From 68ca406930d6380b3be7ada5f15fcf85bfcbd552 Mon Sep 17 00:00:00 2001 From: Len Brown Date: Fri, 19 Feb 2010 00:09:22 -0500 Subject: ACPI: delete the "acpi=ht" boot option acpi=ht was important in 2003 -- before ACPI was universally deployed and enabled by default in the major Linux distributions. At that time, there were a fair number of people who or chose to, or needed to, run with acpi=off, yet also wanted access to Hyper-threading. Today we find that many invocations of "acpi=ht" are accidental, and thus is it possible that it is doing more harm than good. In 2.6.34, we warn on invocation of acpi=ht. In 2.6.35, we delete the boot option. Signed-off-by: Len Brown --- arch/ia64/include/asm/acpi.h | 1 - arch/x86/include/asm/acpi.h | 2 -- arch/x86/kernel/acpi/boot.c | 19 +++---------------- arch/x86/lguest/boot.c | 1 - 4 files changed, 3 insertions(+), 20 deletions(-) (limited to 'arch') diff --git a/arch/ia64/include/asm/acpi.h b/arch/ia64/include/asm/acpi.h index 21adbd7f90f8..837dc82a013e 100644 --- a/arch/ia64/include/asm/acpi.h +++ b/arch/ia64/include/asm/acpi.h @@ -94,7 +94,6 @@ ia64_acpi_release_global_lock (unsigned int *lock) #define acpi_noirq 0 /* ACPI always enabled on IA64 */ #define acpi_pci_disabled 0 /* ACPI PCI always enabled on IA64 */ #define acpi_strict 1 /* no ACPI spec workarounds on IA64 */ -#define acpi_ht 0 /* no HT-only mode on IA64 */ #endif #define acpi_processor_cstate_check(x) (x) /* no idle limits on IA64 :) */ static inline void disable_acpi(void) { } diff --git a/arch/x86/include/asm/acpi.h b/arch/x86/include/asm/acpi.h index 56f462cf22d2..aa2c39d968fc 100644 --- a/arch/x86/include/asm/acpi.h +++ b/arch/x86/include/asm/acpi.h @@ -85,7 +85,6 @@ extern int acpi_ioapic; extern int acpi_noirq; extern int acpi_strict; extern int acpi_disabled; -extern int acpi_ht; extern int acpi_pci_disabled; extern int acpi_skip_timer_override; extern int acpi_use_timer_override; @@ -97,7 +96,6 @@ void acpi_pic_sci_set_trigger(unsigned int, u16); static inline void disable_acpi(void) { acpi_disabled = 1; - acpi_ht = 0; acpi_pci_disabled = 1; acpi_noirq = 1; } diff --git a/arch/x86/kernel/acpi/boot.c b/arch/x86/kernel/acpi/boot.c index 7914ab0ad76e..63bcf39f8f24 100644 --- a/arch/x86/kernel/acpi/boot.c +++ b/arch/x86/kernel/acpi/boot.c @@ -62,7 +62,6 @@ EXPORT_SYMBOL(acpi_disabled); int acpi_noirq; /* skip ACPI IRQ initialization */ int acpi_pci_disabled; /* skip ACPI PCI scan and IRQ initialization */ EXPORT_SYMBOL(acpi_pci_disabled); -int acpi_ht __initdata = 1; /* enable HT */ int acpi_lapic; int acpi_ioapic; @@ -1460,9 +1459,8 @@ void __init acpi_boot_table_init(void) /* * If acpi_disabled, bail out - * One exception: acpi=ht continues far enough to enumerate LAPICs */ - if (acpi_disabled && !acpi_ht) + if (acpi_disabled) return; /* @@ -1493,9 +1491,8 @@ int __init early_acpi_boot_init(void) { /* * If acpi_disabled, bail out - * One exception: acpi=ht continues far enough to enumerate LAPICs */ - if (acpi_disabled && !acpi_ht) + if (acpi_disabled) return 1; /* @@ -1513,9 +1510,8 @@ int __init acpi_boot_init(void) /* * If acpi_disabled, bail out - * One exception: acpi=ht continues far enough to enumerate LAPICs */ - if (acpi_disabled && !acpi_ht) + if (acpi_disabled) return 1; acpi_table_parse(ACPI_SIG_BOOT, acpi_parse_sbf); @@ -1550,21 +1546,12 @@ static int __init parse_acpi(char *arg) /* acpi=force to over-ride black-list */ else if (strcmp(arg, "force") == 0) { acpi_force = 1; - acpi_ht = 1; acpi_disabled = 0; } /* acpi=strict disables out-of-spec workarounds */ else if (strcmp(arg, "strict") == 0) { acpi_strict = 1; } - /* Limit ACPI just to boot-time to enable HT */ - else if (strcmp(arg, "ht") == 0) { - if (!acpi_force) { - printk(KERN_WARNING "acpi=ht will be removed in Linux-2.6.35\n"); - disable_acpi(); - } - acpi_ht = 1; - } /* acpi=rsdt use RSDT instead of XSDT */ else if (strcmp(arg, "rsdt") == 0) { acpi_rsdt_forced = 1; diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c index 7e59dc1d3fc2..8eb9eed60282 100644 --- a/arch/x86/lguest/boot.c +++ b/arch/x86/lguest/boot.c @@ -1391,7 +1391,6 @@ __init void lguest_init(void) #endif #ifdef CONFIG_ACPI acpi_disabled = 1; - acpi_ht = 0; #endif /* -- cgit v1.2.3 From 57283776b2b821ba4d592f61cad04d0293412740 Mon Sep 17 00:00:00 2001 From: Bjorn Helgaas Date: Thu, 11 Mar 2010 12:20:11 -0700 Subject: ACPI: pci_root: pass acpi_pci_root to arch-specific scan The acpi_pci_root structure contains all the individual items (acpi_device, domain, bus number) we pass to pci_acpi_scan_root(), so just pass the single acpi_pci_root pointer directly. This will make it easier to add _CBA support later. For _CBA, we need the entire downstream bus range, not just the base bus number. We have that in the acpi_pci_root structure, so passing the pointer makes it available to the arch-specific code. Signed-off-by: Bjorn Helgaas Reviewed-by: Kenji Kaneshige