[PATCH 0/6] x86, amd: Add basic support for AMD CPU family 15h

[PATCH 0/6] x86, amd: Add basic support for AMD CPU family 15h

[Andreas Herrmann]

Hi,

following a bunch of patches with basic adaptions to support AMD CPU
family 15h.

Patches are against tip/master (v2.6.36-rc6-1544-gc5bf3de).
Please apply.


Thanks,

Andreas

[PATCH 1/6] x86, mtrr: Assume SYS_CFG[Tom2ForceMemTypeWB] exists on all future AMD CPUs

[Andreas Herrmann]

From: Andreas Herrmann <andreas.herrmann3@amd.com>

Instead of adapting the CPU family check in amd_special_default_mtrr()
for each new CPU family assume that all new AMD CPUs support the
necessary bits in SYS_CFG MSR.

Tom2Enabled is architectural (defined in APM Vol.2).
Tom2ForceMemTypeWB is defined in all BKDGs starting with K8 NPT.
In pre K8-NPT BKDG this bit is reserved (read as zero).

W/o this adaption Linux would unnecessarily complain about bad MTRR
settings on every new AMD CPU family, e.g.

[    0.000000] WARNING: BIOS bug: CPU MTRRs don't cover all of memory, losing 4863MB of RAM.

Cc: stable@kernel.org # .32.x, .35.x
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
---
 arch/x86/kernel/cpu/mtrr/cleanup.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/cpu/mtrr/cleanup.c b/arch/x86/kernel/cpu/mtrr/cleanup.c
index c5f59d0..ac140c7 100644
--- a/arch/x86/kernel/cpu/mtrr/cleanup.c
+++ b/arch/x86/kernel/cpu/mtrr/cleanup.c
@@ -827,7 +827,7 @@ int __init amd_special_default_mtrr(void)
 
 	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD)
 		return 0;
-	if (boot_cpu_data.x86 < 0xf || boot_cpu_data.x86 > 0x11)
+	if (boot_cpu_data.x86 < 0xf)
 		return 0;
 	/* In case some hypervisor doesn't pass SYSCFG through: */
 	if (rdmsr_safe(MSR_K8_SYSCFG, &l, &h) < 0)
-- 
1.6.4.4

[PATCH 2/6] x86, nmi: Support NMI watchdog on newer AMD CPU families

[Andreas Herrmann]

From: Andreas Herrmann <andreas.herrmann3@amd.com>

CPU families 0x12, 0x14 and 0x15 support this functionality.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
---
 arch/x86/kernel/cpu/perfctr-watchdog.c |    9 ++++-----
 1 files changed, 4 insertions(+), 5 deletions(-)

diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c
index fb329e9..d9f4ff8 100644
--- a/arch/x86/kernel/cpu/perfctr-watchdog.c
+++ b/arch/x86/kernel/cpu/perfctr-watchdog.c
@@ -700,11 +700,10 @@ static void probe_nmi_watchdog(void)
 {
 	switch (boot_cpu_data.x86_vendor) {
 	case X86_VENDOR_AMD:
-		if (boot_cpu_data.x86 != 6 && boot_cpu_data.x86 != 15 &&
-		    boot_cpu_data.x86 != 16 && boot_cpu_data.x86 != 17)
-			return;
-		wd_ops = &k7_wd_ops;
-		break;
+		if (boot_cpu_data.x86 == 6 ||
+		    (boot_cpu_data.x86 >= 0xf && boot_cpu_data.x86 <= 0x15))
+			wd_ops = &k7_wd_ops;
+		return;
 	case X86_VENDOR_INTEL:
 		/* Work around where perfctr1 doesn't have a working enable
 		 * bit as described in the following errata:
-- 
1.6.4.4

[PATCH 3/6] x86, amd: Add support for CPUID topology extension of AMD CPUs

[Andreas Herrmann]

From: Andreas Herrmann <andreas.herrmann3@amd.com>

Node information (ID, number of internal nodes) is provided via
CPUID Fn8000_001e_ECX.

See AMD CPUID Specification (Publication # 25481, Revision 2.34,
September 2010).

