Skip to contentSkip to conversation

last-minute patches for MCE .31

PATCH In review Hidetoshi Seto 3 participants 25 messages Jun 15, 2009 – Jun 18, 2009 x86

lore
Share & tools
lore
#0 lore raw

cover last-minute patches for MCE .31

Here are 16 patches for bugfixes and cleanups etc. These are against tip/x86/mce3, but today I confirmed that these can be applied on Linus's latest, with no errors.

Please consider applying these for .31

[1] Bugfix
x86, mce: don't init timer if !mce_available x86, mce: call-in should be after updating global_nwo

These bugs are not serious, but should be fixed asap.

[2] Misc (requested by Ingo)
x86, mce: add __read_mostly
x86, mce: rename static variables around trigger x86, mce: sysfs entries for new options

Minor improvements (cleanups), and an enhancement for a feature newly introduced in this merge window (i.e. run-time disablement of cmci).

[3] Cleanups
x86, mce: unify mce.h
x86, mce: make mce_disabled boolean
x86, mce: unify smp_thermal_interrupt, prepare p4 x86, mce: unify smp_thermal_interrupt, prepare mce_intel_64 x86, mce: unify smp_thermal_interrupt, prepare x86, mce: unify smp_thermal_interrupt
x86, mce: squash mce_intel.c into therm_throt.c x86, mce: remove intel_set_thermal_handler()
x86, mce: remove therm_throt.h
x86, mce: mce.h cleanup
x86, mce: rename

There are 3 goals:

[3-1] unify mce.h
There are 2 files named mce.h:

        arch/x86/include/asm/mce.h
        arch/x86/kernel/cpu/mcheck/mce.h
   The latter is small, and in it:
        #include <asm/mce.h>

so we can have a unified one.

[3-2] finish unifying thermal codes
Now codes for thermal interrupt is distributed for multiple files. We can gather them in one therm_throt.c file.

 arch/x86/include/asm/mce.h                |   53 +++++++++------
 arch/x86/include/asm/therm_throt.h        |    9 ---
 arch/x86/kernel/cpu/mcheck/Makefile       |    7 +-
 arch/x86/kernel/cpu/mcheck/mce_intel.c    |   73 --------------------
 arch/x86/kernel/cpu/mcheck/mce_intel_64.c |   21 ------
 arch/x86/kernel/cpu/mcheck/p4.c           |   45 ------------
 arch/x86/kernel/cpu/mcheck/therm_throt.c  |  106 +++++++++++++++++++++++++++-
 7 files changed, 137 insertions(+), 177 deletions(-)
 delete mode 100644 arch/x86/include/asm/therm_throt.h
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel.c
   This makes Makefile clean & simple:
   before:
     obj-y                            =  mce.o therm_throt.o
     obj-$(CONFIG_X86_MCE_P4THERMAL)  += mce_intel.o
     obj-$(CONFIG_X86_MCE_INTEL)      += mce_intel_64.o mce_intel.o
   after:
     obj-y                            =  mce.o
     obj-$(CONFIG_X86_MCE_INTEL)      += mce_intel_64.o
     obj-$(CONFIG_X86_THERMAL_VECTOR) += therm_throt.o

[3-3] rename
Rename files that are no longer 64bit specific:

        mce_amd_64.c    => mce_amd.c
        mce_intel_64.c  => mce_intel.c

Thanks,
H.Seto

===

Hidetoshi Seto (16):
x86, mce: don't init timer if !mce_available
x86, mce: call-in should be after updating global_nwo x86, mce: add __read_mostly
x86, mce: rename static variables around trigger x86, mce: sysfs entries for new options
x86, mce: unify mce.h
x86, mce: make mce_disabled boolean
x86, mce: unify smp_thermal_interrupt, prepare p4 x86, mce: unify smp_thermal_interrupt, prepare mce_intel_64 x86, mce: unify smp_thermal_interrupt, prepare x86, mce: unify smp_thermal_interrupt
x86, mce: squash mce_intel.c into therm_throt.c x86, mce: remove intel_set_thermal_handler()
x86, mce: remove therm_throt.h
x86, mce: mce.h cleanup
x86, mce: rename

 arch/x86/include/asm/mce.h                |   63 ++-
 arch/x86/include/asm/therm_throt.h        |    9 -
 arch/x86/kernel/cpu/mcheck/Makefile       |    9 +-
 arch/x86/kernel/cpu/mcheck/k7.c           |    3 +-
 arch/x86/kernel/cpu/mcheck/mce.c          |  216 ++++++---
 arch/x86/kernel/cpu/mcheck/mce.h          |   38 --
 arch/x86/kernel/cpu/mcheck/mce_amd.c      |  703 +++++++++++++++++++++++++++++
 arch/x86/kernel/cpu/mcheck/mce_amd_64.c   |  703 -----------------------------
 arch/x86/kernel/cpu/mcheck/mce_intel.c    |  249 ++++++++--
 arch/x86/kernel/cpu/mcheck/mce_intel_64.c |  248 ----------
 arch/x86/kernel/cpu/mcheck/non-fatal.c    |    3 +-
 arch/x86/kernel/cpu/mcheck/p4.c           |   48 +--
 arch/x86/kernel/cpu/mcheck/p5.c           |   15 +-
 arch/x86/kernel/cpu/mcheck/p6.c           |    3 +-
 arch/x86/kernel/cpu/mcheck/therm_throt.c  |  106 +++++-
 arch/x86/kernel/cpu/mcheck/winchip.c      |    3 +-
 arch/x86/kernel/traps.c                   |    3 +-
 17 files changed, 1220 insertions(+), 1202 deletions(-)
 delete mode 100644 arch/x86/include/asm/therm_throt.h
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce.h
 create mode 100644 arch/x86/kernel/cpu/mcheck/mce_amd.c
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_amd_64.c
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel_64.c
#1 lore raw

PATCH 1/16 x86, mce: don't init timer if !mce_available

In mce_cpu_restart, mce_init_timer is called unconditionally. If !mce_available (e.g. mce is disabled), there are no useful work for timer. Stop running it.

---
 arch/x86/kernel/cpu/mcheck/mce.c |    5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)
arch/x86/kernel/cpu/mcheck/mce.c +3 -2
index fabba15..6f9db11 100644
@@ -1609,8 +1609,9 @@ static int mce_resume(struct sys_device *dev)
 static void mce_cpu_restart(void *data)
 {
 	del_timer_sync(&__get_cpu_var(mce_timer));
-	if (mce_available(&current_cpu_data))
-		mce_init();
+	if (!mce_available(&current_cpu_data))
+		return;
+	mce_init();
 	mce_init_timer();
 }
 
Signature · 1 line
1.6.3
#2 lore raw

PATCH 2/16 x86, mce: call-in should be after updating global_nwo

In reply to Hidetoshi Seto

At the beginning of Monarch synchronization, processors wait until all of them have entered the exception handler and then check the global_nwo to determine if any of them saw a fatal event.

However since current code does call-in before updating global_nwo, it might happen that the global_nwo does not reflect some of local nwo at the time. This might break printing corrected errors not handled yet on panic.

---
 arch/x86/kernel/cpu/mcheck/mce.c |   65 ++++++++++++++++++--------------------
 1 files changed, 31 insertions(+), 34 deletions(-)
arch/x86/kernel/cpu/mcheck/mce.c +31 -34
index 6f9db11..84b2630 100644
@@ -691,18 +691,18 @@ static atomic_t global_nwo;
  * in the entry order.
  * TBD double check parallel CPU hotunplug
  */
-static int mce_start(int no_way_out, int *order)
+static int mce_start(int *no_way_out)
 {
-	int nwo;
+	int order;
 	int cpus = num_online_cpus();
 	u64 timeout = (u64)monarch_timeout * NSEC_PER_USEC;
 
-	if (!timeout) {
-		*order = -1;
-		return no_way_out;
-	}
+	if (!timeout)
+		return -1;
 
-	atomic_add(no_way_out, &global_nwo);
+	atomic_add(*no_way_out, &global_nwo);
+
+	order = atomic_add_return(1, &mce_callin);
 
 	/*
 	 * Wait for everyone.
@@ -710,40 +710,38 @@ static int mce_start(int no_way_out, int *order)
 	while (atomic_read(&mce_callin) != cpus) {
 		if (mce_timed_out(&timeout)) {
 			atomic_set(&global_nwo, 0);
-			*order = -1;
-			return no_way_out;
+			return -1;
 		}
 		ndelay(SPINUNIT);
 	}
 
-	/*
-	 * Cache the global no_way_out state.
-	 */
-	nwo = atomic_read(&global_nwo);
-
-	/*
-	 * Monarch starts executing now, the others wait.
-	 */
-	if (*order == 1) {
+	if (order == 1) {
+		/*
+		 * Monarch: Starts executing now, the others wait.
+		 */
 		atomic_set(&mce_executing, 1);
-		return nwo;
+	} else {
+		/*
+		 * Subject: Now start the scanning loop one by one in
+		 * the original callin order.
+		 * This way when there are any shared banks it will be
+		 * only seen by one CPU before cleared, avoiding duplicates.
+		 */
+		while (atomic_read(&mce_executing) < order) {
+			if (mce_timed_out(&timeout)) {
+				atomic_set(&global_nwo, 0);
+				return -1;
+			}
+			ndelay(SPINUNIT);
+		}
 	}
 
 	/*
-	 * Now start the scanning loop one by one
-	 * in the original callin order.
-	 * This way when there are any shared banks it will
-	 * be only seen by one CPU before cleared, avoiding duplicates.
+	 * Cache the global no_way_out state.
 	 */
-	while (atomic_read(&mce_executing) < *order) {
-		if (mce_timed_out(&timeout)) {
-			atomic_set(&global_nwo, 0);
-			*order = -1;
-			return no_way_out;
-		}
-		ndelay(SPINUNIT);
-	}
-	return nwo;
+	*no_way_out = atomic_read(&global_nwo);
+
+	return order;
 }
 
 /*
@@ -887,7 +885,6 @@ void do_machine_check(struct pt_regs *regs, long error_code)
 	if (!banks)
 		goto out;
 
-	order = atomic_add_return(1, &mce_callin);
 	mce_setup(&m);
 
 	m.mcgstatus = mce_rdmsrl(MSR_IA32_MCG_STATUS);
@@ -909,7 +906,7 @@ void do_machine_check(struct pt_regs *regs, long error_code)
 	 * This way we don't report duplicated events on shared banks
 	 * because the first one to see it will clear it.
 	 */
-	no_way_out = mce_start(no_way_out, &order);
+	order = mce_start(&no_way_out);
 	for (i = 0; i < banks; i++) {
 		__clear_bit(i, toclear);
 		if (!bank[i])
Signature · 1 line
1.6.3
#6 lore raw

PATCH 3/16 x86, mce: add __read_mostly

In reply to Hidetoshi Seto

As recommended by Ingo Molnar.

---
 arch/x86/kernel/cpu/mcheck/mce.c |   26 +++++++++++++-------------
 1 files changed, 13 insertions(+), 13 deletions(-)
arch/x86/kernel/cpu/mcheck/mce.c +13 -13
index 84b2630..d66f9f1 100644
@@ -57,7 +57,7 @@ static void unexpected_machine_check(struct pt_regs *regs, long error_code)
 void (*machine_check_vector)(struct pt_regs *, long error_code) =
 						unexpected_machine_check;
 
-int				mce_disabled;
+int mce_disabled __read_mostly;
 
 #ifdef CONFIG_X86_NEW_MCE
 
@@ -76,19 +76,19 @@ DEFINE_PER_CPU(unsigned, mce_exception_count);
  *   2: SIGBUS or log uncorrected errors (if possible), log corrected errors
  *   3: never panic or SIGBUS, log all errors (for testing only)
  */
-static int			tolerant = 1;
-static int			banks;
-static u64			*bank;
-static unsigned long		notify_user;
-static int			rip_msr;
-static int			mce_bootlog = -1;
-static int			monarch_timeout = -1;
-static int			mce_panic_timeout;
-static int			mce_dont_log_ce;
-int				mce_cmci_disabled;
-int				mce_ignore_ce;
-int				mce_ser;
+static int			tolerant		__read_mostly = 1;
+static int			banks			__read_mostly;
+static u64			*bank			__read_mostly;
+static int			rip_msr			__read_mostly;
+static int			mce_bootlog		__read_mostly = -1;
+static int			monarch_timeout		__read_mostly = -1;
+static int			mce_panic_timeout	__read_mostly;
+static int			mce_dont_log_ce		__read_mostly;
+int				mce_cmci_disabled	__read_mostly;
+int				mce_ignore_ce		__read_mostly;
+int				mce_ser			__read_mostly;
 
+static unsigned long		notify_user;
 static char			trigger[128];
 static char			*trigger_argv[2] = { trigger, NULL };
 
Signature · 1 line
1.6.3
#7 lore raw

PATCH 4/16 x86, mce: rename static variables around trigger

In reply to Hidetoshi Seto

"trigger" is not straight forward name for valiable that holds name of user mode helper program which triggered by machine check events.

This patch renames this valiable and kins to more recognizable names.

trigger => mce_helper
trigger_argv => mce_helper_argv
notify_user => mce_need_notify

No functional changes.

---
 arch/x86/kernel/cpu/mcheck/mce.c |   27 ++++++++++++++-------------
 1 files changed, 14 insertions(+), 13 deletions(-)
arch/x86/kernel/cpu/mcheck/mce.c +14 -13
index d66f9f1..e145229 100644
@@ -88,9 +88,10 @@ int				mce_cmci_disabled	__read_mostly;
 int				mce_ignore_ce		__read_mostly;
 int				mce_ser			__read_mostly;
 
-static unsigned long		notify_user;
-static char			trigger[128];
-static char			*trigger_argv[2] = { trigger, NULL };
+/* User mode helper program triggered by machine check event */
+static unsigned long		mce_need_notify;
+static char			mce_helper[128];
+static char			*mce_helper_argv[2] = { mce_helper, NULL };
 
 static unsigned long		dont_init_banks;
 
@@ -180,7 +181,7 @@ void mce_log(struct mce *mce)
 	wmb();
 
 	mce->finished = 1;
-	set_bit(0, &notify_user);
+	set_bit(0, &mce_need_notify);
 }
 
 static void print_mce(struct mce *m)
@@ -1115,7 +1116,7 @@ static void mcheck_timer(unsigned long data)
 
 static void mce_do_trigger(struct work_struct *work)
 {
-	call_usermodehelper(trigger, trigger_argv, NULL, UMH_NO_WAIT);
+	call_usermodehelper(mce_helper, mce_helper_argv, NULL, UMH_NO_WAIT);
 }
 
 static DECLARE_WORK(mce_trigger_work, mce_do_trigger);
@@ -1132,7 +1133,7 @@ int mce_notify_irq(void)
 
 	clear_thread_flag(TIF_MCE_NOTIFY);
 
-	if (test_and_clear_bit(0, &notify_user)) {
+	if (test_and_clear_bit(0, &mce_need_notify)) {
 		wake_up_interruptible(&mce_wait);
 
 		/*
@@ -1140,7 +1141,7 @@ int mce_notify_irq(void)
 		 * work_pending is always cleared before the function is
 		 * executed.
 		 */
-		if (trigger[0] && !work_pending(&mce_trigger_work))
+		if (mce_helper[0] && !work_pending(&mce_trigger_work))
 			schedule_work(&mce_trigger_work);
 
 		if (__ratelimit(&ratelimit))
@@ -1657,9 +1658,9 @@ static ssize_t set_bank(struct sys_device *s, struct sysdev_attribute *attr,
 static ssize_t
 show_trigger(struct sys_device *s, struct sysdev_attribute *attr, char *buf)
 {
-	strcpy(buf, trigger);
+	strcpy(buf, mce_helper);
 	strcat(buf, "\n");
-	return strlen(trigger) + 1;
+	return strlen(mce_helper) + 1;
 }
 
 static ssize_t set_trigger(struct sys_device *s, struct sysdev_attribute *attr,
@@ -1668,10 +1669,10 @@ static ssize_t set_trigger(struct sys_device *s, struct sysdev_attribute *attr,
 	char *p;
 	int len;
 
-	strncpy(trigger, buf, sizeof(trigger));
-	trigger[sizeof(trigger)-1] = 0;
-	len = strlen(trigger);
-	p = strchr(trigger, '\n');
+	strncpy(mce_helper, buf, sizeof(mce_helper));
+	mce_helper[sizeof(mce_helper)-1] = 0;
+	len = strlen(mce_helper);
+	p = strchr(mce_helper, '\n');
 
 	if (*p)
 		*p = 0;
Signature · 1 line
1.6.3
#8 lore raw

PATCH 5/16 x86, mce: sysfs entries for new options

In reply to Hidetoshi Seto

Add sysfs interface for admins who want to tweak these options without rebooting the system.

---
 arch/x86/kernel/cpu/mcheck/mce.c |   84 +++++++++++++++++++++++++++++++++++++-
 1 files changed, 83 insertions(+), 1 deletions(-)
arch/x86/kernel/cpu/mcheck/mce.c +83 -1
index e145229..6b465a6 100644
@@ -1619,6 +1619,26 @@ static void mce_restart(void)
 	on_each_cpu(mce_cpu_restart, NULL, 1);
 }
 
+/* Toggle features for corrected errors */
+static void mce_disable_ce(void *all)
+{
+	if (!mce_available(&current_cpu_data))
+		return;
+	if (all)
+		del_timer_sync(&__get_cpu_var(mce_timer));
+	cmci_clear();
+}
+
+static void mce_enable_ce(void *all)
+{
+	if (!mce_available(&current_cpu_data))
+		return;
+	cmci_reenable();
+	cmci_recheck();
+	if (all)
+		mce_init_timer();
+}
+
 static struct sysdev_class mce_sysclass = {
 	.suspend	= mce_suspend,
 	.shutdown	= mce_shutdown,
@@ -1680,6 +1700,52 @@ static ssize_t set_trigger(struct sys_device *s, struct sysdev_attribute *attr,
 	return len;
 }
 
