From 91467bdf6e53058af13fd255375d6634ba0c70e0 Mon Sep 17 00:00:00 2001 From: Bernhard Walle Date: Fri, 18 Jul 2008 19:07:53 +0200 Subject: x86: move dma32_reserve_bootmem() after reserve_crashkernel() On a x86-64 machine (nothing special I could encounter) I had the problem that crashkernel reservation with the usual "64M@16M" failed. While debugging that, I encountered that dma32_reserve_bootmem() reserves a memory region which is in that area. Because dma32_reserve_bootmem() does not rely on a specific offset but crashkernel does, it makes sense to move the dma32_reserve_bootmem() reservation down a bit. I tested that patch and it works without problems. I don't see any negative effects of that move, but maybe I oversaw something ... While we strictly don't need that patch in 2.6.27 because we have the automatic, dynamic offset detection, it makes sense to also include it here because: - it's easier to get it in -stable then, - many people are still used to the 'crashkernel=...@16M' syntax, - not everybody may be using a reloatable kernel. Signed-off-by: Bernhard Walle Cc: kexec@lists.infradead.org Cc: vgoyal@redhat.com Cc: akpm@linux-foundation.org Cc: Bernhard Walle Cc: yhlu.kernel@gmail.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 531b55b8e81a..74d110ef2690 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -792,10 +792,6 @@ void __init setup_arch(char **cmdline_p) initmem_init(0, max_pfn); -#ifdef CONFIG_X86_64 - dma32_reserve_bootmem(); -#endif - #ifdef CONFIG_ACPI_SLEEP /* * Reserve low memory region for sleep support. @@ -810,6 +806,15 @@ void __init setup_arch(char **cmdline_p) #endif reserve_crashkernel(); +#ifdef CONFIG_X86_64 + /* + * dma32_reserve_bootmem() allocates bootmem which may conflict + * with the crashkernel command line, so do that after + * reserve_crashkernel() + */ + dma32_reserve_bootmem(); +#endif + reserve_ibft_region(); #ifdef CONFIG_KVM_CLOCK -- cgit v1.2.1 From 6ffac1e90a17ea0aded5c581204397421eec91b6 Mon Sep 17 00:00:00 2001 From: Suresh Siddha Date: Thu, 24 Jul 2008 18:07:56 -0700 Subject: x64, fpu: fix possible FPU leakage in error conditions On Thu, Jul 24, 2008 at 03:43:44PM -0700, Linus Torvalds wrote: > So how about this patch as a starting point? This is the RightThing(tm) to > do regardless, and if it then makes it easier to do some other cleanups, > we should do it first. What do you think? restore_fpu_checking() calls init_fpu() in error conditions. While this is wrong(as our main intention is to clear the fpu state of the thread), this was benign before commit 92d140e21f1 ("x86: fix taking DNA during 64bit sigreturn"). Post commit 92d140e21f1, live FPU registers may not belong to this process at this error scenario. In the error condition for restore_fpu_checking() (especially during the 64bit signal return), we are doing init_fpu(), which saves the live FPU register state (possibly belonging to some other process context) into the thread struct (through unlazy_fpu() in init_fpu()). This is wrong and can leak the FPU data. For the signal handler restore error condition in restore_i387(), clear the fpu state present in the thread struct(before ultimately sending a SIGSEGV for badframe). For the paranoid error condition check in math_state_restore(), send a SIGSEGV, if we fail to restore the state. Signed-off-by: Suresh Siddha Cc: Cc: Linus Torvalds Signed-off-by: Ingo Molnar --- arch/x86/kernel/signal_64.c | 11 ++++++++++- arch/x86/kernel/traps_64.c | 9 ++++++++- 2 files changed, 18 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/signal_64.c b/arch/x86/kernel/signal_64.c index b45ef8ddd651..ca316b5b742c 100644 --- a/arch/x86/kernel/signal_64.c +++ b/arch/x86/kernel/signal_64.c @@ -104,7 +104,16 @@ 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/traps_64.c b/arch/x86/kernel/traps_64.c index 3f18d73f420c..513caaca7115 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++; } -- cgit v1.2.1 From 8978b74253280d59e97cf49a3ec2c0cbccd5b801 Mon Sep 17 00:00:00 2001 From: FUJITA Tomonori Date: Tue, 29 Jul 2008 13:38:53 +0900 Subject: generic, x86: fix add iommu_num_pages helper function This IOMMU helper function doesn't work for some architectures: http://marc.info/?l=linux-kernel&m=121699304403202&w=2 It also breaks POWER and SPARC builds: http://marc.info/?l=linux-kernel&m=121730388001890&w=2 Currently, only x86 IOMMUs use this so let's move it to x86 for now. Reported-by: Stephen Rothwell Signed-off-by: FUJITA Tomonori Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-dma.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-dma.c b/arch/x86/kernel/pci-dma.c index 8dbffb846de9..87d4d6964ec2 100644 --- a/arch/x86/kernel/pci-dma.c +++ b/arch/x86/kernel/pci-dma.c @@ -123,6 +123,14 @@ void __init pci_iommu_alloc(void) pci_swiotlb_init(); } + +unsigned long iommu_num_pages(unsigned long addr, unsigned long len) +{ + unsigned long size = roundup((addr & ~PAGE_MASK) + len, PAGE_SIZE); + + return size >> PAGE_SHIFT; +} +EXPORT_SYMBOL(iommu_num_pages); #endif /* -- cgit v1.2.1 From 0d39741a27d86d305cc75ba626392be410dcbab9 Mon Sep 17 00:00:00 2001 From: Jack Steiner Date: Tue, 29 Jul 2008 22:34:01 -0700 Subject: GRU Driver: export is_uv_system(), zap_page_range() & follow_page() Exports needed by the GRU driver. Signed-off-by: Jack Steiner Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/genapic_64.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genapic_64.c b/arch/x86/kernel/genapic_64.c index 1fa8be5bd217..eaff0bbb1444 100644 --- a/arch/x86/kernel/genapic_64.c +++ b/arch/x86/kernel/genapic_64.c @@ -99,3 +99,4 @@ int is_uv_system(void) { return uv_system_type != UV_NONE; } +EXPORT_SYMBOL_GPL(is_uv_system); -- cgit v1.2.1 From e0d22d03c06c4e2c194d7010bc1e4a972199f156 Mon Sep 17 00:00:00 2001 From: Krzysztof Helt Date: Thu, 31 Jul 2008 23:43:44 +0200 Subject: x86: fdiv bug detection fix The fdiv detection code writes s32 integer into the boot_cpu_data.fdiv_bug. However, the boot_cpu_data.fdiv_bug is only char (s8) field so the detection overwrites already set fields for other bugs, e.g. the f00f bug field. Use local s32 variable to receive result. This is a partial fix to Bugzilla #9928 - fixes wrong information about the f00f bug (tested) and probably for coma bug (I have no cpu to test this). Signed-off-by: Krzysztof Helt Cc: Andrew Morton Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/bugs.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/bugs.c b/arch/x86/kernel/cpu/bugs.c index c9b58a806e85..c8e315f1aa83 100644 --- a/arch/x86/kernel/cpu/bugs.c +++ b/arch/x86/kernel/cpu/bugs.c @@ -50,6 +50,8 @@ static double __initdata y = 3145727.0; */ static void __init check_fpu(void) { + s32 fdiv_bug; + if (!boot_cpu_data.hard_math) { #ifndef CONFIG_MATH_EMULATION printk(KERN_EMERG "No coprocessor found and no math emulation present.\n"); @@ -74,8 +76,10 @@ static void __init check_fpu(void) "fistpl %0\n\t" "fwait\n\t" "fninit" - : "=m" (*&boot_cpu_data.fdiv_bug) + : "=m" (*&fdiv_bug) : "m" (*&x), "m" (*&y)); + + boot_cpu_data.fdiv_bug = fdiv_bug; if (boot_cpu_data.fdiv_bug) printk("Hmm, FPU with FDIV bug.\n"); } -- cgit v1.2.1 From ec983f7060cd73e14cdd3edd910339127a8a4e96 Mon Sep 17 00:00:00 2001 From: Dave Jones Date: Wed, 30 Jul 2008 12:05:03 -0400 Subject: [CPUFREQ] Remove EXPERIMENTAL annotation from VIA C7 powersaver kconfig. This has been pretty solid, and doesn't see much change at all. Noticed by Harald Welte. Signed-off-by: Dave Jones --- arch/x86/kernel/cpu/cpufreq/Kconfig | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/Kconfig b/arch/x86/kernel/cpu/cpufreq/Kconfig index cb7a5715596d..efae3b22a0ff 100644 --- a/arch/x86/kernel/cpu/cpufreq/Kconfig +++ b/arch/x86/kernel/cpu/cpufreq/Kconfig @@ -235,9 +235,9 @@ config X86_LONGHAUL If in doubt, say N. config X86_E_POWERSAVER - tristate "VIA C7 Enhanced PowerSaver (EXPERIMENTAL)" + tristate "VIA C7 Enhanced PowerSaver" select CPU_FREQ_TABLE - depends on X86_32 && EXPERIMENTAL + depends on X86_32 help This adds the CPUFreq driver for VIA C7 processors. -- cgit v1.2.1 From 460f5ef2835dbc33825f611f408eb09c29be4b85 Mon Sep 17 00:00:00 2001 From: Dave Jones Date: Wed, 30 Jul 2008 13:01:42 -0400 Subject: [CPUFREQ] Fix warning in elanfreq arch/x86/kernel/cpu/cpufreq/elanfreq.c:47:26: warning: symbol 'elan_multiplier' was not declared. Should it be static? Yes, yes it should. Signed-off-by: Dave Jones --- arch/x86/kernel/cpu/cpufreq/elanfreq.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/elanfreq.c b/arch/x86/kernel/cpu/cpufreq/elanfreq.c index 94619c22f563..e4a4bf870e94 100644 --- a/arch/x86/kernel/cpu/cpufreq/elanfreq.c +++ b/arch/x86/kernel/cpu/cpufreq/elanfreq.c @@ -44,7 +44,7 @@ struct s_elan_multiplier { * It is important that the frequencies * are listed in ascending order here! */ -struct s_elan_multiplier elan_multiplier[] = { +static struct s_elan_multiplier elan_multiplier[] = { {1000, 0x02, 0x18}, {2000, 0x02, 0x10}, {4000, 0x02, 0x08}, -- cgit v1.2.1 From 23431b495fcbbde6d09aba30019b6a224d543cf9 Mon Sep 17 00:00:00 2001 From: Mark Langsdorf Date: Thu, 31 Jul 2008 12:39:05 -0500 Subject: [CPUFREQ][1/2] whitespace fix for powernow-k8 Trivial whitespace fix for powernow-k8. Signed-off-by: Mark Langsdorf Signed-off-by: Dave Jones --- arch/x86/kernel/cpu/cpufreq/powernow-k8.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index c45ca6d4dce1..84bb395038d8 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -66,7 +66,6 @@ static u32 find_freq_from_fid(u32 fid) return 800 + (fid * 100); } - /* Return a frequency in KHz, given an input fid */ static u32 find_khz_freq_from_fid(u32 fid) { @@ -78,7 +77,6 @@ static u32 find_khz_freq_from_pstate(struct cpufreq_frequency_table *data, u32 p return data[pstate].frequency; } - /* Return the vco fid for an input fid * * Each "low" fid has corresponding "high" fid, and you can get to "low" fids @@ -166,7 +164,6 @@ static void fidvid_msr_init(void) wrmsr(MSR_FIDVID_CTL, lo, hi); } - /* write the new fid value along with the other control fields to the msr */ static int write_new_fid(struct powernow_k8_data *data, u32 fid) { -- cgit v1.2.1 From 34ae7f35a21694aa5cb8829dc5142c39d73d6ba0 Mon Sep 17 00:00:00 2001 From: Mark Langsdorf Date: Thu, 31 Jul 2008 12:39:12 -0500 Subject: [CPUFREQ][2/2] preregister support for powernow-k8 This patch provides support for the _PSD ACPI object in the Powernow-k8 driver. Although it looks like an invasive patch, most of it is simply the consequence of turning the static acpi_performance_data structure into a pointer. AMD has tested it on several machines over the past few days without issue. [trivial checkpatch warnings fixed up by davej] [X86_POWERNOW_K8_ACPI=n buildfix from Randy Dunlap] Signed-off-by: Mark Langsdorf Tested-by: Frank Arnold Signed-off-by: Randy Dunlap Signed-off-by: Dave Jones --- arch/x86/kernel/cpu/cpufreq/powernow-k8.c | 109 ++++++++++++++++++++---------- arch/x86/kernel/cpu/cpufreq/powernow-k8.h | 3 +- 2 files changed, 75 insertions(+), 37 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index 84bb395038d8..4e7271999a74 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c @@ -737,44 +737,63 @@ 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; + 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; } static int powernow_k8_cpu_init_acpi(struct powernow_k8_data *data) { struct cpufreq_frequency_table *powernow_table; int ret_val; + int cpu = 0; - if (acpi_processor_register_performance(&data->acpi_data, data->cpu)) { + data->acpi_data = percpu_ptr(acpi_perf_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; @@ -787,12 +806,12 @@ 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); @@ -800,16 +819,31 @@ 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; } @@ -821,10 +855,10 @@ 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"); @@ -840,7 +874,7 @@ 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; } @@ -849,16 +883,16 @@ 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); @@ -899,10 +933,10 @@ 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; } @@ -912,11 +946,12 @@ 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; } @@ -1101,7 +1136,7 @@ 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; + cpumask_t oldmask = CPU_MASK_ALL; int rc; if (!cpu_online(pol->cpu)) @@ -1174,10 +1209,7 @@ static int __cpuinit powernowk8_cpu_init(struct cpufreq_policy *pol) /* run on any CPU again */ set_cpus_allowed_ptr(current, &oldmask); - if (cpu_family == CPU_HW_PSTATE) - pol->cpus = cpumask_of_cpu(pol->cpu); - else - pol->cpus = per_cpu(cpu_core_map, pol->cpu); + pol->cpus = data->starting_core_affinity; data->available_cores = &(pol->cpus); /* Take a crude guess here. @@ -1300,6 +1332,7 @@ 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(), @@ -1316,6 +1349,10 @@ 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 and Mark Langsdorf "); diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h index ab48cfed4d96..a62612cd4be8 100644 --- a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h +++ b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h @@ -33,12 +33,13 @@ 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; }; -- cgit v1.2.1 From 31343d8a5079cda57ffd539fcf4f00cea344fe98 Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Fri, 8 Aug 2008 12:15:57 -0700 Subject: x86: Fix broken VMI in 2.6.27-rc.. The lowmem mapping table created by VMI need not depend on max_low_pfn at all. Instead we now create an extra large mapping which covers all possible lowmem instead of the physical ram that is actually available. This allows the vmi initialization to be done before max_low_pfn could be computed. We also move the vmi_init code very early in the boot process so that nobody accidentally breaks the fixmap dependancy. Signed-off-by: Alok N Kataria Acked-by: Zachary Amsden Signed-off-by: H. Peter Anvin --- arch/x86/kernel/setup.c | 16 ++++++++-------- arch/x86/kernel/vmi_32.c | 3 ++- 2 files changed, 10 insertions(+), 9 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 2d888586385d..6e5823b9e765 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -604,6 +604,14 @@ 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; @@ -817,14 +825,6 @@ void __init setup_arch(char **cmdline_p) kvmclock_init(); #endif -#if defined(CONFIG_VMI) && defined(CONFIG_X86_32) - /* - * Must be after max_low_pfn is determined, and before kernel - * pagetables are setup. - */ - vmi_init(); -#endif - paravirt_pagetable_setup_start(swapper_pg_dir); paging_init(); paravirt_pagetable_setup_done(swapper_pg_dir); diff --git a/arch/x86/kernel/vmi_32.c b/arch/x86/kernel/vmi_32.c index 0a1b1a9d922d..6ca515d6db54 100644 --- a/arch/x86/kernel/vmi_32.c +++ b/arch/x86/kernel/vmi_32.c @@ -37,6 +37,7 @@ #include #include #include +#include /* Convenient for calling VMI functions indirectly in the ROM */ typedef u32 __attribute__((regparm(1))) (VROMFUNC)(void); @@ -683,7 +684,7 @@ void vmi_bringup(void) { /* We must establish the lowmem mapping for MMU ops to work */ if (vmi_ops.set_linear_mapping) - vmi_ops.set_linear_mapping(0, (void *)__PAGE_OFFSET, max_low_pfn, 0); + vmi_ops.set_linear_mapping(0, (void *)__PAGE_OFFSET, MAXMEM_PFN, 0); } /* -- cgit v1.2.1 From d388e5fdc461344d04307a3fa83862b9ed429647 Mon Sep 17 00:00:00 2001 From: "Eric W. Biederman" Date: Sat, 9 Aug 2008 15:09:02 -0700 Subject: x86: Restore proper vector locking during cpu hotplug Having cpu_online_map change during assign_irq_vector can result in some really nasty and weird things happening. The one that bit me last time was accessing non existent per cpu memory for non existent cpus. This locking was removed in a sloppy x86_64 and x86_32 merge patch. Guys can we please try and avoid subtly breaking x86 when we are merging files together? Signed-off-by: Eric W. Biederman Signed-off-by: H. Peter Anvin --- arch/x86/kernel/io_apic_32.c | 6 +----- arch/x86/kernel/io_apic_64.c | 25 +++++++++++++++---------- arch/x86/kernel/smpboot.c | 12 +++++++++--- 3 files changed, 25 insertions(+), 18 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_apic_32.c b/arch/x86/kernel/io_apic_32.c index de9aa0e3a9c5..09cddb57bec4 100644 --- a/arch/x86/kernel/io_apic_32.c +++ b/arch/x86/kernel/io_apic_32.c @@ -57,7 +57,7 @@ atomic_t irq_mis_count; static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; static DEFINE_SPINLOCK(ioapic_lock); -static DEFINE_SPINLOCK(vector_lock); +DEFINE_SPINLOCK(vector_lock); int timer_through_8259 __initdata; @@ -1209,10 +1209,6 @@ static int assign_irq_vector(int irq) return vector; } -void setup_vector_irq(int cpu) -{ -} - static struct irq_chip ioapic_chip; #define IOAPIC_AUTO -1 diff --git a/arch/x86/kernel/io_apic_64.c b/arch/x86/kernel/io_apic_64.c index 8269434d1707..61a83b70c18f 100644 --- a/arch/x86/kernel/io_apic_64.c +++ b/arch/x86/kernel/io_apic_64.c @@ -101,7 +101,7 @@ int timer_through_8259 __initdata; static struct { int pin, apic; } ioapic_i8259 = { -1, -1 }; static DEFINE_SPINLOCK(ioapic_lock); -DEFINE_SPINLOCK(vector_lock); +static DEFINE_SPINLOCK(vector_lock); /* * # of IRQ routing registers @@ -697,6 +697,19 @@ static int pin_2_irq(int idx, int apic, int pin) return irq; } +void lock_vector_lock(void) +{ + /* Used to the online set of cpus does not change + * during assign_irq_vector. + */ + spin_lock(&vector_lock); +} + +void unlock_vector_lock(void) +{ + spin_unlock(&vector_lock); +} + static int __assign_irq_vector(int irq, cpumask_t mask) { /* @@ -802,7 +815,7 @@ static void __clear_irq_vector(int irq) cpus_clear(cfg->domain); } -static void __setup_vector_irq(int cpu) +void __setup_vector_irq(int cpu) { /* Initialize vector_irq on a new cpu */ /* This function must be called with vector_lock held */ @@ -825,14 +838,6 @@ static void __setup_vector_irq(int cpu) } } -void setup_vector_irq(int cpu) -{ - spin_lock(&vector_lock); - __setup_vector_irq(smp_processor_id()); - spin_unlock(&vector_lock); -} - - static struct irq_chip ioapic_chip; static void ioapic_register_intr(int irq, unsigned long trigger) diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 332512767f4f..da10f07fc59c 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -326,12 +326,16 @@ static void __cpuinit start_secondary(void *unused) * for which cpus receive the IPI. Holding this * lock helps us to not include this cpu in a currently in progress * smp_call_function(). + * + * We need to hold vector_lock so there the set of online cpus + * does not change while we are assigning vectors to cpus. Holding + * this lock ensures we don't half assign or remove an irq from a cpu. */ ipi_call_lock_irq(); -#ifdef CONFIG_X86_IO_APIC - setup_vector_irq(smp_processor_id()); -#endif + lock_vector_lock(); + __setup_vector_irq(smp_processor_id()); cpu_set(smp_processor_id(), cpu_online_map); + unlock_vector_lock(); ipi_call_unlock_irq(); per_cpu(cpu_state, smp_processor_id()) = CPU_ONLINE; @@ -1336,7 +1340,9 @@ int __cpu_disable(void) remove_siblinginfo(cpu); /* It's now safe to remove this processor from the online map */ + lock_vector_lock(); remove_cpu_from_maps(cpu); + unlock_vector_lock(); fixup_irqs(cpu_online_map); return 0; } -- cgit v1.2.1 From eeb0d7d113895556db473ff1e638803d7d49bff9 Mon Sep 17 00:00:00 2001 From: Rene Herman Date: Mon, 11 Aug 2008 17:44:57 +0200 Subject: x86, debug: tone down arch/x86/kernel/mpparse.c debugging printk commit 11a62a056093a7f25f1595fbd8bd5f93559572b6 turns some formerly nopped debugging printks in arch/x86/kernel/mppparse.c into regular ones. The one at the top of smp_scan_config() in particular also prints on !CONFIG_SMP/CONFIG_X86_LOCAL_APIC kernels and UP machines without anything resembling MP tables which makes their lowly UP owners wonder... Turn the former Dprintk()s into apic_printk()s instead meaning that their printing is dependent on passing the apic=verbose (or =debug) command line param. On 32-bit, "apic" is a __setup() param which isn't early enough for this code and therefore needs a followup changing it into an early_param(). On 64-bit, it already is. Signed-off-by: Rene Herman Cc: Andrew Morton Cc: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/mpparse.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index 6ae005ccaed8..678090508a62 100644 --- a/arch/x86/kernel/mpparse.c +++ b/arch/x86/kernel/mpparse.c @@ -83,7 +83,7 @@ static void __init MP_bus_info(struct mpc_config_bus *m) if (x86_quirks->mpc_oem_bus_info) x86_quirks->mpc_oem_bus_info(m, str); else - printk(KERN_INFO "Bus #%d is %s\n", m->mpc_busid, str); + apic_printk(APIC_VERBOSE, "Bus #%d is %s\n", m->mpc_busid, str); #if MAX_MP_BUSSES < 256 if (m->mpc_busid >= MAX_MP_BUSSES) { @@ -154,7 +154,7 @@ static void __init MP_ioapic_info(struct mpc_config_ioapic *m) static void print_MP_intsrc_info(struct mpc_config_intsrc *m) { - printk(KERN_CONT "Int: type %d, pol %d, trig %d, bus %02x," + apic_printk(APIC_VERBOSE, "Int: type %d, pol %d, trig %d, bus %02x," " IRQ %02x, APIC ID %x, APIC INT %02x\n", m->mpc_irqtype, m->mpc_irqflag & 3, (m->mpc_irqflag >> 2) & 3, m->mpc_srcbus, @@ -163,7 +163,7 @@ static void print_MP_intsrc_info(struct mpc_config_intsrc *m) static void __init print_mp_irq_info(struct mp_config_intsrc *mp_irq) { - printk(KERN_CONT "Int: type %d, pol %d, trig %d, bus %02x," + apic_printk(APIC_VERBOSE, "Int: type %d, pol %d, trig %d, bus %02x," " IRQ %02x, APIC ID %x, APIC INT %02x\n", mp_irq->mp_irqtype, mp_irq->mp_irqflag & 3, (mp_irq->mp_irqflag >> 2) & 3, mp_irq->mp_srcbus, @@ -235,7 +235,7 @@ static void __init MP_intsrc_info(struct mpc_config_intsrc *m) static void __init MP_lintsrc_info(struct mpc_config_lintsrc *m) { - printk(KERN_INFO "Lint: type %d, pol %d, trig %d, bus %02x," + apic_printk(APIC_VERBOSE, "Lint: type %d, pol %d, trig %d, bus %02x," " IRQ %02x, APIC ID %x, APIC LINT %02x\n", m->mpc_irqtype, m->mpc_irqflag & 3, (m->mpc_irqflag >> 2) & 3, m->mpc_srcbusid, @@ -695,7 +695,8 @@ static int __init smp_scan_config(unsigned long base, unsigned long length, unsigned int *bp = phys_to_virt(base); struct intel_mp_floating *mpf; - printk(KERN_DEBUG "Scan SMP from %p for %ld bytes.\n", bp, length); + apic_printk(APIC_VERBOSE, "Scan SMP from %p for %ld bytes.\n", + bp, length); BUILD_BUG_ON(sizeof(*mpf) != 16); while (length > 0) { -- cgit v1.2.1 From fb6bef8002d54a9c3062abc281893ec7c896d3ce Mon Sep 17 00:00:00 2001 From: Rene Herman Date: Mon, 11 Aug 2008 17:45:53 +0200 Subject: x86: make "apic" an early_param() on 32-bit On 32-bit, "apic" is a __setup() param meaning it is parsed rather late in the game. Make it an early_param() for apic_printk() use by arch/x86/kernel/mpparse.c. On 64-bit, it already is an early_param(). Signed-off-by: Rene Herman Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index d6c898358371..f432d4833a12 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1726,9 +1726,9 @@ static int __init apic_set_verbosity(char *str) apic_verbosity = APIC_DEBUG; else if (strcmp("verbose", str) == 0) apic_verbosity = APIC_VERBOSE; - return 1; + return 0; } -__setup("apic=", apic_set_verbosity); +early_param("apic", apic_set_verbosity); static int __init lapic_insert_resource(void) { -- cgit v1.2.1 From 9b0094f7f289d752868aae1a3984517d712cbb22 Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Thu, 7 Aug 2008 15:14:55 -0700 Subject: x86, pci-calgary: fix function declaration Fix function declaration: linux-next-20080807/arch/x86/kernel/pci-calgary_64.c:1353:36: warning: non-ANSI function declaration of function 'get_tce_space_from_tar' Signed-off-by: Randy Dunlap Acked-by: Acked-by: Muli Ben-Yehuda Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-calgary_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index b67a4b1d4eae..02d19328525d 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -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() +static void get_tce_space_from_tar(void) { int bus; void __iomem *target; -- cgit v1.2.1 From 48d97cb65e62a5f1122ac2cf1149800d4f4693e8 Mon Sep 17 00:00:00 2001 From: Rene Herman Date: Mon, 11 Aug 2008 19:20:17 +0200 Subject: x86: make "apic" an early_param() on 32-bit, NULL check Cyrill Gorcunov observed: > you turned it into early_param so now it's NULL injecting vulnerabled. > Could you please add checking for NULL str param? fix that. Also, change the name of 'str' into 'arg', to make it more apparent that this is an optional argument that can be NULL, not a string parameter that is empty when unset. Reported-by: Cyrill Gorcunov Signed-off-by: Rene Herman Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index f432d4833a12..039a8d4aaf62 100644 --- a/arch/x86/kernel/apic_32.c +++ b/arch/x86/kernel/apic_32.c @@ -1720,12 +1720,16 @@ static int __init parse_lapic_timer_c2_ok(char *arg) } early_param("lapic_timer_c2_ok", parse_lapic_timer_c2_ok); -static int __init apic_set_verbosity(char *str) +static int __init apic_set_verbosity(char *arg) { - if (strcmp("debug", str) == 0) + if (!arg) + return -EINVAL; + + if (strcmp(arg, "debug") == 0) apic_verbosity = APIC_DEBUG; - else if (strcmp("verbose", str) == 0) + else if (strcmp(arg, "verbose") == 0) apic_verbosity = APIC_VERBOSE; + return 0; } early_param("apic", apic_set_verbosity); -- cgit v1.2.1 From b74548e76a0eab1f29546e7c5a589429c069a680 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Mon, 11 Aug 2008 13:36:04 -0700 Subject: x86: fix 2.6.27rc1 cannot boot more than 8CPUs Jeff Chua reported that booting a !bigsmp kernel on a 16-way box hangs silently. this is a long-standing issue, smp start AP cpu could check the apic id >=8 etc before trying to start it. achieve this by moving the def_to_bigsmp check later and skip the apicid id > 8 [ mingo@elte.hu: clean up the message that is printed. ] Reported-by: "Jeff Chua" Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar arch/x86/kernel/setup.c | 6 ------ arch/x86/kernel/smpboot.c | 10 ++++++++++ 2 files changed, 10 insertions(+), 6 deletions(-) --- arch/x86/kernel/setup.c | 6 ------ arch/x86/kernel/smpboot.c | 10 ++++++++++ 2 files changed, 10 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 6e5823b9e765..68b48e3fbcbd 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -861,12 +861,6 @@ void __init setup_arch(char **cmdline_p) init_apic_mappings(); ioapic_init_mappings(); -#if defined(CONFIG_SMP) && defined(CONFIG_X86_PC) && defined(CONFIG_X86_32) - if (def_to_bigsmp) - printk(KERN_WARNING "More than 8 CPUs detected and " - "CONFIG_X86_PC cannot handle it.