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
---
 arch/x86/kernel/cpu/amd.c |   50 +++++++++++++++++++++++---------------------
 1 files changed, 26 insertions(+), 24 deletions(-)

diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 7b875fd..c10c2cc 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -253,37 +253,41 @@ static int __cpuinit nearby_node(int apicid)
 #endif
 
 /*
- * Fixup core topology information for AMD multi-node processors.
- * Assumption: Number of cores in each internal node is the same.
+ * Fixup core topology information for
+ * (1) AMD multi-node processors
+ *     Assumption: Number of cores in each internal node is the same.
  */
 #ifdef CONFIG_X86_HT
-static void __cpuinit amd_fixup_dcm(struct cpuinfo_x86 *c)
+static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 {
-	unsigned long long value;
 	u32 nodes, cores_per_node;
+	u8 node_id;
+	unsigned long long value;
 	int cpu = smp_processor_id();
 
-	if (!cpu_has(c, X86_FEATURE_NODEID_MSR))
-		return;
-
-	/* fixup topology information only once for a core */
-	if (cpu_has(c, X86_FEATURE_AMD_DCM))
+	/* get information required for multi-node processors */
+	if (cpu_has(c, X86_FEATURE_TOPOEXT)) {
+		value = cpuid_ecx(0x8000001e);
+		nodes = ((value >> 8) & 7) + 1;
+		node_id = value & 7;
+	} else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) {
+		rdmsrl(MSR_FAM10H_NODE_ID, value);
+		nodes = ((value >> 3) & 7) + 1;
+		node_id = value & 7;
+	} else
 		return;
 
-	rdmsrl(MSR_FAM10H_NODE_ID, value);
+	/* fixup multi-node processor information */
+	if (nodes > 1) {
+		set_cpu_cap(c, X86_FEATURE_AMD_DCM);
+		cores_per_node = c->x86_max_cores / nodes;
 
-	nodes = ((value >> 3) & 7) + 1;
-	if (nodes == 1)
-		return;
-
-	set_cpu_cap(c, X86_FEATURE_AMD_DCM);
-	cores_per_node = c->x86_max_cores / nodes;
+		/* store NodeID, use llc_shared_map to store sibling info */
+		per_cpu(cpu_llc_id, cpu) = node_id;
 
-	/* store NodeID, use llc_shared_map to store sibling info */
-	per_cpu(cpu_llc_id, cpu) = value & 7;
-
-	/* fixup core id to be in range from 0 to (cores_per_node - 1) */
-	c->cpu_core_id = c->cpu_core_id % cores_per_node;
+		/* core id to be in range from 0 to (cores_per_node - 1) */
+		c->cpu_core_id = c->cpu_core_id % cores_per_node;
+	}
 }
 #endif
 
@@ -304,9 +308,7 @@ static void __cpuinit amd_detect_cmp(struct cpuinfo_x86 *c)
 	c->phys_proc_id = c->initial_apicid >> bits;
 	/* use socket ID also for last level cache */
 	per_cpu(cpu_llc_id, cpu) = c->phys_proc_id;
-	/* fixup topology information on multi-node processors */
-	if ((c->x86 == 0x10) && (c->x86_model == 9))
-		amd_fixup_dcm(c);
+	amd_get_topology(c);
 #endif
 }
 
-- 
1.6.4.4

[PATCH 4/6] x86, amd: Extract compute unit information for AMD CPUs

[Andreas Herrmann]

From: Andreas Herrmann <andreas.herrmann3@amd.com>

Get compute unit information from CPUID Fn8000_001E_EBX.
(See AMD CPUID Specification - publication # 25481, revision 2.34,
September 2010.)

Note that each core on a compute unit still has a core_id of its own.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
---
 arch/x86/include/asm/processor.h |    2 ++
 arch/x86/kernel/cpu/amd.c        |   20 +++++++++++++++-----
 2 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index b871129..cae9c3c 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -110,6 +110,8 @@ struct cpuinfo_x86 {
 	u16			phys_proc_id;
 	/* Core id: */
 	u16			cpu_core_id;
+	/* Compute unit id */
+	u8			compute_unit_id;
 	/* Index into per_cpu list: */
 	u16			cpu_index;
 #endif
diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index c10c2cc..9e093f8 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -256,21 +256,29 @@ static int __cpuinit nearby_node(int apicid)
  * Fixup core topology information for
  * (1) AMD multi-node processors
  *     Assumption: Number of cores in each internal node is the same.
+ * (2) AMD processors supporting compute units
  */
 #ifdef CONFIG_X86_HT
 static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 {
-	u32 nodes, cores_per_node;
+	u32 nodes;
 	u8 node_id;
-	unsigned long long value;
 	int cpu = smp_processor_id();
 