+static ssize_t set_ignore_ce(struct sys_device *s,
+			     struct sysdev_attribute *attr,
+			     const char *buf, size_t size)
+{
+	u64 new;
+
+	if (strict_strtoull(buf, 0, &new) < 0)
+		return -EINVAL;
+
+	if (mce_ignore_ce ^ !!new) {
+		if (new) {
+			/* disable ce features */
+			on_each_cpu(mce_disable_ce, (void *)1, 1);
+			mce_ignore_ce = 1;
+		} else {
+			/* enable ce features */
+			mce_ignore_ce = 0;
+			on_each_cpu(mce_enable_ce, (void *)1, 1);
+		}
+	}
+	return size;
+}
+
+static ssize_t set_cmci_disabled(struct sys_device *s,
+				 struct sysdev_attribute *attr,
+				 const char *buf, size_t size)
+{
+	u64 new;
+
+	if (strict_strtoull(buf, 0, &new) < 0)
+		return -EINVAL;
+
+	if (mce_cmci_disabled ^ !!new) {
+		if (new) {
+			/* disable cmci */
+			on_each_cpu(mce_disable_ce, NULL, 1);
+			mce_cmci_disabled = 1;
+		} else {
+			/* enable cmci */
+			mce_cmci_disabled = 0;
+			on_each_cpu(mce_enable_ce, NULL, 1);
+		}
+	}
+	return size;
+}
+
 static ssize_t store_int_with_restart(struct sys_device *s,
 				      struct sysdev_attribute *attr,
 				      const char *buf, size_t size)
@@ -1692,6 +1758,7 @@ static ssize_t store_int_with_restart(struct sys_device *s,
 static SYSDEV_ATTR(trigger, 0644, show_trigger, set_trigger);
 static SYSDEV_INT_ATTR(tolerant, 0644, tolerant);
 static SYSDEV_INT_ATTR(monarch_timeout, 0644, monarch_timeout);
+static SYSDEV_INT_ATTR(dont_log_ce, 0644, mce_dont_log_ce);
 
 static struct sysdev_ext_attribute attr_check_interval = {
 	_SYSDEV_ATTR(check_interval, 0644, sysdev_show_int,
@@ -1699,9 +1766,24 @@ static struct sysdev_ext_attribute attr_check_interval = {
 	&check_interval
 };
 
+static struct sysdev_ext_attribute attr_ignore_ce = {
+	_SYSDEV_ATTR(ignore_ce, 0644, sysdev_show_int, set_ignore_ce),
+	&mce_ignore_ce
+};
+
+static struct sysdev_ext_attribute attr_cmci_disabled = {
+	_SYSDEV_ATTR(ignore_ce, 0644, sysdev_show_int, set_cmci_disabled),
+	&mce_cmci_disabled
+};
+
 static struct sysdev_attribute *mce_attrs[] = {
-	&attr_tolerant.attr, &attr_check_interval.attr, &attr_trigger,
+	&attr_tolerant.attr,
+	&attr_check_interval.attr,
+	&attr_trigger,
 	&attr_monarch_timeout.attr,
+	&attr_dont_log_ce.attr,
+	&attr_ignore_ce.attr,
+	&attr_cmci_disabled.attr,
 	NULL
 };
 
Signature · 1 line
1.6.3
#9 lore raw

PATCH 6/16 x86, mce: unify mce.h

In reply to Hidetoshi Seto

There are 2 headers:

	arch/x86/include/asm/mce.h
	arch/x86/kernel/cpu/mcheck/mce.h
and in the latter small header:
	#include <asm/mce.h>

This patch move all contents in the latter header into the former, and fix all files using the latter to include the former instead.

---
 arch/x86/include/asm/mce.h                |   36 +++++++++++++++++++++++++-
 arch/x86/kernel/cpu/mcheck/k7.c           |    3 +-
 arch/x86/kernel/cpu/mcheck/mce.c          |    1 -
 arch/x86/kernel/cpu/mcheck/mce.h          |   38 -----------------------------
 arch/x86/kernel/cpu/mcheck/mce_intel.c    |    3 +-
 arch/x86/kernel/cpu/mcheck/mce_intel_64.c |    2 -
 arch/x86/kernel/cpu/mcheck/non-fatal.c    |    3 +-
 arch/x86/kernel/cpu/mcheck/p4.c           |    3 +-
 arch/x86/kernel/cpu/mcheck/p5.c           |    3 +-
 arch/x86/kernel/cpu/mcheck/p6.c           |    3 +-
 arch/x86/kernel/cpu/mcheck/winchip.c      |    3 +-
 arch/x86/kernel/traps.c                   |    3 +-
 12 files changed, 42 insertions(+), 59 deletions(-)
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce.h
arch/x86/include/asm/mce.h +34 -2
index 540a466..aae6fe2 100644
@@ -102,10 +102,42 @@ struct mce_log {
 
 #ifdef __KERNEL__
 
+#include <linux/percpu.h>
+#include <linux/init.h>
+#include <asm/atomic.h>
+
 extern int mce_disabled;
 
-#include <asm/atomic.h>
-#include <linux/percpu.h>
+#ifdef CONFIG_X86_OLD_MCE
+void amd_mcheck_init(struct cpuinfo_x86 *c);
+void intel_p4_mcheck_init(struct cpuinfo_x86 *c);
+void intel_p6_mcheck_init(struct cpuinfo_x86 *c);
+#endif
+
+#ifdef CONFIG_X86_ANCIENT_MCE
+void intel_p5_mcheck_init(struct cpuinfo_x86 *c);
+void winchip_mcheck_init(struct cpuinfo_x86 *c);
+extern int mce_p5_enable;
+static inline int mce_p5_enabled(void) { return mce_p5_enable; }
+static inline void enable_p5_mce(void) { mce_p5_enable = 1; }
+#else
+static inline void intel_p5_mcheck_init(struct cpuinfo_x86 *c) {}
+static inline void winchip_mcheck_init(struct cpuinfo_x86 *c) {}
+static inline int mce_p5_enabled(void) { return 0; }
+static inline void enable_p5_mce(void) { }
+#endif
+
+/* Call the installed machine check handler for this CPU setup. */
+extern void (*machine_check_vector)(struct pt_regs *, long error_code);
+
+#ifdef CONFIG_X86_OLD_MCE
+extern int nr_mce_banks;
+extern void intel_set_thermal_handler(void);
+#else
+static inline void intel_set_thermal_handler(void) { }
+#endif
+
+void intel_init_thermal(struct cpuinfo_x86 *c);
 
 void mce_setup(struct mce *m);
 void mce_log(struct mce *m);
arch/x86/kernel/cpu/mcheck/k7.c +1 -2
index 89e5104..b945d5d 100644
@@ -10,10 +10,9 @@
 
 #include <asm/processor.h>
 #include <asm/system.h>
+#include <asm/mce.h>
 #include <asm/msr.h>
 
-#include "mce.h"
-
 /* Machine Check Handler For AMD Athlon/Duron: */
 static void k7_machine_check(struct pt_regs *regs, long error_code)
 {
arch/x86/kernel/cpu/mcheck/mce.c +0 -1
index 6b465a6..c184dc7 100644
@@ -44,7 +44,6 @@
 #include <asm/msr.h>
 
 #include "mce-internal.h"
-#include "mce.h"
 
 /* Handle unconfigured int18 (should never happen) */
 static void unexpected_machine_check(struct pt_regs *regs, long error_code)
arch/x86/kernel/cpu/mcheck/mce.h deleted +0 -38
deleted file mode 100644
index 84a552b..0000000
@@ -1,38 +0,0 @@
-#include <linux/init.h>
-#include <asm/mce.h>
-
-#ifdef CONFIG_X86_OLD_MCE
-void amd_mcheck_init(struct cpuinfo_x86 *c);
-void intel_p4_mcheck_init(struct cpuinfo_x86 *c);
-void intel_p6_mcheck_init(struct cpuinfo_x86 *c);
-#endif
-
-#ifdef CONFIG_X86_ANCIENT_MCE
-void intel_p5_mcheck_init(struct cpuinfo_x86 *c);
-void winchip_mcheck_init(struct cpuinfo_x86 *c);
-extern int mce_p5_enable;
-static inline int mce_p5_enabled(void) { return mce_p5_enable; }
-static inline void enable_p5_mce(void) { mce_p5_enable = 1; }
-#else
-static inline void intel_p5_mcheck_init(struct cpuinfo_x86 *c) {}
-static inline void winchip_mcheck_init(struct cpuinfo_x86 *c) {}
-static inline int mce_p5_enabled(void) { return 0; }
-static inline void enable_p5_mce(void) { }
-#endif
-
-/* Call the installed machine check handler for this CPU setup. */
-extern void (*machine_check_vector)(struct pt_regs *, long error_code);
-
-#ifdef CONFIG_X86_OLD_MCE
-
-extern int nr_mce_banks;
-
-void intel_set_thermal_handler(void);
-
-#else
-
-static inline void intel_set_thermal_handler(void) { }
-
-#endif
-
-void intel_init_thermal(struct cpuinfo_x86 *c);
arch/x86/kernel/cpu/mcheck/mce_intel.c +1 -2
index 2b011d2..475478b 100644
@@ -11,10 +11,9 @@
 #include <asm/processor.h>
 #include <asm/system.h>
 #include <asm/apic.h>
+#include <asm/mce.h>
 #include <asm/msr.h>
 
