = (void *)child->thread.ds_area_msr; + child->thread.bts_ovfl_signal = sig; } - if (cfg.flags & PTRACE_BTS_O_SIGNAL) - ret = ds_set_overflow(ds, DS_O_SIGNAL); - else - ret = ds_set_overflow(ds, DS_O_WRAP); - if (ret < 0) + error = -EINVAL; + if (!child->thread.ds_ctx && cfg.flags) goto errout; if (cfg.flags & PTRACE_BTS_O_TRACE) - child->thread.debugctlmsr |= ds_debugctl_mask(); + child->thread.debugctlmsr |= bts_cfg.debugctl_mask; else - child->thread.debugctlmsr &= ~ds_debugctl_mask(); + child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; if (cfg.flags & PTRACE_BTS_O_SCHED) set_tsk_thread_flag(child, TIF_BTS_TRACE_TS); else clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); - ret = sizeof(cfg); + error = sizeof(cfg); out: if (child->thread.debugctlmsr) @@ -702,10 +777,10 @@ out: else clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); - return ret; + return error; errout: - child->thread.debugctlmsr &= ~ds_debugctl_mask(); + child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); goto out; } @@ -714,29 +789,40 @@ static int ptrace_bts_status(struct task_struct *child, long cfg_size, struct ptrace_bts_config __user *ucfg) { - void *ds = (void *)child->thread.ds_area_msr; struct ptrace_bts_config cfg; + size_t end; + const void *base, *max; + int error; if (cfg_size < sizeof(cfg)) return -EIO; - memset(&cfg, 0, sizeof(cfg)); + error = ds_get_bts_end(child, &end); + if (error < 0) + return error; - if (ds) { - cfg.size = ds_get_bts_size(ds); + error = ds_access_bts(child, /* index = */ 0, &base); + if (error < 0) + return error; - if (ds_get_overflow(ds) == DS_O_SIGNAL) - cfg.flags |= PTRACE_BTS_O_SIGNAL; + error = ds_access_bts(child, /* index = */ end, &max); + if (error < 0) + return error; - if (test_tsk_thread_flag(child, TIF_DEBUGCTLMSR) && - child->thread.debugctlmsr & ds_debugctl_mask()) - cfg.flags |= PTRACE_BTS_O_TRACE; + memset(&cfg, 0, sizeof(cfg)); + cfg.size = (max - base); + cfg.signal = child->thread.bts_ovfl_signal; + cfg.bts_size = sizeof(struct bts_struct); - if (test_tsk_thread_flag(child, TIF_BTS_TRACE_TS)) - cfg.flags |= PTRACE_BTS_O_SCHED; - } + if (cfg.signal) + cfg.flags |= PTRACE_BTS_O_SIGNAL; - cfg.bts_size = sizeof(struct bts_struct); + if (test_tsk_thread_flag(child, TIF_DEBUGCTLMSR) && + child->thread.debugctlmsr & bts_cfg.debugctl_mask) + cfg.flags |= PTRACE_BTS_O_TRACE; + + if (test_tsk_thread_flag(child, TIF_BTS_TRACE_TS)) + cfg.flags |= PTRACE_BTS_O_SCHED; if (copy_to_user(ucfg, &cfg, sizeof(cfg))) return -EFAULT; @@ -744,89 +830,38 @@ static int ptrace_bts_status(struct task_struct *child, return sizeof(cfg); } - static int ptrace_bts_write_record(struct task_struct *child, const struct bts_struct *in) { - int retval; + unsigned char bts_record[BTS_MAX_RECORD_SIZE]; - if (!child->thread.ds_area_msr) - return -ENXIO; + BUG_ON(BTS_MAX_RECORD_SIZE < bts_cfg.sizeof_bts); - retval = ds_write_bts((void *)child->thread.ds_area_msr, in); - if (retval) - return retval; + memset(bts_record, 0, bts_cfg.sizeof_bts); + switch (in->qualifier) { + case BTS_INVALID: + break; - return sizeof(*in); -} + case BTS_BRANCH: + bts_set(bts_record, bts_from, in->variant.lbr.from_ip); + bts_set(bts_record, bts_to, in->variant.lbr.to_ip); + break; -static int ptrace_bts_realloc(struct task_struct *child, - int size, int reduce_size) -{ - unsigned long rlim, vm; - int ret, old_size; + case BTS_TASK_ARRIVES: + case BTS_TASK_DEPARTS: + bts_set(bts_record, bts_from, bts_escape); + bts_set(bts_record, bts_qual, in->qualifier); + bts_set(bts_record, bts_jiffies, in->variant.jiffies); + break; - if (size < 0) + default: return -EINVAL; - - old_size = ds_get_bts_size((void *)child->thread.ds_area_msr); - if (old_size < 0) - return old_size; - - ret = ds_free((void **)&child->thread.ds_area_msr); - if (ret < 0) - goto out; - - size >>= PAGE_SHIFT; - old_size >>= PAGE_SHIFT; - - current->mm->total_vm -= old_size; - current->mm->locked_vm -= old_size; - - if (size == 0) - goto out; - - rlim = current->signal->rlim[RLIMIT_AS].rlim_cur >> PAGE_SHIFT; - vm = current->mm->total_vm + size; - if (rlim < vm) { - ret = -ENOMEM; - - if (!reduce_size) - goto out; - - size = rlim - current->mm->total_vm; - if (size <= 0) - goto out; } - rlim = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur >> PAGE_SHIFT; - vm = current->mm->locked_vm + size; - if (rlim < vm) { - ret = -ENOMEM; - - if (!reduce_size) - goto out; - - size = rlim - current->mm->locked_vm; - if (size <= 0) - goto out; - } - - ret = ds_allocate((void **)&child->thread.ds_area_msr, - size << PAGE_SHIFT); - if (ret < 0) - goto out; - - current->mm->total_vm += size; - current->mm->locked_vm += size; - -out: - if (child->thread.ds_area_msr) - set_tsk_thread_flag(child, TIF_DS_AREA_MSR); - else - clear_tsk_thread_flag(child, TIF_DS_AREA_MSR); - - return ret; + /* The writing task will be the switched-to task on a context + * switch. It needs to write into the switched-from task's BTS + * buffer. */ + return ds_unchecked_write_bts(child, bts_record, bts_cfg.sizeof_bts); } void ptrace_bts_take_timestamp(struct task_struct *tsk, @@ -839,7 +874,66 @@ void ptrace_bts_take_timestamp(struct task_struct *tsk, ptrace_bts_write_record(tsk, &rec); } -#endif /* X86_BTS */ + +static const struct bts_configuration bts_cfg_netburst = { + .sizeof_bts = sizeof(long) * 3, + .sizeof_field = sizeof(long), + .debugctl_mask = (1<<2)|(1<<3)|(1<<5) +}; + +static const struct bts_configuration bts_cfg_pentium_m = { + .sizeof_bts = sizeof(long) * 3, + .sizeof_field = sizeof(long), + .debugctl_mask = (1<<6)|(1<<7) +}; + +static const struct bts_configuration bts_cfg_core2 = { + .sizeof_bts = 8 * 3, + .sizeof_field = 8, + .debugctl_mask = (1<<6)|(1<<7)|(1<<9) +}; + +static inline void bts_configure(const struct bts_configuration *cfg) +{ + bts_cfg = *cfg; +} + +void __cpuinit ptrace_bts_init_intel(struct cpuinfo_x86 *c) +{ + switch (c->x86) { + case 0x6: + switch (c->x86_model) { + case 0xD: + case 0xE: /* Pentium M */ + bts_configure(&bts_cfg_pentium_m); + break; + case 0xF: /* Core2 */ + case 0x1C: /* Atom */ + bts_configure(&bts_cfg_core2); + break; + default: + /* sorry, don't know about them */ + break; + } + break; + case 0xF: + switch (c->x86_model) { + case 0x0: + case 0x1: + case 0x2: /* Netburst */ + bts_configure(&bts_cfg_netburst); + break; + default: + /* sorry, don't know about them */ + break; + } + break; + default: + /* sorry, don't know about them */ + break; + } +} +#endif /* CONFIG_X86_PTRACE_BTS */ /* * Called by kernel/ptrace.c when detaching.. @@ -852,15 +946,15 @@ void ptrace_disable(struct task_struct *child) #ifdef TIF_SYSCALL_EMU clear_tsk_thread_flag(child, TIF_SYSCALL_EMU); #endif - if (child->thread.ds_area_msr) { -#ifdef X86_BTS - ptrace_bts_realloc(child, 0, 0); -#endif - child->thread.debugctlmsr &= ~ds_debugctl_mask(); - if (!child->thread.debugctlmsr) - clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); - clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); - } +#ifdef CONFIG_X86_PTRACE_BTS + (void)ds_release_bts(child); + + child->thread.debugctlmsr &= ~bts_cfg.debugctl_mask; + if (!child->thread.debugctlmsr) + clear_tsk_thread_flag(child, TIF_DEBUGCTLMSR); + + clear_tsk_thread_flag(child, TIF_BTS_TRACE_TS); +#endif /* CONFIG_X86_PTRACE_BTS */ } #if defined CONFIG_X86_32 || defined CONFIG_IA32_EMULATION @@ -980,7 +1074,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) /* * These bits need more cooking - not enabled yet: */ -#ifdef X86_BTS +#ifdef CONFIG_X86_PTRACE_BTS case PTRACE_BTS_CONFIG: ret = ptrace_bts_config (child, data, (struct ptrace_bts_config __user *)addr); @@ -992,7 +1086,7 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) break; case PTRACE_BTS_SIZE: - ret = ptrace_bts_get_size(child); + ret = ds_get_bts_index(child, /* pos = */ NULL); break; case PTRACE_BTS_GET: @@ -1001,14 +1095,14 @@ long arch_ptrace(struct task_struct *child, long request, long addr, long data) break; case PTRACE_BTS_CLEAR: - ret = ptrace_bts_clear(child); + ret = ds_clear_bts(child); break; case PTRACE_BTS_DRAIN: ret = ptrace_bts_drain (child, data, (struct bts_struct __user *) addr); break; -#endif +#endif /* CONFIG_X86_PTRACE_BTS */ default: ret = ptrace_request(child, request, addr, data); diff --git a/arch/x86/kernel/relocate_kernel_32.S b/arch/x86/kernel/relocate_kernel_32.S index 703310a9902..6f50664b2ba 100644 --- a/arch/x86/kernel/relocate_kernel_32.S +++ b/arch/x86/kernel/relocate_kernel_32.S @@ -20,10 +20,11 @@ #define PAGE_ATTR (_PAGE_PRESENT | _PAGE_RW | _PAGE_ACCESSED | _PAGE_DIRTY) #define PAE_PGD_ATTR (_PAGE_PRESENT) -/* control_page + PAGE_SIZE/2 ~ control_page + PAGE_SIZE * 3/4 are - * used to save some data for jumping back +/* control_page + KEXEC_CONTROL_CODE_MAX_SIZE + * ~ control_page + PAGE_SIZE are used as data storage and stack for + * jumping back */ -#define DATA(offset) (PAGE_SIZE/2+(offset)) +#define DATA(offset) (KEXEC_CONTROL_CODE_MAX_SIZE+(offset)) /* Minimal CPU state */ #define ESP DATA(0x0) @@ -376,3 +377,6 @@ swap_pages: popl %ebx popl %ebp ret + + .globl kexec_control_code_size +.set kexec_control_code_size, . - relocate_kernel diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 68b48e3fbcb..9838f2539df 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -445,7 +445,7 @@ static void __init reserve_early_setup_data(void) * @size: Size of the crashkernel memory to reserve. * Returns the base address on success, and -1ULL on failure. */ -unsigned long long find_and_reserve_crashkernel(unsigned long long size) +unsigned long long __init find_and_reserve_crashkernel(unsigned long long size) { const unsigned long long alignment = 16<<20; /* 16M */ unsigned long long start = 0LL; @@ -604,14 +604,6 @@ void __init setup_arch(char **cmdline_p) early_cpu_init(); early_ioremap_init(); -#if defined(CONFIG_VMI) && defined(CONFIG_X86_32) - /* - * Must be before kernel pagetables are setup - * or fixmap area is touched. - */ - vmi_init(); -#endif - ROOT_DEV = old_decode_dev(boot_params.hdr.root_dev); screen_info = boot_params.screen_info; edid_info = boot_params.edid_info; @@ -678,6 +670,18 @@ void __init setup_arch(char **cmdline_p) parse_early_param(); +#ifdef CONFIG_X86_64 + check_efer(); +#endif + +#if defined(CONFIG_VMI) && defined(CONFIG_X86_32) + /* + * Must be before kernel pagetables are setup + * or fixmap area is touched. + */ + vmi_init(); +#endif + /* after early param, so could get panic from serial */ reserve_early_setup_data(); @@ -738,7 +742,6 @@ void __init setup_arch(char **cmdline_p) #else num_physpages = max_pfn; - check_efer(); /* How many end-of-memory variables you have, grandma! */ /* need this before calling reserve_initrd */ diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index 02b02583f5b..2f146405005 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -105,7 +105,16 @@ static inline int restore_i387(struct _fpstate __user *buf) clts(); task_thread_info(current)->status |= TS_USEDFPU; } - return restore_fpu_checking((__force struct i387_fxsave_struct *)buf); + err = restore_fpu_checking((__force struct i387_fxsave_struct *)buf); + if (unlikely(err)) { + /* + * Encountered an error while doing the restore from the + * user buffer, clear the fpu state. + */ + clear_fpu(tsk); + clear_used_math(); + } + return err; } /* diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 6112c363234..a66c93550a0 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -756,6 +756,14 @@ static void __cpuinit do_fork_idle(struct work_struct *work) } #ifdef CONFIG_X86_64 + +/* __ref because it's safe to call free_bootmem when after_bootmem == 0. */ +static void __ref free_bootmem_pda(struct x8664_pda *oldpda) +{ + if (!after_bootmem) + free_bootmem((unsigned long)oldpda, sizeof(*oldpda)); +} + /* * Allocate node local memory for the AP pda. * @@ -784,8 +792,7 @@ int __cpuinit get_local_pda(int cpu) if (oldpda) { memcpy(newpda, oldpda, size); - if (!after_bootmem) - free_bootmem((unsigned long)oldpda, size); + free_bootmem_pda(oldpda); } newpda->in_bootmem = 0; @@ -994,17 +1001,7 @@ int __cpuinit native_cpu_up(unsigned int cpu) flush_tlb_all(); low_mappings = 1; -#ifdef CONFIG_X86_PC - if (def_to_bigsmp && apicid > 8) { - printk(KERN_WARNING - "More than 8 CPUs detected - skipping them.\n" - "Use CONFIG_X86_GENERICARCH and CONFIG_X86_BIGSMP.\n"); - err = -1; - } else - err = do_boot_cpu(apicid, cpu); -#else err = do_boot_cpu(apicid, cpu); -#endif zap_low_mappings(); low_mappings = 0; @@ -1058,6 +1055,34 @@ static __init void disable_smp(void) static int __init smp_sanity_check(unsigned max_cpus) { preempt_disable(); + +#if defined(CONFIG_X86_PC) && defined(CONFIG_X86_32) + if (def_to_bigsmp && nr_cpu_ids > 8) { + unsigned int cpu; + unsigned nr; + + printk(KERN_WARNING + "More than 8 CPUs detected - skipping them.\n" + "Use CONFIG_X86_GENERICARCH and CONFIG_X86_BIGSMP.\n"); + + nr = 0; + for_each_present_cpu(cpu) { + if (nr >= 8) + cpu_clear(cpu, cpu_present_map); + nr++; + } + + nr = 0; + for_each_possible_cpu(cpu) { + if (nr >= 8) + cpu_clear(cpu, cpu_possible_map); + nr++; + } + + nr_cpu_ids = 8; + } +#endif + if (!physid_isset(hard_smp_processor_id(), phys_cpu_present_map)) { printk(KERN_WARNING "weird, boot CPU (#%d) not listed" "by the BIOS.\n", hard_smp_processor_id()); @@ -1196,6 +1221,9 @@ void __init native_smp_prepare_cpus(unsigned int max_cpus) printk(KERN_INFO "CPU%d: ", 0); print_cpu_info(&cpu_data(0)); setup_boot_clock(); + + if (is_uv_system()) + uv_system_init(); out: preempt_enable(); } @@ -1386,17 +1414,3 @@ void __cpu_die(unsigned int cpu) BUG(); } #endif - -/* - * If the BIOS enumerates physical processors before logical, - * maxcpus=N at enumeration-time can be used to disable HT. - */ -static int __init parse_maxcpus(char *arg) -{ - extern unsigned int maxcpus; - - if (arg) - maxcpus = simple_strtoul(arg, NULL, 0); - return 0; -} -early_param("maxcpus", parse_maxcpus); diff --git a/arch/x86/kernel/smpcommon.c b/arch/x86/kernel/smpcommon.c index 99941b37eca..397e309839d 100644 --- a/arch/x86/kernel/smpcommon.c +++ b/arch/x86/kernel/smpcommon.c @@ -8,18 +8,21 @@ DEFINE_PER_CPU(unsigned long, this_cpu_off); EXPORT_PER_CPU_SYMBOL(this_cpu_off); -/* Initialize the CPU's GDT. This is either the boot CPU doing itself - (still using the master per-cpu area), or a CPU doing it for a - secondary which will soon come up. */ +/* + * Initialize the CPU's GDT. This is either the boot CPU doing itself + * (still using the master per-cpu area), or a CPU doing it for a + * secondary which will soon come up. + */ __cpuinit void init_gdt(int cpu) { - struct desc_struct *gdt = get_cpu_gdt_table(cpu); + struct desc_struct gdt; - pack_descriptor(&gdt[GDT_ENTRY_PERCPU], - __per_cpu_offset[cpu], 0xFFFFF, + pack_descriptor(&gdt, __per_cpu_offset[cpu], 0xFFFFF, 0x2 | DESCTYPE_S, 0x8); + gdt.s = 1; - gdt[GDT_ENTRY_PERCPU].s = 1; + write_gdt_entry(get_cpu_gdt_table(cpu), + GDT_ENTRY_PERCPU, &gdt, DESCTYPE_S); per_cpu(this_cpu_off, cpu) = __per_cpu_offset[cpu]; per_cpu(cpu_number, cpu) = cpu; diff --git a/arch/x86/kernel/syscall_64.c b/arch/x86/kernel/syscall_64.c index 170d43c1748..3d1be4f0fac 100644 --- a/arch/x86/kernel/syscall_64.c +++ b/arch/x86/kernel/syscall_64.c @@ -8,12 +8,12 @@ #define __NO_STUBS #define __SYSCALL(nr, sym) extern asmlinkage void sym(void) ; -#undef _ASM_X86_64_UNISTD_H_ +#undef ASM_X86__UNISTD_64_H #include <asm/unistd_64.h> #undef __SYSCALL #define __SYSCALL(nr, sym) [nr] = sym, -#undef _ASM_X86_64_UNISTD_H_ +#undef ASM_X86__UNISTD_64_H typedef void (*sys_call_ptr_t)(void); diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index d0fbb7712ab..8b8c0d6640f 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -17,6 +17,7 @@ #include <asm/genapic.h> #include <asm/idle.h> #include <asm/tsc.h> +#include <asm/irq_vectors.h> #include <mach_apic.h> @@ -783,7 +784,7 @@ static int __init uv_bau_init(void) uv_init_blade(blade, node, cur_cpu); cur_cpu += uv_blade_nr_possible_cpus(blade); } - set_intr_gate(UV_BAU_MESSAGE, uv_bau_message_intr1); + alloc_intr_gate(UV_BAU_MESSAGE, uv_bau_message_intr1); uv_enable_timeouts(); return 0; diff --git a/arch/x86/kernel/traps_64.c b/arch/x86/kernel/traps_64.c index 3f18d73f420..513caaca711 100644 --- a/arch/x86/kernel/traps_64.c +++ b/arch/x86/kernel/traps_64.c @@ -1131,7 +1131,14 @@ asmlinkage void math_state_restore(void) } clts(); /* Allow maths ops (or we recurse) */ - restore_fpu_checking(&me->thread.xstate->fxsave); + /* + * Paranoid restore. send a SIGSEGV if we fail to restore the state. + */ + if (unlikely(restore_fpu_checking(&me->thread.xstate->fxsave))) { + stts(); + force_sig(SIGSEGV, me); + return; + } task_thread_info(me)->status |= TS_USEDFPU; me->fpu_counter++; } diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 7603c055390..8f98e9de1b8 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -104,7 +104,7 @@ __setup("notsc", notsc_setup); /* * Read TSC and the reference counters. Take care of SMI disturbance */ -static u64 __init tsc_read_refs(u64 *pm, u64 *hpet) +static u64 tsc_read_refs(u64 *pm, u64 *hpet) { u64 t1, t2; int i; @@ -122,80 +122,216 @@ static u64 __init tsc_read_refs(u64 *pm, u64 *hpet) return ULLONG_MAX; } -/** - * native_calibrate_tsc - calibrate the tsc on boot +/* + * Try to calibrate the TSC against the Programmable + * Interrupt Timer and return the frequency of the TSC + * in kHz. + * + * Return ULONG_MAX on failure to calibrate. */ -unsigned long native_calibrate_tsc(void) +static unsigned long pit_calibrate_tsc(void) { - unsigned long flags; - u64 tsc1, tsc2, tr1, tr2, delta, pm1, pm2, hpet1, hpet2; - int hpet = is_hpet_enabled(); - unsigned int tsc_khz_val = 0; - - local_irq_save(flags); - - tsc1 = tsc_read_refs(&pm1, hpet ? &hpet1 : NULL); + u64 tsc, t1, t2, delta; + unsigned long tscmin, tscmax; + int pitcnt; + /* Set the Gate high, disable speaker */ outb((inb(0x61) & ~0x02) | 0x01, 0x61); + /* + * Setup CTC channel 2* for mode 0, (interrupt on terminal + * count mode), binary count. Set the latch register to 50ms + * (LSB then MSB) to begin countdown. + */ outb(0xb0, 0x43); outb((CLOCK_TICK_RATE / (1000 / 50)) & 0xff, 0x42); outb((CLOCK_TICK_RATE / (1000 / 50)) >> 8, 0x42); - tr1 = get_cycles(); - while ((inb(0x61) & 0x20) == 0); - tr2 = get_cycles(); - tsc2 = tsc_read_refs(&pm2, hpet ? &hpet2 : NULL); + tsc = t1 = t2 = get_cycles(); - local_irq_restore(flags); + pitcnt = 0; + tscmax = 0; + tscmin = ULONG_MAX; + while ((inb(0x61) & 0x20) == 0) { + t2 = get_cycles(); + delta = t2 - tsc; + tsc = t2; + if ((unsigned long) delta < tscmin) + tscmin = (unsigned int) delta; + if ((unsigned long) delta > tscmax) + tscmax = (unsigned int) delta; + pitcnt++; + } /* - * Preset the result with the raw and inaccurate PIT - * calibration value + * Sanity checks: + * + * If we were not able to read the PIT more than 5000 + * times, then we have been hit by a massive SMI + * + * If the maximum is 10 times larger than the minimum, + * then we got hit by an SMI as well. */ - delta = (tr2 - tr1); + if (pitcnt < 5000 || tscmax > 10 * tscmin) + return ULONG_MAX; + + /* Calculate the PIT value */ + delta = t2 - t1; do_div(delta, 50); - tsc_khz_val = delta; + return delta; +} + + +/** + * native_calibrate_tsc - calibrate the tsc on boot + */ +unsigned long native_calibrate_tsc(void) +{ + u64 tsc1, tsc2, delta, pm1, pm2, hpet1, hpet2; + unsigned long tsc_pit_min = ULONG_MAX, tsc_ref_min = ULONG_MAX; + unsigned long flags; + int hpet = is_hpet_enabled(), i; + + /* + * Run 5 calibration loops to get the lowest frequency value + * (the best estimate). We use two different calibration modes + * here: + * + * 1) PIT loop. We set the PIT Channel 2 to oneshot mode and + * load a timeout of 50ms. We read the time right after we + * started the timer and wait until the PIT count down reaches + * zero. In each wait loop iteration we read the TSC and check + * the delta to the previous read. We keep track of the min + * and max values of that delta. The delta is mostly defined + * by the IO time of the PIT access, so we can detect when a + * SMI/SMM disturbance happend between the two reads. If the + * maximum time is significantly larger than the minimum time, + * then we discard the result and have another try. + * + * 2) Reference counter. If available we use the HPET or the + * PMTIMER as a reference to check the sanity of that value. + * We use separate TSC readouts and check inside of the + * reference read for a SMI/SMM disturbance. We dicard + * disturbed values here as well. We do that around the PIT + * calibration delay loop as we have to wait for a certain + * amount of time anyway. + */ + for (i = 0; i < 5; i++) { + unsigned long tsc_pit_khz; + + /* + * Read the start value and the reference count of + * hpet/pmtimer when available. Then do the PIT + * calibration, which will take at least 50ms, and + * read the end value. + */ + local_irq_save(flags); + tsc1 = tsc_read_refs(&pm1, hpet ? &hpet1 : NULL); + tsc_pit_khz = pit_calibrate_tsc(); + tsc2 = tsc_read_refs(&pm2, hpet ? &hpet2 : NULL); + local_irq_restore(flags); + + /* Pick the lowest PIT TSC calibration so far */ + tsc_pit_min = min(tsc_pit_min, tsc_pit_khz); + + /* hpet or pmtimer available ? */ + if (!hpet && !pm1 && !pm2) + continue; + + /* Check, whether the sampling was disturbed by an SMI */ + if (tsc1 == ULLONG_MAX || tsc2 == ULLONG_MAX) + continue; + + tsc2 = (tsc2 - tsc1) * 1000000LL; + + if (hpet) { + if (hpet2 < hpet1) + hpet2 += 0x100000000ULL; + hpet2 -= hpet1; + tsc1 = ((u64)hpet2 * hpet_readl(HPET_PERIOD)); + do_div(tsc1, 1000000); + } else { + if (pm2 < pm1) + pm2 += (u64)ACPI_PM_OVRRUN; + pm2 -= pm1; + tsc1 = pm2 * 1000000000LL; + do_div(tsc1, PMTMR_TICKS_PER_SEC); + } + + do_div(tsc2, tsc1); + tsc_ref_min = min(tsc_ref_min, (unsigned long) tsc2); + } + + /* + * Now check the results. + */ + if (tsc_pit_min == ULONG_MAX) { + /* PIT gave no useful value */ + printk(KERN_WARNING "TSC: Unable to calibrate against PIT\n"); + + /* We don't have an alternative source, disable TSC */ + if (!hpet && !pm1 && !pm2) { + printk("TSC: No reference (HPET/PMTIMER) available\n"); + return 0; + } + + /* The alternative source failed as well, disable TSC */ + if (tsc_ref_min == ULONG_MAX) { + printk(KERN_WARNING "TSC: HPET/PMTIMER calibration " + "failed due to SMI disturbance.\n"); + return 0; + } + + /* Use the alternative source */ + printk(KERN_INFO "TSC: using %s reference calibration\n", + hpet ? "HPET" : "PMTIMER"); + + return tsc_ref_min; + } - /* hpet or pmtimer available ? */ + /* We don't have an alternative source, use the PIT calibration value */ if (!hpet && !pm1 && !pm2) { - printk(KERN_INFO "TSC calibrated against PIT\n"); - goto out; + printk(KERN_INFO "TSC: Using PIT calibration value\n"); + return tsc_pit_min; } - /* Check, whether the sampling was disturbed by an SMI */ - if (tsc1 == ULLONG_MAX || tsc2 == ULLONG_MAX) { - printk(KERN_WARNING "TSC calibration disturbed by SMI, " - "using PIT calibration result\n"); - goto out; + /* The alternative source failed, use the PIT calibration value */ + if (tsc_ref_min == ULONG_MAX) { + printk(KERN_WARNING "TSC: HPET/PMTIMER calibration failed due " + "to SMI disturbance. Using PIT calibration\n"); + return tsc_pit_min; } - tsc2 = (tsc2 - tsc1) * 1000000LL; - - if (hpet) { - printk(KERN_INFO "TSC calibrated against HPET\n"); - if (hpet2 < hpet1) - hpet2 += 0x100000000ULL; - hpet2 -= hpet1; - tsc1 = ((u64)hpet2 * hpet_readl(HPET_PERIOD)); - do_div(tsc1, 1000000); - } else { - printk(KERN_INFO "TSC calibrated against PM_TIMER\n"); - if (pm2 < pm1) - pm2 += (u64)ACPI_PM_OVRRUN; - pm2 -= pm1; - tsc1 = pm2 * 1000000000LL; - do_div(tsc1, PMTMR_TICKS_PER_SEC); + /* Check the reference deviation */ + delta = ((u64) tsc_pit_min) * 100; + do_div(delta, tsc_ref_min); + + /* + * If both calibration results are inside a 5% window, the we + * use the lower frequency of those as it is probably the + * closest estimate. + */ + if (delta >= 95 && delta <= 105) { + printk(KERN_INFO "TSC: PIT calibration confirmed by %s.\n", + hpet ? "HPET" : "PMTIMER"); + printk(KERN_INFO "TSC: using %s calibration value\n", + tsc_pit_min <= tsc_ref_min ? "PIT" : + hpet ? "HPET" : "PMTIMER"); + return tsc_pit_min <= tsc_ref_min ? tsc_pit_min : tsc_ref_min; } - do_div(tsc2, tsc1); - tsc_khz_val = tsc2; + printk(KERN_WARNING "TSC: PIT calibration deviates from %s: %lu %lu.\n", + hpet ? "HPET" : "PMTIMER", tsc_pit_min, tsc_ref_min); -out: - return tsc_khz_val; + /* + * The calibration values differ too much. In doubt, we use + * the PIT value as we know that there are PMTIMERs around + * running at double speed. + */ + printk(KERN_INFO "TSC: Using PIT calibration value\n"); + return tsc_pit_min; } - #ifdef CONFIG_X86_32 /* Only called from the Powernow K7 cpu freq driver */ int recalibrate_cpu_khz(void) @@ -314,7 +450,7 @@ static int time_cpufreq_notifier(struct notifier_block *nb, unsigned long val, mark_tsc_unstable("cpufreq changes"); } - set_cyc2ns_scale(tsc_khz_ref, freq->cpu); + set_cyc2ns_scale(tsc_khz, freq->cpu); return 0; } @@ -325,6 +461,10 @@ static struct notifier_block time_cpufreq_notifier_block = { static int __init cpufreq_tsc(void) { + if (!cpu_has_tsc) + return 0; + if (boot_cpu_has(X86_FEATURE_CONSTANT_TSC)) + return 0; cpufreq_register_notifier(&time_cpufreq_notifier_block, CPUFREQ_TRANSITION_NOTIFIER); return 0; diff --git a/arch/x86/kernel/tsc_sync.c b/arch/x86/kernel/tsc_sync.c index 0577825cf89..9ffb01c31c4 100644 --- a/arch/x86/kernel/tsc_sync.c +++ b/arch/x86/kernel/tsc_sync.c @@ -88,11 +88,9 @@ static __cpuinit void check_tsc_warp(void) __raw_spin_unlock(&sync_lock); } } - if (!(now-start)) { - printk("Warning: zero tsc calibration delta: %Ld [max: %Ld]\n", + WARN(!(now-start), + "Warning: zero tsc calibration delta: %Ld [max: %Ld]\n", now-start, end-start); - WARN_ON(1); - } } /* diff --git a/arch/x86/kernel/visws_quirks.c b/arch/x86/kernel/visws_quirks.c index 41e01b145c4..594ef47f0a6 100644 --- a/arch/x86/kernel/visws_quirks.c +++ b/arch/x86/kernel/visws_quirks.c @@ -184,8 +184,6 @@ static int __init visws_get_smp_config(unsigned int early) return 1; } -extern unsigned int __cpuinitdata maxcpus; - /* * The Visual Workstation is Intel MP compliant in the hardware * sense, but it doesn't have a BIOS(-configuration table). @@ -244,8 +242,8 @@ static int __init visws_find_smp_config(unsigned int reserve) ncpus = CO_CPU_MAX; } - if (ncpus > maxcpus) - ncpus = maxcpus; + if (ncpus > setup_max_cpus) + ncpus = setup_max_cpus; #ifdef CONFIG_X86_LOCAL_APIC smp_found_config = 1; diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c index 6ca515d6db5..edfb09f3047 100644 --- a/arch/x86/kernel/vmi_32.c +++ b/arch/x86/kernel/vmi_32.c @@ -235,7 +235,7 @@ static void vmi_write_ldt_entry(struct desc_struct *dt, int entry, const void *desc) { u32 *ldt_entry = (u32 *)desc; - vmi_ops.write_idt_entry(dt, entry, ldt_entry[0], ldt_entry[1]); + vmi_ops.write_ldt_entry(dt, entry, ldt_entry[0], ldt_entry[1]); } static void vmi_load_sp0(struct tss_struct *tss, diff --git a/arch/x86/kernel/vmlinux_32.lds.S b/arch/x86/kernel/vmlinux_32.lds.S index cdb2363697d..af5bdad8460 100644 --- a/arch/x86/kernel/vmlinux_32.lds.S +++ b/arch/x86/kernel/vmlinux_32.lds.S @@ -209,3 +209,11 @@ SECTIONS DWARF_DEBUG } + +#ifdef CONFIG_KEXEC +/* Link time checks */ +#include <asm/kexec.h> + +ASSERT(kexec_control_code_size <= KEXEC_CONTROL_CODE_MAX_SIZE, + "kexec control code size is too big") +#endif diff --git a/arch/x86/kernel/vsmp_64.c b/arch/x86/kernel/vsmp_64.c index 0c029e8959c..7766d36983f 100644 --- a/arch/x86/kernel/vsmp_64.c +++ b/arch/x86/kernel/vsmp_64.c @@ -61,7 +61,7 @@ static void vsmp_irq_enable(void) native_restore_fl((flags | X86_EFLAGS_IF) & (~X86_EFLAGS_AC)); } -static unsigned __init vsmp_patch(u8 type, u16 clobbers, void *ibuf, +static unsigned __init_or_module vsmp_patch(u8 type, u16 clobbers, void *ibuf, unsigned long addr, unsigned len) { switch (type) { |