From 0b8255e660a0c229ebfe8f9fde12a8d4d34c50e0 Mon Sep 17 00:00:00 2001
From: Suresh Siddha <suresh.b.siddha@intel.com>
Date: Mon, 21 May 2012 16:58:02 -0700
Subject: x86/x2apic/cluster: Use all the members of one cluster specified in
 the smp_affinity mask for the interrupt destination

If the HW implements round-robin interrupt delivery, this
enables multiple cpu's (which are part of the user specified
interrupt smp_affinity mask and belong to the same x2apic
cluster) to service the interrupt.

Also if the platform supports Power Aware Interrupt Routing,
then this enables the interrupt to be routed to an idle cpu or a
busy cpu depending on the perf/power bias tunable.

We are now grouping all the cpu's in a cluster to one vector
domain. So that will limit the total number of interrupt sources
handled by Linux. Previously we support "cpu-count *
available-vectors-per-cpu" interrupt sources but this will now
reduce to "cpu-count/16 * available-vectors-per-cpu".

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: yinghai@kernel.org
Cc: gorcunov@openvz.org
Cc: agordeev@redhat.com
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1337644682-19854-2-git-send-email-suresh.b.siddha@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/apic/x2apic_cluster.c | 56 ++++++++++++++++++++++++-----------
 1 file changed, 39 insertions(+), 17 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index ff35cff0e1a..90d999c7f2e 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -98,34 +98,47 @@ static void x2apic_send_IPI_all(int vector)
 
 static unsigned int x2apic_cpu_mask_to_apicid(const struct cpumask *cpumask)
 {
-	/*
-	 * We're using fixed IRQ delivery, can only return one logical APIC ID.
-	 * May as well be the first.
-	 */
 	int cpu = cpumask_first(cpumask);
+	u32 dest = 0;
+	int i;
 
-	if ((unsigned)cpu < nr_cpu_ids)
-		return per_cpu(x86_cpu_to_logical_apicid, cpu);
-	else
+	if (cpu > nr_cpu_ids)
 		return BAD_APICID;
+
+	for_each_cpu_and(i, cpumask, per_cpu(cpus_in_cluster, cpu))
+		dest |= per_cpu(x86_cpu_to_logical_apicid, i);
+
+	return dest;
 }
 
 static unsigned int
 x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			      const struct cpumask *andmask)
 {
-	int cpu;
+	u32 dest = 0;
+	u16 cluster;
+	int i;
 
-	/*
-	 * We're using fixed IRQ delivery, can only return one logical APIC ID.
-	 * May as well be the first.
-	 */
-	for_each_cpu_and(cpu, cpumask, andmask) {
-		if (cpumask_test_cpu(cpu, cpu_online_mask))
-			break;
+	for_each_cpu_and(i, cpumask, andmask) {
+		if (!cpumask_test_cpu(i, cpu_online_mask))
+			continue;
+		dest = per_cpu(x86_cpu_to_logical_apicid, i);
+		cluster = x2apic_cluster(i);
+		break;
 	}
 
-	return per_cpu(x86_cpu_to_logical_apicid, cpu);
+	if (!dest)
+		return BAD_APICID;
+
+	for_each_cpu_and(i, cpumask, andmask) {
+		if (!cpumask_test_cpu(i, cpu_online_mask))
+			continue;
+		if (cluster != x2apic_cluster(i))
+			continue;
+		dest |= per_cpu(x86_cpu_to_logical_apicid, i);
+	}
+
+	return dest;
 }
 
 static void init_x2apic_ldr(void)
@@ -208,6 +221,15 @@ static int x2apic_cluster_probe(void)
 		return 0;
 }
 
+/*
+ * Each x2apic cluster is an allocation domain.
+ */
+static void cluster_vector_allocation_domain(int cpu, struct cpumask *retmask)
+{
+	cpumask_clear(retmask);
+	cpumask_copy(retmask, per_cpu(cpus_in_cluster, cpu));
+}
+
 static struct apic apic_x2apic_cluster = {
 
 	.name				= "cluster x2apic",
@@ -225,7 +247,7 @@ static struct apic apic_x2apic_cluster = {
 	.check_apicid_used		= NULL,
 	.check_apicid_present		= NULL,
 
-	.vector_allocation_domain	= x2apic_vector_allocation_domain,
+	.vector_allocation_domain	= cluster_vector_allocation_domain,
 	.init_apic_ldr			= init_x2apic_ldr,
 
 	.ioapic_phys_id_map		= NULL,
-- 
cgit v1.2.3-70-g09d2


From 49d0c7a0a425a89190b7c3b1445faba9eb227bec Mon Sep 17 00:00:00 2001
From: Alexander Gordeev <agordeev@redhat.com>
Date: Tue, 5 Jun 2012 13:23:15 +0200
Subject: x86/apic: Trivial whitespace fixes

Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/20120605112310.GA11443@dhcp-26-207.brq.redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/apic/apic_flat_64.c   | 2 +-
 arch/x86/kernel/apic/io_apic.c        | 4 ++--
 arch/x86/kernel/apic/x2apic_cluster.c | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/kernel/apic/apic_flat_64.c b/arch/x86/kernel/apic/apic_flat_64.c
index 0e881c46e8c..de279b32ceb 100644
--- a/arch/x86/kernel/apic/apic_flat_64.c
+++ b/arch/x86/kernel/apic/apic_flat_64.c
@@ -92,7 +92,7 @@ static void flat_send_IPI_mask(const struct cpumask *cpumask, int vector)
 }
 
 static void
- flat_send_IPI_mask_allbutself(const struct cpumask *cpumask, int vector)
+flat_send_IPI_mask_allbutself(const struct cpumask *cpumask, int vector)
 {
 	unsigned long mask = cpumask_bits(cpumask)[0];
 	int cpu = smp_processor_id();
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 910a3118438..74c569791e7 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1363,7 +1363,7 @@ static void setup_ioapic_irq(unsigned int irq, struct irq_cfg *cfg,
 		    cfg->vector, irq, attr->trigger, attr->polarity, dest);
 
 	if (setup_ioapic_entry(irq, &entry, dest, cfg->vector, attr)) {
-		pr_warn("Failed to setup ioapic entry for ioapic  %d, pin %d\n",
+		pr_warn("Failed to setup ioapic entry for ioapic %d, pin %d\n",
 			mpc_ioapic_id(attr->ioapic), attr->ioapic_pin);
 		__clear_irq_vector(irq, cfg);
 
@@ -1466,7 +1466,7 @@ void setup_IO_APIC_irq_extra(u32 gsi)
  * Set up the timer pin, possibly with the 8259A-master behind.
  */
 static void __init setup_timer_IRQ0_pin(unsigned int ioapic_idx,
-					 unsigned int pin, int vector)
+					unsigned int pin, int vector)
 {
 	struct IO_APIC_route_entry entry;
 
diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index 90d999c7f2e..2919e45d30c 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -81,7 +81,7 @@ static void x2apic_send_IPI_mask(const struct cpumask *mask, int vector)
 }
 
 static void
- x2apic_send_IPI_mask_allbutself(const struct cpumask *mask, int vector)
+x2apic_send_IPI_mask_allbutself(const struct cpumask *mask, int vector)
 {
 	__x2apic_send_IPI_mask(mask, vector, APIC_DEST_ALLBUT);
 }
-- 
cgit v1.2.3-70-g09d2


From bf721d3a3bc7a731add45c8078b142b494ab413e Mon Sep 17 00:00:00 2001
From: Alexander Gordeev <agordeev@redhat.com>
Date: Tue, 5 Jun 2012 13:23:29 +0200
Subject: x86/apic: Factor out default target_cpus() operation

Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/20120605112324.GA11449@dhcp-26-207.brq.redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/include/asm/apic.h           |  5 +++++
 arch/x86/include/asm/x2apic.h         |  9 ---------
 arch/x86/kernel/apic/apic_flat_64.c   | 14 ++------------
 arch/x86/kernel/apic/apic_numachip.c  |  7 +------
 arch/x86/kernel/apic/bigsmp_32.c      | 11 +----------
 arch/x86/kernel/apic/x2apic_cluster.c |  2 +-
 arch/x86/kernel/apic/x2apic_phys.c    |  2 +-
 arch/x86/kernel/apic/x2apic_uv_x.c    |  7 +------
 8 files changed, 12 insertions(+), 45 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index eaff4790ed9..fc38195d640 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -537,6 +537,11 @@ static inline const struct cpumask *default_target_cpus(void)
 #endif
 }
 
+static inline const struct cpumask *online_target_cpus(void)
+{
+	return cpu_online_mask;
+}
+
 DECLARE_EARLY_PER_CPU(u16, x86_bios_cpu_apicid);
 
 
diff --git a/arch/x86/include/asm/x2apic.h b/arch/x86/include/asm/x2apic.h
index 7a5a832a99b..f90f0a587c6 100644
--- a/arch/x86/include/asm/x2apic.h
+++ b/arch/x86/include/asm/x2apic.h
@@ -9,15 +9,6 @@
 #include <asm/ipi.h>
 #include <linux/cpumask.h>
 
-/*
- * Need to use more than cpu 0, because we need more vectors
- * when MSI-X are used.
- */
-static const struct cpumask *x2apic_target_cpus(void)
-{
-	return cpu_online_mask;
-}
-
 static int x2apic_apic_id_valid(int apicid)
 {
 	return 1;
diff --git a/arch/x86/kernel/apic/apic_flat_64.c b/arch/x86/kernel/apic/apic_flat_64.c
index de279b32ceb..61ac1afeff0 100644
--- a/arch/x86/kernel/apic/apic_flat_64.c
+++ b/arch/x86/kernel/apic/apic_flat_64.c
@@ -36,11 +36,6 @@ static int flat_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
 	return 1;
 }
 
-static const struct cpumask *flat_target_cpus(void)
-{
-	return cpu_online_mask;
-}
-
 static void flat_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	/* Careful. Some cpus do not strictly honor the set of cpus
@@ -186,7 +181,7 @@ static struct apic apic_flat =  {
 	.irq_delivery_mode		= dest_LowestPrio,
 	.irq_dest_mode			= 1, /* logical */
 