-#include "mce.h"
-
 void intel_init_thermal(struct cpuinfo_x86 *c)
 {
 	unsigned int cpu = smp_processor_id();
arch/x86/kernel/cpu/mcheck/mce_intel_64.c +0 -2
index f2ef695..c548111 100644
@@ -16,8 +16,6 @@
 #include <asm/idle.h>
 #include <asm/therm_throt.h>
 
-#include "mce.h"
-
 asmlinkage void smp_thermal_interrupt(void)
 {
 	__u64 msr_val;
arch/x86/kernel/cpu/mcheck/non-fatal.c +1 -2
index 70b7104..f5f2d6f 100644
@@ -17,10 +17,9 @@
 
 #include <asm/processor.h>
 #include <asm/system.h>
+#include <asm/mce.h>
 #include <asm/msr.h>
 
-#include "mce.h"
-
 static int		firstbank;
 
 #define MCE_RATE	(15*HZ)	/* timer rate is 15s */
arch/x86/kernel/cpu/mcheck/p4.c +1 -2
index 82cee10..8d3e40e 100644
@@ -12,10 +12,9 @@
 #include <asm/processor.h>
 #include <asm/system.h>
 #include <asm/apic.h>
+#include <asm/mce.h>
 #include <asm/msr.h>
 
-#include "mce.h"
-
 /* as supported by the P4/Xeon family */
 struct intel_mce_extended_msrs {
 	u32 eax;
arch/x86/kernel/cpu/mcheck/p5.c +1 -2
index 015f481..747853f 100644
@@ -10,10 +10,9 @@
 
 #include <asm/processor.h>
 #include <asm/system.h>
+#include <asm/mce.h>
 #include <asm/msr.h>
 
-#include "mce.h"
-
 /* By default disabled */
 int		mce_p5_enable;
 
arch/x86/kernel/cpu/mcheck/p6.c +1 -2
index 43c24e6..01e4f81 100644
@@ -10,10 +10,9 @@
 
 #include <asm/processor.h>
 #include <asm/system.h>
+#include <asm/mce.h>
 #include <asm/msr.h>
 
-#include "mce.h"
-
 /* Machine Check Handler For PII/PIII */
 static void intel_machine_check(struct pt_regs *regs, long error_code)
 {
arch/x86/kernel/cpu/mcheck/winchip.c +1 -2
index 81b0248..54060f5 100644
@@ -9,10 +9,9 @@
 
 #include <asm/processor.h>
 #include <asm/system.h>
+#include <asm/mce.h>
 #include <asm/msr.h>
 
-#include "mce.h"
-
 /* Machine check handler for WinChip C6: */
 static void winchip_machine_check(struct pt_regs *regs, long error_code)
 {
arch/x86/kernel/traps.c +1 -2
index 1e1e27b..71f9c74 100644
@@ -53,6 +53,7 @@
 #include <asm/traps.h>
 #include <asm/desc.h>
 #include <asm/i387.h>
+#include <asm/mce.h>
 
 #include <asm/mach_traps.h>
 
@@ -64,8 +65,6 @@
 #include <asm/setup.h>
 #include <asm/traps.h>
 
-#include "cpu/mcheck/mce.h"
-
 asmlinkage int system_call(void);
 
 /* Do we ignore FPU interrupts ? */
Signature · 1 line
1.6.3
#10 lore raw

PATCH 7/16 x86, mce: make mce_disabled boolean

In reply to Hidetoshi Seto

The mce_disabled on 32bit is a tristate variable [1,0,-1], while 64bit version is boolean [0,1].
This patch makes mce_disabled always boolean, and use mce_p5_enabled to indicate the third state instead.

---
 arch/x86/include/asm/mce.h       |    8 +++-----
 arch/x86/kernel/cpu/mcheck/mce.c |    8 +++-----
 arch/x86/kernel/cpu/mcheck/p5.c  |   12 +++++-------
 3 files changed, 11 insertions(+), 17 deletions(-)
arch/x86/include/asm/mce.h +3 -5
index aae6fe2..6568cde 100644
@@ -107,6 +107,7 @@ struct mce_log {
 #include <asm/atomic.h>
 
 extern int mce_disabled;
+extern int mce_p5_enabled;
 
 #ifdef CONFIG_X86_OLD_MCE
 void amd_mcheck_init(struct cpuinfo_x86 *c);
@@ -117,14 +118,11 @@ void intel_p6_mcheck_init(struct cpuinfo_x86 *c);
 #ifdef CONFIG_X86_ANCIENT_MCE
 void intel_p5_mcheck_init(struct cpuinfo_x86 *c);
 void winchip_mcheck_init(struct cpuinfo_x86 *c);
-extern int mce_p5_enable;
-static inline int mce_p5_enabled(void) { return mce_p5_enable; }
-static inline void enable_p5_mce(void) { mce_p5_enable = 1; }
+static inline void enable_p5_mce(void) { mce_p5_enabled = 1; }
 #else
 static inline void intel_p5_mcheck_init(struct cpuinfo_x86 *c) {}
 static inline void winchip_mcheck_init(struct cpuinfo_x86 *c) {}
-static inline int mce_p5_enabled(void) { return 0; }
-static inline void enable_p5_mce(void) { }
+static inline void enable_p5_mce(void) {}
 #endif
 
 /* Call the installed machine check handler for this CPU setup. */
arch/x86/kernel/cpu/mcheck/mce.c +3 -5
index c184dc7..e0d5af0 100644
@@ -1279,8 +1279,7 @@ static void __cpuinit mce_ancient_init(struct cpuinfo_x86 *c)
 		return;
 	switch (c->x86_vendor) {
 	case X86_VENDOR_INTEL:
-		if (mce_p5_enabled())
-			intel_p5_mcheck_init(c);
+		intel_p5_mcheck_init(c);
 		break;
 	case X86_VENDOR_CENTAUR:
 		winchip_mcheck_init(c);
@@ -1995,7 +1994,7 @@ EXPORT_SYMBOL_GPL(nr_mce_banks);	/* non-fatal.o */
 /* This has to be run for each processor */
 void mcheck_init(struct cpuinfo_x86 *c)
 {
-	if (mce_disabled == 1)
+	if (mce_disabled)
 		return;
 
 	switch (c->x86_vendor) {
@@ -2025,10 +2024,9 @@ void mcheck_init(struct cpuinfo_x86 *c)
 
 static int __init mcheck_enable(char *str)
 {
-	mce_disabled = -1;
+	mce_p5_enabled = 1;
 	return 1;
 }
-
 __setup("mce", mcheck_enable);
 
 #endif /* CONFIG_X86_OLD_MCE */
arch/x86/kernel/cpu/mcheck/p5.c +5 -7
index 747853f..5c0e653 100644
@@ -14,7 +14,7 @@
 #include <asm/msr.h>
 
 /* By default disabled */
-int		mce_p5_enable;
+int mce_p5_enabled __read_mostly;
 
 /* Machine check handler for Pentium class Intel CPUs: */
 static void pentium_machine_check(struct pt_regs *regs, long error_code)
@@ -42,15 +42,13 @@ void intel_p5_mcheck_init(struct cpuinfo_x86 *c)
 {
 	u32 l, h;
 
-	/* Check for MCE support: */
-	if (!cpu_has(c, X86_FEATURE_MCE))
+	/* Default P5 to off as its often misconnected: */
+	if (!mce_p5_enabled)
 		return;
 
-#ifdef CONFIG_X86_OLD_MCE
-	/* Default P5 to off as its often misconnected: */
-	if (mce_disabled != -1)
+	/* Check for MCE support: */
+	if (!cpu_has(c, X86_FEATURE_MCE))
 		return;
-#endif
 
 	machine_check_vector = pentium_machine_check;
 	/* Make sure the vector pointer is visible before we enable MCEs: */
Signature · 1 line
1.6.3
#11 lore raw

PATCH 8/16 x86, mce: unify smp_thermal_interrupt, prepare p4

In reply to Hidetoshi Seto

Remove unused argument regs from handlers, and use inc_irq_stat.

---
 arch/x86/kernel/cpu/mcheck/p4.c |   11 +++++------
 1 files changed, 5 insertions(+), 6 deletions(-)
arch/x86/kernel/cpu/mcheck/p4.c +5 -6
index 8d3e40e..ddeed6e 100644
@@ -35,7 +35,7 @@ static int mce_num_extended_msrs;
 
 #ifdef CONFIG_X86_MCE_P4THERMAL
 
-static void unexpected_thermal_interrupt(struct pt_regs *regs)
+static void unexpected_thermal_interrupt(void)
 {
 	printk(KERN_ERR "CPU%d: Unexpected LVT TMR interrupt!\n",
 			smp_processor_id());
@@ -43,7 +43,7 @@ static void unexpected_thermal_interrupt(struct pt_regs *regs)
 }
 
 /* P4/Xeon Thermal transition interrupt handler: */
-static void intel_thermal_interrupt(struct pt_regs *regs)
+static void intel_thermal_interrupt(void)
 {
 	__u64 msr_val;
 
@@ -54,14 +54,13 @@ static void intel_thermal_interrupt(struct pt_regs *regs)
 }
 
 /* Thermal interrupt handler for this CPU setup: */
-static void (*vendor_thermal_interrupt)(struct pt_regs *regs) =
-						unexpected_thermal_interrupt;
+static void (*vendor_thermal_interrupt)(void) = unexpected_thermal_interrupt;
 
 void smp_thermal_interrupt(struct pt_regs *regs)
 {
 	irq_enter();
-	vendor_thermal_interrupt(regs);
-	__get_cpu_var(irq_stat).irq_thermal_count++;
+	vendor_thermal_interrupt();
+	inc_irq_stat(irq_thermal_count);
 	irq_exit();
 }
 
Signature · 1 line
1.6.3
#12 lore raw

PATCH 9/16 x86, mce: unify smp_thermal_interrupt, prepare mce_intel_64

In reply to Hidetoshi Seto

Divide it into two functions.