\nUse " - "CONFIG_X86_GENERICARCH or CONFIG_X86_BIGSMP.\n"); -#endif kvm_guest_init(); e820_reserve_resources(); diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index da10f07fc59c..91055d7fc1b0 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -994,7 +994,17 @@ 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; -- cgit v1.2.1 From c9d08f0860d47ed6a3fe91d0f19335086179be7b Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Tue, 12 Aug 2008 23:23:03 +0200 Subject: x86: fix 2 section mismatch warnings - map_high() WARNING: vmlinux.o(.text+0x14cf8): Section mismatch in reference from the function map_high() to the function .init.text:init_extra_mapping_uc() The function map_high() references the function __init init_extra_mapping_uc(). This is often because map_high lacks a __init annotation or the annotation of init_extra_mapping_uc is wrong. WARNING: vmlinux.o(.text+0x14d05): Section mismatch in reference from the function map_high() to the function .init.text:init_extra_mapping_wb() The function map_high() references the function __init init_extra_mapping_wb(). This is often because map_high lacks a __init annotation or the annotation of init_extra_mapping_wb is wrong. map_high is called only from __init functions (map_*_high) and calls 2 __init_functions (init_extra_mapping_*) Signed-off-by: Marcin Slusarz Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 2cfcbded888a..2d7e307c7779 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; -- cgit v1.2.1 From 6b3560229d3b6be7443fa9f9c6502e660bcfef5f Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Tue, 12 Aug 2008 23:23:05 +0200 Subject: x86: fix 2 section mismatch warnings - find_and_reserve_crashkernel WARNING: vmlinux.o(.text+0xcd1f): Section mismatch in reference from the function find_and_reserve_crashkernel() to the function .init.text:find_e820_area() The function find_and_reserve_crashkernel() references the function __init find_e820_area(). This is often because find_and_reserve_crashkernel lacks a __init annotation or the annotation of find_e820_area is wrong. WARNING: vmlinux.o(.text+0xcd38): Section mismatch in reference from the function find_and_reserve_crashkernel() to the function .init.text:reserve_bootmem_generic() The function find_and_reserve_crashkernel() references the function __init reserve_bootmem_generic(). This is often because find_and_reserve_crashkernel lacks a __init annotation or the annotation of reserve_bootmem_generic is wrong. find_and_reserve_crashkernel is called from __init function (reserve_crashkernel) and calls 2 __init functions (find_e820_area, reserve_bootmem_generic), so mark it __init Signed-off-by: Marcin Slusarz Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 68b48e3fbcbd..a4656adab53b 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; -- cgit v1.2.1 From a726c6009e6eba4acfccf8b683854866eeabb184 Mon Sep 17 00:00:00 2001 From: John Keller Date: Tue, 29 Jul 2008 14:34:16 -0500 Subject: x86: allow MMCONFIG above 4GB on x86_64 SGI UV will have MMCFG base addresses that are greater than 4GB (32 bits). v2: Use CONFIG_RESOURCES_64BIT instead of CONFIG_X86_64. v3: Create a flag, that is set by platform specific code, to disable the > 4GB check. Signed-off-by: John Keller Cc: jpk@sgi.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/boot.c | 16 +++++++++++++++- 1 file changed, 15 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index fa88a1d71290..bfd10fd211cd 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); -- cgit v1.2.1 From 23b49c19f6946cc33392a1fc75dd788dd4a90fb7 Mon Sep 17 00:00:00 2001 From: Max Krasnyansky Date: Mon, 11 Aug 2008 14:55:31 -0700 Subject: x86: resurrect proper handling of maxcpus= kernel option (v2) For some reason we had two parsers registered for maxcpus=. One in init/main.c and another in arch/x86/smpboot.c. So I nuked the one in arch/x86. Also 64-bit kernels used to handle maxcpus= as documented in Documentation/cpu-hotplug.txt. CPUs with 'id > maxcpus' are initialized but not booted. 32-bit version for some reason ignored them even though all the infrastructure for booting them later is there. In the current mainline both 64 and 32 bit versions are broken. This patch restores the correct behaviour. I've tested x86_64 version on 4- and 8- way Core2 and 2-way Opteron based machines. Various config combinations SMP, !SMP, CPU_HOTPLUG, !CPU_HOTPLUG. Booted with maxcpus=1 and maxcpus=4, etc. Everything is working as expected. So far we've received two reports from different people confirming that 32-bit version also works fine, both on dual core laptops and 16way server machines. [v2: This version fixes visws breakage pointed out by Ingo.] Signed-off-by: Max Krasnyansky Cc: lizf@cn.fujitsu.com Cc: jeff.chua.linux@gmail.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/apic_32.c | 8 -------- arch/x86/kernel/apic_64.c | 7 ------- arch/x86/kernel/smpboot.c | 14 -------------- arch/x86/kernel/visws_quirks.c | 6 ++---- 4 files changed, 2 insertions(+), 33 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/apic_32.c b/arch/x86/kernel/apic_32.c index 039a8d4aaf62..f88bd0d982b0 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 7f1f030da7ee..446c062e831c 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/smpboot.c b/arch/x86/kernel/smpboot.c index 91055d7fc1b0..e25287e4a85f 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -1386,17 +1386,3 @@ 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/visws_quirks.c b/arch/x86/kernel/visws_quirks.c index 41e01b145c48..594ef47f0a63 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; -- cgit v1.2.1 From a58f03b07539f6575adaa011712fa139c9343742 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 14 Aug 2008 02:16:30 -0700 Subject: x86: check bigsmp in smp_sanity_check instead of cpu_up clear bits for cpu nr > 8. This allows us to boot the full range of possible CPUs that the supported APIC model will allow. Previously we'd hang or boot up with less than 8 CPUs. Signed-off-by: Yinghai Lu Tested-by: Jeff Chua Signed-off-by: Ingo Molnar --- arch/x86/kernel/smpboot.c | 38 ++++++++++++++++++++++++++++---------- 1 file changed, 28 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index e25287e4a85f..a8fb8a980fae 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -994,17 +994,7 @@ 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 +1048,34 @@ 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()); -- cgit v1.2.1 From a6825f1c1fa83b1e92b6715ee5771a4d6524d3b9 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Thu, 14 Aug 2008 12:17:06 +0200 Subject: x86: hpet: workaround SB700 BIOS 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. Based on a quirk patch from: crane cai Signed-off-by: Thomas Gleixner --- arch/x86/kernel/hpet.c | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index ad2b15a1334d..59fd3b6b1303 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -359,6 +359,7 @@ static int hpet_clocksource_register(void) int __init hpet_enable(void) { unsigned long id; + int i; if (!is_hpet_capable()) return 0; @@ -369,6 +370,29 @@ 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; -- cgit v1.2.1 From 967060d00d7ab8e992963a966cd3d18156c02d55 Mon Sep 17 00:00:00 2001 From: "Darrick J. Wong" Date: Thu, 14 Aug 2008 15:43:33 -0700 Subject: x86, msr: fix NULL pointer deref due to msr_open on nonexistent CPUs msr_open tests for someone trying to open a device for a nonexistent CPU. However, the function always returns 0, not ret like it should, hence userspace can BUG the kernel trivially. This bug was introduced by the cdev lock_kernel pushdown patch last May. The BUG can be reproduced with these commands: # mknod fubar c 202 8 <-- pick a number less than NR_CPUS that is not the number of an online CPU # cat fubar Signed-off-by: Darrick J. Wong Signed-off-by: Ingo Molnar --- arch/x86/kernel/msr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c index 9fd809552447..e43938086885 100644 --- a/arch/x86/kernel/msr.c +++ b/arch/x86/kernel/msr.c @@ -131,7 +131,7 @@ static int msr_open(struct inode *inode, struct file *file) ret = -EIO; /* MSR not supported */ out: unlock_kernel(); - return 0; + return ret; } /* -- cgit v1.2.1 From 519c31bacf78a969efa8d2e55ed8862848f28590 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 14 Aug 2008 19:55:15 +0200 Subject: x86, AMD IOMMU: use status bit instead of memory write-back for completion wait Signed-off-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/amd_iommu.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 22d7d050905d..028e945c68ad 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -101,16 +101,13 @@ 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, 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; 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; @@ -122,9 +119,15 @@ static int iommu_completion_wait(struct amd_iommu *iommu) 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"); -- cgit v1.2.1 From 9f5f5fb35d2934fe7dc0cb019854a030efd10cd7 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 14 Aug 2008 19:55:16 +0200 Subject: x86, AMD IOMMU: initialize device table properly This patch adds device table initializations which forbids memory accesses for devices per default and disables all page faults. Signed-off-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/amd_iommu_init.c | 18 ++++++++++++++++++ 1 file changed, 18 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index d9a9da597e79..ceba33811537 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -800,6 +800,21 @@ static int __init init_memory_definitions(struct acpi_table_header *table) return 0; } +/* + * 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 */ -- cgit v1.2.1 From 8a456695c5020d6317f9c7af190999e9414b0d3e Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 14 Aug 2008 19:55:17 +0200 Subject: x86m AMD IOMMU: cleanup: replace LOW_U32 macro with generic lower_32_bits Signed-off-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/amd_iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index 028e945c68ad..de39e1f2ede5 100644 --- a/arch/x86/kernel/amd_iommu.c +++ b/arch/x86/kernel/amd_iommu.c @@ -164,7 +164,7 @@ static int iommu_queue_inv_iommu_pages(struct amd_iommu *iommu, 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; -- cgit v1.2.1 From 129d6aba444d1e99d4cbfb9866a4652912426b65 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Thu, 14 Aug 2008 19:55:18 +0200 Subject: x86, AMD IOMMU: initialize dma_ops after sysfs registration If sysfs registration fails all memory used by IOMMU is freed. This happens after dma_ops initialization and the functions will access the freed memory then. Fix this by initializing dma_ops after the sysfs registration. Signed-off-by: Joerg Roedel Signed-off-by: Ingo Molnar --- arch/x86/kernel/amd_iommu_init.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu_init.c b/arch/x86/kernel/amd_iommu_init.c index ceba33811537..a69cc0f52042 100644 --- a/arch/x86/kernel/amd_iommu_init.c +++ b/arch/x86/kernel/amd_iommu_init.c @@ -972,15 +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; -- cgit v1.2.1 From dcc984166870150709f0c645b521a47becd9a047 Mon Sep 17 00:00:00 2001 From: Aristeu Rozanski Date: Thu, 14 Aug 2008 16:32:15 -0400 Subject: x86, perfctr: don't use CCCR_OVF_PMI1 on Pentium 4Ds Currently, setup_p4_watchdog() use CCCR_OVF_PMI1 to enable the counter overflow interrupts to the second logical core. But this bit doesn't work on Pentium 4 Ds (model 4, stepping 4) and this patch avoids its use on these processors. Tested on 4 different machines that have this specific model with success. Signed-off-by: Aristeu Rozanski Cc: jvillalovos@redhat.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perfctr-watchdog.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c index de7439f82b92..05cc22dbd4ff 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) -- cgit v1.2.1 From 394a15051c33f2b18e72f42283b36a9388fa414b Mon Sep 17 00:00:00 2001 From: Mark Langsdorf Date: Thu, 14 Aug 2008 09:11:26 -0500 Subject: x86: invalidate caches before going into suspend MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit When a CPU core is shut down, all of its caches need to be flushed to prevent stale data from causing errors if the core is resumed. Current Linux suspend code performs an assignment after the flush, which can add dirty data back to the cache.  On some AMD platforms, additional speculative reads have caused crashes on resume because of this dirty data. Relocate the cache flush to be the very last thing done before halting.  Tie into an assembly line so the compile will not reorder it.  Add some documentation explaining what is going on and why we're doing this. Signed-off-by: Mark Langsdorf Acked-by: Mark Borden Acked-by: Michael Hohmuth Signed-off-by: Ingo Molnar --- arch/x86/kernel/process_32.c | 5 ++--- arch/x86/kernel/process_64.c | 5 ++--- 2 files changed, 4 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/process_32.c b/arch/x86/kernel/process_32.c index 53bc653ed5ca..3b7a1ddcc0bc 100644 --- a/arch/x86/kernel/process_32.c +++ b/arch/x86/kernel/process_32.c @@ -95,7 +95,6 @@ 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; @@ -104,8 +103,8 @@ 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) diff --git a/arch/x86/kernel/process_64.c b/arch/x86/kernel/process_64.c index 3fb62a7d9a16..71553b664e2a 100644 --- a/arch/x86/kernel/process_64.c +++ b/arch/x86/kernel/process_64.c @@ -93,14 +93,13 @@ DECLARE_PER_CPU(int, cpu_state); static inline void play_dead(void) { idle_task_exit(); - wbinvd(); 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) -- cgit v1.2.1 From 15636668449d4135ac77a79715ba430a81aed16d Mon Sep 17 00:00:00 2001 From: Aristeu Rozanski Date: Fri, 15 Aug 2008 08:36:14 -0400 Subject: x86, NMI: fix watchdog failure message > it just won't work at boot time - the second logic unit will be stuck: > > Booting processor 1/2 APIC 0x1 > Initializing CPU#1 > Calibrating delay using timer specific routine.. 5586.12 BogoMIPS (lpj=2793063) > CPU: Trace cache: 12K uops, L1 D cache: 16K > CPU: L2 cache: 1024K > CPU: Physical Processor ID: 0 > CPU: Processor Core ID: 1 > CPU1: Thermal monitoring enabled (TM1) > Intel(R) Pentium(R) D CPU 2.80GHz stepping 04 > Brought up 2 CPUs > testing NMI watchdog ... <4>WARNING: CPU#1: NMI appears to be stuck (0->0)! while at it... - fix that newline Signed-off-by: Aristeu Rozanski Cc: jvillalo@redhat.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/nmi.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c index ac6d51222e7d..919473ad4a29 100644 --- a/arch/x86/kernel/nmi.c +++ b/arch/x86/kernel/nmi.c @@ -142,11 +142,15 @@ int __init check_nmi_watchdog(void) if (!per_cpu(wd_enabled, cpu)) continue; if (get_nmi_count(cpu) - prev_nmi_count[cpu] <= 5) { + printk("\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 " + "linux-kernel@vger.kernel.org and attach " + "the output of 'dmesg' command.\n"); per_cpu(wd_enabled, cpu) = 0; atomic_dec(&nmi_active); } -- cgit v1.2.1 From 8bb851900f5d0a79d3fddac808cc670d9894ef67 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 15 Aug 2008 15:34:32 +0200 Subject: x86, nmi: clean UP NMI watchdog failure message clean up the failure message - and redirect people to bugzilla instead of lkml. Signed-off-by: Ingo Molnar --- arch/x86/kernel/nmi.c | 32 +++++++++++++++++++------------- 1 file changed, 19 insertions(+), 13 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/nmi.c b/arch/x86/kernel/nmi.c index 919473ad4a29..abb78a2cc4ad 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; @@ -141,19 +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("\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 " - "linux-kernel@vger.kernel.org and attach " - "the output of 'dmesg' command.\n"); - 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)) { -- cgit v1.2.1 From 9744f5a32853642f8ed0749a1c9ed8cf9c9c9dc4 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 3 Aug 2008 19:25:48 +0200 Subject: x86, acpi: cleanup, temp_stack is used only when CONFIG_SMP is set fix: arch/x86/kernel/acpi/sleep.c:24: warning: 'temp_stack' defined but not used [ Sven Wegener : fix build bug ] Signed-off-by: Marcin Slusarz Acked-by: Pavel Machek Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/sleep.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index fa2161d5003b..81e5ab6542d8 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 -- cgit v1.2.1 From 0d5cdc97e242a5589e5dca23277675f4b4482490 Mon Sep 17 00:00:00 2001 From: Jens Rottmann Date: Mon, 4 Aug 2008 14:40:16 +0200 Subject: x86, geode-mfgpt: check IRQ before using MFGPT as clocksource Adds a simple IRQ autodetection to the AMD Geode MFGPT driver, and more importantly, adds some checks, if IRQs can actually be received on the chosen line. This fixes cases where MFGPT is selected as clocksource though not producing any ticks, so the kernel simply starves during boot. Signed-off-by: Jens Rottmann Cc: Andres Salomon Cc: linux-geode@bombadil.infradead.org Cc: Jordan Crouse Signed-off-by: Ingo Molnar --- arch/x86/kernel/mfgpt_32.c | 52 +++++++++++++++++++++++++++++++++------------- 1 file changed, 37 insertions(+), 15 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/mfgpt_32.c b/arch/x86/kernel/mfgpt_32.c index 07c0f828f488..3b599518c322 100644 --- a/arch/x86/kernel/mfgpt_32.c +++ b/arch/x86/kernel/mfgpt_32.c @@ -33,6 +33,8 @@ #include #include +#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; -- cgit v1.2.1 From 66d4bdf22b8652cda215e2653c8bbec7a767ed57 Mon Sep 17 00:00:00 2001 From: Jan Beulich Date: Thu, 31 Jul 2008 16:48:31 +0100 Subject: x86-64: fix overlap of modules and fixmap areas Plus add a build time check so this doesn't go unnoticed again. Signed-off-by: Jan Beulich Signed-off-by: Ingo Molnar --- arch/x86/kernel/head64.c | 1 + 1 file changed, 1 insertion(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/head64.c b/arch/x86/kernel/head64.c index 1b318e903bf6..9bfc4d72fb2e 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(); -- cgit v1.2.1 From 163f6876f5c3ff8215e900b93779e960a56b3694 Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Fri, 15 Aug 2008 00:40:22 -0700 Subject: kexec jump: rename KEXEC_CONTROL_CODE_SIZE to KEXEC_CONTROL_PAGE_SIZE Rename KEXEC_CONTROL_CODE_SIZE to KEXEC_CONTROL_PAGE_SIZE, because control page is used for not only code on some platform. For example in kexec jump, it is used for data and stack too. [akpm@linux-foundation.org: unbreak powerpc and arm, finish conversion] Signed-off-by: Huang Ying Cc: Pavel Machek Cc: "Rafael J. Wysocki" Cc: "Eric W. Biederman" Cc: Vivek Goyal Cc: Ingo Molnar Cc: Russell King Cc: Benjamin Herrenschmidt Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/machine_kexec_32.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c index 9fe478d98406..466450167dea 100644 --- a/arch/x86/kernel/machine_kexec_32.c +++ b/arch/x86/kernel/machine_kexec_32.c @@ -78,7 +78,7 @@ 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. * -- cgit v1.2.1 From fb45daa69d287b394eca1619b3fadff7c0215c71 Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Fri, 15 Aug 2008 00:40:23 -0700 Subject: kexec jump: check code size in control page Kexec/Kexec-jump require code size in control page is less than PAGE_SIZE/2. This patch add link-time checking for this. ASSERT() of ld link script is used as the link-time checking mechanism. [akpm@linux-foundation.org: build fix] Signed-off-by: Huang Ying Cc: Pavel Machek Cc: "Rafael J. Wysocki" Cc: "Eric W. Biederman" Acked-by: Vivek Goyal Cc: Ingo Molnar Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/machine_kexec_32.c | 2 +- arch/x86/kernel/relocate_kernel_32.S | 10 +++++++--- arch/x86/kernel/vmlinux_32.lds.S | 8 ++++++++ 3 files changed, 16 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c index 466450167dea..5c8e7735c896 100644 --- a/arch/x86/kernel/machine_kexec_32.c +++ b/arch/x86/kernel/machine_kexec_32.c @@ -138,7 +138,7 @@ 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); diff --git a/arch/x86/kernel/relocate_kernel_32.S b/arch/x86/kernel/relocate_kernel_32.S index 703310a99023..6f50664b2ba5 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/vmlinux_32.lds.S b/arch/x86/kernel/vmlinux_32.lds.S index cdb2363697d2..af5bdad84604 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 + +ASSERT(kexec_control_code_size <= KEXEC_CONTROL_CODE_MAX_SIZE, + "kexec control code size is too big") +#endif -- cgit v1.2.1 From 3122c331190e9d1622bf1c8cf6ce3b17cca67c9e Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Fri, 15 Aug 2008 00:40:26 -0700 Subject: kexec jump: fix for ftrace Ftrace depends on some processor state that we destroyed during kexec and restored by restore_processor_state(). So save_processor_state() and restore_processor_state() are moved into machine_kexec() and ftrace is restored after restore_processor_state(). Signed-off-by: Huang Ying Cc: Pavel Machek Cc: "Rafael J. Wysocki" Cc: "Eric W. Biederman" Cc: Vivek Goyal Cc: Ingo Molnar Cc: Steven Rostedt Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- arch/x86/kernel/machine_kexec_32.c | 16 +++++++++++++++- 1 file changed, 15 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/machine_kexec_32.c b/arch/x86/kernel/machine_kexec_32.c index 5c8e7735c896..0732adba05ca 100644 --- a/arch/x86/kernel/machine_kexec_32.c +++ b/arch/x86/kernel/machine_kexec_32.c @@ -12,6 +12,7 @@ #include #include #include +#include #include #include @@ -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(); @@ -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) -- cgit v1.2.1 From fc0091b3c86396afc8e6c273aff21671cf882ee1 Mon Sep 17 00:00:00 2001 From: Alex Nixon Date: Fri, 15 Aug 2008 17:21:14 +0100 Subject: x86: change init_gdt to update the gdt via write_gdt, rather than a direct write. By writing directly, a memory access violation can occur whilst hotplugging a CPU if the entry was previously marked read-only. Signed-off-by: Alex Nixon Cc: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar --- arch/x86/kernel/smpcommon.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/smpcommon.c b/arch/x86/kernel/smpcommon.c index 99941b37eca0..397e309839dd 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; -- cgit v1.2.1 From d554d9a4295dd0595d12eeccbc55d1f495b15176 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Mon, 11 Aug 2008 00:07:44 +0200 Subject: x86, tsc: fix section mismatch warning WARNING: vmlinux.o(.text+0x7950): Section mismatch in reference from the function native_calibrate_tsc() to the function .init.text:tsc_read_refs() The function native_calibrate_tsc() references the function __init tsc_read_refs(). This is often because native_calibrate_tsc lacks a __init annotation or the annotation of tsc_read_refs is wrong. tsc_read_refs is called from native_calibrate_tsc which is not __init and native_calibrate_tsc cannot be marked __init Signed-off-by: Marcin Slusarz Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/tsc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 7603c0553909..46af71676738 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; -- cgit v1.2.1 From 67d0c9ebdc9f5f356657146b4e862b2d745e9e78 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Mon, 11 Aug 2008 00:12:37 +0200 Subject: x86: fix MP_processor_info section mismatch warning WARNING: arch/x86/kernel/built-in.o(.cpuinit.text+0x1fe7): Section mismatch in reference from the function MP_processor_info() to the variable .init.data:x86_quirks The function __cpuinit MP_processor_info() references a variable __initdata x86_quirks. If x86_quirks is only used by MP_processor_info then annotate x86_quirks with a matching annotation. MP_processor_info uses x86_quirks which is __init and is used only from smp_read_mpc and construct_default_ISA_mptable which are __init Signed-off-by: Marcin Slusarz Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/mpparse.