-	.target_cpus			= flat_target_cpus,
+	.target_cpus			= online_target_cpus,
 	.disable_esr			= 0,
 	.dest_logical			= APIC_DEST_LOGICAL,
 	.check_apicid_used		= NULL,
@@ -262,11 +257,6 @@ static int physflat_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
 	return 0;
 }
 
-static const struct cpumask *physflat_target_cpus(void)
-{
-	return cpu_online_mask;
-}
-
 static void physflat_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	cpumask_clear(retmask);
@@ -345,7 +335,7 @@ static struct apic apic_physflat =  {
 	.irq_delivery_mode		= dest_Fixed,
 	.irq_dest_mode			= 0, /* physical */
 
-	.target_cpus			= physflat_target_cpus,
+	.target_cpus			= online_target_cpus,
 	.disable_esr			= 0,
 	.dest_logical			= 0,
 	.check_apicid_used		= NULL,
diff --git a/arch/x86/kernel/apic/apic_numachip.c b/arch/x86/kernel/apic/apic_numachip.c
index 6ec6d5d297c..3255a60fcc9 100644
--- a/arch/x86/kernel/apic/apic_numachip.c
+++ b/arch/x86/kernel/apic/apic_numachip.c
@@ -72,11 +72,6 @@ static int numachip_phys_pkg_id(int initial_apic_id, int index_msb)
 	return initial_apic_id >> index_msb;
 }
 
-static const struct cpumask *numachip_target_cpus(void)
-{
-	return cpu_online_mask;
-}
-
 static void numachip_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	cpumask_clear(retmask);
@@ -253,7 +248,7 @@ static struct apic apic_numachip __refconst = {
 	.irq_delivery_mode		= dest_Fixed,
 	.irq_dest_mode			= 0, /* physical */
 
-	.target_cpus			= numachip_target_cpus,
+	.target_cpus			= online_target_cpus,
 	.disable_esr			= 0,
 	.dest_logical			= 0,
 	.check_apicid_used		= NULL,
diff --git a/arch/x86/kernel/apic/bigsmp_32.c b/arch/x86/kernel/apic/bigsmp_32.c
index 31fbdbfbf96..c288e81e00f 100644
--- a/arch/x86/kernel/apic/bigsmp_32.c
+++ b/arch/x86/kernel/apic/bigsmp_32.c
@@ -26,15 +26,6 @@ static int bigsmp_apic_id_registered(void)
 	return 1;
 }
 
-static const struct cpumask *bigsmp_target_cpus(void)
-{
-#ifdef CONFIG_SMP
-	return cpu_online_mask;
-#else
-	return cpumask_of(0);
-#endif
-}
-
 static unsigned long bigsmp_check_apicid_used(physid_mask_t *map, int apicid)
 {
 	return 0;
@@ -205,7 +196,7 @@ static struct apic apic_bigsmp = {
 	/* phys delivery to target CPU: */
 	.irq_dest_mode			= 0,
 
-	.target_cpus			= bigsmp_target_cpus,
+	.target_cpus			= default_target_cpus,
 	.disable_esr			= 1,
 	.dest_logical			= 0,
 	.check_apicid_used		= bigsmp_check_apicid_used,
diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index 2919e45d30c..612622c47df 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -241,7 +241,7 @@ static struct apic apic_x2apic_cluster = {
 	.irq_delivery_mode		= dest_LowestPrio,
 	.irq_dest_mode			= 1, /* logical */
 
-	.target_cpus			= x2apic_target_cpus,
+	.target_cpus			= online_target_cpus,
 	.disable_esr			= 0,
 	.dest_logical			= APIC_DEST_LOGICAL,
 	.check_apicid_used		= NULL,
diff --git a/arch/x86/kernel/apic/x2apic_phys.c b/arch/x86/kernel/apic/x2apic_phys.c
index 93b25706f17..b1a8b39e3c3 100644
--- a/arch/x86/kernel/apic/x2apic_phys.c
+++ b/arch/x86/kernel/apic/x2apic_phys.c
@@ -140,7 +140,7 @@ static struct apic apic_x2apic_phys = {
 	.irq_delivery_mode		= dest_Fixed,
 	.irq_dest_mode			= 0, /* physical */
 
-	.target_cpus			= x2apic_target_cpus,
+	.target_cpus			= online_target_cpus,
 	.disable_esr			= 0,
 	.dest_logical			= 0,
 	.check_apicid_used		= NULL,
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index c6d03f7a440..16efb92bfea 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -185,11 +185,6 @@ EXPORT_SYMBOL_GPL(uv_possible_blades);
 unsigned long sn_rtc_cycles_per_second;
 EXPORT_SYMBOL(sn_rtc_cycles_per_second);
 
-static const struct cpumask *uv_target_cpus(void)
-{
-	return cpu_online_mask;
-}
-
 static void uv_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	cpumask_clear(retmask);
@@ -362,7 +357,7 @@ static struct apic __refdata apic_x2apic_uv_x = {
 	.irq_delivery_mode		= dest_Fixed,
 	.irq_dest_mode			= 0, /* physical */
 
-	.target_cpus			= uv_target_cpus,
+	.target_cpus			= online_target_cpus,
 	.disable_esr			= 0,
 	.dest_logical			= APIC_DEST_LOGICAL,
 	.check_apicid_used		= NULL,
-- 
cgit v1.2.3-70-g09d2


From ff164324123c0fe181d8de7dadcc7b3fbe25f2cf Mon Sep 17 00:00:00 2001
From: Alexander Gordeev <agordeev@redhat.com>
Date: Thu, 7 Jun 2012 15:15:59 +0200
Subject: x86/apic: Make cpu_mask_to_apicid() operations return error code

Current cpu_mask_to_apicid() and cpu_mask_to_apicid_and()
implementations have few shortcomings:

1. A value returned by cpu_mask_to_apicid() is written to
hardware registers unconditionally. Should BAD_APICID get ever
returned it will be written to a hardware too. But the value of
BAD_APICID is not universal across all hardware in all modes and
might cause unexpected results, i.e. interrupts might get routed
to CPUs that are not configured to receive it.

2. Because the value of BAD_APICID is not universal it is
counter- intuitive to return it for a hardware where it does not
make sense (i.e. x2apic).

3. cpu_mask_to_apicid_and() operation is thought as an
complement to cpu_mask_to_apicid() that only applies a AND mask
on top of a cpumask being passed. Yet, as consequence of 18374d8
commit the two operations are inconsistent in that of:
  cpu_mask_to_apicid() should not get a offline CPU with the cpumask
  cpu_mask_to_apicid_and() should not fail and return BAD_APICID
These limitations are impossible to realize just from looking at
the operations prototypes.

Most of these shortcomings are resolved by returning a error
code instead of BAD_APICID. As the result, faults are reported
back early rather than possibilities to cause a unexpected
behaviour exist (in case of [1]).

The only exception is setup_timer_IRQ0_pin() routine. Although
obviously controversial to this fix, its existing behaviour is
preserved to not break the fragile check_timer() and would
better addressed in a separate fix.

Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Acked-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/20120607131559.GF4759@dhcp-26-207.brq.redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/include/asm/apic.h           | 44 ++++++++++++------
 arch/x86/kernel/apic/apic.c           | 33 +++++++------
 arch/x86/kernel/apic/es7000_32.c      | 21 +++++----
 arch/x86/kernel/apic/io_apic.c        | 88 +++++++++++++++++++++++------------
 arch/x86/kernel/apic/numaq_32.c       | 14 ++++--
 arch/x86/kernel/apic/summit_32.c      | 22 +++++----
 arch/x86/kernel/apic/x2apic_cluster.c | 24 ++++++----
 arch/x86/kernel/apic/x2apic_uv_x.c    | 27 +++++++----
 arch/x86/platform/uv/uv_irq.c         |  7 ++-
 drivers/iommu/intel_irq_remapping.c   | 13 ++++--
 10 files changed, 188 insertions(+), 105 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index e3fecd50d5c..ae91f9c7e36 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -331,9 +331,11 @@ struct apic {
 	unsigned long (*set_apic_id)(unsigned int id);
 	unsigned long apic_id_mask;
 
-	unsigned int (*cpu_mask_to_apicid)(const struct cpumask *cpumask);
-	unsigned int (*cpu_mask_to_apicid_and)(const struct cpumask *cpumask,
-					       const struct cpumask *andmask);
+	int (*cpu_mask_to_apicid)(const struct cpumask *cpumask,
+				  unsigned int *apicid);
+	int (*cpu_mask_to_apicid_and)(const struct cpumask *cpumask,
+				      const struct cpumask *andmask,
+				      unsigned int *apicid);
 
 	/* ipi */
 	void (*send_IPI_mask)(const struct cpumask *mask, int vector);
@@ -591,29 +593,45 @@ static inline int default_phys_pkg_id(int cpuid_apic, int index_msb)
 
 #endif
 
-static inline unsigned int
-flat_cpu_mask_to_apicid(const struct cpumask *cpumask)
+static inline int
+__flat_cpu_mask_to_apicid(unsigned long cpu_mask, unsigned int *apicid)
 {
-	return cpumask_bits(cpumask)[0] & APIC_ALL_CPUS;
+	cpu_mask &= APIC_ALL_CPUS;
+	if (likely(cpu_mask)) {
+		*apicid = (unsigned int)cpu_mask;
+		return 0;
+	} else {
+		return -EINVAL;
+	}
 }
 
-static inline unsigned int
+static inline int
+flat_cpu_mask_to_apicid(const struct cpumask *cpumask,
+			unsigned int *apicid)
+{
+	return __flat_cpu_mask_to_apicid(cpumask_bits(cpumask)[0], apicid);
+}
+
+static inline int
 flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
-			    const struct cpumask *andmask)
+			    const struct cpumask *andmask,
+			    unsigned int *apicid)
 {
 	unsigned long mask1 = cpumask_bits(cpumask)[0];
 	unsigned long mask2 = cpumask_bits(andmask)[0];
 	unsigned long mask3 = cpumask_bits(cpu_online_mask)[0];
 
-	return (unsigned int)(mask1 & mask2 & mask3);
+	return __flat_cpu_mask_to_apicid(mask1 & mask2 & mask3, apicid);
 }
 
-extern unsigned int
-default_cpu_mask_to_apicid(const struct cpumask *cpumask);
+extern int
+default_cpu_mask_to_apicid(const struct cpumask *cpumask,
+			   unsigned int *apicid);
 
-extern unsigned int
+extern int
 default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
-			       const struct cpumask *andmask);
+			       const struct cpumask *andmask,
+			       unsigned int *apicid);
 
 static inline bool
 flat_vector_allocation_domain(int cpu, struct cpumask *retmask)
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 96a2608252f..b8d92606f84 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -2123,24 +2123,26 @@ void default_init_apic_ldr(void)
 	apic_write(APIC_LDR, val);
 }
 
-unsigned int default_cpu_mask_to_apicid(const struct cpumask *cpumask)
+static inline int __default_cpu_to_apicid(int cpu, unsigned int *apicid)
 {
-	int cpu;
-
-	/*
-	 * We're using fixed IRQ delivery, can only return one phys APIC ID.
-	 * May as well be the first.
-	 */
-	cpu = cpumask_first(cpumask);
-	if (likely((unsigned)cpu < nr_cpu_ids))
-		return per_cpu(x86_cpu_to_apicid, cpu);
+	if (likely((unsigned int)cpu < nr_cpu_ids)) {
+		*apicid = per_cpu(x86_cpu_to_apicid, cpu);
+		return 0;
+	} else {
+		return -EINVAL;
+	}
+}
 
-	return BAD_APICID;
+int default_cpu_mask_to_apicid(const struct cpumask *cpumask,
+			       unsigned int *apicid)
+{
+	int cpu = cpumask_first(cpumask);
+	return __default_cpu_to_apicid(cpu, apicid);
 }
 
-unsigned int
-default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
-			       const struct cpumask *andmask)
+int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
+				   const struct cpumask *andmask,
+				   unsigned int *apicid)
 {
 	int cpu;
 
@@ -2148,7 +2150,8 @@ default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 		if (cpumask_test_cpu(cpu, cpu_online_mask))
 			break;
 	}
-	return per_cpu(x86_cpu_to_apicid, cpu);
+
+	return __default_cpu_to_apicid(cpu, apicid);
 }
 
 /*
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index 3c42865757e..515ebb00a9f 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -525,7 +525,8 @@ static int es7000_check_phys_apicid_present(int cpu_physical_apicid)
 	return 1;
 }
 
-static unsigned int es7000_cpu_mask_to_apicid(const struct cpumask *cpumask)
+static int
+es7000_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *dest_id)
 {
 	unsigned int round = 0;
 	int cpu, uninitialized_var(apicid);
@@ -539,31 +540,33 @@ static unsigned int es7000_cpu_mask_to_apicid(const struct cpumask *cpumask)
 		if (round && APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
 			WARN(1, "Not a valid mask!");
 
-			return BAD_APICID;
+			return -EINVAL;
 		}
 		apicid = new_apicid;
 		round++;
 	}
-	return apicid;
+	*dest_id = apicid;
+	return 0;
 }
 
-static unsigned int
+static int
 es7000_cpu_mask_to_apicid_and(const struct cpumask *inmask,
-			      const struct cpumask *andmask)
+			      const struct cpumask *andmask,
+			      unsigned int *apicid)
 {
-	int apicid = early_per_cpu(x86_cpu_to_logical_apicid, 0);
+	*apicid = early_per_cpu(x86_cpu_to_logical_apicid, 0);
 	cpumask_var_t cpumask;
 
 	if (!alloc_cpumask_var(&cpumask, GFP_ATOMIC))
-		return apicid;
+		return 0;
 
 	cpumask_and(cpumask, inmask, andmask);
 	cpumask_and(cpumask, cpumask, cpu_online_mask);
-	apicid = es7000_cpu_mask_to_apicid(cpumask);
+	es7000_cpu_mask_to_apicid(cpumask, apicid);
 
 	free_cpumask_var(cpumask);
 
-	return apicid;
+	return 0;
 }
 
 static int es7000_phys_pkg_id(int cpuid_apic, int index_msb)
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 4061a7dee5c..0deb773404e 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1359,7 +1359,14 @@ static void setup_ioapic_irq(unsigned int irq, struct irq_cfg *cfg,
 	if (assign_irq_vector(irq, cfg, apic->target_cpus()))
 		return;
 
-	dest = apic->cpu_mask_to_apicid_and(cfg->domain, apic->target_cpus());
+	if (apic->cpu_mask_to_apicid_and(cfg->domain, apic->target_cpus(),
+					 &dest)) {
+		pr_warn("Failed to obtain apicid for ioapic %d, pin %d\n",
+			mpc_ioapic_id(attr->ioapic), attr->ioapic_pin);
+		__clear_irq_vector(irq, cfg);
+
+		return;
+	}
 
 	apic_printk(APIC_VERBOSE,KERN_DEBUG
 		    "IOAPIC[%d]: Set routing entry (%d-%d -> 0x%x -> "
@@ -1474,6 +1481,7 @@ static void __init setup_timer_IRQ0_pin(unsigned int ioapic_idx,
 					unsigned int pin, int vector)
 {
 	struct IO_APIC_route_entry entry;
+	unsigned int dest;
 
 	if (irq_remapping_enabled)
 		return;
@@ -1484,9 +1492,12 @@ static void __init setup_timer_IRQ0_pin(unsigned int ioapic_idx,
 	 * We use logical delivery to get the timer IRQ
 	 * to the first CPU.
 	 */
+	if (unlikely(apic->cpu_mask_to_apicid(apic->target_cpus(), &dest)))
+		dest = BAD_APICID;
+
 	entry.dest_mode = apic->irq_dest_mode;
 	entry.mask = 0;			/* don't mask IRQ for edge */
-	entry.dest = apic->cpu_mask_to_apicid(apic->target_cpus());
+	entry.dest = dest;
 	entry.delivery_mode = apic->irq_delivery_mode;
 	entry.polarity = 0;
 	entry.trigger = 0;
@@ -2245,16 +2256,25 @@ int __ioapic_set_affinity(struct irq_data *data, const struct cpumask *mask,
 			  unsigned int *dest_id)
 {
 	struct irq_cfg *cfg = data->chip_data;
+	unsigned int irq = data->irq;
+	int err;
 
 	if (!cpumask_intersects(mask, cpu_online_mask))
-		return -1;
+		return -EINVAL;
 
-	if (assign_irq_vector(data->irq, data->chip_data, mask))
-		return -1;
+	err = assign_irq_vector(irq, cfg, mask);
+	if (err)
+		return err;
+
+	err = apic->cpu_mask_to_apicid_and(mask, cfg->domain, dest_id);
+	if (err) {
+		if (assign_irq_vector(irq, cfg, data->affinity))
+			pr_err("Failed to recover vector for irq %d\n", irq);
+		return err;
+	}
 
 	cpumask_copy(data->affinity, mask);
 
-	*dest_id = apic->cpu_mask_to_apicid_and(mask, cfg->domain);
 	return 0;
 }
 
@@ -3040,7 +3060,10 @@ static int msi_compose_msg(struct pci_dev *pdev, unsigned int irq,
 	if (err)
 		return err;
 
-	dest = apic->cpu_mask_to_apicid_and(cfg->domain, apic->target_cpus());
+	err = apic->cpu_mask_to_apicid_and(cfg->domain,
+					   apic->target_cpus(), &dest);
+	if (err)
+		return err;
 
 	if (irq_remapped(cfg)) {
 		compose_remapped_msi_msg(pdev, irq, dest, msg, hpet_id);
@@ -3361,6 +3384,8 @@ static struct irq_chip ht_irq_chip = {
 int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev)
 {
 	struct irq_cfg *cfg;
+	struct ht_irq_msg msg;
+	unsigned dest;
 	int err;
 
 	if (disable_apic)
@@ -3368,36 +3393,37 @@ int arch_setup_ht_irq(unsigned int irq, struct pci_dev *dev)
 
 	cfg = irq_cfg(irq);
 	err = assign_irq_vector(irq, cfg, apic->target_cpus());
-	if (!err) {
-		struct ht_irq_msg msg;
-		unsigned dest;
+	if (err)
+		return err;
 
-		dest = apic->cpu_mask_to_apicid_and(cfg->domain,
-						    apic->target_cpus());
+	err = apic->cpu_mask_to_apicid_and(cfg->domain,
+					   apic->target_cpus(), &dest);
+	if (err)
+		return err;
 
-		msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest);
+	msg.address_hi = HT_IRQ_HIGH_DEST_ID(dest);
 
-		msg.address_lo =
-			HT_IRQ_LOW_BASE |
-			HT_IRQ_LOW_DEST_ID(dest) |
-			HT_IRQ_LOW_VECTOR(cfg->vector) |
-			((apic->irq_dest_mode == 0) ?
-				HT_IRQ_LOW_DM_PHYSICAL :
-				HT_IRQ_LOW_DM_LOGICAL) |
-			HT_IRQ_LOW_RQEOI_EDGE |
-			((apic->irq_delivery_mode != dest_LowestPrio) ?
-				HT_IRQ_LOW_MT_FIXED :
-				HT_IRQ_LOW_MT_ARBITRATED) |
-			HT_IRQ_LOW_IRQ_MASKED;
+	msg.address_lo =
+		HT_IRQ_LOW_BASE |
+		HT_IRQ_LOW_DEST_ID(dest) |
+		HT_IRQ_LOW_VECTOR(cfg->vector) |
+		((apic->irq_dest_mode == 0) ?
+			HT_IRQ_LOW_DM_PHYSICAL :
+			HT_IRQ_LOW_DM_LOGICAL) |
+		HT_IRQ_LOW_RQEOI_EDGE |
+		((apic->irq_delivery_mode != dest_LowestPrio) ?
+			HT_IRQ_LOW_MT_FIXED :
+			HT_IRQ_LOW_MT_ARBITRATED) |
+		HT_IRQ_LOW_IRQ_MASKED;
 
-		write_ht_irq_msg(irq, &msg);
+	write_ht_irq_msg(irq, &msg);
 
-		irq_set_chip_and_handler_name(irq, &ht_irq_chip,
-					      handle_edge_irq, "edge");
+	irq_set_chip_and_handler_name(irq, &ht_irq_chip,
+				      handle_edge_irq, "edge");
 
-		dev_printk(KERN_DEBUG, &dev->dev, "irq %d for HT\n", irq);
-	}
-	return err;
+	dev_printk(KERN_DEBUG, &dev->dev, "irq %d for HT\n", irq);
+
+	return 0;
 }
 #endif /* CONFIG_HT_IRQ */
 
diff --git a/arch/x86/kernel/apic/numaq_32.c b/arch/x86/kernel/apic/numaq_32.c
index eb2d466fd81..2b55514c328 100644
--- a/arch/x86/kernel/apic/numaq_32.c
+++ b/arch/x86/kernel/apic/numaq_32.c
@@ -406,16 +406,20 @@ static inline int numaq_check_phys_apicid_present(int phys_apicid)
  * We use physical apicids here, not logical, so just return the default
  * physical broadcast to stop people from breaking us
  */
-static unsigned int numaq_cpu_mask_to_apicid(const struct cpumask *cpumask)
+static int
+numaq_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *apicid)
 {
-	return 0x0F;
+	*apicid = 0x0F;
+	return 0;
 }
 
-static inline unsigned int
+static int
 numaq_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
-			     const struct cpumask *andmask)
+			     const struct cpumask *andmask,
+			     unsigned int *apicid)
 {
-	return 0x0F;
+	*apicid = 0x0F;
+	return 0;
 }
 
 /* No NUMA-Q box has a HT CPU, but it can't hurt to use the default code. */
diff --git a/arch/x86/kernel/apic/summit_32.c b/arch/x86/kernel/apic/summit_32.c
index 35d254c1fec..5766d84f12d 100644
--- a/arch/x86/kernel/apic/summit_32.c
+++ b/arch/x86/kernel/apic/summit_32.c
@@ -263,7 +263,8 @@ static int summit_check_phys_apicid_present(int physical_apicid)
 	return 1;
 }
 
-static unsigned int summit_cpu_mask_to_apicid(const struct cpumask *cpumask)
+static int
+summit_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *dest_id)
 {
 	unsigned int round = 0;
 	int cpu, apicid = 0;
@@ -276,30 +277,33 @@ static unsigned int summit_cpu_mask_to_apicid(const struct cpumask *cpumask)
 
 		if (round && APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
 			printk("%s: Not a valid mask!\n", __func__);
-			return BAD_APICID;
+			return -EINVAL;
 		}
 		apicid |= new_apicid;
 		round++;
 	}
-	return apicid;
+	*dest_id = apicid;
+	return 0;
 }
 
-static unsigned int summit_cpu_mask_to_apicid_and(const struct cpumask *inmask,
-			      const struct cpumask *andmask)
+static int
+summit_cpu_mask_to_apicid_and(const struct cpumask *inmask,
+			      const struct cpumask *andmask,
+			      unsigned int *apicid)
 {
-	int apicid = early_per_cpu(x86_cpu_to_logical_apicid, 0);
+	*apicid = early_per_cpu(x86_cpu_to_logical_apicid, 0);
 	cpumask_var_t cpumask;
 
 	if (!alloc_cpumask_var(&cpumask, GFP_ATOMIC))
-		return apicid;
+		return 0;
 
 	cpumask_and(cpumask, inmask, andmask);
 	cpumask_and(cpumask, cpumask, cpu_online_mask);
-	apicid = summit_cpu_mask_to_apicid(cpumask);
+	summit_cpu_mask_to_apicid(cpumask, apicid);
 
 	free_cpumask_var(cpumask);
 
-	return apicid;
+	return 0;
 }
 
 /*
diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index 612622c47df..5f86f79335f 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -96,24 +96,26 @@ static void x2apic_send_IPI_all(int vector)
 	__x2apic_send_IPI_mask(cpu_online_mask, vector, APIC_DEST_ALLINC);
 }
 
-static unsigned int x2apic_cpu_mask_to_apicid(const struct cpumask *cpumask)
+static int
+x2apic_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *apicid)
 {
 	int cpu = cpumask_first(cpumask);
-	u32 dest = 0;
 	int i;
 
-	if (cpu > nr_cpu_ids)
-		return BAD_APICID;
+	if (cpu >= nr_cpu_ids)
+		return -EINVAL;
 
+	*apicid = 0;
 	for_each_cpu_and(i, cpumask, per_cpu(cpus_in_cluster, cpu))
-		dest |= per_cpu(x86_cpu_to_logical_apicid, i);
+		*apicid |= per_cpu(x86_cpu_to_logical_apicid, i);
 
-	return dest;
+	return 0;
 }
 
-static unsigned int
+static int
 x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
-			      const struct cpumask *andmask)
+			      const struct cpumask *andmask,
+			      unsigned int *apicid)
 {
 	u32 dest = 0;
 	u16 cluster;
@@ -128,7 +130,7 @@ x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 	}
 
 	if (!dest)
-		return BAD_APICID;
+		return -EINVAL;
 
 	for_each_cpu_and(i, cpumask, andmask) {
 		if (!cpumask_test_cpu(i, cpu_online_mask))
@@ -138,7 +140,9 @@ x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 		dest |= per_cpu(x86_cpu_to_logical_apicid, i);
 	}
 
-	return dest;
+	*apicid = dest;
+
+	return 0;
 }
 
 static void init_x2apic_ldr(void)
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index df89a7d7874..2f3030fef31 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -269,23 +269,31 @@ static void uv_init_apic_ldr(void)
 {
 }
 
-static unsigned int uv_cpu_mask_to_apicid(const struct cpumask *cpumask)
+static inline int __uv_cpu_to_apicid(int cpu, unsigned int *apicid)
+{
+	if (likely((unsigned int)cpu < nr_cpu_ids)) {
+		*apicid = per_cpu(x86_cpu_to_apicid, cpu) | uv_apicid_hibits;
+		return 0;
+	} else {
+		return -EINVAL;
+	}
+}
+
+static int
+uv_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *apicid)
 {
 	/*
 	 * We're using fixed IRQ delivery, can only return one phys APIC ID.
 	 * May as well be the first.
 	 */
 	int cpu = cpumask_first(cpumask);
-
-	if ((unsigned)cpu < nr_cpu_ids)
-		return per_cpu(x86_cpu_to_apicid, cpu) | uv_apicid_hibits;
-	else
-		return BAD_APICID;
+	return __uv_cpu_to_apicid(cpu, apicid);
 }
 
-static unsigned int
+static int
 uv_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
-			  const struct cpumask *andmask)
+			  const struct cpumask *andmask,
+			  unsigned int *apicid)
 {
 	int cpu;
 
@@ -297,7 +305,8 @@ uv_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 		if (cpumask_test_cpu(cpu, cpu_online_mask))
 			break;
 	}
-	return per_cpu(x86_cpu_to_apicid, cpu) | uv_apicid_hibits;
+
+	return __uv_cpu_to_apicid(cpu, apicid);
 }
 
 static unsigned int x2apic_get_apic_id(unsigned long x)
diff --git a/arch/x86/platform/uv/uv_irq.c b/arch/x86/platform/uv/uv_irq.c
index f25c2765a5c..dd1ff39a464 100644
--- a/arch/x86/platform/uv/uv_irq.c
+++ b/arch/x86/platform/uv/uv_irq.c
@@ -135,6 +135,7 @@ arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade,
 	unsigned long mmr_value;
 	struct uv_IO_APIC_route_entry *entry;
 	int mmr_pnode, err;
+	unsigned int dest;
 
 	BUILD_BUG_ON(sizeof(struct uv_IO_APIC_route_entry) !=
 			sizeof(unsigned long));
@@ -143,6 +144,10 @@ arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade,
 	if (err != 0)
 		return err;
 
+	err = apic->cpu_mask_to_apicid(eligible_cpu, &dest);
+	if (err != 0)
+		return err;
+
 	if (limit == UV_AFFINITY_CPU)
 		irq_set_status_flags(irq, IRQ_NO_BALANCING);
 	else
@@ -159,7 +164,7 @@ arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade,
 	entry->polarity		= 0;
 	entry->trigger		= 0;
 	entry->mask		= 0;
-	entry->dest		= apic->cpu_mask_to_apicid(eligible_cpu);
+	entry->dest		= dest;
 
 	mmr_pnode = uv_blade_to_pnode(mmr_blade);
 	uv_write_global_mmr64(mmr_pnode, mmr_offset, mmr_value);
diff --git a/drivers/iommu/intel_irq_remapping.c b/drivers/iommu/intel_irq_remapping.c
index 6d347064b8b..dafbad06390 100644
--- a/drivers/iommu/intel_irq_remapping.c
+++ b/drivers/iommu/intel_irq_remapping.c
@@ -924,6 +924,7 @@ intel_ioapic_set_affinity(struct irq_data *data, const struct cpumask *mask,
 	struct irq_cfg *cfg = data->chip_data;
 	unsigned int dest, irq = data->irq;
 	struct irte irte;
+	int err;
 
 	if (!cpumask_intersects(mask, cpu_online_mask))
 		return -EINVAL;
@@ -931,10 +932,16 @@ intel_ioapic_set_affinity(struct irq_data *data, const struct cpumask *mask,
 	if (get_irte(irq, &irte))
 		return -EBUSY;
 
-	if (assign_irq_vector(irq, cfg, mask))
-		return -EBUSY;
+	err = assign_irq_vector(irq, cfg, mask);
+	if (err)
+		return err;
 
-	dest = apic->cpu_mask_to_apicid_and(cfg->domain, mask);
+	err = apic->cpu_mask_to_apicid_and(cfg->domain, mask, &dest);
+	if (err) {
+		if (assign_irq_vector(irq, cfg, data->affinity));
+			pr_err("Failed to recover vector for irq %d\n", irq);
+		return err;
+	}
 
 	irte.vector = cfg->vector;
 	irte.dest_id = IRTE_DEST(dest);
-- 
cgit v1.2.3-70-g09d2


From 4988a40c3981212fa8c64da68722affc1cb6697a Mon Sep 17 00:00:00 2001
From: Alexander Gordeev <agordeev@redhat.com>
Date: Thu, 7 Jun 2012 15:16:25 +0200
Subject: x86/apic: Make cpu_mask_to_apicid() operations check cpu_online_mask

Currently cpu_mask_to_apicid() should not get a offline CPU with
the cpumask. Otherwise some apic drivers might try to access
non-existent per-cpu variables (i.e. x2apic). In that regard
cpu_mask_to_apicid() and cpu_mask_to_apicid_and() operations are
inconsistent.

This fix makes the two operations do not rely on calling
functions and always return the apicid for only online CPUs. As
result, the meaning and implementations of cpu_mask_to_apicid()
and cpu_mask_to_apicid_and() operations become straight.

Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Acked-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/20120607131624.GG4759@dhcp-26-207.brq.redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/include/asm/apic.h           | 6 ++----
 arch/x86/kernel/apic/apic.c           | 2 +-
 arch/x86/kernel/apic/es7000_32.c      | 3 +--
 arch/x86/kernel/apic/summit_32.c      | 3 +--
 arch/x86/kernel/apic/x2apic_cluster.c | 2 +-
 arch/x86/kernel/apic/x2apic_uv_x.c    | 2 +-
 6 files changed, 7 insertions(+), 11 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index ae91f9c7e36..1ed3eead203 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -596,7 +596,7 @@ static inline int default_phys_pkg_id(int cpuid_apic, int index_msb)
 static inline int
 __flat_cpu_mask_to_apicid(unsigned long cpu_mask, unsigned int *apicid)
 {
-	cpu_mask &= APIC_ALL_CPUS;
+	cpu_mask = cpu_mask & APIC_ALL_CPUS & cpumask_bits(cpu_online_mask)[0];
 	if (likely(cpu_mask)) {
 		*apicid = (unsigned int)cpu_mask;
 		return 0;
@@ -619,9 +619,7 @@ flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 {
 	unsigned long mask1 = cpumask_bits(cpumask)[0];
 	unsigned long mask2 = cpumask_bits(andmask)[0];
-	unsigned long mask3 = cpumask_bits(cpu_online_mask)[0];
-
-	return __flat_cpu_mask_to_apicid(mask1 & mask2 & mask3, apicid);
+	return __flat_cpu_mask_to_apicid(mask1 & mask2, apicid);
 }
 
 extern int
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index b8d92606f84..7e9bbe73bc5 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -2136,7 +2136,7 @@ static inline int __default_cpu_to_apicid(int cpu, unsigned int *apicid)
 int default_cpu_mask_to_apicid(const struct cpumask *cpumask,
 			       unsigned int *apicid)
 {
-	int cpu = cpumask_first(cpumask);
+	int cpu = cpumask_first_and(cpumask, cpu_online_mask);
 	return __default_cpu_to_apicid(cpu, apicid);
 }
 
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index 515ebb00a9f..b35cfb9b696 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -534,7 +534,7 @@ es7000_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *dest_id)
 	/*
 	 * The cpus in the mask must all be on the apic cluster.
 	 */
-	for_each_cpu(cpu, cpumask) {
+	for_each_cpu_and(cpu, cpumask, cpu_online_mask) {
 		int new_apicid = early_per_cpu(x86_cpu_to_logical_apicid, cpu);
 
 		if (round && APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
@@ -561,7 +561,6 @@ es7000_cpu_mask_to_apicid_and(const struct cpumask *inmask,
 		return 0;
 
 	cpumask_and(cpumask, inmask, andmask);
-	cpumask_and(cpumask, cpumask, cpu_online_mask);
 	es7000_cpu_mask_to_apicid(cpumask, apicid);
 
 	free_cpumask_var(cpumask);
diff --git a/arch/x86/kernel/apic/summit_32.c b/arch/x86/kernel/apic/summit_32.c
index 5766d84f12d..79d360f6729 100644
--- a/arch/x86/kernel/apic/summit_32.c
+++ b/arch/x86/kernel/apic/summit_32.c
@@ -272,7 +272,7 @@ summit_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *dest_id)
 	/*
 	 * The cpus in the mask must all be on the apic cluster.
 	 */
-	for_each_cpu(cpu, cpumask) {
+	for_each_cpu_and(cpu, cpumask, cpu_online_mask) {
 		int new_apicid = early_per_cpu(x86_cpu_to_logical_apicid, cpu);
 
 		if (round && APIC_CLUSTER(apicid) != APIC_CLUSTER(new_apicid)) {
@@ -298,7 +298,6 @@ summit_cpu_mask_to_apicid_and(const struct cpumask *inmask,
 		return 0;
 
 	cpumask_and(cpumask, inmask, andmask);
-	cpumask_and(cpumask, cpumask, cpu_online_mask);
 	summit_cpu_mask_to_apicid(cpumask, apicid);
 
 	free_cpumask_var(cpumask);
diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index 5f86f79335f..23a46cf5b6f 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -99,7 +99,7 @@ static void x2apic_send_IPI_all(int vector)
 static int
 x2apic_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *apicid)
 {
-	int cpu = cpumask_first(cpumask);
+	int cpu = cpumask_first_and(cpumask, cpu_online_mask);
 	int i;
 
 	if (cpu >= nr_cpu_ids)
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index 2f3030fef31..307aa076bd6 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -286,7 +286,7 @@ uv_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *apicid)
 	 * We're using fixed IRQ delivery, can only return one phys APIC ID.
 	 * May as well be the first.
 	 */
-	int cpu = cpumask_first(cpumask);
+	int cpu = cpumask_first_and(cpumask, cpu_online_mask);
 	return __uv_cpu_to_apicid(cpu, apicid);
 }
 
-- 
cgit v1.2.3-70-g09d2


From cac4afbc3da58d9e5701b34bd4c1f11ea13328d4 Mon Sep 17 00:00:00 2001
From: Alexander Gordeev <agordeev@redhat.com>
Date: Thu, 14 Jun 2012 12:39:34 +0200
Subject: x86/x2apic/cluster: Vector_allocation_domain() should return a value

Since commit 8637e38 ("x86/apic: Avoid useless scanning thru a
cpumask in assign_irq_vector()") vector_allocation_domain()
operation indicates if a cpumask is dynamic or static. This
update fixes the oversight and makes the operation to return a
value.

Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Cc: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/20120614103933.GJ3383@dhcp-26-207.brq.redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/apic/x2apic_cluster.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index 23a46cf5b6f..1885a73b7f3 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -228,10 +228,11 @@ static int x2apic_cluster_probe(void)
 /*
  * Each x2apic cluster is an allocation domain.
  */
-static void cluster_vector_allocation_domain(int cpu, struct cpumask *retmask)
+static bool cluster_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	cpumask_clear(retmask);
 	cpumask_copy(retmask, per_cpu(cpus_in_cluster, cpu));
+	return true;
 }
 
 static struct apic apic_x2apic_cluster = {
-- 
cgit v1.2.3-70-g09d2


From a5a391561bc25898ba1a702a0c4b028aa5b11ce9 Mon Sep 17 00:00:00 2001
From: Alexander Gordeev <agordeev@redhat.com>
Date: Thu, 14 Jun 2012 09:49:35 +0200
Subject: x86/apic: Eliminate cpu_mask_to_apicid() operation

Since there are only two locations where cpu_mask_to_apicid() is
called from, remove the operation and use only
cpu_mask_to_apicid_and() instead.

Signed-off-by: Alexander Gordeev <agordeev@redhat.com>
Suggested-and-acked-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/20120614074935.GE3383@dhcp-26-207.brq.redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/include/asm/apic.h           | 33 ++++++++-------------------------
 arch/x86/kernel/apic/apic.c           | 24 ++++++------------------
 arch/x86/kernel/apic/apic_flat_64.c   |  2 --
 arch/x86/kernel/apic/apic_noop.c      |  1 -
 arch/x86/kernel/apic/apic_numachip.c  |  1 -
 arch/x86/kernel/apic/bigsmp_32.c      |  1 -
 arch/x86/kernel/apic/es7000_32.c      |  4 +---
 arch/x86/kernel/apic/io_apic.c        |  3 ++-
 arch/x86/kernel/apic/numaq_32.c       |  8 --------
 arch/x86/kernel/apic/probe_32.c       |  1 -
 arch/x86/kernel/apic/summit_32.c      |  3 +--
 arch/x86/kernel/apic/x2apic_cluster.c | 17 -----------------
 arch/x86/kernel/apic/x2apic_phys.c    |  1 -
 arch/x86/kernel/apic/x2apic_uv_x.c    | 29 ++++++-----------------------
 arch/x86/platform/uv/uv_irq.c         |  2 +-
 15 files changed, 25 insertions(+), 105 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index 1ed3eead203..eec240e1209 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -331,8 +331,6 @@ struct apic {
 	unsigned long (*set_apic_id)(unsigned int id);
 	unsigned long apic_id_mask;
 
-	int (*cpu_mask_to_apicid)(const struct cpumask *cpumask,
-				  unsigned int *apicid);
 	int (*cpu_mask_to_apicid_and)(const struct cpumask *cpumask,
 				      const struct cpumask *andmask,
 				      unsigned int *apicid);
@@ -594,9 +592,15 @@ static inline int default_phys_pkg_id(int cpuid_apic, int index_msb)
 #endif
 
 static inline int
-__flat_cpu_mask_to_apicid(unsigned long cpu_mask, unsigned int *apicid)
+flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
+			    const struct cpumask *andmask,
+			    unsigned int *apicid)
 {
-	cpu_mask = cpu_mask & APIC_ALL_CPUS & cpumask_bits(cpu_online_mask)[0];
+	unsigned long cpu_mask = cpumask_bits(cpumask)[0] &
+				 cpumask_bits(andmask)[0] &
+				 cpumask_bits(cpu_online_mask)[0] &
+				 APIC_ALL_CPUS;
+
 	if (likely(cpu_mask)) {
 		*apicid = (unsigned int)cpu_mask;
 		return 0;
@@ -605,27 +609,6 @@ __flat_cpu_mask_to_apicid(unsigned long cpu_mask, unsigned int *apicid)
 	}
 }
 