Signed-off-by: Len Brown --- arch/ia64/pci/pci.c | 5 ++++- arch/x86/pci/acpi.c | 5 ++++- drivers/acpi/pci_root.c | 2 +- include/acpi/acpi_drivers.h | 3 +-- 4 files changed, 10 insertions(+), 5 deletions(-) (limited to 'arch') diff --git a/arch/ia64/pci/pci.c b/arch/ia64/pci/pci.c index 64aff520b899..aa2533ae7e9e 100644 --- a/arch/ia64/pci/pci.c +++ b/arch/ia64/pci/pci.c @@ -335,8 +335,11 @@ pcibios_setup_root_windows(struct pci_bus *bus, struct pci_controller *ctrl) } struct pci_bus * __devinit -pci_acpi_scan_root(struct acpi_device *device, int domain, int bus) +pci_acpi_scan_root(struct acpi_pci_root *root) { + struct acpi_device *device = root->device; + int domain = root->segment; + int bus = root->secondary.start; struct pci_controller *controller; unsigned int windows = 0; struct pci_bus *pbus; diff --git a/arch/x86/pci/acpi.c b/arch/x86/pci/acpi.c index e31160216efb..0b7882dbe784 100644 --- a/arch/x86/pci/acpi.c +++ b/arch/x86/pci/acpi.c @@ -229,8 +229,11 @@ res_alloc_fail: return; } -struct pci_bus * __devinit pci_acpi_scan_root(struct acpi_device *device, int domain, int busnum) +struct pci_bus * __devinit pci_acpi_scan_root(struct acpi_pci_root *root) { + struct acpi_device *device = root->device; + int domain = root->segment; + int busnum = root->secondary.start; struct pci_bus *bus; struct pci_sysdata *sd; int node; diff --git a/drivers/acpi/pci_root.c b/drivers/acpi/pci_root.c index bf476fea97ab..680450c905b3 100644 --- a/drivers/acpi/pci_root.c +++ b/drivers/acpi/pci_root.c @@ -498,7 +498,7 @@ static int __devinit acpi_pci_root_add(struct acpi_device *device) * PCI namespace does not get created until this call is made (and * thus the root bridge's pci_dev does not exist). */ - root->bus = pci_acpi_scan_root(device, segment, root->secondary.start); + root->bus = pci_acpi_scan_root(root); if (!root->bus) { printk(KERN_ERR PREFIX "Bus %04x:%02x not present in PCI namespace\n", diff --git a/include/acpi/acpi_drivers.h b/include/acpi/acpi_drivers.h index 4f7b44866b76..23d78b4d088b 100644 --- a/include/acpi/acpi_drivers.h +++ b/include/acpi/acpi_drivers.h @@ -104,8 +104,7 @@ int acpi_pci_bind_root(struct acpi_device *device); /* Arch-defined function to add a bus to the system */ -struct pci_bus *pci_acpi_scan_root(struct acpi_device *device, int domain, - int bus); +struct pci_bus *pci_acpi_scan_root(struct acpi_pci_root *root); void pci_acpi_crs_quirks(void); /* -------------------------------------------------------------------------- -- cgit v1.2.3 From b6dacf63e9fb2e7a1369843d6cef332f76fca6a3 Mon Sep 17 00:00:00 2001 From: Matthew Garrett Date: Tue, 11 May 2010 13:49:25 -0400 Subject: ACPI: Unconditionally set SCI_EN on resume The ACPI spec tells us that the firmware will reenable SCI_EN on resume. Reality disagrees in some cases. The ACPI spec tells us that the only way to set SCI_EN is via an SMM call. https://bugzilla.kernel.org/show_bug.cgi?id=13745 shows us that doing so may break machines. Tracing the ACPI calls made by Windows shows that it unconditionally sets SCI_EN on resume with a direct register write, and therefore the overwhelming probability is that everything is fine with this behaviour. Signed-off-by: Matthew Garrett Tested-by: Rafael J. Wysocki Signed-off-by: Len Brown --- arch/x86/kernel/acpi/sleep.c | 2 - drivers/acpi/sleep.c | 157 +------------------------------------------ include/linux/acpi.h | 1 - 3 files changed, 2 insertions(+), 158 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/acpi/sleep.c b/arch/x86/kernel/acpi/sleep.c index f9961034e557..82e508677b91 100644 --- a/arch/x86/kernel/acpi/sleep.c +++ b/arch/x86/kernel/acpi/sleep.c @@ -162,8 +162,6 @@ static int __init acpi_sleep_setup(char *str) #endif if (strncmp(str, "old_ordering", 12) == 0) acpi_old_suspend_ordering(); - if (strncmp(str, "sci_force_enable", 16) == 0) - acpi_set_sci_en_on_resume(); str = strchr(str, ','); if (str != NULL) str += strspn(str, ", \t"); diff --git a/drivers/acpi/sleep.c b/drivers/acpi/sleep.c index baa76bbf244a..4ab2275b4461 100644 --- a/drivers/acpi/sleep.c +++ b/drivers/acpi/sleep.c @@ -80,22 +80,6 @@ static int acpi_sleep_prepare(u32 acpi_state) #ifdef CONFIG_ACPI_SLEEP static u32 acpi_target_sleep_state = ACPI_STATE_S0; -/* - * According to the ACPI specification the BIOS should make sure that ACPI is - * enabled and SCI_EN bit is set on wake-up from S1 - S3 sleep states. Still, - * some BIOSes don't do that and therefore we use acpi_enable() to enable ACPI - * on such systems during resume. Unfortunately that doesn't help in - * particularly pathological cases in which SCI_EN has to be set directly on - * resume, although the specification states very clearly that this flag is - * owned by the hardware. The set_sci_en_on_resume variable will be set in such - * cases. - */ -static bool set_sci_en_on_resume; - -void __init acpi_set_sci_en_on_resume(void) -{ - set_sci_en_on_resume = true; -} /* * ACPI 1.0 wants us to execute _PTS before suspending devices, so we allow the @@ -253,11 +237,8 @@ static int acpi_suspend_enter(suspend_state_t pm_state) break; } - /* If ACPI is not enabled by the BIOS, we need to enable it here. */ - if (set_sci_en_on_resume) - acpi_write_bit_register(ACPI_BITREG_SCI_ENABLE, 1); - else - acpi_enable(); + /* This violates the spec but is required for bug compatibility. */ + acpi_write_bit_register(ACPI_BITREG_SCI_ENABLE, 1); /* Reprogram control registers and execute _BFS */ acpi_leave_sleep_state_prep(acpi_state); @@ -346,12 +327,6 @@ static int __init init_old_suspend_ordering(const struct dmi_system_id *d) return 0; } -static int __init init_set_sci_en_on_resume(const struct dmi_system_id *d) -{ - set_sci_en_on_resume = true; - return 0; -} - static struct dmi_system_id __initdata acpisleep_dmi_table[] = { { .callback = init_old_suspend_ordering, @@ -370,22 +345,6 @@ static struct dmi_system_id __initdata acpisleep_dmi_table[] = { }, }, { - .callback = init_set_sci_en_on_resume, - .ident = "Apple MacBook 1,1", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Apple Computer, Inc."), - DMI_MATCH(DMI_PRODUCT_NAME, "MacBook1,1"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Apple MacMini 1,1", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Apple Computer, Inc."), - DMI_MATCH(DMI_PRODUCT_NAME, "Macmini1,1"), - }, - }, - { .callback = init_old_suspend_ordering, .ident = "Asus Pundit P1-AH2 (M2N8L motherboard)", .matches = { @@ -394,94 +353,6 @@ static struct dmi_system_id __initdata acpisleep_dmi_table[] = { }, }, { - .callback = init_set_sci_en_on_resume, - .ident = "Toshiba Satellite L300", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "TOSHIBA"), - DMI_MATCH(DMI_PRODUCT_NAME, "Satellite L300"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Hewlett-Packard HP G7000 Notebook PC", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"), - DMI_MATCH(DMI_PRODUCT_NAME, "HP G7000 Notebook PC"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Hewlett-Packard HP Pavilion dv3 Notebook PC", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"), - DMI_MATCH(DMI_PRODUCT_NAME, "HP Pavilion dv3 Notebook PC"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Hewlett-Packard Pavilion dv4", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"), - DMI_MATCH(DMI_PRODUCT_NAME, "HP Pavilion dv4"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Hewlett-Packard Pavilion dv7", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"), - DMI_MATCH(DMI_PRODUCT_NAME, "HP Pavilion dv7"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Hewlett-Packard Compaq Presario C700 Notebook PC", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"), - DMI_MATCH(DMI_PRODUCT_NAME, "Compaq Presario C700 Notebook PC"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Hewlett-Packard Compaq Presario CQ40 Notebook PC", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Hewlett-Packard"), - DMI_MATCH(DMI_PRODUCT_NAME, "Compaq Presario CQ40 Notebook PC"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Lenovo ThinkPad T410", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"), - DMI_MATCH(DMI_PRODUCT_VERSION, "ThinkPad T410"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Lenovo ThinkPad T510", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"), - DMI_MATCH(DMI_PRODUCT_VERSION, "ThinkPad T510"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Lenovo ThinkPad W510", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"), - DMI_MATCH(DMI_PRODUCT_VERSION, "ThinkPad W510"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Lenovo ThinkPad X201[s]", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "LENOVO"), - DMI_MATCH(DMI_PRODUCT_VERSION, "ThinkPad X201"), - }, - }, - { .callback = init_old_suspend_ordering, .ident = "Panasonic CF51-2L", .matches = { @@ -490,30 +361,6 @@ static struct dmi_system_id __initdata acpisleep_dmi_table[] = { DMI_MATCH(DMI_BOARD_NAME, "CF51-2L"), }, }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Dell Studio 1558", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), - DMI_MATCH(DMI_PRODUCT_NAME, "Studio 1558"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Dell Studio 1557", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), - DMI_MATCH(DMI_PRODUCT_NAME, "Studio 1557"), - }, - }, - { - .callback = init_set_sci_en_on_resume, - .ident = "Dell Studio 1555", - .matches = { - DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."), - DMI_MATCH(DMI_PRODUCT_NAME, "Studio 1555"), - }, - }, {}, }; #endif /* CONFIG_SUSPEND */ diff --git a/include/linux/acpi.h b/include/linux/acpi.h index b926afe8c03e..87ca4913294c 100644 --- a/include/linux/acpi.h +++ b/include/linux/acpi.h @@ -251,7 +251,6 @@ int acpi_check_mem_region(resource_size_t start, resource_size_t n, void __init acpi_no_s4_hw_signature(void); void __init