---
 arch/x86/kernel/cpu/mcheck/mce_intel_64.c |   14 ++++++++------
 1 files changed, 8 insertions(+), 6 deletions(-)
arch/x86/kernel/cpu/mcheck/mce_intel_64.c +8 -6
index c548111..a5232b2 100644
@@ -16,19 +16,21 @@
 #include <asm/idle.h>
 #include <asm/therm_throt.h>
 
-asmlinkage void smp_thermal_interrupt(void)
+static void intel_thermal_interrupt(void)
 {
 	__u64 msr_val;
 
-	ack_APIC_irq();
-
-	exit_idle();
-	irq_enter();
-
 	rdmsrl(MSR_IA32_THERM_STATUS, msr_val);
 	if (therm_throt_process(msr_val & THERM_STATUS_PROCHOT))
 		mce_log_therm_throt_event(msr_val);
+}
 
+asmlinkage void smp_thermal_interrupt(void)
+{
+	ack_APIC_irq();
+	exit_idle();
+	irq_enter();
+	intel_thermal_interrupt();
 	inc_irq_stat(irq_thermal_count);
 	irq_exit();
 }
Signature · 1 line
1.6.3
#13 lore raw

PATCH 10/16 x86, mce: unify smp_thermal_interrupt, prepare

In reply to Hidetoshi Seto

Let them in same shape.

---
 arch/x86/include/asm/mce.h                |   25 ++++++++++++++-----------
 arch/x86/kernel/cpu/mcheck/mce_intel_64.c |   20 ++++++++++++++++++--
 arch/x86/kernel/cpu/mcheck/p4.c           |   10 ++++++----
 3 files changed, 38 insertions(+), 17 deletions(-)
arch/x86/include/asm/mce.h +14 -11
index 6568cde..3bc827c 100644
@@ -110,6 +110,7 @@ extern int mce_disabled;
 extern int mce_p5_enabled;
 
 #ifdef CONFIG_X86_OLD_MCE
+extern int nr_mce_banks;
 void amd_mcheck_init(struct cpuinfo_x86 *c);
 void intel_p4_mcheck_init(struct cpuinfo_x86 *c);
 void intel_p6_mcheck_init(struct cpuinfo_x86 *c);
@@ -128,15 +129,6 @@ static inline void enable_p5_mce(void) {}
 /* Call the installed machine check handler for this CPU setup. */
 extern void (*machine_check_vector)(struct pt_regs *, long error_code);
 
-#ifdef CONFIG_X86_OLD_MCE
-extern int nr_mce_banks;
-extern void intel_set_thermal_handler(void);
-#else
-static inline void intel_set_thermal_handler(void) { }
-#endif
-
-void intel_init_thermal(struct cpuinfo_x86 *c);
-
 void mce_setup(struct mce *m);
 void mce_log(struct mce *m);
 DECLARE_PER_CPU(struct sys_device, mce_dev);
@@ -175,8 +167,6 @@ int mce_available(struct cpuinfo_x86 *c);
 DECLARE_PER_CPU(unsigned, mce_exception_count);
 DECLARE_PER_CPU(unsigned, mce_poll_count);
 
-void mce_log_therm_throt_event(__u64 status);
-
 extern atomic_t mce_entry;
 
 void do_machine_check(struct pt_regs *, long);
@@ -205,5 +195,18 @@ void mcheck_init(struct cpuinfo_x86 *c);
 
 extern void (*mce_threshold_vector)(void);
 
+/*
+ * Thermal handler
+ */
+
+void intel_set_thermal_handler(void);
+void intel_init_thermal(struct cpuinfo_x86 *c);
+
+#ifdef CONFIG_X86_NEW_MCE
+void mce_log_therm_throt_event(__u64 status);
+#else
+static inline void mce_log_therm_throt_event(__u64 status) {}
+#endif
+
 #endif /* __KERNEL__ */
 #endif /* _ASM_X86_MCE_H */
arch/x86/kernel/cpu/mcheck/mce_intel_64.c +18 -2
index a5232b2..922e3a4 100644
@@ -16,6 +16,14 @@
 #include <asm/idle.h>
 #include <asm/therm_throt.h>
 
+static void unexpected_thermal_interrupt(void)
+{
+	printk(KERN_ERR "CPU%d: Unexpected LVT TMR interrupt!\n",
+			smp_processor_id());
+	add_taint(TAINT_MACHINE_CHECK);
+}
+
+/* P4/Xeon Thermal transition interrupt handler: */
 static void intel_thermal_interrupt(void)
 {
 	__u64 msr_val;
@@ -25,14 +33,22 @@ static void intel_thermal_interrupt(void)
 		mce_log_therm_throt_event(msr_val);
 }
 
+/* Thermal interrupt handler for this CPU setup: */
+static void (*vendor_thermal_interrupt)(void) = unexpected_thermal_interrupt;
+
 asmlinkage void smp_thermal_interrupt(void)
 {
-	ack_APIC_irq();
 	exit_idle();
 	irq_enter();
-	intel_thermal_interrupt();
 	inc_irq_stat(irq_thermal_count);
+	intel_thermal_interrupt();
 	irq_exit();
+	ack_APIC_irq();
+}
+
+void intel_set_thermal_handler(void)
+{
+	vendor_thermal_interrupt = intel_thermal_interrupt;
 }
 
 /*
arch/x86/kernel/cpu/mcheck/p4.c +6 -4
index ddeed6e..75313f5 100644
@@ -12,6 +12,7 @@
 #include <asm/processor.h>
 #include <asm/system.h>
 #include <asm/apic.h>
+#include <asm/idle.h>
 #include <asm/mce.h>
 #include <asm/msr.h>
 
@@ -47,10 +48,9 @@ static void intel_thermal_interrupt(void)
 {
 	__u64 msr_val;
 
-	ack_APIC_irq();
-
 	rdmsrl(MSR_IA32_THERM_STATUS, msr_val);
-	therm_throt_process(msr_val & THERM_STATUS_PROCHOT);
+	if (therm_throt_process(msr_val & THERM_STATUS_PROCHOT))
+		mce_log_therm_throt_event(msr_val);
 }
 
 /* Thermal interrupt handler for this CPU setup: */
@@ -58,10 +58,12 @@ static void (*vendor_thermal_interrupt)(void) = unexpected_thermal_interrupt;
 
 void smp_thermal_interrupt(struct pt_regs *regs)
 {
+	exit_idle();
 	irq_enter();
-	vendor_thermal_interrupt();
 	inc_irq_stat(irq_thermal_count);
+	vendor_thermal_interrupt();
 	irq_exit();
+	ack_APIC_irq();
 }
 
 void intel_set_thermal_handler(void)
Signature · 1 line
1.6.3
#14 lore raw

PATCH 11/16 x86, mce: unify smp_thermal_interrupt

In reply to Hidetoshi Seto

Put common functions into therm_throt.c, modify Makefile.

unexpected_thermal_interrupt
intel_thermal_interrupt
smp_thermal_interrupt
intel_set_thermal_handler

---
 arch/x86/kernel/cpu/mcheck/Makefile       |    7 ++--
 arch/x86/kernel/cpu/mcheck/mce_intel_64.c |   38 ------------------------
 arch/x86/kernel/cpu/mcheck/p4.c           |   45 -----------------------------
 arch/x86/kernel/cpu/mcheck/therm_throt.c  |   40 +++++++++++++++++++++++++-
 4 files changed, 43 insertions(+), 87 deletions(-)
arch/x86/kernel/cpu/mcheck/Makefile +4 -3
index 45004fa..53df57d 100644
@@ -1,11 +1,12 @@
-obj-y				=  mce.o therm_throt.o
+obj-y				=  mce.o
 
 obj-$(CONFIG_X86_NEW_MCE)	+= mce-severity.o
 obj-$(CONFIG_X86_OLD_MCE)	+= k7.o p4.o p6.o
 obj-$(CONFIG_X86_ANCIENT_MCE)	+= winchip.o p5.o
-obj-$(CONFIG_X86_MCE_P4THERMAL)	+= mce_intel.o
-obj-$(CONFIG_X86_MCE_INTEL)	+= mce_intel_64.o mce_intel.o
+obj-$(CONFIG_X86_MCE_INTEL)	+= mce_intel_64.o
 obj-$(CONFIG_X86_MCE_AMD)	+= mce_amd_64.o
 obj-$(CONFIG_X86_MCE_NONFATAL)	+= non-fatal.o
 obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o
 obj-$(CONFIG_X86_MCE_INJECT)	+= mce-inject.o
+
+obj-$(CONFIG_X86_THERMAL_VECTOR) += therm_throt.o mce_intel.o
arch/x86/kernel/cpu/mcheck/mce_intel_64.c +0 -38
index 922e3a4..3b7a057 100644
@@ -9,48 +9,10 @@
 #include <linux/interrupt.h>
 #include <linux/percpu.h>
 #include <asm/processor.h>
-#include <asm/apic.h>
 #include <asm/msr.h>
 #include <asm/mce.h>
-#include <asm/hw_irq.h>
-#include <asm/idle.h>
 #include <asm/therm_throt.h>
 