-static inline int
-flat_cpu_mask_to_apicid(const struct cpumask *cpumask,
-			unsigned int *apicid)
-{
-	return __flat_cpu_mask_to_apicid(cpumask_bits(cpumask)[0], apicid);
-}
-
-static inline int
-flat_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
-			    const struct cpumask *andmask,
-			    unsigned int *apicid)
-{
-	unsigned long mask1 = cpumask_bits(cpumask)[0];
-	unsigned long mask2 = cpumask_bits(andmask)[0];
-	return __flat_cpu_mask_to_apicid(mask1 & mask2, apicid);
-}
-
-extern int
-default_cpu_mask_to_apicid(const struct cpumask *cpumask,
-			   unsigned int *apicid);
-
 extern int
 default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			       const struct cpumask *andmask,
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 7e9bbe73bc5..048a4f806d4 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -2123,23 +2123,6 @@ void default_init_apic_ldr(void)
 	apic_write(APIC_LDR, val);
 }
 
-static inline int __default_cpu_to_apicid(int cpu, unsigned int *apicid)
-{
-	if (likely((unsigned int)cpu < nr_cpu_ids)) {
-		*apicid = per_cpu(x86_cpu_to_apicid, cpu);
-		return 0;
-	} else {
-		return -EINVAL;
-	}
-}
-
-int default_cpu_mask_to_apicid(const struct cpumask *cpumask,
-			       unsigned int *apicid)
-{
-	int cpu = cpumask_first_and(cpumask, cpu_online_mask);
-	return __default_cpu_to_apicid(cpu, apicid);
-}
-
 int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 				   const struct cpumask *andmask,
 				   unsigned int *apicid)
@@ -2151,7 +2134,12 @@ int default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			break;
 	}
 
-	return __default_cpu_to_apicid(cpu, apicid);
+	if (likely((unsigned int)cpu < nr_cpu_ids)) {
+		*apicid = per_cpu(x86_cpu_to_apicid, cpu);
+		return 0;
+	} else {
+		return -EINVAL;
+	}
 }
 
 /*
diff --git a/arch/x86/kernel/apic/apic_flat_64.c b/arch/x86/kernel/apic/apic_flat_64.c
index bddc92566d0..00c77cf78e9 100644
--- a/arch/x86/kernel/apic/apic_flat_64.c
+++ b/arch/x86/kernel/apic/apic_flat_64.c
@@ -191,7 +191,6 @@ static struct apic apic_flat =  {
 	.set_apic_id			= set_apic_id,
 	.apic_id_mask			= 0xFFu << 24,
 
-	.cpu_mask_to_apicid		= flat_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= flat_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= flat_send_IPI_mask,
@@ -308,7 +307,6 @@ static struct apic apic_physflat =  {
 	.set_apic_id			= set_apic_id,
 	.apic_id_mask			= 0xFFu << 24,
 
-	.cpu_mask_to_apicid		= default_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= default_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= physflat_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/apic_noop.c b/arch/x86/kernel/apic/apic_noop.c
index ac9edf247b1..65c07fc630a 100644
--- a/arch/x86/kernel/apic/apic_noop.c
+++ b/arch/x86/kernel/apic/apic_noop.c
@@ -159,7 +159,6 @@ struct apic apic_noop = {
 	.set_apic_id			= NULL,
 	.apic_id_mask			= 0x0F << 24,
 
-	.cpu_mask_to_apicid		= flat_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= flat_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= noop_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/apic_numachip.c b/arch/x86/kernel/apic/apic_numachip.c
index c028132ad35..bc552cff257 100644
--- a/arch/x86/kernel/apic/apic_numachip.c
+++ b/arch/x86/kernel/apic/apic_numachip.c
@@ -234,7 +234,6 @@ static struct apic apic_numachip __refconst = {
 	.set_apic_id			= set_apic_id,
 	.apic_id_mask			= 0xffU << 24,
 
-	.cpu_mask_to_apicid		= default_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= default_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= numachip_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/bigsmp_32.c b/arch/x86/kernel/apic/bigsmp_32.c
index df342fe4d6a..d50e3640d5a 100644
--- a/arch/x86/kernel/apic/bigsmp_32.c
+++ b/arch/x86/kernel/apic/bigsmp_32.c
@@ -188,7 +188,6 @@ static struct apic apic_bigsmp = {
 	.set_apic_id			= NULL,
 	.apic_id_mask			= 0xFF << 24,
 
-	.cpu_mask_to_apicid		= default_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= default_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= bigsmp_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/es7000_32.c b/arch/x86/kernel/apic/es7000_32.c
index b35cfb9b696..2c5317ea1b8 100644
--- a/arch/x86/kernel/apic/es7000_32.c
+++ b/arch/x86/kernel/apic/es7000_32.c
@@ -525,7 +525,7 @@ static int es7000_check_phys_apicid_present(int cpu_physical_apicid)
 	return 1;
 }
 
-static int
+static inline int
 es7000_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *dest_id)
 {
 	unsigned int round = 0;
@@ -643,7 +643,6 @@ static struct apic __refdata apic_es7000_cluster = {
 	.set_apic_id			= NULL,
 	.apic_id_mask			= 0xFF << 24,
 
-	.cpu_mask_to_apicid		= es7000_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= es7000_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= es7000_send_IPI_mask,
@@ -710,7 +709,6 @@ static struct apic __refdata apic_es7000 = {
 	.set_apic_id			= NULL,
 	.apic_id_mask			= 0xFF << 24,
 
-	.cpu_mask_to_apicid		= es7000_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= es7000_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= es7000_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 0deb773404e..0540f083f45 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1492,7 +1492,8 @@ static void __init setup_timer_IRQ0_pin(unsigned int ioapic_idx,
 	 * We use logical delivery to get the timer IRQ
 	 * to the first CPU.
 	 */
-	if (unlikely(apic->cpu_mask_to_apicid(apic->target_cpus(), &dest)))
+	if (unlikely(apic->cpu_mask_to_apicid_and(apic->target_cpus(),
+						  apic->target_cpus(), &dest)))
 		dest = BAD_APICID;
 
 	entry.dest_mode = apic->irq_dest_mode;
diff --git a/arch/x86/kernel/apic/numaq_32.c b/arch/x86/kernel/apic/numaq_32.c
index 2b55514c328..d661ee95cab 100644
--- a/arch/x86/kernel/apic/numaq_32.c
+++ b/arch/x86/kernel/apic/numaq_32.c
@@ -406,13 +406,6 @@ static inline int numaq_check_phys_apicid_present(int phys_apicid)
  * We use physical apicids here, not logical, so just return the default
  * physical broadcast to stop people from breaking us
  */
-static int
-numaq_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *apicid)
-{
-	*apicid = 0x0F;
-	return 0;
-}
-
 static int
 numaq_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			     const struct cpumask *andmask,
@@ -499,7 +492,6 @@ static struct apic __refdata apic_numaq = {
 	.set_apic_id			= NULL,
 	.apic_id_mask			= 0x0F << 24,
 
-	.cpu_mask_to_apicid		= numaq_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= numaq_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= numaq_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/probe_32.c b/arch/x86/kernel/apic/probe_32.c
index 2c6f003b2e4..eef6bcd1bf1 100644
--- a/arch/x86/kernel/apic/probe_32.c
+++ b/arch/x86/kernel/apic/probe_32.c
@@ -108,7 +108,6 @@ static struct apic apic_default = {
 	.set_apic_id			= NULL,
 	.apic_id_mask			= 0x0F << 24,
 
-	.cpu_mask_to_apicid		= flat_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= flat_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= default_send_IPI_mask_logical,
diff --git a/arch/x86/kernel/apic/summit_32.c b/arch/x86/kernel/apic/summit_32.c
index 79d360f6729..bbad180f289 100644
--- a/arch/x86/kernel/apic/summit_32.c
+++ b/arch/x86/kernel/apic/summit_32.c
@@ -263,7 +263,7 @@ static int summit_check_phys_apicid_present(int physical_apicid)
 	return 1;
 }
 
-static int
+static inline int
 summit_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *dest_id)
 {
 	unsigned int round = 0;
@@ -516,7 +516,6 @@ static struct apic apic_summit = {
 	.set_apic_id			= NULL,
 	.apic_id_mask			= 0xFF << 24,
 
-	.cpu_mask_to_apicid		= summit_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= summit_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= summit_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index 1885a73b7f3..943d03fc6fc 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -96,22 +96,6 @@ static void x2apic_send_IPI_all(int vector)
 	__x2apic_send_IPI_mask(cpu_online_mask, vector, APIC_DEST_ALLINC);
 }
 
-static int
-x2apic_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *apicid)
-{
-	int cpu = cpumask_first_and(cpumask, cpu_online_mask);
-	int i;
-
-	if (cpu >= nr_cpu_ids)
-		return -EINVAL;
-
-	*apicid = 0;
-	for_each_cpu_and(i, cpumask, per_cpu(cpus_in_cluster, cpu))
-		*apicid |= per_cpu(x86_cpu_to_logical_apicid, i);
-
-	return 0;
-}
-
 static int
 x2apic_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			      const struct cpumask *andmask,
@@ -270,7 +254,6 @@ static struct apic apic_x2apic_cluster = {
 	.set_apic_id			= x2apic_set_apic_id,
 	.apic_id_mask			= 0xFFFFFFFFu,
 
-	.cpu_mask_to_apicid		= x2apic_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= x2apic_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= x2apic_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/x2apic_phys.c b/arch/x86/kernel/apic/x2apic_phys.c
index f109388a0e8..e03a1e180e8 100644
--- a/arch/x86/kernel/apic/x2apic_phys.c
+++ b/arch/x86/kernel/apic/x2apic_phys.c
@@ -123,7 +123,6 @@ static struct apic apic_x2apic_phys = {
 	.set_apic_id			= x2apic_set_apic_id,
 	.apic_id_mask			= 0xFFFFFFFFu,
 
-	.cpu_mask_to_apicid		= default_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= default_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= x2apic_send_IPI_mask,
diff --git a/arch/x86/kernel/apic/x2apic_uv_x.c b/arch/x86/kernel/apic/x2apic_uv_x.c
index 307aa076bd6..026de0114d1 100644
--- a/arch/x86/kernel/apic/x2apic_uv_x.c
+++ b/arch/x86/kernel/apic/x2apic_uv_x.c
@@ -269,27 +269,6 @@ static void uv_init_apic_ldr(void)
 {
 }
 
