x86/ras: Move AMD MCE injection module to arch/x86/
Subsystems (4)
Borislav Petkov
maintainercover x86/ras: Move AMD MCE injection module to arch/x86/
From: Borislav Petkov <[email protected]>
Hi all,
so this has been on my mind for a while now and the recent pull request sdebacle imply accelerated it. The advantage of moving it closer to MCE code is that improvements/additions to the MCE code can automatically get injection functionality added for testing.
With that in place, injection of threshold and deferred errors will get added.
Still RFC as there's more RAS stuff pending review currently.
Thanks.
Borislav Petkov (4):
RAS: Add a menuconfig option with descriptive text x86/mce: Rename rcu_dereference_check_mce() to mce_log_get_idx_check() x86/mce: Add a wrapper around mce_log() for injection x86/ras: Move AMD MCE injector to arch/x86/ras/
arch/x86/Makefile | 2 ++
arch/x86/kernel/cpu/mcheck/mce-internal.h | 2 ++
arch/x86/kernel/cpu/mcheck/mce.c | 15 ++++++++---
arch/x86/ras/Kconfig | 11 +++++++++
arch/x86/ras/Makefile | 2 ++
{drivers/edac => arch/x86/ras}/mce_amd_inj.c | 6 ++---
drivers/edac/Kconfig | 10 --------
drivers/edac/Makefile | 1 -
drivers/ras/Kconfig | 37 ++++++++++++++++++++++++++--
9 files changed, 66 insertions(+), 20 deletions(-)
create mode 100644 arch/x86/ras/Kconfig
create mode 100644 arch/x86/ras/Makefile
rename {drivers/edac => arch/x86/ras}/mce_amd_inj.c (98%)Signature · 1 line
2.5.0.rc2.28.g6003e7f
Borislav Petkov
maintainerRFC PATCH 1/4 RAS: Add a menuconfig option with descriptive text
From: Borislav Petkov <[email protected]>
Text taken a previous patch from "Gong Chen" <[email protected]>.
--- drivers/ras/Kconfig | 34 ++++++++++++++++++++++++++++++++-- 1 file changed, 32 insertions(+), 2 deletions(-)
drivers/ras/Kconfig +32 -2
@@ -1,2 +1,32 @@ -config RAS - bool +menuconfig RAS + bool "Reliability, Availability, Serviceability features" + help + Reliability, availability, and serviceability (RAS) is a computer + hardware engineering term. Computers designed with higher levels + of RAS have a multitude of features that protect data integrity + and help them stay available for long periods of time without + failure. + + Reliability can be defined as the probability that the system will + produce correct outputs up to some given time. Reliability is + enhanced by features that help to avoid, detect and repair hardware + faults. + + Availability is the probability a system is operational at a given + time, i.e. the amount of time a device is actually operating as the + percentage of total time it should be operating. + + Serviceability or maintainability is the simplicity and speed with + which a system can be repaired or maintained; if the time to repair + a failed system increases, then availability will decrease. + + Note that Reliability and Availability are distinct concepts: + Reliability is a measure of the ability of a system to function + correctly, including avoiding data corruption, whereas Availability + measures how often it is available for use, even though it may not + be functioning correctly. For example, a server may run forever and + so have ideal availability, but may be unreliable, with frequent + data corruption. + +if RAS +endif
Signature · 1 line
2.5.0.rc2.28.g6003e7f
Borislav Petkov
maintainerRFC PATCH 2/4 x86/mce: Rename rcu_dereference_check_mce() to mce_log_get_idx_check()
In reply to Borislav PetkovFrom: Borislav Petkov <[email protected]>
The "rcu_" prefix misleads for it being a proper RCU interface which is not. It basically checks whether we're preemptible or holding the chrdev_read mutex.
Rename it accordingly.