-static void unexpected_thermal_interrupt(void)
-{
-	printk(KERN_ERR "CPU%d: Unexpected LVT TMR interrupt!\n",
-			smp_processor_id());
-	add_taint(TAINT_MACHINE_CHECK);
-}
-
-/* P4/Xeon Thermal transition interrupt handler: */
-static void intel_thermal_interrupt(void)
-{
-	__u64 msr_val;
-
-	rdmsrl(MSR_IA32_THERM_STATUS, msr_val);
-	if (therm_throt_process(msr_val & THERM_STATUS_PROCHOT))
-		mce_log_therm_throt_event(msr_val);
-}
-
-/* Thermal interrupt handler for this CPU setup: */
-static void (*vendor_thermal_interrupt)(void) = unexpected_thermal_interrupt;
-
-asmlinkage void smp_thermal_interrupt(void)
-{
-	exit_idle();
-	irq_enter();
-	inc_irq_stat(irq_thermal_count);
-	intel_thermal_interrupt();
-	irq_exit();
-	ack_APIC_irq();
-}
-
-void intel_set_thermal_handler(void)
-{
-	vendor_thermal_interrupt = intel_thermal_interrupt;
-}
-
 /*
  * Support for Intel Correct Machine Check Interrupts. This allows
  * the CPU to raise an interrupt when a corrected machine check happened.
arch/x86/kernel/cpu/mcheck/p4.c +0 -45

63 lines not on this page · raw message

arch/x86/kernel/cpu/mcheck/therm_throt.c +39 -1
index 7b1ae2e..b3792b1 100644
@@ -13,6 +13,7 @@
  * Credits: Adapted from Zwane Mwaikambo's original code in mce_intel.c.
  *          Inspired by Ross Biro's and Al Borchers' counter code.
  */
+#include <linux/interrupt.h>
 #include <linux/notifier.h>
 #include <linux/jiffies.h>
 #include <linux/percpu.h>
@@ -20,6 +21,8 @@
 #include <linux/cpu.h>
 
 #include <asm/therm_throt.h>
+#include <asm/idle.h>
+#include <asm/mce.h>
 
 /* How long to wait between reporting thermal events */
 #define CHECK_INTERVAL		(300 * HZ)
@@ -186,6 +189,41 @@ static __init int thermal_throttle_init_device(void)
 
 	return 0;
 }
-
 device_initcall(thermal_throttle_init_device);
+
 #endif /* CONFIG_SYSFS */
+
+/* Thermal transition interrupt handler */
+void intel_thermal_interrupt(void)
+{
+	__u64 msr_val;
+
+	rdmsrl(MSR_IA32_THERM_STATUS, msr_val);
+	if (therm_throt_process(msr_val & THERM_STATUS_PROCHOT))
+		mce_log_therm_throt_event(msr_val);
+}
+
+static void unexpected_thermal_interrupt(void)
+{
+	printk(KERN_ERR "CPU%d: Unexpected LVT TMR interrupt!\n",
+			smp_processor_id());
+	add_taint(TAINT_MACHINE_CHECK);
+}
+
+static void (*smp_thermal_vector)(void) = unexpected_thermal_interrupt;
+
+asmlinkage void smp_thermal_interrupt(struct pt_regs *regs)
+{
+	exit_idle();
+	irq_enter();
+	inc_irq_stat(irq_thermal_count);
+	smp_thermal_vector();
+	irq_exit();
+	/* Ack only at the end to avoid potential reentry */
+	ack_APIC_irq();
+}
+
+void intel_set_thermal_handler(void)
+{
+	smp_thermal_vector = intel_thermal_interrupt;
+}
Signature · 1 line
1.6.3
#3 lore raw

Re: [PATCH 02/16] x86, mce: call-in should be after updating global_nwo

In reply to Hidetoshi Seto

2009/6/15 Hidetoshi Seto <[email protected]>:

Quoted 8 lines

At the beginning of Monarch synchronization, processors wait until all of them have entered the exception handler and then check the global_nwo to determine if any of them saw a fatal event.

However since current code does call-in before updating global_nwo, it might happen that the global_nwo does not reflect some of local nwo at the time.  This might break printing corrected errors not handled yet on panic.

This is almost same as the patch that I just posted. I think this is because it trigger a failure in mce-test suite, and we both find the failure.

But your patch lacks memory barrier between mce_callin and global_nwo updating/reading. Atomic operations does not imply memory barrier.

Best Regards,
Huang Ying

#15 lore raw

PATCH 12/16 x86, mce: squash mce_intel.c into therm_throt.c

In reply to Hidetoshi Seto

move intel_init_thermal() into therm_throt.c

---
 arch/x86/kernel/cpu/mcheck/Makefile      |    2 +-
 arch/x86/kernel/cpu/mcheck/mce_intel.c   |   73 ------------------------------
 arch/x86/kernel/cpu/mcheck/therm_throt.c |   66 +++++++++++++++++++++++++++
 3 files changed, 67 insertions(+), 74 deletions(-)
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel.c
arch/x86/kernel/cpu/mcheck/Makefile +1 -1
index 53df57d..659564e 100644
@@ -9,4 +9,4 @@ obj-$(CONFIG_X86_MCE_NONFATAL)	+= non-fatal.o
 obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o
 obj-$(CONFIG_X86_MCE_INJECT)	+= mce-inject.o
 
-obj-$(CONFIG_X86_THERMAL_VECTOR) += therm_throt.o mce_intel.o
+obj-$(CONFIG_X86_THERMAL_VECTOR) += therm_throt.o
arch/x86/kernel/cpu/mcheck/mce_intel.c deleted +0 -73

73 lines not on this page · raw message

arch/x86/kernel/cpu/mcheck/therm_throt.c +66 -0

82 lines not on this page · raw message

Signature · 1 line
1.6.3
#21 lore raw

PATCH 13/16 x86, mce: remove intel_set_thermal_handler()

In reply to Hidetoshi Seto

and make intel_thermal_interrupt() static.

---
 arch/x86/include/asm/mce.h               |    1 -
 arch/x86/kernel/cpu/mcheck/therm_throt.c |    9 ++-------
 2 files changed, 2 insertions(+), 8 deletions(-)
arch/x86/include/asm/mce.h +0 -1
index 3bc827c..365a594 100644
@@ -199,7 +199,6 @@ extern void (*mce_threshold_vector)(void);
  * Thermal handler
  */
 
-void intel_set_thermal_handler(void);
 void intel_init_thermal(struct cpuinfo_x86 *c);
 
 #ifdef CONFIG_X86_NEW_MCE
arch/x86/kernel/cpu/mcheck/therm_throt.c +2 -7
index 7a508aa..7c7944c 100644
@@ -202,7 +202,7 @@ device_initcall(thermal_throttle_init_device);
 #endif /* CONFIG_SYSFS */
 
 /* Thermal transition interrupt handler */
-void intel_thermal_interrupt(void)
+static void intel_thermal_interrupt(void)
 {
 	__u64 msr_val;
 
@@ -231,11 +231,6 @@ asmlinkage void smp_thermal_interrupt(struct pt_regs *regs)
 	ack_APIC_irq();
 }
 
-void intel_set_thermal_handler(void)
-{
-	smp_thermal_vector = intel_thermal_interrupt;
-}
-
 void intel_init_thermal(struct cpuinfo_x86 *c)
 {
 	unsigned int cpu = smp_processor_id();
@@ -278,7 +273,7 @@ void intel_init_thermal(struct cpuinfo_x86 *c)
 	wrmsr(MSR_IA32_THERM_INTERRUPT,
 		l | (THERM_INT_LOW_ENABLE | THERM_INT_HIGH_ENABLE), h);
 
-	intel_set_thermal_handler();
+	smp_thermal_vector = intel_thermal_interrupt;
 
 	rdmsr(MSR_IA32_MISC_ENABLE, l, h);
 	wrmsr(MSR_IA32_MISC_ENABLE, l | MSR_IA32_MISC_ENABLE_TM1, h);
Signature · 1 line
1.6.3
#22 lore raw

PATCH 14/16 x86, mce: remove therm_throt.h

In reply to Hidetoshi Seto

Now all symbols in the header are static. Remove the header.

---
 arch/x86/include/asm/therm_throt.h        |    9 ---------
 arch/x86/kernel/cpu/mcheck/mce_intel_64.c |    1 -
 arch/x86/kernel/cpu/mcheck/p4.c           |    1 -
 arch/x86/kernel/cpu/mcheck/therm_throt.c  |    5 ++---
 4 files changed, 2 insertions(+), 14 deletions(-)
 delete mode 100644 arch/x86/include/asm/therm_throt.h
arch/x86/include/asm/therm_throt.h deleted +0 -9
deleted file mode 100644
index c62349e..0000000
@@ -1,9 +0,0 @@
-#ifndef _ASM_X86_THERM_THROT_H
-#define _ASM_X86_THERM_THROT_H
-
-#include <asm/atomic.h>
-
-extern atomic_t therm_throt_en;
-int therm_throt_process(int curr);
-
-#endif /* _ASM_X86_THERM_THROT_H */
arch/x86/kernel/cpu/mcheck/mce_intel_64.c +0 -1
index 3b7a057..663a88e 100644
@@ -11,7 +11,6 @@
 #include <asm/processor.h>
 #include <asm/msr.h>
 #include <asm/mce.h>
-#include <asm/therm_throt.h>
 
 /*
  * Support for Intel Correct Machine Check Interrupts. This allows
arch/x86/kernel/cpu/mcheck/p4.c +0 -1
index 76c5f03..4482aea 100644
@@ -6,7 +6,6 @@
 #include <linux/init.h>
 #include <linux/smp.h>
 
-#include <asm/therm_throt.h>
 #include <asm/processor.h>
 #include <asm/mce.h>
 #include <asm/msr.h>
arch/x86/kernel/cpu/mcheck/therm_throt.c +2 -3
index 7c7944c..bff8dd1 100644
@@ -24,7 +24,6 @@
 #include <linux/smp.h>
 #include <linux/cpu.h>
 
-#include <asm/therm_throt.h>
 #include <asm/processor.h>
 #include <asm/system.h>
 #include <asm/apic.h>
@@ -38,7 +37,7 @@
 static DEFINE_PER_CPU(__u64, next_check) = INITIAL_JIFFIES;
 static DEFINE_PER_CPU(unsigned long, thermal_throttle_count);
 
-atomic_t therm_throt_en		= ATOMIC_INIT(0);
+static atomic_t therm_throt_en		= ATOMIC_INIT(0);
 
 #ifdef CONFIG_SYSFS
 #define define_therm_throt_sysdev_one_ro(_name)				\
@@ -93,7 +92,7 @@ static struct attribute_group thermal_throttle_attr_group = {
  *          1 : Event should be logged further, and a message has been
  *              printed to the syslog.
  */
-int therm_throt_process(int curr)
+static int therm_throt_process(int curr)
 {
 	unsigned int cpu = smp_processor_id();
 	__u64 tmp_jiffs = get_jiffies_64();
Signature · 1 line
1.6.3
#23 lore raw

PATCH 15/16 x86, mce: mce.h cleanup

In reply to Hidetoshi Seto

Reorder definitions.

- static inline dummy mcheck_init() for !CONFIG_X86_MCE
- gather defs for exception, threshold handler

---
 arch/x86/include/asm/mce.h |   29 ++++++++++++++++++-----------
 1 files changed, 18 insertions(+), 11 deletions(-)
arch/x86/include/asm/mce.h +18 -11
index 365a594..5cdd8d1 100644
@@ -109,6 +109,12 @@ struct mce_log {
 extern int mce_disabled;
 extern int mce_p5_enabled;
 
+#ifdef CONFIG_X86_MCE
+void mcheck_init(struct cpuinfo_x86 *c);
+#else
+static inline void mcheck_init(struct cpuinfo_x86 *c) {}
+#endif
+
 #ifdef CONFIG_X86_OLD_MCE
 extern int nr_mce_banks;
 void amd_mcheck_init(struct cpuinfo_x86 *c);
@@ -126,13 +132,9 @@ static inline void winchip_mcheck_init(struct cpuinfo_x86 *c) {}
 static inline void enable_p5_mce(void) {}
 #endif
 
-/* Call the installed machine check handler for this CPU setup. */
-extern void (*machine_check_vector)(struct pt_regs *, long error_code);
-
 void mce_setup(struct mce *m);
 void mce_log(struct mce *m);
 DECLARE_PER_CPU(struct sys_device, mce_dev);
-extern void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu);
 
 /*
  * To support more than 128 would need to escape the predefined
@@ -169,8 +171,6 @@ DECLARE_PER_CPU(unsigned, mce_poll_count);
 
 extern atomic_t mce_entry;
 
-void do_machine_check(struct pt_regs *, long);
-
 typedef DECLARE_BITMAP(mce_banks_t, MAX_NR_BANKS);
 DECLARE_PER_CPU(mce_banks_t, mce_poll_banks);
 
@@ -187,13 +187,20 @@ void mce_notify_process(void);
 DECLARE_PER_CPU(struct mce, injectm);
 extern struct file_operations mce_chrdev_ops;
 
-#ifdef CONFIG_X86_MCE
-void mcheck_init(struct cpuinfo_x86 *c);
-#else
-#define mcheck_init(c) do { } while (0)
-#endif
+/*
+ * Exception handler
+ */
+
+/* Call the installed machine check handler for this CPU setup. */
+extern void (*machine_check_vector)(struct pt_regs *, long error_code);
+void do_machine_check(struct pt_regs *, long);
+
+/*
+ * Threshold handler
+ */
 
 extern void (*mce_threshold_vector)(void);
+extern void (*threshold_cpu_callback)(unsigned long action, unsigned int cpu);
 
 /*
  * Thermal handler
Signature · 1 line
1.6.3
#24 lore raw

PATCH 16/16 x86, mce: rename

In reply to Hidetoshi Seto

Rename files that are no longer 64bit specific:

	mce_amd_64.c	=> mce_amd.c
	mce_intel_64.c	=> mce_intel.c
---
 arch/x86/kernel/cpu/mcheck/Makefile       |    4 +-
 arch/x86/kernel/cpu/mcheck/mce_amd.c      |  703 +++++++++++++++++++++++++++++
 arch/x86/kernel/cpu/mcheck/mce_amd_64.c   |  703 -----------------------------
 arch/x86/kernel/cpu/mcheck/mce_intel.c    |  225 +++++++++
 arch/x86/kernel/cpu/mcheck/mce_intel_64.c |  225 ---------
 5 files changed, 930 insertions(+), 930 deletions(-)
 create mode 100644 arch/x86/kernel/cpu/mcheck/mce_amd.c
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_amd_64.c
 create mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel.c
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel_64.c
arch/x86/kernel/cpu/mcheck/Makefile +2 -2
index 659564e..188a1ca 100644
@@ -3,8 +3,8 @@ obj-y				=  mce.o
 obj-$(CONFIG_X86_NEW_MCE)	+= mce-severity.o
 obj-$(CONFIG_X86_OLD_MCE)	+= k7.o p4.o p6.o
 obj-$(CONFIG_X86_ANCIENT_MCE)	+= winchip.o p5.o
-obj-$(CONFIG_X86_MCE_INTEL)	+= mce_intel_64.o
-obj-$(CONFIG_X86_MCE_AMD)	+= mce_amd_64.o
+obj-$(CONFIG_X86_MCE_INTEL)	+= mce_intel.o
+obj-$(CONFIG_X86_MCE_AMD)	+= mce_amd.o
 obj-$(CONFIG_X86_MCE_NONFATAL)	+= non-fatal.o
 obj-$(CONFIG_X86_MCE_THRESHOLD) += threshold.o
 obj-$(CONFIG_X86_MCE_INJECT)	+= mce-inject.o
arch/x86/kernel/cpu/mcheck/mce_amd.c new file +703 -0

703 lines not on this page · raw message

arch/x86/kernel/cpu/mcheck/mce_amd_64.c deleted +0 -703

703 lines not on this page · raw message

arch/x86/kernel/cpu/mcheck/mce_intel.c new file +225 -0

225 lines not on this page · raw message

arch/x86/kernel/cpu/mcheck/mce_intel_64.c deleted +0 -225

225 lines not on this page · raw message

Signature · 1 line
1.6.3
#4 lore raw

Re: [PATCH 02/16] x86, mce: call-in should be after updating global_nwo

In reply to huang ying
Quoted 5 lines · huang ying

2009/6/15 Hidetoshi Seto <[email protected]>:

Quoted 4 lines

At the beginning of Monarch synchronization, processors wait until all of them have entered the exception handler and then check the global_nwo to determine if any of them saw a fatal event.

Quoted 8 lines
Quoted 4 lines

However since current code does call-in before updating global_nwo, it might happen that the global_nwo does not reflect some of local nwo at the time. This might break printing corrected errors not handled yet on panic.

This is almost same as the patch that I just posted. I think this is because it trigger a failure in mce-test suite, and we both find the failure.

That's right.

Quoted 2 lines

But your patch lacks memory barrier between mce_callin and global_nwo updating/reading. Atomic operations does not imply memory barrier.

You are right. I agree that your patch is better fix.

It seems my patch includes cleanup for mce_start(), so I'll extract it and make another patch soon.

Thanks,
H.Seto

#5 lore raw

PATCH 1/1 x86, mce: cleanup mce_start()

In reply to huang ying

[ my 02/16 should be replaced by combination of Huang's fix and this cleanup ]

Simplify interface of mce_start().

- no_way_out = mce_start(no_way_out, &order);
+ order = mce_start(&no_way_out);

Now Monarch and Subjects share same exit(return) in usual path.

