Date:	Thu, 30 Jul 2009 16:52:30 +0800
From:	Huang Ying <ying.huang@...el.com>
To:	"H. Peter Anvin" <hpa@...or.com>
Cc:	Ingo Molnar <mingo@...e.hu>, Andi Kleen <ak@...ux.intel.com>,
	linux-kernel@...r.kernel.org, Huang Ying <ying.huang@...el.com>
Subject: [PATCH -v4 1/4] x86: MCE: Support specifying context for software mce injection

The cpu context is specified via the new mce.inject_flags fields.
This allows more realistic machine check testing in different
situations. "RANDOM" context is implemented via NMI broadcasting to
add randomization to testing.

AK: Fix NMI broadcasting check. Fix 32-bit building. Some race
fixes. Move to module. Various changes

Signed-off-by: Huang Ying <ying.huang@...el.com>
Signed-off-by: Andi Kleen <ak@...ux.intel.com>


ChangeLog:

v3:

- Re-based on latest x86-tip.git/mce4

- Fix 32-bit building

v2:

- Re-base on latest x86-tip.git/mce3
---
 arch/x86/include/asm/mce.h              |   11 ++-
 arch/x86/kernel/cpu/mcheck/mce-inject.c |  156 +++++++++++++++++++++++++------
 2 files changed, 135 insertions(+), 32 deletions(-)

diff --git a/arch/x86/include/asm/mce.h b/arch/x86/include/asm/mce.h
index 5cdd8d1..7196cdc 100644
--- a/arch/x86/include/asm/mce.h
+++ b/arch/x86/include/asm/mce.h
@@ -38,6 +38,13 @@
 #define MCM_ADDR_MEM	 3	/* memory address */
 #define MCM_ADDR_GENERIC 7	/* generic */
 
+#define MCJ_CTX_MASK		3
+#define MCJ_CTX(flags)		((flags) & MCJ_CTX_MASK)
+#define MCJ_CTX_RANDOM		0    /* inject context: random */
+#define MCJ_CTX_PROCESS		1    /* inject context: process */
+#define MCJ_CTX_IRQ		2    /* inject context: IRQ */
+#define MCJ_NMI_BROADCAST	4    /* do NMI broadcasting */
+
 /* Fields are zero when not available */
 struct mce {
 	__u64 status;
@@ -48,8 +55,8 @@ struct mce {
 	__u64 tsc;	/* cpu time stamp counter */
 	__u64 time;	/* wall time_t when error was detected */
 	__u8  cpuvendor;	/* cpu vendor as encoded in system.h */
-	__u8  pad1;
-	__u16 pad2;
+	__u8  inject_flags;	/* software inject flags */
+	__u16  pad;
 	__u32 cpuid;	/* CPUID 1 EAX */
 	__u8  cs;		/* code segment */
 	__u8  bank;	/* machine check bank */
diff --git a/arch/x86/kernel/cpu/mcheck/mce-inject.c b/arch/x86/kernel/cpu/mcheck/mce-inject.c
index a3a235a..ad5d927 100644
--- a/arch/x86/kernel/cpu/mcheck/mce-inject.c
+++ b/arch/x86/kernel/cpu/mcheck/mce-inject.c
@@ -18,7 +18,12 @@
 #include <linux/string.h>
 #include <linux/fs.h>
 #include <linux/smp.h>
+#include <linux/notifier.h>
+#include <linux/kdebug.h>
+#include <linux/cpu.h>
+#include <linux/sched.h>
 #include <asm/mce.h>
+#include <asm/apic.h>
 
 /* Update fake mce registers on current CPU. */
 static void inject_mce(struct mce *m)
@@ -39,44 +44,141 @@ static void inject_mce(struct mce *m)
 	i->finished = 1;
 }
 
-struct delayed_mce {
-	struct timer_list timer;
-	struct mce m;
+static void raise_corrected(struct mce *m)
+{
+	unsigned long flags;
+	mce_banks_t b;
+
+	memset(&b, 0xff, sizeof(mce_banks_t));
+	local_irq_save(flags);
+	machine_check_poll(0, &b);
+	local_irq_restore(flags);
+	m->finished = 0;
+}
+
+static void raise_uncorrected(struct mce *m, struct pt_regs *pregs)
+{
+	struct pt_regs regs;
+	unsigned long flags;
+
+	if (!pregs) {
+		memset(&regs, 0, sizeof(struct pt_regs));
+		regs.ip = m->ip;
+		regs.cs = m->cs;
+		pregs = &regs;
+	}
+	/* in mcheck exeception handler, irq will be disabled */
+	local_irq_save(flags);
+	do_machine_check(pregs, 0);
+	local_irq_restore(flags);
+	m->finished = 0;
+}
+
+static cpumask_t mce_inject_cpumask;
+
+static int mce_raise_notify(struct notifier_block *self,
+			    unsigned long val, void *data)
+{
+	struct die_args *args = (struct die_args *)data;
+	int cpu = smp_processor_id();
+	struct mce *m = &__get_cpu_var(injectm);
+	if (val != DIE_NMI_IPI || !cpu_isset(cpu, mce_inject_cpumask))
+		return NOTIFY_DONE;
+	cpu_clear(cpu, mce_inject_cpumask);
+	if (m->status & MCI_STATUS_UC)
+		raise_uncorrected(m, args->regs);
+	else if (m->status)
+		raise_corrected(m);
+	return NOTIFY_STOP;
+}
+
+static struct notifier_block mce_raise_nb = {
+	.notifier_call = mce_raise_notify,
+	.priority = 1000,
 };
 