 	/* get information required for multi-node processors */
 	if (cpu_has(c, X86_FEATURE_TOPOEXT)) {
-		value = cpuid_ecx(0x8000001e);
-		nodes = ((value >> 8) & 7) + 1;
-		node_id = value & 7;
+		u32 eax, ebx, ecx, edx;
+
+		cpuid(0x8000001e, &eax, &ebx, &ecx, &edx);
+		nodes = ((ecx >> 8) & 7) + 1;
+		node_id = ecx & 7;
+
+		/* get compute unit information */
+		smp_num_siblings = ((ebx >> 8) & 3) + 1;
+		c->compute_unit_id = ebx & 0xff;
 	} else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) {
+		u64 value;
+
 		rdmsrl(MSR_FAM10H_NODE_ID, value);
 		nodes = ((value >> 3) & 7) + 1;
 		node_id = value & 7;
@@ -279,6 +287,8 @@ static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 
 	/* fixup multi-node processor information */
 	if (nodes > 1) {
+		u32 cores_per_node;
+
 		set_cpu_cap(c, X86_FEATURE_AMD_DCM);
 		cores_per_node = c->x86_max_cores / nodes;
 
-- 
1.6.4.4

[PATCH 5/6] x86, amd: Use compute unit information to determine thread siblings

[Andreas Herrmann]

From: Andreas Herrmann <andreas.herrmann3@amd.com>

This information is vital for different load balancing policies.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
---
 arch/x86/kernel/smpboot.c |   28 ++++++++++++++++++++--------
 1 files changed, 20 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index d3484f1..2ced73b 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -396,6 +396,19 @@ void __cpuinit smp_store_cpu_info(int id)
 		identify_secondary_cpu(c);
 }
 
+static void __cpuinit link_thread_siblings(int cpu1, int cpu2)
+{
+	struct cpuinfo_x86 *c1 = &cpu_data(cpu1);
+	struct cpuinfo_x86 *c2 = &cpu_data(cpu2);
+
+	cpumask_set_cpu(cpu1, cpu_sibling_mask(cpu2));
+	cpumask_set_cpu(cpu2, cpu_sibling_mask(cpu1));
+	cpumask_set_cpu(cpu1, cpu_core_mask(cpu2));
+	cpumask_set_cpu(cpu2, cpu_core_mask(cpu1));
+	cpumask_set_cpu(cpu1, c2->llc_shared_map);
+	cpumask_set_cpu(cpu2, c1->llc_shared_map);
+}
+
 
 void __cpuinit set_cpu_sibling_map(int cpu)
 {
@@ -408,14 +421,13 @@ void __cpuinit set_cpu_sibling_map(int cpu)
 		for_each_cpu(i, cpu_sibling_setup_mask) {
 			struct cpuinfo_x86 *o = &cpu_data(i);
 
-			if (c->phys_proc_id == o->phys_proc_id &&
-			    c->cpu_core_id == o->cpu_core_id) {
-				cpumask_set_cpu(i, cpu_sibling_mask(cpu));
-				cpumask_set_cpu(cpu, cpu_sibling_mask(i));
-				cpumask_set_cpu(i, cpu_core_mask(cpu));
-				cpumask_set_cpu(cpu, cpu_core_mask(i));
-				cpumask_set_cpu(i, c->llc_shared_map);
-				cpumask_set_cpu(cpu, o->llc_shared_map);
+			if (cpu_has(c, X86_FEATURE_TOPOEXT)) {
+				if (c->phys_proc_id == o->phys_proc_id &&
+				    c->compute_unit_id == o->compute_unit_id)
+					link_thread_siblings(cpu, i);
+			} else if (c->phys_proc_id == o->phys_proc_id &&
+				   c->cpu_core_id == o->cpu_core_id) {
+				link_thread_siblings(cpu, i);
 			}
 		}
 	} else {
-- 
1.6.4.4

[PATCH 6/6] x86, amd_nb: Enable GART support for AMD family 0x15 CPUs

[Andreas Herrmann]

From: Andreas Herrmann <andreas.herrmann3@amd.com>

AMD CPU family 0x15 still supports GART for compatibility reasons.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
---
 arch/x86/kernel/amd_nb.c |    4 +++-
 include/linux/pci_ids.h  |    1 +
 2 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/amd_nb.c b/arch/x86/kernel/amd_nb.c
index 4ffc38d..8f6463d 100644
--- a/arch/x86/kernel/amd_nb.c
+++ b/arch/x86/kernel/amd_nb.c
@@ -15,6 +15,7 @@ static u32 *flush_words;
 struct pci_device_id k8_nb_ids[] = {
 	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_K8_NB_MISC) },
 	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_10H_NB_MISC) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_15H_NB_MISC) },
 	{}
 };
 EXPORT_SYMBOL(k8_nb_ids);
@@ -45,7 +46,8 @@ int cache_k8_northbridges(void)
 		k8_northbridges.num++;
 
 	/* some CPU families (e.g. family 0x11) do not support GART */
-	if (boot_cpu_data.x86 == 0xf || boot_cpu_data.x86 == 0x10)
+	if (boot_cpu_data.x86 == 0xf || boot_cpu_data.x86 == 0x10 ||
+	    boot_cpu_data.x86 == 0x15)
 		k8_northbridges.gart_supported = 1;
 
 	k8_northbridges.nb_misc = kmalloc((k8_northbridges.num + 1) *
diff --git a/include/linux/pci_ids.h b/include/linux/pci_ids.h
index 570fdde..2615c37 100644
--- a/include/linux/pci_ids.h
+++ b/include/linux/pci_ids.h
@@ -517,6 +517,7 @@
 #define PCI_DEVICE_ID_AMD_11H_NB_DRAM	0x1302
 #define PCI_DEVICE_ID_AMD_11H_NB_MISC	0x1303
 #define PCI_DEVICE_ID_AMD_11H_NB_LINK	0x1304
+#define PCI_DEVICE_ID_AMD_15H_NB_MISC	0x1603
 #define PCI_DEVICE_ID_AMD_LANCE		0x2000
 #define PCI_DEVICE_ID_AMD_LANCE_HOME	0x2001
 #define PCI_DEVICE_ID_AMD_SCSI		0x2020
-- 
1.6.4.4

[tip:x86/amd-nb] x86, mtrr: Assume SYS_CFG[Tom2ForceMemTypeWB] exists on all future AMD CPUs

[tip-bot for Andreas Herrmann]

Commit-ID:  3fdbf004c1706480a7c7fac3c9d836fa6df20d7d
Gitweb:     http://git.kernel.org/tip/3fdbf004c1706480a7c7fac3c9d836fa6df20d7d
Author:     Andreas Herrmann <andreas.herrmann3@amd.com>
AuthorDate: Thu, 30 Sep 2010 14:32:35 +0200
Committer:  H. Peter Anvin <hpa@linux.intel.com>
CommitDate: Fri, 1 Oct 2010 16:18:31 -0700

x86, mtrr: Assume SYS_CFG[Tom2ForceMemTypeWB] exists on all future AMD CPUs

Instead of adapting the CPU family check in amd_special_default_mtrr()
for each new CPU family assume that all new AMD CPUs support the
necessary bits in SYS_CFG MSR.

Tom2Enabled is architectural (defined in APM Vol.2).
Tom2ForceMemTypeWB is defined in all BKDGs starting with K8 NPT.
In pre K8-NPT BKDG this bit is reserved (read as zero).

W/o this adaption Linux would unnecessarily complain about bad MTRR
settings on every new AMD CPU family, e.g.

[    0.000000] WARNING: BIOS bug: CPU MTRRs don't cover all of memory, losing 4863MB of RAM.

Cc: stable@kernel.org # .32.x, .35.x
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100930123235.GB20545@loge.amd.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
---
 arch/x86/kernel/cpu/mtrr/cleanup.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/cpu/mtrr/cleanup.c b/arch/x86/kernel/cpu/mtrr/cleanup.c
index c5f59d0..ac140c7 100644
--- a/arch/x86/kernel/cpu/mtrr/cleanup.c
+++ b/arch/x86/kernel/cpu/mtrr/cleanup.c
@@ -827,7 +827,7 @@ int __init amd_special_default_mtrr(void)
 
 	if (boot_cpu_data.x86_vendor != X86_VENDOR_AMD)
 		return 0;
-	if (boot_cpu_data.x86 < 0xf || boot_cpu_data.x86 > 0x11)
+	if (boot_cpu_data.x86 < 0xf)
 		return 0;
 	/* In case some hypervisor doesn't pass SYSCFG through: */
 	if (rdmsr_safe(MSR_K8_SYSCFG, &l, &h) < 0)

[tip:x86/amd-nb] x86, nmi: Support NMI watchdog on newer AMD CPU families

[tip-bot for Andreas Herrmann]

Commit-ID:  420b13b60a3e5c5dcc6ec290e131cf5fbc603d94
Gitweb:     http://git.kernel.org/tip/420b13b60a3e5c5dcc6ec290e131cf5fbc603d94
Author:     Andreas Herrmann <andreas.herrmann3@amd.com>
AuthorDate: Thu, 30 Sep 2010 14:33:58 +0200
Committer:  H. Peter Anvin <hpa@linux.intel.com>
CommitDate: Fri, 1 Oct 2010 16:18:32 -0700

x86, nmi: Support NMI watchdog on newer AMD CPU families

CPU families 0x12, 0x14 and 0x15 support this functionality.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100930123357.GC20545@loge.amd.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
---
 arch/x86/kernel/cpu/perfctr-watchdog.c |    9 ++++-----
 1 files changed, 4 insertions(+), 5 deletions(-)

diff --git a/arch/x86/kernel/cpu/perfctr-watchdog.c b/arch/x86/kernel/cpu/perfctr-watchdog.c
index fb329e9..d9f4ff8 100644
--- a/arch/x86/kernel/cpu/perfctr-watchdog.c
+++ b/arch/x86/kernel/cpu/perfctr-watchdog.c
@@ -700,11 +700,10 @@ static void probe_nmi_watchdog(void)
 {
 	switch (boot_cpu_data.x86_vendor) {
 	case X86_VENDOR_AMD:
-		if (boot_cpu_data.x86 != 6 && boot_cpu_data.x86 != 15 &&
-		    boot_cpu_data.x86 != 16 && boot_cpu_data.x86 != 17)
-			return;
-		wd_ops = &k7_wd_ops;
-		break;
+		if (boot_cpu_data.x86 == 6 ||
+		    (boot_cpu_data.x86 >= 0xf && boot_cpu_data.x86 <= 0x15))
+			wd_ops = &k7_wd_ops;
+		return;
 	case X86_VENDOR_INTEL:
 		/* Work around where perfctr1 doesn't have a working enable
 		 * bit as described in the following errata:

[tip:x86/amd-nb] x86, amd: Add support for CPUID topology extension of AMD CPUs

[tip-bot for Andreas Herrmann]

Commit-ID:  23588c38a84c9175c6668789b64ffba4651e5c6a
Gitweb:     http://git.kernel.org/tip/23588c38a84c9175c6668789b64ffba4651e5c6a
Author:     Andreas Herrmann <andreas.herrmann3@amd.com>
AuthorDate: Thu, 30 Sep 2010 14:36:28 +0200
Committer:  H. Peter Anvin <hpa@linux.intel.com>
CommitDate: Fri, 1 Oct 2010 16:18:32 -0700

x86, amd: Add support for CPUID topology extension of AMD CPUs

Node information (ID, number of internal nodes) is provided via
CPUID Fn8000_001e_ECX.

See AMD CPUID Specification (Publication # 25481, Revision 2.34,
September 2010).

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100930123628.GD20545@loge.amd.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
---
 arch/x86/kernel/cpu/amd.c |   50 +++++++++++++++++++++++---------------------
 1 files changed, 26 insertions(+), 24 deletions(-)

diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 0f0ace5..7e6a37d 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -253,37 +253,41 @@ static int __cpuinit nearby_node(int apicid)
 #endif
 
 /*
- * Fixup core topology information for AMD multi-node processors.
- * Assumption: Number of cores in each internal node is the same.
+ * Fixup core topology information for
+ * (1) AMD multi-node processors
+ *     Assumption: Number of cores in each internal node is the same.
  */
 #ifdef CONFIG_X86_HT
-static void __cpuinit amd_fixup_dcm(struct cpuinfo_x86 *c)
+static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 {
-	unsigned long long value;
 	u32 nodes, cores_per_node;
+	u8 node_id;
+	unsigned long long value;
 	int cpu = smp_processor_id();
 
-	if (!cpu_has(c, X86_FEATURE_NODEID_MSR))
-		return;
-
-	/* fixup topology information only once for a core */
-	if (cpu_has(c, X86_FEATURE_AMD_DCM))
+	/* get information required for multi-node processors */
+	if (cpu_has(c, X86_FEATURE_TOPOEXT)) {
+		value = cpuid_ecx(0x8000001e);
+		nodes = ((value >> 8) & 7) + 1;
+		node_id = value & 7;
+	} else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) {
+		rdmsrl(MSR_FAM10H_NODE_ID, value);
+		nodes = ((value >> 3) & 7) + 1;
+		node_id = value & 7;
+	} else
 		return;
 