-static inline int __uv_cpu_to_apicid(int cpu, unsigned int *apicid)
-{
-	if (likely((unsigned int)cpu < nr_cpu_ids)) {
-		*apicid = per_cpu(x86_cpu_to_apicid, cpu) | uv_apicid_hibits;
-		return 0;
-	} else {
-		return -EINVAL;
-	}
-}
-
-static int
-uv_cpu_mask_to_apicid(const struct cpumask *cpumask, unsigned int *apicid)
-{
-	/*
-	 * We're using fixed IRQ delivery, can only return one phys APIC ID.
-	 * May as well be the first.
-	 */
-	int cpu = cpumask_first_and(cpumask, cpu_online_mask);
-	return __uv_cpu_to_apicid(cpu, apicid);
-}
-
 static int
 uv_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			  const struct cpumask *andmask,
@@ -306,7 +285,12 @@ uv_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			break;
 	}
 
-	return __uv_cpu_to_apicid(cpu, apicid);
+	if (likely((unsigned int)cpu < nr_cpu_ids)) {
+		*apicid = per_cpu(x86_cpu_to_apicid, cpu) | uv_apicid_hibits;
+		return 0;
+	} else {
+		return -EINVAL;
+	}
 }
 
 static unsigned int x2apic_get_apic_id(unsigned long x)
@@ -384,7 +368,6 @@ static struct apic __refdata apic_x2apic_uv_x = {
 	.set_apic_id			= set_apic_id,
 	.apic_id_mask			= 0xFFFFFFFFu,
 
-	.cpu_mask_to_apicid		= uv_cpu_mask_to_apicid,
 	.cpu_mask_to_apicid_and		= uv_cpu_mask_to_apicid_and,
 
 	.send_IPI_mask			= uv_send_IPI_mask,
diff --git a/arch/x86/platform/uv/uv_irq.c b/arch/x86/platform/uv/uv_irq.c
index dd1ff39a464..a67c7a6bac7 100644
--- a/arch/x86/platform/uv/uv_irq.c
+++ b/arch/x86/platform/uv/uv_irq.c
@@ -144,7 +144,7 @@ arch_enable_uv_irq(char *irq_name, unsigned int irq, int cpu, int mmr_blade,
 	if (err != 0)
 		return err;
 
-	err = apic->cpu_mask_to_apicid(eligible_cpu, &dest);
+	err = apic->cpu_mask_to_apicid_and(eligible_cpu, eligible_cpu, &dest);
 	if (err != 0)
 		return err;
 
-- 
cgit v1.2.3-70-g09d2


From b39f25a849d7677a7dbf183f2483fd41c201a5ce Mon Sep 17 00:00:00 2001
From: Suresh Siddha <suresh.b.siddha@intel.com>
Date: Mon, 25 Jun 2012 13:38:27 -0700
Subject: x86/apic: Optimize cpu traversal in __assign_irq_vector() using
 domain membership

Currently __assign_irq_vector() goes through each cpu in the
specified mask until it finds a free vector in all the cpu's
that are part of the same interrupt domain. We visit all the
interrupt domain sibling cpus to reserve the free vector. So,
when we fail to find a free vector in an interrupt domain, it is
safe to continue our search with a cpu belonging to a new
interrupt domain. No need to go through each cpu, if the domain
containing that cpu is already visited.

Use the irq_cfg's old_domain to track the visited domains and
optimize the cpu traversal while finding a free vector in the
given cpumask.

NOTE: We can also optimize the search by using for_each_cpu() and
skip the current cpu, if it is not the first cpu in the mask
returned by the vector_allocation_domain(). But re-using the
cfg->old_domain to track the visited domains will be slightly
faster.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Alexander Gordeev <agordeev@redhat.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Link: http://lkml.kernel.org/r/1340656709-11423-2-git-send-email-suresh.b.siddha@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/include/asm/apic.h           |  8 +++-----
 arch/x86/kernel/apic/apic_noop.c      |  3 +--
 arch/x86/kernel/apic/io_apic.c        | 15 ++++++++-------
 arch/x86/kernel/apic/x2apic_cluster.c |  3 +--
 arch/x86/kernel/vsmp_64.c             |  3 +--
 5 files changed, 14 insertions(+), 18 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index eec240e1209..8bebeb8952f 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -306,7 +306,7 @@ struct apic {
 	unsigned long (*check_apicid_used)(physid_mask_t *map, int apicid);
 	unsigned long (*check_apicid_present)(int apicid);
 
-	bool (*vector_allocation_domain)(int cpu, struct cpumask *retmask);
+	void (*vector_allocation_domain)(int cpu, struct cpumask *retmask);
 	void (*init_apic_ldr)(void);
 
 	void (*ioapic_phys_id_map)(physid_mask_t *phys_map, physid_mask_t *retmap);
@@ -614,7 +614,7 @@ default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			       const struct cpumask *andmask,
 			       unsigned int *apicid);
 
-static inline bool
+static inline void
 flat_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	/* Careful. Some cpus do not strictly honor the set of cpus
@@ -627,14 +627,12 @@ flat_vector_allocation_domain(int cpu, struct cpumask *retmask)
 	 */
 	cpumask_clear(retmask);
 	cpumask_bits(retmask)[0] = APIC_ALL_CPUS;
-	return false;
 }
 
-static inline bool
+static inline void
 default_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	cpumask_copy(retmask, cpumask_of(cpu));
-	return true;
 }
 
 static inline unsigned long default_check_apicid_used(physid_mask_t *map, int apicid)
diff --git a/arch/x86/kernel/apic/apic_noop.c b/arch/x86/kernel/apic/apic_noop.c
index 65c07fc630a..08c337bc49f 100644
--- a/arch/x86/kernel/apic/apic_noop.c
+++ b/arch/x86/kernel/apic/apic_noop.c
@@ -100,12 +100,11 @@ static unsigned long noop_check_apicid_present(int bit)
 	return physid_isset(bit, phys_cpu_present_map);
 }
 
-static bool noop_vector_allocation_domain(int cpu, struct cpumask *retmask)
+static void noop_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	if (cpu != 0)
 		pr_warning("APIC: Vector allocated for non-BSP cpu\n");
 	cpumask_copy(retmask, cpumask_of(cpu));
-	return true;
 }
 
 static u32 noop_apic_read(u32 reg)
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index a951ef7decb..8a08f09aa50 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1134,12 +1134,13 @@ __assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask)
 
 	/* Only try and allocate irqs on cpus that are present */
 	err = -ENOSPC;
-	for_each_cpu_and(cpu, mask, cpu_online_mask) {
+	cpumask_clear(cfg->old_domain);
+	cpu = cpumask_first_and(mask, cpu_online_mask);
+	while (cpu < nr_cpu_ids) {
 		int new_cpu;
 		int vector, offset;
-		bool more_domains;
 
-		more_domains = apic->vector_allocation_domain(cpu, tmp_mask);
+		apic->vector_allocation_domain(cpu, tmp_mask);
 
 		if (cpumask_subset(tmp_mask, cfg->domain)) {
 			free_cpumask_var(tmp_mask);
@@ -1156,10 +1157,10 @@ next:
 		}
 
 		if (unlikely(current_vector == vector)) {
-			if (more_domains)
-				continue;
-			else
-				break;
+			cpumask_or(cfg->old_domain, cfg->old_domain, tmp_mask);
+			cpumask_andnot(tmp_mask, mask, cfg->old_domain);
+			cpu = cpumask_first_and(tmp_mask, cpu_online_mask);
+			continue;
 		}
 
 		if (test_bit(vector, used_vectors))
diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index 943d03fc6fc..b5d889b5659 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -212,11 +212,10 @@ static int x2apic_cluster_probe(void)
 /*
  * Each x2apic cluster is an allocation domain.
  */
-static bool cluster_vector_allocation_domain(int cpu, struct cpumask *retmask)
+static void cluster_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	cpumask_clear(retmask);
 	cpumask_copy(retmask, per_cpu(cpus_in_cluster, cpu));
-	return true;
 }
 
 static struct apic apic_x2apic_cluster = {
diff --git a/arch/x86/kernel/vsmp_64.c b/arch/x86/kernel/vsmp_64.c
index fa5adb7c228..3f0285ac00f 100644
--- a/arch/x86/kernel/vsmp_64.c
+++ b/arch/x86/kernel/vsmp_64.c
@@ -208,10 +208,9 @@ static int apicid_phys_pkg_id(int initial_apic_id, int index_msb)
  * In vSMP, all cpus should be capable of handling interrupts, regardless of
  * the APIC used.
  */
-static bool fill_vector_allocation_domain(int cpu, struct cpumask *retmask)
+static void fill_vector_allocation_domain(int cpu, struct cpumask *retmask)
 {
 	cpumask_setall(retmask);
-	return false;
 }
 
 static void vsmp_apic_post_init(void)
-- 
cgit v1.2.3-70-g09d2


From 1ac322d0b169c95ce34d55b3ed6d40ce1a5f3a02 Mon Sep 17 00:00:00 2001
From: Suresh Siddha <suresh.b.siddha@intel.com>
Date: Mon, 25 Jun 2012 13:38:28 -0700
Subject: x86/apic/x2apic: Limit the vector reservation to the user specified
 mask

For the x2apic cluster mode, vector for an interrupt is
currently reserved on all the cpu's that are part of the x2apic
cluster. But the interrupts will be routed only to the cluster
(derived from the first cpu in the mask) members specified in
the mask. So there is no need to reserve the vector in the
unused cluster members.

Modify __assign_irq_vector() to reserve the vectors based on the
user specified irq destination mask. If the new mask is a proper
subset of the currently used mask, cleanup the vector allocation
on the unused cpu members.

Also, allow the apic driver to tune the vector domain based on
the affinity mask (which in most cases is the user-specified
mask).

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Alexander Gordeev <agordeev@redhat.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Link: http://lkml.kernel.org/r/1340656709-11423-3-git-send-email-suresh.b.siddha@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/include/asm/apic.h           |  9 ++++++---
 arch/x86/kernel/apic/apic_noop.c      |  3 ++-
 arch/x86/kernel/apic/io_apic.c        | 31 +++++++++++++++----------------
 arch/x86/kernel/apic/x2apic_cluster.c |  6 +++---
 arch/x86/kernel/vsmp_64.c             |  3 ++-
 5 files changed, 28 insertions(+), 24 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/include/asm/apic.h b/arch/x86/include/asm/apic.h
index 8bebeb8952f..88093c1d44f 100644
--- a/arch/x86/include/asm/apic.h
+++ b/arch/x86/include/asm/apic.h
@@ -306,7 +306,8 @@ struct apic {
 	unsigned long (*check_apicid_used)(physid_mask_t *map, int apicid);
 	unsigned long (*check_apicid_present)(int apicid);
 
-	void (*vector_allocation_domain)(int cpu, struct cpumask *retmask);
+	void (*vector_allocation_domain)(int cpu, struct cpumask *retmask,
+					 const struct cpumask *mask);
 	void (*init_apic_ldr)(void);
 
 	void (*ioapic_phys_id_map)(physid_mask_t *phys_map, physid_mask_t *retmap);
@@ -615,7 +616,8 @@ default_cpu_mask_to_apicid_and(const struct cpumask *cpumask,
 			       unsigned int *apicid);
 
 static inline void
-flat_vector_allocation_domain(int cpu, struct cpumask *retmask)
+flat_vector_allocation_domain(int cpu, struct cpumask *retmask,
+			      const struct cpumask *mask)
 {
 	/* Careful. Some cpus do not strictly honor the set of cpus
 	 * specified in the interrupt destination when using lowest
@@ -630,7 +632,8 @@ flat_vector_allocation_domain(int cpu, struct cpumask *retmask)
 }
 
 static inline void
-default_vector_allocation_domain(int cpu, struct cpumask *retmask)
+default_vector_allocation_domain(int cpu, struct cpumask *retmask,
+				 const struct cpumask *mask)
 {
 	cpumask_copy(retmask, cpumask_of(cpu));
 }
diff --git a/arch/x86/kernel/apic/apic_noop.c b/arch/x86/kernel/apic/apic_noop.c
index 08c337bc49f..e145f28b409 100644
--- a/arch/x86/kernel/apic/apic_noop.c
+++ b/arch/x86/kernel/apic/apic_noop.c
@@ -100,7 +100,8 @@ static unsigned long noop_check_apicid_present(int bit)
 	return physid_isset(bit, phys_cpu_present_map);
 }
 
-static void noop_vector_allocation_domain(int cpu, struct cpumask *retmask)
+static void noop_vector_allocation_domain(int cpu, struct cpumask *retmask,
+					  const struct cpumask *mask)
 {
 	if (cpu != 0)
 		pr_warning("APIC: Vector allocated for non-BSP cpu\n");
diff --git a/arch/x86/kernel/apic/io_apic.c b/arch/x86/kernel/apic/io_apic.c
index 8a08f09aa50..9684f963bef 100644
--- a/arch/x86/kernel/apic/io_apic.c
+++ b/arch/x86/kernel/apic/io_apic.c
@@ -1113,7 +1113,6 @@ __assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask)
 	 */
 	static int current_vector = FIRST_EXTERNAL_VECTOR + VECTOR_OFFSET_START;
 	static int current_offset = VECTOR_OFFSET_START % 16;
-	unsigned int old_vector;
 	int cpu, err;
 	cpumask_var_t tmp_mask;
 
@@ -1123,28 +1122,28 @@ __assign_irq_vector(int irq, struct irq_cfg *cfg, const struct cpumask *mask)
 	if (!alloc_cpumask_var(&tmp_mask, GFP_ATOMIC))
 		return -ENOMEM;
 
-	old_vector = cfg->vector;
-	if (old_vector) {
-		cpumask_and(tmp_mask, mask, cpu_online_mask);
-		if (cpumask_subset(tmp_mask, cfg->domain)) {
-			free_cpumask_var(tmp_mask);
-			return 0;
-		}
-	}
-
 	/* Only try and allocate irqs on cpus that are present */
 	err = -ENOSPC;
 	cpumask_clear(cfg->old_domain);
 	cpu = cpumask_first_and(mask, cpu_online_mask);
 	while (cpu < nr_cpu_ids) {
-		int new_cpu;
-		int vector, offset;
+		int new_cpu, vector, offset;
 
-		apic->vector_allocation_domain(cpu, tmp_mask);
+		apic->vector_allocation_domain(cpu, tmp_mask, mask);
 
 		if (cpumask_subset(tmp_mask, cfg->domain)) {
-			free_cpumask_var(tmp_mask);
-			return 0;
+			err = 0;
+			if (cpumask_equal(tmp_mask, cfg->domain))
+				break;
+			/*
+			 * New cpumask using the vector is a proper subset of
+			 * the current in use mask. So cleanup the vector
+			 * allocation for the members that are not used anymore.
+			 */
+			cpumask_andnot(cfg->old_domain, cfg->domain, tmp_mask);
+			cfg->move_in_progress = 1;
+			cpumask_and(cfg->domain, cfg->domain, tmp_mask);
+			break;
 		}
 
 		vector = current_vector;
@@ -1172,7 +1171,7 @@ next:
 		/* Found one! */
 		current_vector = vector;
 		current_offset = offset;
-		if (old_vector) {
+		if (cfg->vector) {
 			cfg->move_in_progress = 1;
 			cpumask_copy(cfg->old_domain, cfg->domain);
 		}
diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index b5d889b5659..bde78d0098a 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -212,10 +212,10 @@ static int x2apic_cluster_probe(void)
 /*
  * Each x2apic cluster is an allocation domain.
  */
-static void cluster_vector_allocation_domain(int cpu, struct cpumask *retmask)
+static void cluster_vector_allocation_domain(int cpu, struct cpumask *retmask,
+					     const struct cpumask *mask)
 {
-	cpumask_clear(retmask);
-	cpumask_copy(retmask, per_cpu(cpus_in_cluster, cpu));
+	cpumask_and(retmask, mask, per_cpu(cpus_in_cluster, cpu));
 }
 
 static struct apic apic_x2apic_cluster = {
diff --git a/arch/x86/kernel/vsmp_64.c b/arch/x86/kernel/vsmp_64.c
index 3f0285ac00f..992f890283e 100644
--- a/arch/x86/kernel/vsmp_64.c
+++ b/arch/x86/kernel/vsmp_64.c
@@ -208,7 +208,8 @@ static int apicid_phys_pkg_id(int initial_apic_id, int index_msb)
  * In vSMP, all cpus should be capable of handling interrupts, regardless of
  * the APIC used.
  */
-static void fill_vector_allocation_domain(int cpu, struct cpumask *retmask)
+static void fill_vector_allocation_domain(int cpu, struct cpumask *retmask,
+					  const struct cpumask *mask)
 {
 	cpumask_setall(retmask);
 }
-- 
cgit v1.2.3-70-g09d2


From d872818dbbeed1bccf58c7f8c7db432154c802f9 Mon Sep 17 00:00:00 2001
From: Suresh Siddha <suresh.b.siddha@intel.com>
Date: Mon, 25 Jun 2012 13:38:29 -0700
Subject: x86/apic/x2apic: Use multiple cluster members for the irq destination
 only with the explicit affinity

During boot or driver load etc, interrupt destination is setup
using default target cpu's. Later the user (irqbalance etc) or
the driver (irq_set_affinity/ irq_set_affinity_hint) can request
the interrupt to be migrated to some specific set of cpu's.

In the x2apic cluster routing, for the default scenario use
single cpu as the interrupt destination and when there is an
explicit interrupt affinity request, route the interrupt to
multiple members of a x2apic cluster specified in the cpumask of
the migration request.

This will minmize the vector pressure when there are lot of
interrupt sources and relatively few x2apic clusters (for
example a single socket server). This will allow the performance
critical interrupts to be routed to multiple cpu's in the x2apic
cluster (irqbalance for example uses the cache siblings etc
while specifying the interrupt destination) and allow
non-critical interrupts to be serviced by a single logical cpu.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Alexander Gordeev <agordeev@redhat.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Link: http://lkml.kernel.org/r/1340656709-11423-4-git-send-email-suresh.b.siddha@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 arch/x86/kernel/apic/x2apic_cluster.c | 21 +++++++++++++++++++--
 1 file changed, 19 insertions(+), 2 deletions(-)

(limited to 'arch/x86/kernel/apic/x2apic_cluster.c')

diff --git a/arch/x86/kernel/apic/x2apic_cluster.c b/arch/x86/kernel/apic/x2apic_cluster.c
index bde78d0098a..c88baa4ff0e 100644
--- a/arch/x86/kernel/apic/x2apic_cluster.c
+++ b/arch/x86/kernel/apic/x2apic_cluster.c
@@ -209,13 +209,30 @@ static int x2apic_cluster_probe(void)
 		return 0;
 }
 
+static const struct cpumask *x2apic_cluster_target_cpus(void)
+{
+	return cpu_all_mask;
+}
+
 /*
  * Each x2apic cluster is an allocation domain.
  */
 static void cluster_vector_allocation_domain(int cpu, struct cpumask *retmask,
 					     const struct cpumask *mask)
 {
-	cpumask_and(retmask, mask, per_cpu(cpus_in_cluster, cpu));
+	/*
+	 * To minimize vector pressure, default case of boot, device bringup
+	 * etc will use a single cpu for the interrupt destination.
+	 *
+	 * On explicit migration requests coming from irqbalance etc,
+	 * interrupts will be routed to the x2apic cluster (cluster-id
+	 * derived from the first cpu in the mask) members specified
+	 * in the mask.
+	 */
+	if (mask == x2apic_cluster_target_cpus())
+		cpumask_copy(retmask, cpumask_of(cpu));
+	else
+		cpumask_and(retmask, mask, per_cpu(cpus_in_cluster, cpu));
 }
 
 static struct apic apic_x2apic_cluster = {
@@ -229,7 +246,7 @@ static struct apic apic_x2apic_cluster = {
 	.irq_delivery_mode		= dest_LowestPrio,
 	.irq_dest_mode			= 1, /* logical */
 
-	.target_cpus			= online_target_cpus,
+	.target_cpus			= x2apic_cluster_target_cpus,
 	.disable_esr			= 0,
 	.dest_logical			= APIC_DEST_LOGICAL,
 	.check_apicid_used		= NULL,
-- 
cgit v1.2.3-70-g09d2