 /* Inject mce on current CPU */
-static void raise_mce(unsigned long data)
+static int raise_local(struct mce *m)
 {
-	struct delayed_mce *dm = (struct delayed_mce *)data;
-	struct mce *m = &dm->m;
+	int context = MCJ_CTX(m->inject_flags);
+	int ret = 0;
 	int cpu = m->extcpu;
 
-	inject_mce(m);
 	if (m->status & MCI_STATUS_UC) {
-		struct pt_regs regs;
-		memset(&regs, 0, sizeof(struct pt_regs));
-		regs.ip = m->ip;
-		regs.cs = m->cs;
 		printk(KERN_INFO "Triggering MCE exception on CPU %d\n", cpu);
-		do_machine_check(&regs, 0);
+		switch (context) {
+		case MCJ_CTX_IRQ:
+			/*
+			 * Could do more to fake interrupts like
+			 * calling irq_enter, but the necessary
+			 * machinery isn't exported currently.
+			 */
+			/*FALL THROUGH*/
+		case MCJ_CTX_PROCESS:
+			raise_uncorrected(m, NULL);
+			break;
+		default:
+			printk(KERN_INFO "Invalid MCE context\n");
+			ret = -EINVAL;
+		}
 		printk(KERN_INFO "MCE exception done on CPU %d\n", cpu);
-	} else {
-		mce_banks_t b;
-		memset(&b, 0xff, sizeof(mce_banks_t));
+	} else if (m->status) {
 		printk(KERN_INFO "Starting machine check poll CPU %d\n", cpu);
-		machine_check_poll(0, &b);
+		raise_corrected(m);
 		mce_notify_irq();
-		printk(KERN_INFO "Finished machine check poll on CPU %d\n",
-		       cpu);
-	}
-	kfree(dm);
+		printk(KERN_INFO "Machine check poll done on CPU %d\n", cpu);
+	} else
+		m->finished = 0;
+
+	return ret;
+}
+
+static void raise_mce(struct mce *m)
+{
+	int context = MCJ_CTX(m->inject_flags);
+
+	inject_mce(m);
+
+	if (context == MCJ_CTX_RANDOM)
+		return;
+
+#ifdef CONFIG_X86_LOCAL_APIC
+	if (m->inject_flags & MCJ_NMI_BROADCAST) {
+		unsigned long start;
+		int cpu;
+		get_online_cpus();
+		mce_inject_cpumask = cpu_online_map;
+		cpu_clear(get_cpu(), mce_inject_cpumask);
+		for_each_online_cpu(cpu) {
+			struct mce *mcpu = &per_cpu(injectm, cpu);
+			if (!mcpu->finished ||
+			    MCJ_CTX(mcpu->inject_flags) != MCJ_CTX_RANDOM)
+				cpu_clear(cpu, mce_inject_cpumask);
+		}
+		if (!cpus_empty(mce_inject_cpumask))
+			apic->send_IPI_mask(&mce_inject_cpumask, NMI_VECTOR);
+		start = jiffies;
+		while (!cpus_empty(mce_inject_cpumask)) {
+			if (!time_before(jiffies, start + 2*HZ)) {
+				printk(KERN_ERR
+				"Timeout waiting for mce inject NMI %lx\n",
+					*cpus_addr(mce_inject_cpumask));
+				break;
+			}
+			cpu_relax();
+		}
+		raise_local(m);
+		put_cpu();
+		put_online_cpus();
+	} else
+#endif
+		raise_local(m);
 }
 
 /* Error injection interface */
 static ssize_t mce_write(struct file *filp, const char __user *ubuf,
 			 size_t usize, loff_t *off)
 {
-	struct delayed_mce *dm;
 	struct mce m;
 
 	if (!capable(CAP_SYS_ADMIN))
@@ -96,19 +198,12 @@ static ssize_t mce_write(struct file *filp, const char __user *ubuf,
 	if (m.extcpu >= num_possible_cpus() || !cpu_online(m.extcpu))
 		return -EINVAL;
 
-	dm = kmalloc(sizeof(struct delayed_mce), GFP_KERNEL);
-	if (!dm)
-		return -ENOMEM;
-
 	/*
 	 * Need to give user space some time to set everything up,
 	 * so do it a jiffie or two later everywhere.
-	 * Should we use a hrtimer here for better synchronization?
 	 */
-	memcpy(&dm->m, &m, sizeof(struct mce));
-	setup_timer(&dm->timer, raise_mce, (unsigned long)dm);
-	dm->timer.expires = jiffies + 2;
-	add_timer_on(&dm->timer, m.extcpu);
+	schedule_timeout(2);
+	raise_mce(&m);
 	return usize;
 }
 
@@ -116,6 +211,7 @@ static int inject_init(void)
 {
 	printk(KERN_INFO "Machine check injector initialized\n");
 	mce_chrdev_ops.write = mce_write;
+	register_die_notifier(&mce_raise_nb);
 	return 0;
 }
 
-- 
1.6.3.3

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists