[PATCH v2 0/2] fix "perf top" soft lockups under Xen

[PATCH v2 0/2] fix "perf top" soft lockups under Xen

[Lin Ming]

v2: 
- fix compile issues if no CONFIG_SMP, Konrad Rzeszutek Wilk
- move inc_irq_stat after irq_work_run

These 2 patches fixed the "perf top" soft lockups under Xen
reported by Steven at: https://lkml.org/lkml/2012/2/9/506

Both Steven and I tested it and "perf top" works well now.

The soft lockup code path is:

__irq_work_queue
  arch_irq_work_raise
    apic->send_IPI_self(IRQ_WORK_VECTOR);
      apic_send_IPI_self
        __default_send_IPI_shortcut
          __xapic_wait_icr_idle

static inline void __xapic_wait_icr_idle(void)
{
        while (native_apic_mem_read(APIC_ICR) & APIC_ICR_BUSY)
                cpu_relax();
}
The lockup happens at above while looop.

The cause is that Xen has not implemented the APIC IPI interface yet.
Xen has IPI interface: xen_send_IPI_one, but it's only used in
xen_smp_send_reschedule, xen_smp_send_call_function_ipi and
xen_smp_send_call_function_single_ipi, etc.

So we need to implement Xen's APIC IPI interface as Ben's patch does.
And implement Xen's IRQ_WORK_VECTOR handler.

Ben Guthro (1):
      xen: implement apic ipi interface

Lin Ming (1):
      xen: implement IRQ_WORK_VECTOR handler

 arch/x86/include/asm/xen/events.h |    1 +
 arch/x86/xen/enlighten.c          |    9 +++
 arch/x86/xen/smp.c                |  111 +++++++++++++++++++++++++++++++++++-
 arch/x86/xen/smp.h                |   12 ++++
 4 files changed, 129 insertions(+), 4 deletions(-)

[PATCH v2 1/2] xen: implement apic ipi interface

[Lin Ming]

From: Ben Guthro <ben@guthro.net>

Map native ipi vector to xen vector.
Implement apic ipi interface with xen_send_IPI_one.

Tested-by: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Ben Guthro <ben@guthro.net>
Signed-off-by: Lin Ming <mlin@ss.pku.edu.cn>
---
 arch/x86/xen/enlighten.c |    9 +++++
 arch/x86/xen/smp.c       |   81 +++++++++++++++++++++++++++++++++++++++++++--
 arch/x86/xen/smp.h       |   12 +++++++
 3 files changed, 98 insertions(+), 4 deletions(-)
 create mode 100644 arch/x86/xen/smp.h

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 4f51beb..1ed61c2 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -74,6 +74,7 @@
 
 #include "xen-ops.h"
 #include "mmu.h"
+#include "smp.h"
 #include "multicalls.h"
 
 EXPORT_SYMBOL_GPL(hypercall_page);
@@ -849,6 +850,14 @@ static void set_xen_basic_apic_ops(void)
 	apic->icr_write = xen_apic_icr_write;
 	apic->wait_icr_idle = xen_apic_wait_icr_idle;
 	apic->safe_wait_icr_idle = xen_safe_apic_wait_icr_idle;
+
+#ifdef CONFIG_SMP
+	apic->send_IPI_allbutself = xen_send_IPI_allbutself;
+	apic->send_IPI_mask_allbutself = xen_send_IPI_mask_allbutself;
+	apic->send_IPI_mask = xen_send_IPI_mask;
+	apic->send_IPI_all = xen_send_IPI_all;
+	apic->send_IPI_self = xen_send_IPI_self;
+#endif
 }
 
 #endif
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 5fac691..2dc6628 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -465,8 +465,8 @@ static void xen_smp_send_reschedule(int cpu)
 	xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR);
 }
 
-static void xen_send_IPI_mask(const struct cpumask *mask,
-			      enum ipi_vector vector)
+static void __xen_send_IPI_mask(const struct cpumask *mask,
+			      int vector)
 {
 	unsigned cpu;
 
@@ -478,7 +478,7 @@ static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
 {
 	int cpu;
 
-	xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
+	__xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
 
 	/* Make sure other vcpus get a chance to run if they need to. */
 	for_each_cpu(cpu, mask) {
@@ -491,10 +491,83 @@ static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
 
 static void xen_smp_send_call_function_single_ipi(int cpu)
 {
-	xen_send_IPI_mask(cpumask_of(cpu),
+	__xen_send_IPI_mask(cpumask_of(cpu),
 			  XEN_CALL_FUNCTION_SINGLE_VECTOR);
 }
 
+static inline int xen_map_vector(int vector)
+{
+	int xen_vector;
+
+	switch (vector) {
+	case RESCHEDULE_VECTOR:
+		xen_vector = XEN_RESCHEDULE_VECTOR;
+		break;
+	case CALL_FUNCTION_VECTOR:
+		xen_vector = XEN_CALL_FUNCTION_VECTOR;
+		break;
+	case CALL_FUNCTION_SINGLE_VECTOR:
+		xen_vector = XEN_CALL_FUNCTION_SINGLE_VECTOR;
+		break;
+	default:
+		xen_vector = -1;
+		printk(KERN_ERR "xen: vector 0x%x is not implemented\n",
+			vector);
+	}
+
+	return xen_vector;
+}
+
+void xen_send_IPI_mask(const struct cpumask *mask,
+			      int vector)
+{
+	int xen_vector = xen_map_vector(vector);
+
+	if (xen_vector >= 0)
+		__xen_send_IPI_mask(mask, xen_vector);
+}
+
+void xen_send_IPI_all(int vector)
+{
+	int xen_vector = xen_map_vector(vector);
+
+	if (xen_vector >= 0)
+		__xen_send_IPI_mask(cpu_online_mask, xen_vector);
+}
+
+void xen_send_IPI_self(int vector)
+{
+	int xen_vector = xen_map_vector(vector);
+
+	if (xen_vector >= 0)
+		xen_send_IPI_one(smp_processor_id(), xen_vector);
+}
+
+void xen_send_IPI_mask_allbutself(const struct cpumask *mask,
+				int vector)
+{
+	unsigned cpu;
+	unsigned int this_cpu = smp_processor_id();
+
+	if (!(num_online_cpus() > 1))
+		return;
+
+	for_each_cpu_and(cpu, mask, cpu_online_mask) {
+		if (this_cpu == cpu)
+			continue;
+
+		xen_smp_send_call_function_single_ipi(cpu);
+	}
+}
+
+void xen_send_IPI_allbutself(int vector)
+{
+	int xen_vector = xen_map_vector(vector);
+
+	if (xen_vector >= 0)
+		xen_send_IPI_mask_allbutself(cpu_online_mask, xen_vector);
+}
+
 static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id)
 {
 	irq_enter();
diff --git a/arch/x86/xen/smp.h b/arch/x86/xen/smp.h
new file mode 100644
index 0000000..8981a76
--- /dev/null
+++ b/arch/x86/xen/smp.h
@@ -0,0 +1,12 @@
+#ifndef _XEN_SMP_H
+
+extern void xen_send_IPI_mask(const struct cpumask *mask,
+			      int vector);
+extern void xen_send_IPI_mask_allbutself(const struct cpumask *mask,
+				int vector);
+extern void xen_send_IPI_allbutself(int vector);
+extern void physflat_send_IPI_allbutself(int vector);
+extern void xen_send_IPI_all(int vector);
+extern void xen_send_IPI_self(int vector);
+
+#endif
-- 
1.7.2.5

[PATCH v2 1/2] xen: implement apic ipi interface

[Lin Ming]

From: Ben Guthro <ben@guthro.net>

Map native ipi vector to xen vector.
Implement apic ipi interface with xen_send_IPI_one.

Tested-by: Steven Noonan <steven@uplinklabs.net>
Signed-off-by: Ben Guthro <ben@guthro.net>
Signed-off-by: Lin Ming <mlin@ss.pku.edu.cn>
---
 arch/x86/xen/enlighten.c |    9 +++++
 arch/x86/xen/smp.c       |   81 +++++++++++++++++++++++++++++++++++++++++++--
 arch/x86/xen/smp.h       |   12 +++++++
 3 files changed, 98 insertions(+), 4 deletions(-)
 create mode 100644 arch/x86/xen/smp.h

diff --git a/arch/x86/xen/enlighten.c b/arch/x86/xen/enlighten.c
index 4f51beb..1ed61c2 100644
--- a/arch/x86/xen/enlighten.c
+++ b/arch/x86/xen/enlighten.c
@@ -74,6 +74,7 @@
 
 #include "xen-ops.h"
 #include "mmu.h"
+#include "smp.h"
 #include "multicalls.h"
 
 EXPORT_SYMBOL_GPL(hypercall_page);
@@ -849,6 +850,14 @@ static void set_xen_basic_apic_ops(void)
 	apic->icr_write = xen_apic_icr_write;
 	apic->wait_icr_idle = xen_apic_wait_icr_idle;
 	apic->safe_wait_icr_idle = xen_safe_apic_wait_icr_idle;
+
+#ifdef CONFIG_SMP
+	apic->send_IPI_allbutself = xen_send_IPI_allbutself;
+	apic->send_IPI_mask_allbutself = xen_send_IPI_mask_allbutself;
+	apic->send_IPI_mask = xen_send_IPI_mask;
+	apic->send_IPI_all = xen_send_IPI_all;
+	apic->send_IPI_self = xen_send_IPI_self;
+#endif
 }
 
 #endif
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 5fac691..2dc6628 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -465,8 +465,8 @@ static void xen_smp_send_reschedule(int cpu)
 	xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR);
 }
 
-static void xen_send_IPI_mask(const struct cpumask *mask,
-			      enum ipi_vector vector)
+static void __xen_send_IPI_mask(const struct cpumask *mask,
+			      int vector)
 {
 	unsigned cpu;
 
@@ -478,7 +478,7 @@ static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
 {
 	int cpu;
 
-	xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
+	__xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
 
 	/* Make sure other vcpus get a chance to run if they need to. */
 	for_each_cpu(cpu, mask) {
@@ -491,10 +491,83 @@ static void xen_smp_send_call_function_ipi(const struct cpumask *mask)
 
 static void xen_smp_send_call_function_single_ipi(int cpu)
 {
-	xen_send_IPI_mask(cpumask_of(cpu),
+	__xen_send_IPI_mask(cpumask_of(cpu),
 			  XEN_CALL_FUNCTION_SINGLE_VECTOR);
 }
 
+static inline int xen_map_vector(int vector)
+{
+	int xen_vector;
+
+	switch (vector) {
+	case RESCHEDULE_VECTOR:
+		xen_vector = XEN_RESCHEDULE_VECTOR;
+		break;
+	case CALL_FUNCTION_VECTOR:
+		xen_vector = XEN_CALL_FUNCTION_VECTOR;
+		break;
+	case CALL_FUNCTION_SINGLE_VECTOR:
+		xen_vector = XEN_CALL_FUNCTION_SINGLE_VECTOR;
+		break;
+	default:
+		xen_vector = -1;
+		printk(KERN_ERR "xen: vector 0x%x is not implemented\n",
+			vector);
+	}
+
+	return xen_vector;
+}
+
+void xen_send_IPI_mask(const struct cpumask *mask,
+			      int vector)
+{
+	int xen_vector = xen_map_vector(vector);
+
+	if (xen_vector >= 0)
+		__xen_send_IPI_mask(mask, xen_vector);
+}
+
+void xen_send_IPI_all(int vector)
+{
+	int xen_vector = xen_map_vector(vector);
+
+	if (xen_vector >= 0)
+		__xen_send_IPI_mask(cpu_online_mask, xen_vector);
+}
+
+void xen_send_IPI_self(int vector)
+{
+	int xen_vector = xen_map_vector(vector);
+
+	if (xen_vector >= 0)
+		xen_send_IPI_one(smp_processor_id(), xen_vector);
+}
+
+void xen_send_IPI_mask_allbutself(const struct cpumask *mask,
+				int vector)
+{
+	unsigned cpu;
+	unsigned int this_cpu = smp_processor_id();
+
+	if (!(num_online_cpus() > 1))
+		return;
+
+	for_each_cpu_and(cpu, mask, cpu_online_mask) {
+		if (this_cpu == cpu)
+			continue;
+
+		xen_smp_send_call_function_single_ipi(cpu);
+	}
+}
+
+void xen_send_IPI_allbutself(int vector)
+{
+	int xen_vector = xen_map_vector(vector);
+
+	if (xen_vector >= 0)
+		xen_send_IPI_mask_allbutself(cpu_online_mask, xen_vector);
+}
+
 static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id)
 {
 	irq_enter();
diff --git a/arch/x86/xen/smp.h b/arch/x86/xen/smp.h
new file mode 100644
index 0000000..8981a76
--- /dev/null
+++ b/arch/x86/xen/smp.h
@@ -0,0 +1,12 @@
+#ifndef _XEN_SMP_H
+
+extern void xen_send_IPI_mask(const struct cpumask *mask,
+			      int vector);
+extern void xen_send_IPI_mask_allbutself(const struct cpumask *mask,
+				int vector);
+extern void xen_send_IPI_allbutself(int vector);
+extern void physflat_send_IPI_allbutself(int vector);
+extern void xen_send_IPI_all(int vector);
+extern void xen_send_IPI_self(int vector);
+
+#endif
-- 
1.7.2.5

[PATCH v2 2/2] xen: implement IRQ_WORK_VECTOR handler

[Lin Ming]

Signed-off-by: Lin Ming <mlin@ss.pku.edu.cn>
---
 arch/x86/include/asm/xen/events.h |    1 +
 arch/x86/xen/smp.c                |   30 ++++++++++++++++++++++++++++++
 2 files changed, 31 insertions(+), 0 deletions(-)

diff --git a/arch/x86/include/asm/xen/events.h b/arch/x86/include/asm/xen/events.h
index 1df3541..cc146d5 100644
--- a/arch/x86/include/asm/xen/events.h
+++ b/arch/x86/include/asm/xen/events.h
@@ -6,6 +6,7 @@ enum ipi_vector {
 	XEN_CALL_FUNCTION_VECTOR,
 	XEN_CALL_FUNCTION_SINGLE_VECTOR,
 	XEN_SPIN_UNLOCK_VECTOR,
+	XEN_IRQ_WORK_VECTOR,
 
 	XEN_NR_IPIS,
 };
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 2dc6628..3ec3f8e 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -16,6 +16,7 @@
 #include <linux/err.h>
 #include <linux/slab.h>
 #include <linux/smp.h>
+#include <linux/irq_work.h>
 
 #include <asm/paravirt.h>
 #include <asm/desc.h>
@@ -41,10 +42,12 @@ cpumask_var_t xen_cpu_initialized_map;
 static DEFINE_PER_CPU(int, xen_resched_irq);
 static DEFINE_PER_CPU(int, xen_callfunc_irq);
 static DEFINE_PER_CPU(int, xen_callfuncsingle_irq);
+static DEFINE_PER_CPU(int, xen_irq_work);
 static DEFINE_PER_CPU(int, xen_debug_irq) = -1;
 
 static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id);
 static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id);
+static irqreturn_t xen_irq_work_interrupt(int irq, void *dev_id);
 
 /*
  * Reschedule call back.
@@ -143,6 +146,17 @@ static int xen_smp_intr_init(unsigned int cpu)
 		goto fail;
 	per_cpu(xen_callfuncsingle_irq, cpu) = rc;
 
+	callfunc_name = kasprintf(GFP_KERNEL, "irqwork%d", cpu);
+	rc = bind_ipi_to_irqhandler(XEN_IRQ_WORK_VECTOR,
+				    cpu,
+				    xen_irq_work_interrupt,
+				    IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
+				    callfunc_name,
+				    NULL);
+	if (rc < 0)
+		goto fail;
+	per_cpu(xen_irq_work, cpu) = rc;
+
 	return 0;
 
  fail:
@@ -155,6 +169,8 @@ static int xen_smp_intr_init(unsigned int cpu)
 	if (per_cpu(xen_callfuncsingle_irq, cpu) >= 0)
 		unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu),
 				       NULL);
+	if (per_cpu(xen_irq_work, cpu) >= 0)
+		unbind_from_irqhandler(per_cpu(xen_irq_work, cpu), NULL);
 
 	return rc;
 }
@@ -509,6 +525,9 @@ static inline int xen_map_vector(int vector)
 	case CALL_FUNCTION_SINGLE_VECTOR:
 		xen_vector = XEN_CALL_FUNCTION_SINGLE_VECTOR;
 		break;
+	case IRQ_WORK_VECTOR:
+		xen_vector = XEN_IRQ_WORK_VECTOR;
+		break;
 	default:
 		xen_vector = -1;
 		printk(KERN_ERR "xen: vector 0x%x is not implemented\n",
@@ -588,6 +607,16 @@ static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
+static irqreturn_t xen_irq_work_interrupt(int irq, void *dev_id)
+{
+	irq_enter();
+	irq_work_run();
+	inc_irq_stat(apic_irq_work_irqs);
+	irq_exit();
+
+	return IRQ_HANDLED;
+}
+
 static const struct smp_ops xen_smp_ops __initconst = {
 	.smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
 	.smp_prepare_cpus = xen_smp_prepare_cpus,
@@ -634,6 +663,7 @@ static void xen_hvm_cpu_die(unsigned int cpu)
 	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
 	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
 	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
+	unbind_from_irqhandler(per_cpu(xen_irq_work, cpu), NULL);
 	native_cpu_die(cpu);
 }
 
-- 
1.7.2.5

[PATCH v2 2/2] xen: implement IRQ_WORK_VECTOR handler

[Lin Ming]

Signed-off-by: Lin Ming <mlin@ss.pku.edu.cn>
---
 arch/x86/include/asm/xen/events.h |    1 +
 arch/x86/xen/smp.c                |   30 ++++++++++++++++++++++++++++++
 2 files changed, 31 insertions(+), 0 deletions(-)

diff --git a/arch/x86/include/asm/xen/events.h b/arch/x86/include/asm/xen/events.h
index 1df3541..cc146d5 100644
--- a/arch/x86/include/asm/xen/events.h
+++ b/arch/x86/include/asm/xen/events.h
@@ -6,6 +6,7 @@ enum ipi_vector {
 	XEN_CALL_FUNCTION_VECTOR,
 	XEN_CALL_FUNCTION_SINGLE_VECTOR,
 	XEN_SPIN_UNLOCK_VECTOR,
+	XEN_IRQ_WORK_VECTOR,
 
 	XEN_NR_IPIS,
 };
diff --git a/arch/x86/xen/smp.c b/arch/x86/xen/smp.c
index 2dc6628..3ec3f8e 100644
--- a/arch/x86/xen/smp.c
+++ b/arch/x86/xen/smp.c
@@ -16,6 +16,7 @@
 #include <linux/err.h>
 #include <linux/slab.h>
 #include <linux/smp.h>
+#include <linux/irq_work.h>
 
 #include <asm/paravirt.h>
 #include <asm/desc.h>
@@ -41,10 +42,12 @@ cpumask_var_t xen_cpu_initialized_map;
 static DEFINE_PER_CPU(int, xen_resched_irq);
 static DEFINE_PER_CPU(int, xen_callfunc_irq);
 static DEFINE_PER_CPU(int, xen_callfuncsingle_irq);
+static DEFINE_PER_CPU(int, xen_irq_work);
 static DEFINE_PER_CPU(int, xen_debug_irq) = -1;
 
 static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id);
 static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id);
+static irqreturn_t xen_irq_work_interrupt(int irq, void *dev_id);
 
 /*
  * Reschedule call back.
@@ -143,6 +146,17 @@ static int xen_smp_intr_init(unsigned int cpu)
 		goto fail;
 	per_cpu(xen_callfuncsingle_irq, cpu) = rc;
 
+	callfunc_name = kasprintf(GFP_KERNEL, "irqwork%d", cpu);
+	rc = bind_ipi_to_irqhandler(XEN_IRQ_WORK_VECTOR,
+				    cpu,
+				    xen_irq_work_interrupt,
+				    IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
+				    callfunc_name,
+				    NULL);
+	if (rc < 0)
+		goto fail;
+	per_cpu(xen_irq_work, cpu) = rc;
+
 	return 0;
 
  fail:
@@ -155,6 +169,8 @@ static int xen_smp_intr_init(unsigned int cpu)
 	if (per_cpu(xen_callfuncsingle_irq, cpu) >= 0)
 		unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu),
 				       NULL);
+	if (per_cpu(xen_irq_work, cpu) >= 0)
+		unbind_from_irqhandler(per_cpu(xen_irq_work, cpu), NULL);
 
 	return rc;
 }
@@ -509,6 +525,9 @@ static inline int xen_map_vector(int vector)
 	case CALL_FUNCTION_SINGLE_VECTOR:
 		xen_vector = XEN_CALL_FUNCTION_SINGLE_VECTOR;
 		break;
+	case IRQ_WORK_VECTOR:
+		xen_vector = XEN_IRQ_WORK_VECTOR;
+		break;
 	default:
 		xen_vector = -1;
 		printk(KERN_ERR "xen: vector 0x%x is not implemented\n",
@@ -588,6 +607,16 @@ static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
+static irqreturn_t xen_irq_work_interrupt(int irq, void *dev_id)
+{
+	irq_enter();
+	irq_work_run();
+	inc_irq_stat(apic_irq_work_irqs);
+	irq_exit();
+
+	return IRQ_HANDLED;
+}
+
 static const struct smp_ops xen_smp_ops __initconst = {
 	.smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
 	.smp_prepare_cpus = xen_smp_prepare_cpus,
@@ -634,6 +663,7 @@ static void xen_hvm_cpu_die(unsigned int cpu)
 	unbind_from_irqhandler(per_cpu(xen_callfunc_irq, cpu), NULL);
 	unbind_from_irqhandler(per_cpu(xen_debug_irq, cpu), NULL);
 	unbind_from_irqhandler(per_cpu(xen_callfuncsingle_irq, cpu), NULL);
+	unbind_from_irqhandler(per_cpu(xen_irq_work, cpu), NULL);
 	native_cpu_die(cpu);
 }
 
-- 
1.7.2.5

Re: [PATCH v2 0/2] fix "perf top" soft lockups under Xen

[Konrad Rzeszutek Wilk]

On Sat, Apr 21, 2012 at 12:11:03AM +0800, Lin Ming wrote:
> v2: 
> - fix compile issues if no CONFIG_SMP, Konrad Rzeszutek Wilk
> - move inc_irq_stat after irq_work_run

They look good to me (and they work nicely - thanks!)
so putting in the 3.5  queue.

> 
> These 2 patches fixed the "perf top" soft lockups under Xen
> reported by Steven at: https://lkml.org/lkml/2012/2/9/506
> 
> Both Steven and I tested it and "perf top" works well now.
> 
> The soft lockup code path is:
> 
> __irq_work_queue
>   arch_irq_work_raise
>     apic->send_IPI_self(IRQ_WORK_VECTOR);
>       apic_send_IPI_self
>         __default_send_IPI_shortcut
>           __xapic_wait_icr_idle
> 
> static inline void __xapic_wait_icr_idle(void)
> {
>         while (native_apic_mem_read(APIC_ICR) & APIC_ICR_BUSY)
>                 cpu_relax();
> }
> The lockup happens at above while looop.
> 
> The cause is that Xen has not implemented the APIC IPI interface yet.
> Xen has IPI interface: xen_send_IPI_one, but it's only used in
> xen_smp_send_reschedule, xen_smp_send_call_function_ipi and
> xen_smp_send_call_function_single_ipi, etc.
> 
> So we need to implement Xen's APIC IPI interface as Ben's patch does.
> And implement Xen's IRQ_WORK_VECTOR handler.
> 
> Ben Guthro (1):
>       xen: implement apic ipi interface
> 
> Lin Ming (1):
>       xen: implement IRQ_WORK_VECTOR handler
> 
>  arch/x86/include/asm/xen/events.h |    1 +
>  arch/x86/xen/enlighten.c          |    9 +++
>  arch/x86/xen/smp.c                |  111 +++++++++++++++++++++++++++++++++++-
>  arch/x86/xen/smp.h                |   12 ++++
>  4 files changed, 129 insertions(+), 4 deletions(-)

Re: [PATCH v2 0/2] fix "perf top" soft lockups under Xen

[Konrad Rzeszutek Wilk]

On Sat, Apr 21, 2012 at 12:11:03AM +0800, Lin Ming wrote:
> v2: 
> - fix compile issues if no CONFIG_SMP, Konrad Rzeszutek Wilk
> - move inc_irq_stat after irq_work_run

They look good to me (and they work nicely - thanks!)
so putting in the 3.5  queue.

> 
> These 2 patches fixed the "perf top" soft lockups under Xen
> reported by Steven at: https://lkml.org/lkml/2012/2/9/506
> 
> Both Steven and I tested it and "perf top" works well now.
> 
> The soft lockup code path is:
> 
> __irq_work_queue
>   arch_irq_work_raise
>     apic->send_IPI_self(IRQ_WORK_VECTOR);
>       apic_send_IPI_self
>         __default_send_IPI_shortcut
>           __xapic_wait_icr_idle
> 
> static inline void __xapic_wait_icr_idle(void)
> {
>         while (native_apic_mem_read(APIC_ICR) & APIC_ICR_BUSY)
>                 cpu_relax();
> }
> The lockup happens at above while looop.
> 
> The cause is that Xen has not implemented the APIC IPI interface yet.
> Xen has IPI interface: xen_send_IPI_one, but it's only used in
> xen_smp_send_reschedule, xen_smp_send_call_function_ipi and
> xen_smp_send_call_function_single_ipi, etc.
> 
> So we need to implement Xen's APIC IPI interface as Ben's patch does.
> And implement Xen's IRQ_WORK_VECTOR handler.
> 
> Ben Guthro (1):
>       xen: implement apic ipi interface
> 
> Lin Ming (1):
>       xen: implement IRQ_WORK_VECTOR handler
> 
>  arch/x86/include/asm/xen/events.h |    1 +
>  arch/x86/xen/enlighten.c          |    9 +++
>  arch/x86/xen/smp.c                |  111 +++++++++++++++++++++++++++++++++++-
>  arch/x86/xen/smp.h                |   12 ++++
>  4 files changed, 129 insertions(+), 4 deletions(-)