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index 678090508a62..9942fc72b63c 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 = ""; -- cgit v1.2.1 From c72a5efec1193faa2ef34c0bd48d7251a70ec934 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Mon, 11 Aug 2008 00:11:13 +0200 Subject: x86: mmconf: fix section mismatch warning WARNING: arch/x86/kernel/built-in.o(.cpuinit.text+0x1591): Section mismatch in reference from the function init_amd() to the function .init.text:check_enable_amd_mmconf_dmi() The function __cpuinit init_amd() references a function __init check_enable_amd_mmconf_dmi(). If check_enable_amd_mmconf_dmi is only used by init_amd then annotate check_enable_amd_mmconf_dmi with a matching annotation. check_enable_amd_mmconf_dmi is only called from init_amd which is __cpuinit Signed-off-by: Marcin Slusarz Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/mmconf-fam10h_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/mmconf-fam10h_64.c b/arch/x86/kernel/mmconf-fam10h_64.c index fdfdc550b366..efc2f361fe85 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); } -- cgit v1.2.1 From 39e00fe20aaad4326ed5e0e3221451732bc7f679 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Mon, 11 Aug 2008 00:09:38 +0200 Subject: x86: mpparse.c: fix section mismatch warning WARNING: vmlinux.o(.text+0x118f7): Section mismatch in reference from the function construct_ioapic_table() to the function .init.text:MP_bus_info() The function construct_ioapic_table() references the function __init MP_bus_info(). This is often because construct_ioapic_table lacks a __init annotation or the annotation of MP_bus_info is wrong. construct_ioapic_table is called only from construct_default_ISA_mptable which is __init Signed-off-by: Marcin Slusarz Signed-off-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/mpparse.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/mpparse.c b/arch/x86/kernel/mpparse.c index 9942fc72b63c..b3fb430725cb 100644 --- a/arch/x86/kernel/mpparse.c +++ b/arch/x86/kernel/mpparse.c @@ -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) -- cgit v1.2.1 From e532c06f2a835b5cc4f4166f467437d9b09c1d0e Mon Sep 17 00:00:00 2001 From: David Fries Date: Sun, 17 Aug 2008 23:03:40 -0500 Subject: x86: fix i486 suspend to disk CR4 oops arch/x86/power/cpu_32.c __save_processor_state calls read_cr4() only a i486 CPU doesn't have the CR4 register. Trying to read it produces an invalid opcode oops during suspend to disk. Use the safe rc4 reading op instead. If the value to be written is zero the write is skipped. arch/x86/power/hibernate_asm_32.S done: swapped the use of %eax and %ecx to use jecxz for the zero test and jump over store to %cr4. restore_image: s/%ecx/%eax/ to be consistent with done: In addition to __save_processor_state, acpi_save_state_mem, efi_call_phys_prelog, and efi_call_phys_epilog had checks added (acpi restore was in assembly and already had a check for non-zero). There were other reads and writes of CR4, but MCE and virtualization shouldn't be executed on a i486 anyway. Signed-off-by: David Fries Acked-by: H. Peter Anvin Signed-off-by: Ingo Molnar --- arch/x86/kernel/acpi/sleep.c | 2 +- arch/x86/kernel/efi_32.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index 81e5ab6542d8..426e5d91b63a 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -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/efi_32.c b/arch/x86/kernel/efi_32.c index 4b63c8e1f13b..5cab48ee61a4 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 = -- cgit v1.2.1 From d19fbfdfe6a7034c8b6a7062365780485ab5aeaa Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 17 Aug 2008 17:50:51 +0200 Subject: x86: silence section mismatch warning - get_local_pda Take out part of get_local_pda referencing __init function (free_bootmem) to new (static) function marked as __ref. It's safe to do because free_bootmem is called before __init sections are dropped. WARNING: vmlinux.o(.cpuinit.text+0x3cd7): Section mismatch in reference from the function get_local_pda() to the function .init.text:free_bootmem() The function __cpuinit get_local_pda() references a function __init free_bootmem(). If free_bootmem is only used by get_local_pda then annotate free_bootmem with a matching annotation. Signed-off-by: Marcin Slusarz Cc: Mike Travis Signed-off-by: Ingo Molnar --- arch/x86/kernel/smpboot.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index a8fb8a980fae..e139e617f422 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; -- cgit v1.2.1 From f71066624d5d91bf179a1ea25da0800b72e20c60 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Sun, 17 Aug 2008 17:50:52 +0200 Subject: x86, calgary: fix section mismatch warning - get_tce_space_from_tar WARNING: vmlinux.o(.text+0x27032): Section mismatch in reference from the function get_tce_space_from_tar() to the function .init.text:calgary_bus_has_devices() The function get_tce_space_from_tar() references the function __init calgary_bus_has_devices(). This is often because get_tce_space_from_tar lacks a __init annotation or the annotation of calgary_bus_has_devices is wrong. get_tce_space_from_tar is called only from __init function (calgary_init) and calls __init function (calgary_bus_has_devices). So annotate it properly. Signed-off-by: Marcin Slusarz Cc: Chandru Siddalingappa Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-calgary_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index 02d19328525d..218d783ed7a8 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -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; -- cgit v1.2.1 From 8a7c5ef3ba8cca8f9c4f91ddef3d081f517914ae Mon Sep 17 00:00:00 2001 From: Jiri Kosina Date: Tue, 19 Aug 2008 02:13:55 +0200 Subject: x86 iommu: remove unneeded parenthesis The parenthesis in __iommu_queue_command() are not needed when assigning into 'target' variable. Signed-off-by: Jiri Kosina Signed-off-by: Ingo Molnar --- arch/x86/kernel/amd_iommu.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/amd_iommu.c b/arch/x86/kernel/amd_iommu.c index de39e1f2ede5..69b4d060b21c 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); -- cgit v1.2.1 From f607e3a03c90e8c050cb0c12ec9967c2925cc812 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Tue, 19 Aug 2008 13:34:59 -0700 Subject: Revert "[CPUFREQ][2/2] preregister support for powernow-k8" This reverts commit 34ae7f35a21694aa5cb8829dc5142c39d73d6ba0, which has been reported to cause a number of problems. During suspend and resume, it apparently causes a crash in a CPU hotplug notifier to happen, although the exact details are sketchy because of the inability to get good traces during the suspend sequence. See buzilla entries http://bugzilla.kernel.org/show_bug.cgi?id=11296 http://bugzilla.kernel.org/show_bug.cgi?id=11339 for more examples and details. [ Mark: "Revert the patch for now. I'm still looking into getting a reliable reproduction and I do not have a fix at this time." ] Requested-by: Rafael J. Wysocki Acked-by: Mark Langsdorf Acked-by: Dave Jones Signed-off-by: Linus Torvalds --- arch/x86/kernel/cpu/cpufreq/powernow-k8.c | 109 ++++++++++-------------------- arch/x86/kernel/cpu/cpufreq/powernow-k8.h | 3 +- 2 files changed, 37 insertions(+), 75 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.c b/arch/x86/kernel/cpu/cpufreq/powernow-k8.c index 4e7271999a74..84bb395038d8 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 and Mark Langsdorf "); diff --git a/arch/x86/kernel/cpu/cpufreq/powernow-k8.h b/arch/x86/kernel/cpu/cpufreq/powernow-k8.h index a62612cd4be8..ab48cfed4d96 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; }; -- cgit v1.2.1 From c6744955d0ec0cb485c28c51eeb7185e260f6172 Mon Sep 17 00:00:00 2001 From: Samuel Sieb Date: Wed, 6 Aug 2008 22:06:29 -0700 Subject: x86: fix "kernel won't boot on a Cyrix MediaGXm (Geode)" Cyrix MediaGXm/Cx5530 Unicorn Revision 1.19.3B has stopped booting starting at v2.6.22. The reason is this commit: > commit f25f64ed5bd3c2932493681bdfdb483ea707da0a > Author: Juergen Beisert > Date: Sun Jul 22 11:12:38 2007 +0200 > > x86: Replace NSC/Cyrix specific chipset access macros by inlined functions. this commit activated a macro which was dormant before due to (buggy) macro side-effects. I've looked through various datasheets and found that the GXm and GXLV Geode processors don't have an incrementor. Remove the incrementor setup entirely. As the incrementor value differs according to clock speed and we would hope that the BIOS configures it correctly, it is probably the right solution. Cc: Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/cyrix.c | 18 ------------------ 1 file changed, 18 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/cyrix.c b/arch/x86/kernel/cpu/cyrix.c index 3fd7a67bb06a..e710a21bb6e8 100644 --- a/arch/x86/kernel/cpu/cyrix.c +++ b/arch/x86/kernel/cpu/cyrix.c @@ -134,23 +134,6 @@ 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. */ @@ -174,7 +157,6 @@ static void __cpuinit geode_configure(void) set_cx86_memwb(); set_cx86_reorder(); - set_cx86_inc(); local_irq_restore(flags); } -- cgit v1.2.1 From 99dd8713306a89f3e106143581244e550e00a644 Mon Sep 17 00:00:00 2001 From: Cliff Wickman Date: Tue, 19 Aug 2008 12:51:59 -0500 Subject: x86, SGI UV: hardcode the TLB flush interrupt system vector The UV TLB shootdown mechanism needs a system interrupt vector. Its vector had been hardcoded as 200, but needs to moved to the reserved system vector range so that it does not collide with some device vector. This is still temporary until dynamic system IRQ allocation is provided. But it will be needed when real UV hardware becomes available and runs 2.6.27. Signed-off-by: Cliff Wickman Signed-off-by: Ingo Molnar --- arch/x86/kernel/tlb_uv.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tlb_uv.c b/arch/x86/kernel/tlb_uv.c index d0fbb7712ab0..8b8c0d6640fa 100644 --- a/arch/x86/kernel/tlb_uv.c +++ b/arch/x86/kernel/tlb_uv.c @@ -17,6 +17,7 @@ #include #include #include +#include #include @@ -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; -- cgit v1.2.1 From 85d577979a8a94a32ca1d828a91c2eb8300bc9f2 Mon Sep 17 00:00:00 2001 From: David Howells Date: Mon, 18 Aug 2008 11:58:17 +0200 Subject: werror: fix pci calgary Fix an integer comparison always false warning in the PCI Calgary 64 driver. A u8 is being compared to something that's 512 by default, resulting in the following warning: arch/x86/kernel/pci-calgary_64.c:1285: warning: comparison is always false due to limited range of data type This was introduced by patch b34e90b8f0f30151349134f87b5dc6ef75a5218c. Signed-off-by: David Howells Signed-off-by: Ingo Molnar --- arch/x86/kernel/pci-calgary_64.c | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index 218d783ed7a8..363a74ec7a76 100644 --- a/arch/x86/kernel/pci-calgary_64.c +++ b/arch/x86/kernel/pci-calgary_64.c @@ -1269,13 +1269,15 @@ 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 +1298,7 @@ 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; -- cgit v1.2.1 From bde78a79a6eb015f33aa4660df1b06f5135def20 Mon Sep 17 00:00:00 2001 From: Arjan van de Ven Date: Tue, 8 Jul 2008 09:51:56 -0700 Subject: x86: use WARN() in arch/x86/kernel Use WARN() instead of a printk+WARN_ON() pair; this way the message becomes part of the warning section for better reporting/collection. This also allowed the folding of some if()'s into the WARN() Signed-off-by: Arjan van de Ven Cc: akpm@linux-foundation.org Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/mtrr/main.c | 5 +---- arch/x86/kernel/pci-calgary_64.c | 3 +-- arch/x86/kernel/tsc_sync.c | 6 ++---- 3 files changed, 4 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/mtrr/main.c b/arch/x86/kernel/cpu/mtrr/main.c index 6f23969c8faf..b117d7f8a564 100644 --- a/arch/x86/kernel/cpu/mtrr/main.c +++ b/arch/x86/kernel/cpu/mtrr/main.c @@ -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/pci-calgary_64.c b/arch/x86/kernel/pci-calgary_64.c index 363a74ec7a76..dcdac6c826e9 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; } diff --git a/arch/x86/kernel/tsc_sync.c b/arch/x86/kernel/tsc_sync.c index 0577825cf89b..9ffb01c31c40 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); - } } /* -- cgit v1.2.1 From 7946612de2087e163308e26034286fc2dc9dacf1 Mon Sep 17 00:00:00 2001 From: Jeremy Fitzhardinge Date: Wed, 20 Aug 2008 11:31:07 -0700 Subject: x86: export pv_lock_ops non-GPL None of the spinlock API is exported GPL, so there's no reason for pv_lock_ops to be. Signed-off-by: Jeremy Fitzhardinge Signed-off-by: Ingo Molnar Cc: drago01 --- arch/x86/kernel/paravirt.