--- arch/x86/kernel/cpu/mcheck/mce.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
arch/x86/kernel/cpu/mcheck/mce.c +4 -4
@@ -52,11 +52,11 @@static DEFINE_MUTEX (mce_chrdev_read_mutex); -#define rcu_dereference_check_mce (p) \ +#define mce_log_get_idx_check (p) \ ({ \ rcu_lockdep_assert(rcu_read_lock_sched_held() || \ lockdep_is_held(&mce_chrdev_read_mutex), \ - "suspicious rcu_dereference_check_mce() usage"); \ + "suspicious mce_log_get_idx_check() usage"); \ smp_load_acquire(&(p)); \ })
@@ -165,7 +165,7 @@ void mce_log(struct mce *mce) mce->finished =0 ;wmb ();for (;;) { - entry =rcu_dereference_check_mce (mcelog.next); + entry =mce_log_get_idx_check (mcelog.next);for (;;) {/*
@@ -1812,7 +1812,7 @@ static ssize_t mce_chrdev_read(struct file *filp, char __user *ubuf, goto out; } - next =rcu_dereference_check_mce (mcelog.next); + next =mce_log_get_idx_check (mcelog.next);/* Only supports full reads right now */ err = -EINVAL ;
Signature · 1 line
2.5.0.rc2.28.g6003e7f
Re: [RFC PATCH 2/4] x86/mce: Rename rcu_dereference_check_mce() to mce_log_get_idx_check()
Quoted 10 lines · Borislav Petkov
From: Borislav Petkov <[email protected]>
The "
rcu_" prefix misleads for it being a proper RCU interface which is not. It basically checks whether we're preemptible or holding thechrdev_readmutex.Rename it accordingly.
Signed-off-by Borislav Petkov <[email protected]>Cc "Paul E. McKenney" <[email protected]>
Quoted 44 lines
--- arch/x86/kernel/cpu/mcheck/mce.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)arch/x86/kernel/cpu/mcheck/mce.c +4
-4@@ -52,11 +52,11 @@static DEFINE_MUTEX (mce_chrdev_read_mutex); -#define rcu_dereference_check_mce (p) \ +#define mce_log_get_idx_check (p) \ ({ \ rcu_lockdep_assert(rcu_read_lock_sched_held() || \ lockdep_is_held(&mce_chrdev_read_mutex), \ - "suspicious rcu_dereference_check_mce() usage"); \ + "suspicious mce_log_get_idx_check() usage"); \ smp_load_acquire(&(p)); \ })@@ -165,7 +165,7 @@ void mce_log(struct mce *mce) mce->finished =0 ;wmb ();for (;;) { - entry =rcu_dereference_check_mce (mcelog.next); + entry =mce_log_get_idx_check (mcelog.next);for (;;) {/* @@ -1812,7 +1812,7 @@ static ssize_t mce_chrdev_read(struct file *filp, char __user *ubuf, goto out; } - next =rcu_dereference_check_mce (mcelog.next); + next =mce_log_get_idx_check (mcelog.next);/* Only supports full reads right now */ err = -EINVAL ;Signature · 1 line
2.5.0.rc2.28.g6003e7f
Borislav Petkov
maintainerRFC PATCH 3/4 x86/mce: Add a wrapper around mce_log() for injection
In reply to Borislav PetkovFrom: Borislav Petkov <[email protected]>
Will be used by an injector module in a following patch.
--- arch/x86/kernel/cpu/mcheck/mce-internal.h | 2 ++ arch/x86/kernel/cpu/mcheck/mce.c | 7 +++++++ 2 files changed, 9 insertions(+)
arch/x86/kernel/cpu/mcheck/mce-internal.h +2 -0
@@ -79,3 +79,5 @@ static inline int apei_clear_mce(u64 record_id) return -EINVAL; }#endif + +void mce_inject_log (struct mce *m);
arch/x86/kernel/cpu/mcheck/mce.c +7 -0
@@ -199,6 +199,13 @@ void mce_log(struct mce *mce)set_bit (0 , &mce_need_notify); } +void mce_inject_log (struct mce *m) +{ +mutex_lock (&mce_chrdev_read_mutex); +mce_log (m); +mutex_unlock (&mce_chrdev_read_mutex); +} +static struct notifier_block mce_srao_nb;void mce_register_decode_chain (struct notifier_block *nb)
Signature · 1 line
2.5.0.rc2.28.g6003e7f
Borislav Petkov
maintainerRFC PATCH 4/4 x86/ras: Move AMD MCE injector to arch/x86/ras/
In reply to Borislav PetkovFrom: Borislav Petkov <[email protected]>
This is an x86-specific module and would benefit from being closer to the arch code. Move it there. Update copyright while at it.
---
arch/x86/Makefile | 2 ++
arch/x86/ras/Kconfig | 11 +++++++++++
arch/x86/ras/Makefile | 2 ++
{drivers/edac => arch/x86/ras}/mce_amd_inj.c | 6 +++---
drivers/edac/Kconfig | 10 ----------
drivers/edac/Makefile | 1 -
drivers/ras/Kconfig | 3 +++
7 files changed, 21 insertions(+), 14 deletions(-)
create mode 100644 arch/x86/ras/Kconfig
create mode 100644 arch/x86/ras/Makefile
rename {drivers/edac => arch/x86/ras}/mce_amd_inj.c (98%)arch/x86/Makefile +2 -0
@@ -212,6 +212,8 @@ drivers-$(CONFIG_PM) += arch/x86/power/ drivers-$(CONFIG_FB) += arch/x86/video/ +drivers-$(CONFIG_RAS) += arch/x86/ras/ + #### # boot loader support. Several targets are kept for legacy purposes
arch/x86/ras/Kconfig new file +11 -0
@@ -0,0 +1,11 @@ +config AMD_MCE_INJ + tristate "Simple MCE injection interface for AMD processors" + depends on RAS && EDAC_DECODE_MCE && DEBUG_FS + default n + help + This is a simple debugfs interface to inject MCEs and test different + aspects of the MCE handling code. + + WARNING: Do not even assume this interface is staying stable! + +
arch/x86/ras/Makefile new file +2 -0
@@ -0,0 +1,2 @@ +obj-$(CONFIG_AMD_MCE_INJ) += mce_amd_inj.o +
arch/x86/ras/mce_amd_inj.c renamed from drivers/edac/mce_amd_inj.c +3 -3
@@ -6,7 +6,7 @@* This file may be distributed under the terms of the GNU General Public * License version 2. * -* Copyright (c) 2010-14: Borislav Petkov <[email protected]> +* Copyright (c) 2010-15: Borislav Petkov <[email protected]> * Advanced Micro Devices Inc. */
@@ -19,7 +19,7 @@#include <linux/uaccess.h> #include <asm/mce.h> -#include "mce_amd.h" +#include "../kernel/cpu/mcheck/mce-internal.h" /* * Collect all the MCi_XXX settings
@@ -195,7 +195,7 @@ static void do_inject(void) i_mce.status |=MCI_STATUS_MISCV ;if (inj_type ==SW_INJ ) { -amd_decode_mce (NULL ,0 , &i_mce); +mce_inject_log (&i_mce);return ; }
drivers/edac/Kconfig +0 -10
@@ -61,16 +61,6 @@ config EDAC_DECODE_MCE which occur really early upon boot, before the module infrastructure has been initialized. -config EDAC_MCE_INJ - tristate "Simple MCE injection interface" - depends on EDAC_DECODE_MCE && DEBUG_FS - default n - help - This is a simple debugfs interface to inject MCEs and test different - aspects of the MCE handling code. - - WARNING: Do not even assume this interface is staying stable! - config EDAC_MM_EDAC tristate "Main Memory EDAC (Error Detection And Correction) reporting" select RAS
drivers/edac/Makefile +0 -1
@@ -17,7 +17,6 @@ edac_core-y += edac_pci.o edac_pci_sysfs.o endif obj-$(CONFIG_EDAC_GHES) += ghes_edac.o -obj-$(CONFIG_EDAC_MCE_INJ) += mce_amd_inj.o edac_mce_amd-y := mce_amd.o obj-$(CONFIG_EDAC_DECODE_MCE) += edac_mce_amd.o
drivers/ras/Kconfig +3 -0
@@ -29,4 +29,7 @@ menuconfig RAS data corruption. if RAS + +source arch/x86/ras/Kconfig + endif
Signature · 1 line
2.5.0.rc2.28.g6003e7f