---
 arch/x86/kernel/cpu/mcheck/mce.c |   66 ++++++++++++++++++--------------------
 1 files changed, 31 insertions(+), 35 deletions(-)
arch/x86/kernel/cpu/mcheck/mce.c +31 -35
index dda7721..739fd7e 100644
@@ -691,23 +691,21 @@ static atomic_t global_nwo;
  * in the entry order.
  * TBD double check parallel CPU hotunplug
  */
-static int mce_start(int no_way_out, int *order)
+static int mce_start(int *no_way_out)
 {
-	int nwo;
+	int order;
 	int cpus = num_online_cpus();
 	u64 timeout = (u64)monarch_timeout * NSEC_PER_USEC;
 
-	if (!timeout) {
-		*order = -1;
-		return no_way_out;
-	}
+	if (!timeout)
+		return -1;
 
-	atomic_add(no_way_out, &global_nwo);
+	atomic_add(*no_way_out, &global_nwo);
 	/*
 	 * global_nwo should be updated before mce_callin
 	 */
 	smp_wmb();
-	*order = atomic_add_return(1, &mce_callin);
+	order = atomic_add_return(1, &mce_callin);
 
 	/*
 	 * Wait for everyone.
@@ -715,8 +713,7 @@ static int mce_start(int no_way_out, int *order)
 	while (atomic_read(&mce_callin) != cpus) {
 		if (mce_timed_out(&timeout)) {
 			atomic_set(&global_nwo, 0);
-			*order = -1;
-			return no_way_out;
+			return -1;
 		}
 		ndelay(SPINUNIT);
 	}
@@ -725,34 +722,34 @@ static int mce_start(int no_way_out, int *order)
 	 * mce_callin should be read before global_nwo
 	 */
 	smp_rmb();
-	/*
-	 * Cache the global no_way_out state.
-	 */
-	nwo = atomic_read(&global_nwo);
 
-	/*
-	 * Monarch starts executing now, the others wait.
-	 */
-	if (*order == 1) {
+	if (order == 1) {
+		/*
+		 * Monarch: Starts executing now, the others wait.
+		 */
 		atomic_set(&mce_executing, 1);
-		return nwo;
+	} else {
+		/*
+		 * Subject: Now start the scanning loop one by one in
+		 * the original callin order.
+		 * This way when there are any shared banks it will be
+		 * only seen by one CPU before cleared, avoiding duplicates.
+		 */
+		while (atomic_read(&mce_executing) < order) {
+			if (mce_timed_out(&timeout)) {
+				atomic_set(&global_nwo, 0);
+				return -1;
+			}
+			ndelay(SPINUNIT);
+		}
 	}
 
 	/*
-	 * Now start the scanning loop one by one
-	 * in the original callin order.
-	 * This way when there are any shared banks it will
-	 * be only seen by one CPU before cleared, avoiding duplicates.
+	 * Cache the global no_way_out state.
 	 */
-	while (atomic_read(&mce_executing) < *order) {
-		if (mce_timed_out(&timeout)) {
-			atomic_set(&global_nwo, 0);
-			*order = -1;
-			return no_way_out;
-		}
-		ndelay(SPINUNIT);
-	}
-	return nwo;
+	*no_way_out = atomic_read(&global_nwo);
+
+	return order;
 }
 
 /*
@@ -871,8 +868,7 @@ void do_machine_check(struct pt_regs *regs, long error_code)
 	 * Establish sequential order between the CPUs entering the machine
 	 * check handler.
 	 */
-	int order = -1;
-
+	int order;
 	/*
 	 * If no_way_out gets set, there is no safe way to recover from this
 	 * MCE.  If tolerant is cranked up, we'll try anyway.
@@ -917,7 +913,7 @@ void do_machine_check(struct pt_regs *regs, long error_code)
 	 * This way we don't report duplicated events on shared banks
 	 * because the first one to see it will clear it.
 	 */
-	no_way_out = mce_start(no_way_out, &order);
+	order = mce_start(&no_way_out);
 	for (i = 0; i < banks; i++) {
 		__clear_bit(i, toclear);
 		if (!bank[i])
Signature · 1 line
1.6.3
#16 lore raw

Re: [PATCH 12/16] x86, mce: squash mce_intel.c into therm_throt.c

In reply to Hidetoshi Seto

2009/6/15 Hidetoshi Seto <[email protected]>:

Quoted 9 lines

move intel_init_thermal() into therm_throt.c

---
 arch/x86/kernel/cpu/mcheck/Makefile      |    2 +- arch/x86/kernel/cpu/mcheck/mce_intel.c   |   73 ------------------------------ arch/x86/kernel/cpu/mcheck/therm_throt.c |   66 +++++++++++++++++++++++++++

 3 files changed, 67 insertions(+), 74 deletions(-)

 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel.c

Why we need this patch? I think the design principle behind MCE is to separate vendor/cpu specific code into different files. At least #ifdef CONFIG_X86_MCE_INTEL should be used around moved code.

Best Regards,
Huang Ying

#17 lore raw

Re: [PATCH 12/16] x86, mce: squash mce_intel.c into therm_throt.c

Quoted 11 lines · huang ying

2009/6/15 Hidetoshi Seto <[email protected]>:

Quoted 9 lines

move intel_init_thermal() into therm_throt.c

---
 arch/x86/kernel/cpu/mcheck/Makefile      |    2 +-
 arch/x86/kernel/cpu/mcheck/mce_intel.c   |   73 ------------------------------
 arch/x86/kernel/cpu/mcheck/therm_throt.c |   66 +++++++++++++++++++++++++++
 3 files changed, 67 insertions(+), 74 deletions(-)
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel.c
Quoted 3 lines

Why we need this patch? I think the design principle behind MCE is to separate vendor/cpu specific code into different files. At least #ifdef CONFIG_X86_MCE_INTEL should be used around moved code.

Because at this time there are no vendor other than Intel that uses thermal codes here. I don't know much about AMD, whether there are any plan to implement AMD specific thermal codes or not, and why they have to make such specific thing here in future (patent?).

You can pull out "codes may not be shared" from therm_throt.c and put it again into mce_intel.c (that was mce_intel_64.c). But I believe it is not needed now, until another code is available. Or you can rename therm_throt.c to thermal_intel.c or so for a while.

(BTW, now CONFIG_X86_MCE_INTEL enables both of cmci and thermal,
are there possible option to disable thermal while cmci is on?)

Thanks.
H.Seto

#18 lore raw

Re: [PATCH 12/16] x86, mce: squash mce_intel.c into therm_throt.c

Quoted 21 lines · Hidetoshi Seto
Quoted 11 lines · huang ying

2009/6/15 Hidetoshi Seto <[email protected]>:

Quoted 9 lines

move intel_init_thermal() into therm_throt.c

---
 arch/x86/kernel/cpu/mcheck/Makefile      |    2 +- arch/x86/kernel/cpu/mcheck/mce_intel.c   |   73 ------------------------------ arch/x86/kernel/cpu/mcheck/therm_throt.c |   66 +++++++++++++++++++++++++++

 3 files changed, 67 insertions(+), 74 deletions(-)

 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel.c

Quoted 3 lines

Why we need this patch? I think the design principle behind MCE is to separate vendor/cpu specific code into different files. At least #ifdef CONFIG_X86_MCE_INTEL should be used around moved code.

Because at this time there are no vendor other than Intel that uses thermal codes here.  I don't know much about AMD, whether there are any plan to implement AMD specific thermal codes or not, and why they have to make such specific thing here in future (patent?).

Quoted 4 lines

You can pull out "codes may not be shared" from therm_throt.c and put it again into mce_intel.c (that was mce_intel_64.c).  But I believe it is not needed now, until another code is available. Or you can rename therm_throt.c to thermal_intel.c or so for a while.

I think thermal_intel.c is better.

Best Regards,
Huang Ying

#20 lore raw

Re: [PATCH 12/16] x86, mce: squash mce_intel.c into therm_throt.c

In reply to huang ying
Quoted 11 lines · huang ying

2009/6/15 Hidetoshi Seto <[email protected]>:

Quoted 9 lines

move intel_init_thermal() into therm_throt.c

---
 arch/x86/kernel/cpu/mcheck/Makefile      |    2 +-
 arch/x86/kernel/cpu/mcheck/mce_intel.c   |   73 ------------------------------
 arch/x86/kernel/cpu/mcheck/therm_throt.c |   66 +++++++++++++++++++++++++++
 3 files changed, 67 insertions(+), 74 deletions(-)
 delete mode 100644 arch/x86/kernel/cpu/mcheck/mce_intel.c
Quoted 4 lines

Why we need this patch? I think the design principle behind MCE is to separate vendor/cpu specific code into different files. At least #ifdef CONFIG_X86_MCE_INTEL should be used around moved code.

Honestly, such a design principle is pretty broken. mce_intel.c contained two completely orthogonal piece of functionality -- CMCI and thermal -- that really should be in different files (and mce_intel.c probably should be renamed cmci.c). Furthermore, using #ifdef for selecting specific code is considered less clean than using the Makefile for this purpose.

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
#19 lore raw

Re: [PATCH 12/16] x86, mce: squash mce_intel.c into therm_throt.c

In reply to huang ying
Quoted 8 lines · huang ying
Quoted 5 lines

You can pull out "codes may not be shared" from therm_throt.c and put it again into mce_intel.c (that was mce_intel_64.c). But I believe it is not needed now, until another code is available. Or you can rename therm_throt.c to thermal_intel.c or so for a while.

I think thermal_intel.c is better.

Actually intel_thermal.c and intel_cmci.c would probably be the right naming. The mce_ prefix for files in this directory is redundant; so for vendor-specific features the vendor is the higher order organizing factor.

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.