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/paravirt.c b/arch/x86/kernel/paravirt.c index 94da4d52d798..300da17e61cb 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); -- cgit v1.2.1 From b2a6a58ca6a3ddf4e278a53199c5b8fd39adbc0e Mon Sep 17 00:00:00 2001 From: Vegard Nossum Date: Wed, 20 Aug 2008 18:18:26 +0200 Subject: x86: fix BUG: unable to handle kernel paging request (numaq_tsc_disable) This section mismatch: >> Seems to be a section mismatch; init_intel() is __cpuinit while >> numaq_tsc_disable() is __init. Seems to be introduced in: >> >> commit 64898a8bad8c94ad7a4bd5cc86b66edfbb081f4a >> Author: Yinghai Lu >> Date: Sat Jul 19 18:01:16 2008 -0700 >> >> x86: extend and use x86_quirks to clean up NUMAQ code > > Oops, I am wrong about numaq_tsc_disable() being __init. Still, I > believe that Yinghai might be able to say what's really wrong :-) Would lead to this crash: BUG: unable to handle kernel paging request at c08a45f0 IP: [] numaq_tsc_disable+0x0/0x40 Fixed by the patch below. Signed-off-by: Vegard Nossum Signed-off-by: Ingo Molnar --- arch/x86/kernel/numaq_32.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/numaq_32.c b/arch/x86/kernel/numaq_32.c index b8c45610b20a..eecc8c18f010 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; -- cgit v1.2.1 From 8323444b5dba3fe55e56a95d20d8f55c1d6745af Mon Sep 17 00:00:00 2001 From: "venkatesh.pallipadi@intel.com" Date: Wed, 20 Aug 2008 16:45:53 -0700 Subject: x86: PAT Update validate_pat_support for intel CPUs Pentium III and Core Solo/Duo CPUs have an erratum " Page with PAT set to WC while associated MTRR is UC may consolidate to UC " which can result in WC setting in PAT to be ineffective. We will disable PAT on such CPUs, so that we can continue to use MTRR WC setting. Signed-off-by: Venkatesh Pallipadi Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/addon_cpuid_features.c | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/addon_cpuid_features.c b/arch/x86/kernel/cpu/addon_cpuid_features.c index 84a8220a6072..a6ef672adbba 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: -- cgit v1.2.1 From 38cc1c3df77c1bb739a4766788eb9fa49f16ffdf Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 21 Aug 2008 20:24:24 -0700 Subject: x86: work around MTRR mask setting Joshua Hoblitt reported that only 3 GB of his 16 GB of RAM is usable. Booting with mtrr_show showed us the BIOS-initialized MTRR settings - which are all wrong. So the root cause is that the BIOS has not set the mask correctly: > [ 0.429971] MSR00000200: 00000000d0000000 > [ 0.433305] MSR00000201: 0000000ff0000800 > should be ==> [ 0.433305] MSR00000201: 0000003ff0000800 > > [ 0.436638] MSR00000202: 00000000e0000000 > [ 0.439971] MSR00000203: 0000000fe0000800 > should be ==> [ 0.439971] MSR00000203: 0000003fe0000800 > > [ 0.443304] MSR00000204: 0000000000000006 > [ 0.446637] MSR00000205: 0000000c00000800 > should be ==> [ 0.446637] MSR00000205: 0000003c00000800 > > [ 0.449970] MSR00000206: 0000000400000006 > [ 0.453303] MSR00000207: 0000000fe0000800 > should be ==> [ 0.453303] MSR00000207: 0000003fe0000800 > > [ 0.456636] MSR00000208: 0000000420000006 > [ 0.459970] MSR00000209: 0000000ff0000800 > should be ==> [ 0.459970] MSR00000209: 0000003ff0000800 So detect this borkage and add the prefix 111. Signed-off-by: Yinghai Lu Cc: Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/mtrr/generic.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c index 509bd3d9eacd..43102e03e2d1 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 +393,18 @@ 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) { + WARN_ON("mtrr: your BIOS has set up an incorrect mask, fixing it up.\n"); + mask_lo = tmp; + } + } /* This works correctly if size is a power of two, i.e. a contiguous range. */ -- cgit v1.2.1 From 3a6ddd5f18405ca92e004416af8ed44b9c9783d7 Mon Sep 17 00:00:00 2001 From: Alok Kataria Date: Thu, 21 Aug 2008 11:32:26 -0700 Subject: x86: fix VMI for early params while fixing a different bug i moved the call to vmi_init before early params could be parsed. This broke the vmi specific commandline parameters. Fix that, by moving vmi initialization after kernel has got a chance to parse early parameters. Signed-off-by: Alok N Kataria Signed-off-by: Ingo Molnar --- arch/x86/kernel/setup.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index a4656adab53b..362d4e7f2d38 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -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 +670,14 @@ void __init setup_arch(char **cmdline_p) parse_early_param(); +#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(); -- cgit v1.2.1 From c4bd1fdab0deec0f69aeabab22075cb22ac8ad44 Mon Sep 17 00:00:00 2001 From: Marcin Slusarz Date: Thu, 21 Aug 2008 20:49:05 +0200 Subject: x86: fix section mismatch warning - uv_cpu_init WARNING: vmlinux.o(.cpuinit.text+0x3cc4): Section mismatch in reference from the function uv_cpu_init() to the function .init.text:uv_system_init() The function __cpuinit uv_cpu_init() references a function __init uv_system_init(). If uv_system_init is only used by uv_cpu_init then annotate uv_system_init with a matching annotation. uv_system_init was ment to be called only once, so do it from codepath (native_smp_prepare_cpus) which is called once, right before activation of other cpus (smp_init). Note: old code relied on uv_node_to_blade being initialized to 0, but it'a not initialized from anywhere. Signed-off-by: Marcin Slusarz Acked-by: Jack Steiner Signed-off-by: Ingo Molnar --- arch/x86/kernel/genx2apic_uv_x.c | 8 +++++--- arch/x86/kernel/smpboot.c | 3 +++ 2 files changed, 8 insertions(+), 3 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/genx2apic_uv_x.c b/arch/x86/kernel/genx2apic_uv_x.c index 2d7e307c7779..bfa837cb16be 100644 --- a/arch/x86/kernel/genx2apic_uv_x.c +++ b/arch/x86/kernel/genx2apic_uv_x.c @@ -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/smpboot.c b/arch/x86/kernel/smpboot.c index e139e617f422..7985c5b3f916 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -1221,6 +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(); } -- cgit v1.2.1 From 9754a5b840a209bc1f192d59f63e81b698a55ac8 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 22 Aug 2008 08:22:23 +0200 Subject: x86: work around MTRR mask setting, v2 improve the debug printout: - make it actually display something - print it only once would be nice to have a WARN_ONCE() facility, to feed such things to kerneloops.org. Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/mtrr/generic.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/mtrr/generic.c b/arch/x86/kernel/cpu/mtrr/generic.c index 43102e03e2d1..cb7d3b6a80eb 100644 --- a/arch/x86/kernel/cpu/mtrr/generic.c +++ b/arch/x86/kernel/cpu/mtrr/generic.c @@ -401,7 +401,12 @@ static void generic_get_mtrr(unsigned int reg, unsigned long *base, tmp |= ~((1<<(hi - 1)) - 1); if (tmp != mask_lo) { - WARN_ON("mtrr: your BIOS has set up an incorrect mask, fixing it up.\n"); + 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; } } -- cgit v1.2.1 From 8735728ef8dc935c4fb351f913758fdbb62c308d Mon Sep 17 00:00:00 2001 From: "Rafael J. Wysocki" Date: Fri, 22 Aug 2008 22:23:09 +0200 Subject: x86 MCE: Fix CPU hotplug problem with multiple multicore AMD CPUs During CPU hot-remove the sysfs directory created by threshold_create_bank(), defined in arch/x86/kernel/cpu/mcheck/mce_amd_64.c, has to be removed before its parent directory, created by mce_create_device(), defined in arch/x86/kernel/cpu/mcheck/mce_64.c . Moreover, when the CPU in question is hotplugged again, obviously the latter has to be created before the former. At present, the right ordering is not enforced, because all of these operations are carried out by CPU hotplug notifiers which are not appropriately ordered with respect to each other. This leads to serious problems on systems with two or more multicore AMD CPUs, among other things during suspend and hibernation. Fix the problem by placing threshold bank CPU hotplug callbacks in mce_cpu_callback(), so that they are invoked at the right places, if defined. Additionally, use kobject_del() to remove the sysfs directory associated with the kobject created by kobject_create_and_add() in threshold_create_bank(), to prevent the kernel from crashing during CPU hotplug operations on systems with two or more multicore AMD CPUs. This patch fixes bug #11337. Signed-off-by: Rafael J. Wysocki Acked-by: Andi Kleen Tested-by: Mark Langsdorf Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/mcheck/mce_64.c | 5 +++++ arch/x86/kernel/cpu/mcheck/mce_amd_64.c | 18 +++++------------- 2 files changed, 10 insertions(+), 13 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/mcheck/mce_64.c b/arch/x86/kernel/cpu/mcheck/mce_64.c index 65a339678ece..726a5fcdf341 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 88736cadbaa6..5eb390a4b2e9 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; } -- cgit v1.2.1 From 060700b571717c997a2ea5e2049b848fa248ee13 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Sun, 24 Aug 2008 11:52:06 -0700 Subject: x86: do not enable TSC notifier if we don't need it Impact: crash on non-TSC-equipped CPUs Don't enable the TSC notifier if we *either*: 1. don't have a CPU, or 2. have a CPU with constant TSC. In either of those cases, the notifier is either damaging (1) or useless(2). From: Linus Torvalds Signed-off-by: H. Peter Anvin --- arch/x86/kernel/tsc.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 46af71676738..9bed5cae4bdc 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -325,6 +325,10 @@ 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; -- cgit v1.2.1 From 52a8968ce95da8469ba0a9b3e4010fe31caf77a3 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 25 Aug 2008 13:35:06 +0200 Subject: x86: fix cpufreq + sched_clock() regression I noticed that my sched_clock() was slow on a number of machine, so I started looking at cpufreq. The below seems to fix the problem for me. Signed-off-by: Ingo Molnar --- arch/x86/kernel/tsc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 9bed5cae4bdc..8e786b0d665a 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -314,7 +314,7 @@ 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; } -- cgit v1.2.1 From c6f31932d0a1d2b13952f506ebc92675e2d8df80 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 25 Aug 2008 17:27:21 -0700 Subject: x86: msr: propagate errors from smp_call_function_single() Propagate error (-ENXIO) from smp_call_function_single(). These errors can happen when a CPU is unplugged while the MSR driver is open. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/msr.c | 14 ++++++++++---- 1 file changed, 10 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c index e43938086885..9c34a1005dba 100644 --- a/arch/x86/kernel/msr.c +++ b/arch/x86/kernel/msr.c @@ -79,8 +79,11 @@ static ssize_t msr_read(struct file *file, char __user *buf, for (; count; count -= 8) { err = rdmsr_safe_on_cpu(cpu, reg, &data[0], &data[1]); - if (err) - return -EIO; + if (err) { + if (err == -EFAULT) /* Fix idiotic error code */ + err = -EIO; + return err; + } if (copy_to_user(tmp, &data, 8)) return -EFAULT; tmp += 2; @@ -105,8 +108,11 @@ static ssize_t msr_write(struct file *file, const char __user *buf, if (copy_from_user(&data, tmp, 8)) return -EFAULT; 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; + return err; + } tmp += 2; } -- cgit v1.2.1 From 4b46ca701bdcdc19fcf32823f9fcabf8236e4e78 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 25 Aug 2008 17:28:20 -0700 Subject: x86: cpuid: propagate error from smp_call_function_single() Propagate error (-ENXIO) from smp_call_function_single() in the CPUID driver. This can happen when a CPU is unplugged while the CPUID driver is open. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpuid.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpuid.c b/arch/x86/kernel/cpuid.c index 14b11b3be31c..23e8316c8357 100644 --- a/arch/x86/kernel/cpuid.c +++ b/arch/x86/kernel/cpuid.c @@ -89,6 +89,7 @@ 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; + int err; if (count % 16) return -EINVAL; /* Invalid chunk size */ @@ -96,7 +97,9 @@ 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); + err = smp_call_function_single(cpu, cpuid_smp_cpuid, &cmd, 1); + if (err) + return err; if (copy_to_user(tmp, &cmd, 16)) return -EFAULT; tmp += 16; -- cgit v1.2.1 From 85f1cb60157e06d9e8996b02fad9ba6964523d75 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 25 Aug 2008 17:34:27 -0700 Subject: x86: msr: correct return value on partial operations Return the correct return value when the MSR driver partially completes a request (we should return the number of bytes actually read or written, instead of the error code.) Signed-off-by: H. Peter Anvin --- arch/x86/kernel/msr.c | 28 ++++++++++++++++++---------- 1 file changed, 18 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/msr.c b/arch/x86/kernel/msr.c index 9c34a1005dba..2e2af5d18191 100644 --- a/arch/x86/kernel/msr.c +++ b/arch/x86/kernel/msr.c @@ -72,7 +72,8 @@ 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 */ @@ -82,14 +83,17 @@ static ssize_t msr_read(struct file *file, char __user *buf, if (err) { if (err == -EFAULT) /* Fix idiotic error code */ err = -EIO; - return err; + break; + } + if (copy_to_user(tmp, &data, 8)) { + err = -EFAULT; + break; } - if (copy_to_user(tmp, &data, 8)) - return -EFAULT; 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, @@ -99,24 +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) { if (err == -EFAULT) /* Fix idiotic error code */ err = -EIO; - return err; + break; } tmp += 2; + bytes += 8; } - return ((char __user *)tmp) - buf; + return bytes ? bytes : err; } static int msr_open(struct inode *inode, struct file *file) -- cgit v1.2.1 From 9ea2b82ed6265a31f9a84886d74d8a2ef01b27c8 Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 25 Aug 2008 17:35:25 -0700 Subject: x86: cpuid: correct return value on partial operations Return the correct return value when the CPUID driver partially completes a request (we should return the number of bytes actually read or written, instead of the error code.) Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpuid.c | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpuid.c b/arch/x86/kernel/cpuid.c index 23e8316c8357..8e9cd6a8ec12 100644 --- a/arch/x86/kernel/cpuid.c +++ b/arch/x86/kernel/cpuid.c @@ -89,7 +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; - int err; + ssize_t bytes = 0; + int err = 0; if (count % 16) return -EINVAL; /* Invalid chunk size */ @@ -99,14 +100,17 @@ static ssize_t cpuid_read(struct file *file, char __user *buf, cmd.ecx = pos >> 32; err = smp_call_function_single(cpu, cpuid_smp_cpuid, &cmd, 1); if (err) - return err; - if (copy_to_user(tmp, &cmd, 16)) - return -EFAULT; + 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) -- cgit v1.2.1 From fbb16e243887332dd5754e48ffe5b963378f3cd2 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 3 Sep 2008 00:54:47 +0200 Subject: [x86] Fix TSC calibration issues Larry Finger reported at http://lkml.org/lkml/2008/9/1/90: An ancient laptop of mine started throwing errors from b43legacy when I started using 2.6.27 on it. This has been bisected to commit bfc0f59 "x86: merge tsc calibration". The unification of the TSC code adopted mostly the 64bit code, which prefers PMTIMER/HPET over the PIT calibration. Larrys system has an AMD K6 CPU. Such systems are known to have PMTIMER incarnations which run at double speed. This results in a miscalibration of the TSC by factor 0.5. So the resulting calibrated CPU/TSC speed is half of the real CPU speed, which means that the TSC based delay loop will run half the time it should run. That might explain why the b43legacy driver went berserk. On the other hand we know about systems, where the PIT based calibration results in random crap due to heavy SMI/SMM disturbance. On those systems the PMTIMER/HPET based calibration logic with SMI detection shows better results. According to Alok also virtualized systems suffer from the PIT calibration method. The solution is to use a more wreckage aware aproach than the current either/or decision. 1) reimplement the retry loop which was dropped from the 32bit code during the merge. It repeats the calibration and selects the lowest frequency value as this is probably the closest estimate to the real frequency 2) Monitor the delta of the TSC values in the delay loop which waits for the PIT counter to reach zero. If the maximum value is significantly different from the minimum, then we have a pretty safe indicator that the loop was disturbed by an SMI. 3) keep the pmtimer/hpet reference as a backup solution for systems where the SMI disturbance is a permanent point of failure for PIT based calibration 4) do the loop iteration for both methods, record the lowest value and decide after all iterations finished. 5) Set a clear preference to PIT based calibration when the result makes sense. The implementation does the reference calibration based on HPET/PMTIMER around the delay, which is necessary for the PIT anyway, but keeps separate TSC values to ensure the "independency" of the resulting calibration values. Tested on various 32bit/64bit machines including Geode 266Mhz, AMD K6 (affected machine with a double speed pmtimer which I grabbed out of the dump), Pentium class machines and AMD/Intel 64 bit boxen. Bisected-by: Larry Finger Signed-off-by: Thomas Gleixner Tested-by: Larry Finger Signed-off-by: Linus Torvalds --- arch/x86/kernel/tsc.c | 235 ++++++++++++++++++++++++++++++++++++++------------ 1 file changed, 181 insertions(+), 54 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 8e786b0d665a..ac79bd143da8 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -127,75 +127,202 @@ static u64 tsc_read_refs(u64 *pm, u64 *hpet) */ unsigned long native_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; + u64 tsc1, tsc2, tr1, tr2, tsc, delta, pm1, pm2, hpet1, hpet2; + unsigned long tsc_pit_min = ULONG_MAX, tsc_ref_min = ULONG_MAX; + unsigned long flags, tscmin, tscmax; + int hpet = is_hpet_enabled(), pitcnt, i; - local_irq_save(flags); - - tsc1 = tsc_read_refs(&pm1, hpet ? &hpet1 : NULL); - - outb((inb(0x61) & ~0x02) | 0x01, 0x61); - - 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); - - local_irq_restore(flags); + /* + * 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++) { + + tscmin = ULONG_MAX; + tscmax = 0; + pitcnt = 0; + + local_irq_save(flags); + + /* + * Read the start value and the reference count of + * hpet/pmtimer when available: + */ + tsc1 = tsc_read_refs(&pm1, hpet ? &hpet1 : NULL); + + /* 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. + * + * Some devices need a delay here. + */ + outb(0xb0, 0x43); + outb((CLOCK_TICK_RATE / (1000 / 50)) & 0xff, 0x42); + outb((CLOCK_TICK_RATE / (1000 / 50)) >> 8, 0x42); + + tsc = tr1 = tr2 = get_cycles(); + + while ((inb(0x61) & 0x20) == 0) { + tr2 = get_cycles(); + delta = tr2 - tsc; + tsc = tr2; + if ((unsigned int) delta < tscmin) + tscmin = (unsigned int) delta; + if ((unsigned int) delta > tscmax) + tscmax = (unsigned int) delta; + pitcnt++; + } + + /* + * We waited at least 50ms above. Now read + * pmtimer/hpet reference again + */ + tsc2 = tsc_read_refs(&pm2, hpet ? &hpet2 : NULL); + + local_irq_restore(flags); + + /* + * 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. + */ + if (pitcnt > 5000 && tscmax < 10 * tscmin) { + + /* Calculate the PIT value */ + delta = tr2 - tr1; + do_div(delta, 50); + + /* We take the smallest value into account */ + tsc_pit_min = min(tsc_pit_min, (unsigned long) delta); + } + + /* 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); + } /* - * Preset the result with the raw and inaccurate PIT - * calibration value + * Now check the results. */ - delta = (tr2 - tr1); - do_div(delta, 50); - tsc_khz_val = delta; + if (tsc_pit_min == ULONG_MAX) { + /* PIT gave no useful value */ + printk(KERN_WARNING "TSC: PIT calibration failed due to " + "SMI disturbance.\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) -- cgit v1.2.1 From ec0c15afb41fd9ad45b53468b60db50170e22346 Mon Sep 17 00:00:00 2001 From: Linus Torvalds Date: Wed, 3 Sep 2008 07:30:13 -0700 Subject: Split up PIT part of TSC calibration from native_calibrate_tsc The TSC calibration function is still very complicated, but this makes it at least a little bit less so by moving the PIT part out into a helper function of its own. Tested-by: Larry Finger Signed-of-by: Linus Torvalds --- arch/x86/kernel/tsc.c | 132 +++++++++++++++++++++++++++----------------------- 1 file changed, 71 insertions(+), 61 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index ac79bd143da8..346cae5ac423 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -122,15 +122,75 @@ static u64 tsc_read_refs(u64 *pm, u64 *hpet) return ULLONG_MAX; } +/* + * 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. + */ +static unsigned long pit_calibrate_tsc(void) +{ + 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); + + tsc = t1 = t2 = get_cycles(); + + 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++; + } + + /* + * 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. + */ + if (pitcnt < 5000 || tscmax > 10 * tscmin) + return ULONG_MAX; + + /* Calculate the PIT value */ + delta = t2 - t1; + do_div(delta, 50); + return delta; +} + + /** * native_calibrate_tsc - calibrate the tsc on boot */ unsigned long native_calibrate_tsc(void) { - u64 tsc1, tsc2, tr1, tr2, tsc, delta, pm1, pm2, hpet1, hpet2; + u64 tsc1, tsc2, delta, pm1, pm2, hpet1, hpet2; unsigned long tsc_pit_min = ULONG_MAX, tsc_ref_min = ULONG_MAX; - unsigned long flags, tscmin, tscmax; - int hpet = is_hpet_enabled(), pitcnt, i; + unsigned long flags; + int hpet = is_hpet_enabled(), i; /* * Run 5 calibration loops to get the lowest frequency value @@ -157,72 +217,22 @@ unsigned long native_calibrate_tsc(void) * amount of time anyway. */ for (i = 0; i < 5; i++) { - - tscmin = ULONG_MAX; - tscmax = 0; - pitcnt = 0; - - local_irq_save(flags); + unsigned long tsc_pit_khz; /* * Read the start value and the reference count of - * hpet/pmtimer when available: + * 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); - - /* 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. - * - * Some devices need a delay here. - */ - outb(0xb0, 0x43); - outb((CLOCK_TICK_RATE / (1000 / 50)) & 0xff, 0x42); - outb((CLOCK_TICK_RATE / (1000 / 50)) >> 8, 0x42); - - tsc = tr1 = tr2 = get_cycles(); - - while ((inb(0x61) & 0x20) == 0) { - tr2 = get_cycles(); - delta = tr2 - tsc; - tsc = tr2; - if ((unsigned int) delta < tscmin) - tscmin = (unsigned int) delta; - if ((unsigned int) delta > tscmax) - tscmax = (unsigned int) delta; - pitcnt++; - } - - /* - * We waited at least 50ms above. Now read - * pmtimer/hpet reference again - */ + tsc_pit_khz = pit_calibrate_tsc(); tsc2 = tsc_read_refs(&pm2, hpet ? &hpet2 : NULL); - local_irq_restore(flags); - /* - * 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. - */ - if (pitcnt > 5000 && tscmax < 10 * tscmin) { - - /* Calculate the PIT value */ - delta = tr2 - tr1; - do_div(delta, 50); - - /* We take the smallest value into account */ - tsc_pit_min = min(tsc_pit_min, (unsigned long) delta); - } + /* 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) -- cgit v1.2.1 From e6a5652fd156a286faadbf7a4062b5354d4e346e Mon Sep 17 00:00:00 2001 From: Chuck Ebbert Date: Wed, 3 Sep 2008 19:33:14 -0400 Subject: x86: add io delay quirk for Presario F700 Manually adding "io_delay=0xed" fixes system lockups in ioapic mode on this machine. System Information Manufacturer: Hewlett-Packard Product Name: Presario F700 (KA695EA#ABF) Base Board Information Manufacturer: Quanta Product Name: 30D3 Reference: https://bugzilla.redhat.com/show_bug.cgi?id=459546 Signed-off-by: Chuck Ebbert Signed-off-by: H. Peter Anvin --- arch/x86/kernel/io_delay.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/io_delay.c b/arch/x86/kernel/io_delay.c index 1c3a66a67f83..720d2607aacb 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") + } + }, { } }; -- cgit v1.2.1 From de014d617636d6a6bd5aef3b3d1f7f9a35669057 Mon Sep 17 00:00:00 2001 From: Alok N Kataria Date: Wed, 3 Sep 2008 18:18:01 -0700 Subject: x86: Change warning message in TSC calibration. When calibration against PIT fails, the warning that we print is misleading. In a virtualized environment the VM may get descheduled while calibration or, the check in PIT calibration may fail due to other virtualization overheads. The warning message explicitly assumes that calibration failed due to SMI's which may not be the case. Change that to something proper. Signed-off-by: Alok N Kataria Signed-off-by: Linus Torvalds --- arch/x86/kernel/tsc.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c index 346cae5ac423..8f98e9de1b82 100644 --- a/arch/x86/kernel/tsc.c +++ b/arch/x86/kernel/tsc.c @@ -267,8 +267,7 @@ unsigned long native_calibrate_tsc(void) */ if (tsc_pit_min == ULONG_MAX) { /* PIT gave no useful value */ - printk(KERN_WARNING "TSC: PIT calibration failed due to " - "SMI disturbance.\n"); + printk(KERN_WARNING "TSC: Unable to calibrate against PIT\n"); /* We don't have an alternative source, disable TSC */ if (!hpet && !pm1 && !pm2) { -- cgit v1.2.1 From 7cfb0435330364f90f274a26ecdc5f47f738498c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Wed, 3 Sep 2008 21:37:24 +0000 Subject: HPET: make minimum reprogramming delta useful The minimum reprogramming delta was hardcoded in HPET ticks, which is stupid as it does not work with faster running HPETs. The C1E idle patches made this prominent on AMD/RS690 chipsets, where the HPET runs with 25MHz. Set it to 5us which seems to be a reasonable value and fixes the problems on the bug reporters machines. We have a further sanity check now in the clock events, which increases the delta when it is not sufficient. Signed-off-by: Thomas Gleixner Tested-by: Luiz Fernando N. Capitulino Tested-by: Dmitry Nezhevenko Signed-off-by: Ingo Molnar --- arch/x86/kernel/hpet.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 59fd3b6b1303..2256315416d8 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 -- cgit v1.2.1 From b6734c35af028f06772c0b2c836c7d579e6d4dad Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 18 Aug 2008 17:39:32 -0700 Subject: x86: add NOPL as a synthetic CPU feature bit The long noops ("NOPL") are supposed to be detected by family >= 6. Unfortunately, several non-Intel x86 implementations, both hardware and software, don't obey this dictum. Instead, probe for NOPL directly by executing a NOPL instruction and see if we get #UD. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/cpu/common.c | 32 +++++++++++++++++++++++++++++++- arch/x86/kernel/cpu/common_64.c | 36 ++++++++++++++++++++++++++++++++++++ arch/x86/kernel/cpu/feature_names.c | 3 ++- 3 files changed, 69 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 80ab20d4fa39..0785b3c8d043 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -13,6 +13,7 @@ #include #include #include +#include #ifdef CONFIG_X86_LOCAL_APIC #include #include @@ -341,6 +342,35 @@ static void __init early_cpu_detect(void) 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. Hence, probe for it based on first + * principles. + */ +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 generic_identify(struct cpuinfo_x86 *c) { u32 tfms, xlvl; @@ -395,8 +425,8 @@ 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 dd6e3f15017e..c3afba5a81a7 100644 --- a/arch/x86/kernel/cpu/common_64.c +++ b/arch/x86/kernel/cpu/common_64.c @@ -18,6 +18,7 @@ #include #include #include +#include #include #ifdef CONFIG_X86_LOCAL_APIC #include @@ -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); diff --git a/arch/x86/kernel/cpu/feature_names.c b/arch/x86/kernel/cpu/feature_names.c index e43ad4ad4cba..c9017799497c 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) */ -- cgit v1.2.1 From f31d731e4467e61de51d7f6d7115f3b712d9354c Mon Sep 17 00:00:00 2001 From: "H. Peter Anvin" Date: Mon, 18 Aug 2008 17:50:33 -0700 Subject: x86: use X86_FEATURE_NOPL in alternatives Use X86_FEATURE_NOPL to determine if it is safe to use P6 NOPs in alternatives. Also, replace table and loop with simple if statement. Signed-off-by: H. Peter Anvin --- arch/x86/kernel/alternative.c | 36 +++++++++++++----------------------- 1 file changed, 13 insertions(+), 23 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/alternative.c b/arch/x86/kernel/alternative.c index 2763cb37b553..65a0c1b48696 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 */ -- cgit v1.2.1 From f7676254f179eac6b5244a80195ec8ae0e9d4606 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sat, 6 Sep 2008 03:03:32 +0200 Subject: x86: HPET fix moronic 32/64bit thinko We use the HPET only in 32bit mode because: 1) some HPETs are 32bit only 2) on i386 there is no way to read/write the HPET atomic 64bit wide The HPET code unification done by the "moron of the year" did not take into account that unsigned long is different on 32 and 64 bit. This thinko results in a possible endless loop in the clockevents code, when the return comparison fails due to the 64bit/332bit unawareness. unsigned long cnt = (u32) hpet_read() + delta can wrap over 32bit. but the final compare will fail and return -ETIME causing endless loops. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/hpet.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 2256315416d8..801497a16e0e 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -270,15 +270,15 @@ 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; + return (s32)((u32)hpet_readl(HPET_COUNTER) - cnt) >= 0 ? -ETIME : 0; } /* -- cgit v1.2.1 From 72d43d9bc9210d24d09202eaf219eac09e17b339 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sat, 6 Sep 2008 03:06:08 +0200 Subject: x86: HPET: read back compare register before reading counter After fixing the u32 thinko I sill had occasional hickups on ATI chipsets with small deltas. There seems to be a delay between writing the compare register and the transffer to the internal register which triggers the interrupt. Reading back the value makes sure, that it hit the internal match register befor we compare against the counter value. Signed-off-by: Thomas Gleixner --- arch/x86/kernel/hpet.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/hpet.c b/arch/x86/kernel/hpet.c index 801497a16e0e..73deaffadd03 100644 --- a/arch/x86/kernel/hpet.c +++ b/arch/x86/kernel/hpet.c @@ -278,6 +278,13 @@ static int hpet_legacy_next_event(unsigned long delta, cnt += (u32) delta; hpet_writel(cnt, HPET_T0_CMP); + /* + * 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; } -- cgit v1.2.1 From 12cf105cd66d95cf32c73cfa847a50bd1b700f23 Mon Sep 17 00:00:00 2001 From: Krzysztof Helt Date: Thu, 4 Sep 2008 21:09:43 +0200 Subject: x86: delay early cpu initialization until cpuid is done Move early cpu initialization after cpu early get cap so the early cpu initialization can fix up cpu caps. Signed-off-by: Krzysztof Helt Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common.c b/arch/x86/kernel/cpu/common.c index 0785b3c8d043..8aab8517642e 100644 --- a/arch/x86/kernel/cpu/common.c +++ b/arch/x86/kernel/cpu/common.c @@ -335,11 +335,11 @@ 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); } /* -- cgit v1.2.1 From dd786dd12c99634055a9066f25ea957f29991c22 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Thu, 4 Sep 2008 21:09:43 +0200 Subject: x86: move mtrr cpu cap setting early in early_init_xxxx Krzysztof Helt found MTRR is not detected on k6-2 root cause: we moved mtrr_bp_init() early for mtrr trimming, and in early_detect we only read the CPU capability from cpuid, so some cpu doesn't have that bit in cpuid. So we need to add early_init_xxxx to preset those bit before mtrr_bp_init for those earlier cpus. this patch is for v2.6.27 Reported-by: Krzysztof Helt Signed-off-by: Yinghai Lu Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/amd.c | 9 +++++---- arch/x86/kernel/cpu/centaur.c | 11 +++++++++++ arch/x86/kernel/cpu/cyrix.c | 32 ++++++++++++++++++++++++++++---- 3 files changed, 44 insertions(+), 8 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c index cae9cabc3031..18514ed26104 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 e0f45edd6a55..a0534c04d38a 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/cyrix.c b/arch/x86/kernel/cpu/cyrix.c index e710a21bb6e8..898a5a2002ed 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 @@ -161,6 +166,24 @@ static void __cpuinit geode_configure(void) 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) { @@ -416,6 +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, }; -- cgit v1.2.1 From d04ec773d7ca1bbc05a2768be95c1cebe2b07757 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Wed, 6 Aug 2008 10:27:30 +0200 Subject: x86: pda_init(): fix memory leak when using CPU hotplug pda->irqstackptr is allocated whenever a CPU is set online. But it is never freed. This results in a memory leak of 16K for each CPU offline/online cycle. Fix is to allocate pda->irqstackptr only once. Signed-off-by: Andreas Herrmann Cc: akpm@linux-foundation.org Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common_64.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common_64.c b/arch/x86/kernel/cpu/common_64.c index c3afba5a81a7..4f2eeb5652ea 100644 --- a/arch/x86/kernel/cpu/common_64.c +++ b/arch/x86/kernel/cpu/common_64.c @@ -529,17 +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 + -- cgit v1.2.1 From 23952a96ae738277f3139b63d622e22984589031 Mon Sep 17 00:00:00 2001 From: Andreas Herrmann Date: Wed, 6 Aug 2008 10:29:37 +0200 Subject: x86: cpu_init(): fix memory leak when using CPU hotplug Exception stacks are allocated each time a CPU is set online. But the allocated space is never freed. Thus with one CPU hotplug offline/online cycle there is a memory leak of 24K (6 pages) for a CPU. Fix is to allocate exception stacks only once -- when the CPU is set online for the first time. Signed-off-by: Andreas Herrmann Cc: akpm@linux-foundation.org Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/common_64.c | 23 +++++++++++++---------- 1 file changed, 13 insertions(+), 10 deletions(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/cpu/common_64.c b/arch/x86/kernel/cpu/common_64.c index 4f2eeb5652ea..a11f5d4477cd 100644 --- a/arch/x86/kernel/cpu/common_64.c +++ b/arch/x86/kernel/cpu/common_64.c @@ -640,19 +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); -- cgit v1.2.1 From d6be118a97ce51ca84035270f91c2bccecbfac5f Mon Sep 17 00:00:00 2001 From: Prarit Bhargava Date: Tue, 9 Sep 2008 09:56:08 -0400 Subject: x86: fix memmap=exactmap boot argument When using kdump modifying the e820 map is yielding strange results. For example starting with BIOS-provided physical RAM map: BIOS-e820: 0000000000000100 - 0000000000093400 (usable) BIOS-e820: 0000000000093400 - 00000000000a0000 (reserved) BIOS-e820: 0000000000100000 - 000000003fee0000 (usable) BIOS-e820: 000000003fee0000 - 000000003fef3000 (ACPI data) BIOS-e820: 000000003fef3000 - 000000003ff80000 (ACPI NVS) BIOS-e820: 000000003ff80000 - 0000000040000000 (reserved) BIOS-e820: 00000000e0000000 - 00000000f0000000 (reserved) BIOS-e820: 00000000fec00000 - 00000000fec10000 (reserved) BIOS-e820: 00000000fee00000 - 00000000fee01000 (reserved) BIOS-e820: 00000000ff000000 - 0000000100000000 (reserved) and booting with args memmap=exactmap memmap=640K@0K memmap=5228K@16384K memmap=125188K@22252K memmap=76K#1047424K memmap=564K#1047500K resulted in: user-defined physical RAM map: user: 0000000000000000 - 0000000000093400 (usable) user: 0000000000093400 - 00000000000a0000 (reserved) user: 0000000000100000 - 000000003fee0000 (usable) user: 000000003fee0000 - 000000003fef3000 (ACPI data) user: 000000003fef3000 - 000000003ff80000 (ACPI NVS) user: 000000003ff80000 - 0000000040000000 (reserved) user: 00000000e0000000 - 00000000f0000000 (reserved) user: 00000000fec00000 - 00000000fec10000 (reserved) user: 00000000fee00000 - 00000000fee01000 (reserved) user: 00000000ff000000 - 0000000100000000 (reserved) But should have resulted in: user-defined physical RAM map: user: 0000000000000000 - 00000000000a0000 (usable) user: 0000000001000000 - 000000000151b000 (usable) user: 00000000015bb000 - 0000000008ffc000 (usable) user: 000000003fee0000 - 000000003ff80000 (ACPI data) This is happening because of an improper usage of strcmp() in the e820 parsing code. The strcmp() always returns !0 and never resets the value for e820.nr_map and returns an incorrect user-defined map. This patch fixes the problem. Signed-off-by: Prarit Bhargava Signed-off-by: Ingo Molnar --- arch/x86/kernel/e820.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kernel') diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index 9af89078f7bb..66e48aa2dd1b 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 -- cgit v1.2.1