acpi_old_suspend_ordering(void); void __init acpi_s4_no_nvs(void); -void __init acpi_set_sci_en_on_resume(void); #endif /* CONFIG_PM_SLEEP */ struct acpi_osc_context { -- cgit v1.2.3 From d334a49113a4a33109fd24e46073280ecd1bea0d Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Tue, 18 May 2010 14:35:20 +0800 Subject: ACPI, APEI, Generic Hardware Error Source memory error support Generic Hardware Error Source provides a way to report platform hardware errors (such as that from chipset). It works in so called "Firmware First" mode, that is, hardware errors are reported to firmware firstly, then reported to Linux by firmware. This way, some non-standard hardware error registers or non-standard hardware link can be checked by firmware to produce more valuable hardware error information for Linux. Now, only SCI notification type and memory errors are supported. More notification type and hardware error type will be added later. These memory errors are reported to user space through /dev/mcelog via faking a corrected Machine Check, so that the error memory page can be offlined by /sbin/mcelog if the error count for one page is beyond the threshold. On some machines, Machine Check can not report physical address for some corrected memory errors, but GHES can do that. So this simplified GHES is implemented firstly. Signed-off-by: Huang Ying Signed-off-by: Andi Kleen Signed-off-by: Len Brown --- arch/x86/include/asm/mce.h | 8 + arch/x86/kernel/cpu/mcheck/Makefile | 2 + arch/x86/kernel/cpu/mcheck/mce-apei.c | 52 +++++ drivers/acpi/apei/Kconfig | 14 ++ drivers/acpi/apei/Makefile | 1 + drivers/acpi/apei/ghes.c | 427 ++++++++++++++++++++++++++++++++++ 6 files changed, 504 insertions(+) create mode 100644 arch/x86/kernel/cpu/mcheck/mce-apei.c create mode 100644 drivers/acpi/apei/ghes.c (limited to 'arch') diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h index 6c3fdd631ed3..f32a4301c4d4 100644 --- a/arch/x86/include/asm/mce.h +++ b/arch/x86/include/asm/mce.h @@ -225,5 +225,13 @@ extern void mcheck_intel_therm_init(void); static inline void mcheck_intel_therm_init(void) { } #endif +/* + * Used by APEI to report memory error via /dev/mcelog + */ + +struct cper_sec_mem_err; +extern void apei_mce_report_mem_error(int corrected, + struct cper_sec_mem_err *mem_err); + #endif /* __KERNEL__ */ #endif /* _ASM_X86_MCE_H */ diff --git a/arch/x86/kernel/cpu/mcheck/Makefile b/arch/x86/kernel/cpu/mcheck/Makefile index 4ac6d48fe11b..bb34b03af252 100644 --- a/arch/x86/kernel/cpu/mcheck/Makefile +++ b/arch/x86/kernel/cpu/mcheck/Makefile @@ -7,3 +7,5 @@ obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o obj-$(CONFIG_X86_MCE_INJECT) += mce-inject.o obj-$(CONFIG_X86_THERMAL_VECTOR) += therm_throt.o + +obj-$(CONFIG_ACPI_APEI) += mce-apei.o diff --git a/arch/x86/kernel/cpu/mcheck/mce-apei.c b/arch/x86/kernel/cpu/mcheck/mce-apei.c new file mode 100644 index 000000000000..4eccd1fadb14 --- /dev/null +++ b/arch/x86/kernel/cpu/mcheck/mce-apei.c @@ -0,0 +1,52 @@ +/* + * Bridge between MCE and APEI + * + * On some machine, corrected memory errors are reported via APEI + * generic hardware error source (GHES) instead of corrected Machine + * Check. These corrected memory errors can be reported to user space + * through /dev/mcelog via faking a corrected Machine Check, so that + * the error memory page can be offlined by /sbin/mcelog if the error + * count for one page is beyond the threshold. + * + * Copyright 2010 Intel Corp. + * Author: Huang Ying + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include +#include +#include + +#include "mce-internal.h" + +void apei_mce_report_mem_error(int corrected, struct cper_sec_mem_err *mem_err) +{ + struct mce m; + + /* Only corrected MC is reported */ + if (!corrected) + return; + + mce_setup(&m); + m.bank = 1; + /* Fake a memory read corrected error with unknown channel */ + m.status = MCI_STATUS_VAL | MCI_STATUS_EN | MCI_STATUS_ADDRV | 0x9f; + m.addr = mem_err->physical_addr; + mce_log(&m); + mce_notify_irq(); +} +EXPORT_SYMBOL_GPL(apei_mce_report_mem_error); diff --git a/drivers/acpi/apei/Kconfig b/drivers/acpi/apei/Kconfig index 5f0a41c2bc62..f8c668f27b5a 100644 --- a/drivers/acpi/apei/Kconfig +++ b/drivers/acpi/apei/Kconfig @@ -7,6 +7,20 @@ config ACPI_APEI especially. In addition it supports error serialization and error injection. +config ACPI_APEI_GHES + tristate "APEI Generic Hardware Error Source" + depends on ACPI_APEI && X86 + select ACPI_HED + help + Generic Hardware Error Source provides a way to report + platform hardware errors (such as that from chipset). It + works in so called "Firmware First" mode, that is, hardware + errors are reported to firmware firstly, then reported to + Linux by firmware. This way, some non-standard hardware + error registers or non-standard hardware link can be checked + by firmware to produce more valuable hardware error + information for Linux. + config ACPI_APEI_EINJ tristate "APEI Error INJection (EINJ)" depends on ACPI_APEI && DEBUG_FS diff --git a/drivers/acpi/apei/Makefile b/drivers/acpi/apei/Makefile index fef963ec5362..41c61db4c51c 100644 --- a/drivers/acpi/apei/Makefile +++ b/drivers/acpi/apei/Makefile @@ -1,4 +1,5 @@ obj-$(CONFIG_ACPI_APEI) += apei.o +obj-$(CONFIG_ACPI_APEI_GHES) += ghes.o obj-$(CONFIG_ACPI_APEI_EINJ) += einj.o apei-y := apei-base.o hest.o cper.o diff --git a/drivers/acpi/apei/ghes.c b/drivers/acpi/apei/ghes.c new file mode 100644 index 000000000000..fd0cc016a099 --- /dev/null +++ b/drivers/acpi/apei/ghes.c @@ -0,0 +1,427 @@ +/* + * APEI Generic Hardware Error Source support + * + * Generic Hardware Error Source provides a way to report platform + * hardware errors (such as that from chipset). It works in so called + * "Firmware First" mode, that is, hardware errors are reported to + * firmware firstly, then reported to Linux by firmware. This way, + * some non-standard hardware error registers or non-standard hardware + * link can be checked by firmware to produce more hardware error + * information for Linux. + * + * For more information about Generic Hardware Error Source, please + * refer to ACPI Specification version 4.0, section 17.3.2.6 + * + * Now, only SCI notification type and memory errors are + * supported. More notification type and hardware error type will be + * added later. + * + * Copyright 2010 Intel Corp. + * Author: Huang Ying + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License version + * 2 as published by the Free Software Foundation; + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "apei-internal.h" + +#define GHES_PFX "GHES: " + +#define GHES_ESTATUS_MAX_SIZE 65536 + +/* + * One struct ghes is created for each generic hardware error + * source. + * + * It provides the context for APEI hardware error timer/IRQ/SCI/NMI + * handler. Handler for one generic hardware error source is only + * triggered after the previous one is done. So handler can uses + * struct ghes without locking. + * + * estatus: memory buffer for error status block, allocated during + * HEST parsing. + */ +#define GHES_TO_CLEAR 0x0001 + +struct ghes { + struct acpi_hest_generic *generic; + struct acpi_hest_generic_status *estatus; + struct list_head list; + u64 buffer_paddr; + unsigned long flags; +}; + +/* + * Error source lists, one list for each notification method. The + * members in lists are struct ghes. + * + * The list members are only added in HEST parsing and deleted during + * module_exit, that is, single-threaded. So no lock is needed for + * that. + * + * But the mutual exclusion is needed between members adding/deleting + * and timer/IRQ/SCI/NMI handler, which may traverse the list. RCU is + * used for that. + */ +static LIST_HEAD(ghes_sci); + +static struct ghes *ghes_new(struct acpi_hest_generic *generic) +{ + struct ghes *ghes; + unsigned int error_block_length; + int rc; + + ghes = kzalloc(sizeof(*ghes), GFP_KERNEL); + if (!ghes) + return ERR_PTR(-ENOMEM); + ghes->generic = generic; + INIT_LIST_HEAD(&ghes->list); + rc = acpi_pre_map_gar(&generic->error_status_address); + if (rc) + goto err_free; + error_block_length = generic->error_block_length; + if (error_block_length > GHES_ESTATUS_MAX_SIZE) { + pr_warning(FW_WARN GHES_PFX + "Error status block length is too long: %u for " + "generic hardware error source: %d.\n", + error_block_length, generic->header.source_id); + error_block_length = GHES_ESTATUS_MAX_SIZE; + } + ghes->estatus = kmalloc(error_block_length, GFP_KERNEL); + if (!ghes->estatus) { + rc = -ENOMEM; + goto err_unmap; + } + + return ghes; + +err_unmap: + acpi_post_unmap_gar(&generic->error_status_address); +err_free: + kfree(ghes); + return ERR_PTR(rc); +} + +static void ghes_fini(struct ghes *ghes) +{ + kfree(ghes->estatus); + acpi_post_unmap_gar(&ghes->generic->error_status_address); +} + +enum { + GHES_SER_NO = 0x0, + GHES_SER_CORRECTED = 0x1, + GHES_SER_RECOVERABLE = 0x2, + GHES_SER_PANIC = 0x3, +}; + +static inline int ghes_severity(int severity) +{ + switch (severity) { + case CPER_SER_INFORMATIONAL: + return GHES_SER_NO; + case CPER_SER_CORRECTED: + return GHES_SER_CORRECTED; + case CPER_SER_RECOVERABLE: + return GHES_SER_RECOVERABLE; + case CPER_SER_FATAL: + return GHES_SER_PANIC; + default: + /* Unkown, go panic */ + return GHES_SER_PANIC; + } +} + +/* SCI handler run in work queue, so ioremap can be used here */ +static int ghes_copy_tofrom_phys(void *buffer, u64 paddr, u32 len, + int from_phys) +{ + void *vaddr; + + vaddr = ioremap_cache(paddr, len); + if (!vaddr) + return -ENOMEM; + if (from_phys) + memcpy(buffer, vaddr, len); + else + memcpy(vaddr, buffer, len); + iounmap(vaddr); + + return 0; +} + +static int ghes_read_estatus(struct ghes *ghes, int silent) +{ + struct acpi_hest_generic *g = ghes->generic; + u64 buf_paddr; + u32 len; + int rc; + + rc = acpi_atomic_read(&buf_paddr, &g->error_status_address); + if (rc) { + if (!silent && printk_ratelimit()) + pr_warning(FW_WARN GHES_PFX +"Failed to read error status block address for hardware error source: %d.\n", + g->header.source_id); + return -EIO; + } + if (!buf_paddr) + return -ENOENT; + + rc = ghes_copy_tofrom_phys(ghes->estatus, buf_paddr, + sizeof(*ghes->estatus), 1); + if (rc) + return rc; + if (!ghes->estatus->block_status) + return -ENOENT; + + ghes->buffer_paddr = buf_paddr; + ghes->flags |= GHES_TO_CLEAR; + + rc = -EIO; + len = apei_estatus_len(ghes->estatus); + if (len < sizeof(*ghes->estatus)) + goto err_read_block; + if (len > ghes->generic->error_block_length) + goto err_read_block; + if (apei_estatus_check_header(ghes->estatus)) + goto err_read_block; + rc = ghes_copy_tofrom_phys(ghes->estatus + 1, + buf_paddr + sizeof(*ghes->estatus), + len - sizeof(*ghes->estatus), 1); + if (rc) + return rc; + if (apei_estatus_check(ghes->estatus)) + goto err_read_block; + rc = 0; + +err_read_block: + if (rc && !silent) + pr_warning(FW_WARN GHES_PFX + "Failed to read error status block!\n"); + return rc; +} + +static void ghes_clear_estatus(struct ghes *ghes) +{ + ghes->estatus->block_status = 0; + if (!(ghes->flags & GHES_TO_CLEAR)) + return; + ghes_copy_tofrom_phys(ghes->estatus, ghes->buffer_paddr, + sizeof(ghes->estatus->block_status), 0); + ghes->flags &= ~GHES_TO_CLEAR; +} + +static void ghes_do_proc(struct ghes *ghes) +{ + int ser, processed = 0; + struct acpi_hest_generic_data *gdata; + + ser = ghes_severity(ghes->estatus->error_severity); + apei_estatus_for_each_section(ghes->estatus, gdata) { +#ifdef CONFIG_X86_MCE + if (!uuid_le_cmp(*(uuid_le *)gdata->section_type, + CPER_SEC_PLATFORM_MEM)) { + apei_mce_report_mem_error( + ser == GHES_SER_CORRECTED, + (struct cper_sec_mem_err *)(gdata+1)); + processed = 1; + } +#endif + } + + if (!processed && printk_ratelimit()) + pr_warning(GHES_PFX + "Unknown error record from generic hardware error source: %d\n", + ghes->generic->header.source_id); +} + +static int ghes_proc(struct ghes *ghes) +{ + int rc; + + rc = ghes_read_estatus(ghes, 0); + if (rc) + goto out; + ghes_do_proc(ghes); + +out: + ghes_clear_estatus(ghes); + return 0; +} + +static int ghes_notify_sci(struct notifier_block *this, + unsigned long event, void *data) +{ + struct ghes *ghes; + int ret = NOTIFY_DONE; + + rcu_read_lock(); + list_for_each_entry_rcu(ghes, &ghes_sci, list) { + if (!ghes_proc(ghes)) + ret = NOTIFY_OK; + } + rcu_read_unlock(); + + return ret; +} + +static struct notifier_block ghes_notifier_sci = { + .notifier_call = ghes_notify_sci, +}; + +static int hest_ghes_parse(struct acpi_hest_header *hest_hdr, void *data) +{ + struct acpi_hest_generic *generic; + struct ghes *ghes = NULL; + int rc = 0; + + if (hest_hdr->type != ACPI_HEST_TYPE_GENERIC_ERROR) + return 0; + + generic = (struct acpi_hest_generic *)hest_hdr; + if (!generic->enabled) + return 0; + + if (generic->error_block_length < + sizeof(struct acpi_hest_generic_status)) { + pr_warning(FW_BUG GHES_PFX +"Invalid error block length: %u for generic hardware error source: %d\n", + generic->error_block_length, + generic->header.source_id); + goto err; + } + if (generic->records_to_preallocate == 0) { + pr_warning(FW_BUG GHES_PFX +"Invalid records to preallocate: %u for generic hardware error source: %d\n", + generic->records_to_preallocate, + generic->header.source_id); + goto err; + } + ghes = ghes_new(generic); + if (IS_ERR(ghes)) { + rc = PTR_ERR(ghes); + ghes = NULL; + goto err; + } + switch (generic->notify.type) { + case ACPI_HEST_NOTIFY_POLLED: + pr_warning(GHES_PFX +"Generic hardware error source: %d notified via POLL is not supported!\n", + generic->header.source_id); + break; + case ACPI_HEST_NOTIFY_EXTERNAL: + case ACPI_HEST_NOTIFY_LOCAL: + pr_warning(GHES_PFX +"Generic hardware error source: %d notified via IRQ is not supported!\n", + generic->header.source_id); + break; + case ACPI_HEST_NOTIFY_SCI: + if (list_empty(&ghes_sci)) + register_acpi_hed_notifier(&ghes_notifier_sci); + list_add_rcu(&ghes->list, &ghes_sci); + break; + case ACPI_HEST_NOTIFY_NMI: + pr_warning(GHES_PFX +"Generic hardware error source: %d notified via NMI is not supported!\n", + generic->header.source_id); + break; + default: + pr_warning(FW_WARN GHES_PFX + "Unknown notification type: %u for generic hardware error source: %d\n", + generic->notify.type, generic->header.source_id); + break; + } + + return 0; +err: + if (ghes) + ghes_fini(ghes); + return rc; +} + +static void ghes_cleanup(void) +{ + struct ghes *ghes, *nghes; + + if (!list_empty(&ghes_sci)) + unregister_acpi_hed_notifier(&ghes_notifier_sci); + + synchronize_rcu(); + + list_for_each_entry_safe(ghes, nghes, &ghes_sci, list) { + list_del(&ghes->list); + ghes_fini(ghes); + kfree(ghes); + } +} + +static int __init ghes_init(void) +{ + int rc; + + if (acpi_disabled) + return -ENODEV; + + if (hest_disable) { + pr_info(GHES_PFX "HEST is not enabled!\n"); + return -EINVAL; + } + + rc = apei_hest_parse(hest_ghes_parse, NULL); + if (rc) { + pr_err(GHES_PFX + "Error during parsing HEST generic hardware error sources.\n"); + goto err_cleanup; + } + + if (list_empty(&ghes_sci)) { + pr_info(GHES_PFX + "No functional generic hardware error sources.\n"); + rc = -ENODEV; + goto err_cleanup; + } + + pr_info(GHES_PFX + "Generic Hardware Error Source support is initialized.\n"); + + return 0; +err_cleanup: + ghes_cleanup(); + return rc; +} + +static void __exit ghes_exit(void) +{ + ghes_cleanup(); +} + +module_init(ghes_init); +module_exit(ghes_exit); + +MODULE_AUTHOR("Huang Ying"); +MODULE_DESCRIPTION("APEI Generic Hardware Error Source support"); +MODULE_LICENSE("GPL"); -- cgit v1.2.3 From 482908b49ebfa453dd0455910c951c750567c05d Mon Sep 17 00:00:00 2001 From: Huang Ying Date: Tue, 18 May 2010 14:35:22 +0800 Subject: ACPI, APEI, Use ERST for persistent storage of MCE Traditionally, fatal MCE will cause Linux print error log to console then reboot. Because MCE registers will preserve their content after warm reboot, the hardware error can be logged to disk or network after reboot. But system may fail to warm reboot, then you may lose the hardware error log. ERST can help here. Through saving the hardware error log into flash via ERST before go panic, the hardware error log can be gotten from the flash after system boot successful again. The fatal MCE processing procedure with ERST involved is as follow: - Hardware detect error, MCE raised - MCE read MCE registers, check error severity (fatal), prepare error record - Write MCE error record into flash via ERST - Go panic, then trigger system reboot - System reboot, /sbin/mcelog run, it reads /dev/mcelog to check flash for error record of previous boot via ERST, and output and clear them if available - /sbin/mcelog logs error records into disk or network ERST only accepts CPER record format, but there is no pre-defined CPER section can accommodate all information in struct mce, so a customized section type is defined to hold struct mce inside a CPER record as an error section. Signed-off-by: Huang Ying Signed-off-by: Andi Kleen Signed-off-by: Len Brown --- arch/x86/kernel/cpu/mcheck/mce-apei.c | 86 +++++++++++++++++++++++++++++++ arch/x86/kernel/cpu/mcheck/mce-internal.h | 23 +++++++++ arch/x86/kernel/cpu/mcheck/mce.c | 79 ++++++++++++++++++++++++---- 3 files changed, 177 insertions(+), 11 deletions(-) (limited to 'arch') diff --git a/arch/x86/kernel/cpu/mcheck/mce-apei.c b/arch/x86/kernel/cpu/mcheck/mce-apei.c index 4eccd1fadb14..745b54f9be89 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-apei.c +++ b/arch/x86/kernel/cpu/mcheck/mce-apei.c @@ -8,6 +8,9 @@ * the error memory page can be offlined by /sbin/mcelog if the error * count for one page is beyond the threshold. * + * For fatal MCE, save MCE record into persistent storage via ERST, so + * that the MCE record can be logged after reboot via ERST. + * * Copyright 2010 Intel Corp. * Author: Huang Ying * @@ -50,3 +53,86 @@ void apei_mce_report_mem_error(int corrected, struct cper_sec_mem_err *mem_err) mce_notify_irq(); } EXPORT_SYMBOL_GPL(apei_mce_report_mem_error); + +#define CPER_CREATOR_MCE \ + UUID_LE(0x75a574e3, 0x5052, 0x4b29, 0x8a, 0x8e, 0xbe, 0x2c, \ + 0x64, 0x90, 0xb8, 0x9d) +#define CPER_SECTION_TYPE_MCE \ + UUID_LE(0xfe08ffbe, 0x95e4, 0x4be7, 0xbc, 0x73, 0x40, 0x96, \ + 0x04, 0x4a, 0x38, 0xfc) + +/* + * CPER specification (in UEFI specification 2.3 appendix N) requires + * byte-packed. + */ +struct cper_mce_record { + struct cper_record_header hdr; + struct cper_section_descriptor sec_hdr; + struct mce mce; +} __packed; + +int apei_write_mce(struct mce *m) +{ + struct cper_mce_record rcd; + + memset(&rcd, 0, sizeof(rcd)); + memcpy(rcd.hdr.signature, CPER_SIG_RECORD, CPER_SIG_SIZE); + rcd.hdr.revision = CPER_RECORD_REV; + rcd.hdr.signature_end = CPER_SIG_END; + rcd.hdr.section_count = 1; + rcd.hdr.error_severity = CPER_SER_FATAL; + /* timestamp, platform_id, partition_id are all invalid */ + rcd.hdr.validation_bits = 0; + rcd.hdr.record_length = sizeof(rcd); + rcd.hdr.creator_id = CPER_CREATOR_MCE; + rcd.hdr.notification_type = CPER_NOTIFY_MCE; + rcd.hdr.record_id = cper_next_record_id(); + rcd.hdr.flags = CPER_HW_ERROR_FLAGS_PREVERR; + + rcd.sec_hdr.section_offset = (void *)&rcd.mce - (void *)&rcd; + rcd.sec_hdr.section_length = sizeof(rcd.mce); + rcd.sec_hdr.revision = CPER_SEC_REV; + /* fru_id and fru_text is invalid */ + rcd.sec_hdr.validation_bits = 0; + rcd.sec_hdr.flags = CPER_SEC_PRIMARY; + rcd.sec_hdr.section_type = CPER_SECTION_TYPE_MCE; + rcd.sec_hdr.section_severity = CPER_SER_FATAL; + + memcpy(&rcd.mce, m, sizeof(*m)); + + return erst_write(&rcd.hdr); +} + +ssize_t apei_read_mce(struct mce *m, u64 *record_id) +{ + struct cper_mce_record rcd; + ssize_t len; + + len = erst_read_next(&rcd.hdr, sizeof(rcd)); + if (len <= 0) + return len; + /* Can not skip other records in storage via ERST unless clear them */ + else if (len != sizeof(rcd) || + uuid_le_cmp(rcd.hdr.creator_id, CPER_CREATOR_MCE)) { + if (printk_ratelimit()) + pr_warning( + "MCE-APEI: Can not skip the unknown record in ERST"); + return -EIO; + } + + memcpy(m, &rcd.mce, sizeof(*m)); + *record_id = rcd.hdr.record_id; + + return sizeof(*m); +} + +/* Check whether there is record in ERST */ +int apei_check_mce(void) +{ + return erst_get_record_count(); +} + +int apei_clear_mce(u64 record_id) +{ + return erst_clear(record_id); +} diff --git a/arch/x86/kernel/cpu/mcheck/mce-internal.h b/arch/x86/kernel/cpu/mcheck/mce-internal.h index 32996f9fab67..fefcc69ee8b5 100644 --- a/arch/x86/kernel/cpu/mcheck/mce-internal.h +++ b/arch/x86/kernel/cpu/mcheck/mce-internal.h @@ -28,3 +28,26 @@ extern int mce_ser; extern struct mce_bank *mce_banks; +#ifdef CONFIG_ACPI_APEI +int apei_write_mce(struct mce *m); +ssize_t apei_read_mce(struct mce *m, u64 *record_id); +int apei_check_mce(void); +int apei_clear_mce(u64 record_id); +#else +static inline int apei_write_mce(struct mce *m) +{ + return -EINVAL; +} +static inline ssize_t apei_read_mce(struct mce *m, u64 *record_id) +{ + return 0; +} +static inline int apei_check_mce(void) +{ + return 0; +} +static inline int apei_clear_mce(u64 record_id) +{ + return -EINVAL; +} +#endif diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c index 8a6f0afa767e..09535ca9b9d7 100644 --- a/arch/x86/kernel/cpu/mcheck/mce.c +++ b/arch/x86/kernel/cpu/mcheck/mce.c @@ -264,7 +264,7 @@ static void wait_for_panic(void) static void mce_panic(char *msg, struct mce *final, char *exp) { - int i; + int i, apei_err = 0; if (!fake_panic) { /* @@ -287,8 +287,11 @@ static void mce_panic(char *msg, struct mce *final, char *exp) struct mce *m = &mcelog.entry[i]; if (!(m->status & MCI_STATUS_VAL)) continue; - if (!(m->status & MCI_STATUS_UC)) + if (!(m->status & MCI_STATUS_UC)) { print_mce(m); + if (!apei_err) + apei_err = apei_write_mce(m); + } } /* Now print uncorrected but with the final one last */ for (i = 0; i < MCE_LOG_LEN; i++) { @@ -297,11 +300,17 @@ static void mce_panic(char *msg, struct mce *final, char *exp) continue; if (!(m->status & MCI_STATUS_UC)) continue; - if (!final || memcmp(m, final, sizeof(struct mce))) + if (!final || memcmp(m, final, sizeof(struct mce))) { print_mce(m); + if (!apei_err) + apei_err = apei_write_mce(m); + } } - if (final) + if (final) { print_mce(final); + if (!apei_err) + apei_err = apei_write_mce(final); + } if (cpu_missing) printk(KERN_EMERG "Some CPUs didn't answer in synchronization\n"); print_mce_tail(); @@ -1493,6 +1502,43 @@ static void collect_tscs(void *data) rdtscll(cpu_tsc[smp_processor_id()]); } +static int mce_apei_read_done; + +/* Collect MCE record of previous boot in persistent storage via APEI ERST. */ +static int __mce_read_apei(char __user **ubuf, size_t usize) +{ + int rc; + u64 record_id; + struct mce m; + + if (usize < sizeof(struct mce)) + return -EINVAL; + + rc = apei_read_mce(&m, &record_id); + /* Error or no more MCE record */ + if (rc <= 0) { + mce_apei_read_done = 1; + return rc; + } + rc = -EFAULT; + if (copy_to_user(*ubuf, &m, sizeof(struct mce))) + return rc; + /* + * In fact, we should have cleared the record after that has + * been flushed to the disk or sent to network in + * /sbin/mcelog, but we have no interface to support that now, + * so just clear it to avoid duplication. + */ + rc = apei_clear_mce(record_id); + if (rc) { + mce_apei_read_done = 1; + return rc; + } + *ubuf += sizeof(struct mce); + + return 0; +} + static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize, loff_t *off) { @@ -1506,15 +1552,19 @@ static ssize_t mce_read(struct file *filp, char __user *ubuf, size_t usize, return -ENOMEM; mutex_lock(&mce_read_mutex); + + if (!mce_apei_read_done) { + err = __mce_read_apei(&buf, usize); + if (err || buf != ubuf) + goto out; + } + next = rcu_dereference_check_mce(mcelog.next); /* Only supports full reads right now */ - if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce)) { - mutex_unlock(&mce_read_mutex); - kfree(cpu_tsc); - - return -EINVAL; - } + err = -EINVAL; + if (*off != 0 || usize < MCE_LOG_LEN*sizeof(struct mce)) + goto out; err = 0; prev = 0; @@ -1562,10 +1612,15 @@ timeout: memset(&mcelog.entry[i], 0, sizeof(struct mce)); } } + + if (err) + err = -EFAULT; + +out: mutex_unlock(&mce_read_mutex); kfree(cpu_tsc); - return err ? -EFAULT : buf - ubuf; + return err ? err : buf - ubuf; } static unsigned int mce_poll(struct file *file, poll_table *wait) @@ -1573,6 +1628,8 @@ static unsigned int mce_poll(struct file *file, poll_table *wait) poll_wait(file, &mce_wait, wait); if (rcu_dereference_check_mce(mcelog.next)) return POLLIN | POLLRDNORM; + if (!mce_apei_read_done && apei_check_mce()) + return POLLIN | POLLRDNORM; return 0; } -- cgit v1.2.3