-	rdmsrl(MSR_FAM10H_NODE_ID, value);
+	/* fixup multi-node processor information */
+	if (nodes > 1) {
+		set_cpu_cap(c, X86_FEATURE_AMD_DCM);
+		cores_per_node = c->x86_max_cores / nodes;
 
-	nodes = ((value >> 3) & 7) + 1;
-	if (nodes == 1)
-		return;
-
-	set_cpu_cap(c, X86_FEATURE_AMD_DCM);
-	cores_per_node = c->x86_max_cores / nodes;
+		/* store NodeID, use llc_shared_map to store sibling info */
+		per_cpu(cpu_llc_id, cpu) = node_id;
 
-	/* store NodeID, use llc_shared_map to store sibling info */
-	per_cpu(cpu_llc_id, cpu) = value & 7;
-
-	/* fixup core id to be in range from 0 to (cores_per_node - 1) */
-	c->cpu_core_id = c->cpu_core_id % cores_per_node;
+		/* core id to be in range from 0 to (cores_per_node - 1) */
+		c->cpu_core_id = c->cpu_core_id % cores_per_node;
+	}
 }
 #endif
 
@@ -304,9 +308,7 @@ static void __cpuinit amd_detect_cmp(struct cpuinfo_x86 *c)
 	c->phys_proc_id = c->initial_apicid >> bits;
 	/* use socket ID also for last level cache */
 	per_cpu(cpu_llc_id, cpu) = c->phys_proc_id;
-	/* fixup topology information on multi-node processors */
-	if ((c->x86 == 0x10) && (c->x86_model == 9))
-		amd_fixup_dcm(c);
+	amd_get_topology(c);
 #endif
 }
 

[tip:x86/amd-nb] x86, amd: Extract compute unit information for AMD CPUs

[tip-bot for Andreas Herrmann]

Commit-ID:  6057b4d331f19a3ea51aec463ea7839c128b3227
Gitweb:     http://git.kernel.org/tip/6057b4d331f19a3ea51aec463ea7839c128b3227
Author:     Andreas Herrmann <andreas.herrmann3@amd.com>
AuthorDate: Thu, 30 Sep 2010 14:38:57 +0200
Committer:  H. Peter Anvin <hpa@linux.intel.com>
CommitDate: Fri, 1 Oct 2010 16:18:32 -0700

x86, amd: Extract compute unit information for AMD CPUs

Get compute unit information from CPUID Fn8000_001E_EBX.
(See AMD CPUID Specification - publication # 25481, revision 2.34,
September 2010.)

Note that each core on a compute unit still has a core_id of its own.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100930123857.GE20545@loge.amd.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
---
 arch/x86/include/asm/processor.h |    2 ++
 arch/x86/kernel/cpu/amd.c        |   20 +++++++++++++++-----
 2 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/arch/x86/include/asm/processor.h b/arch/x86/include/asm/processor.h
index 325b7bd..69e80c2 100644
--- a/arch/x86/include/asm/processor.h
+++ b/arch/x86/include/asm/processor.h
@@ -110,6 +110,8 @@ struct cpuinfo_x86 {
 	u16			phys_proc_id;
 	/* Core id: */
 	u16			cpu_core_id;
+	/* Compute unit id */
+	u8			compute_unit_id;
 	/* Index into per_cpu list: */
 	u16			cpu_index;
 #endif
diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 7e6a37d..70168ab 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -256,21 +256,29 @@ static int __cpuinit nearby_node(int apicid)
  * Fixup core topology information for
  * (1) AMD multi-node processors
  *     Assumption: Number of cores in each internal node is the same.
+ * (2) AMD processors supporting compute units
  */
 #ifdef CONFIG_X86_HT
 static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 {
-	u32 nodes, cores_per_node;
+	u32 nodes;
 	u8 node_id;
-	unsigned long long value;
 	int cpu = smp_processor_id();
 
 	/* get information required for multi-node processors */
 	if (cpu_has(c, X86_FEATURE_TOPOEXT)) {
-		value = cpuid_ecx(0x8000001e);
-		nodes = ((value >> 8) & 7) + 1;
-		node_id = value & 7;
+		u32 eax, ebx, ecx, edx;
+
+		cpuid(0x8000001e, &eax, &ebx, &ecx, &edx);
+		nodes = ((ecx >> 8) & 7) + 1;
+		node_id = ecx & 7;
+
+		/* get compute unit information */
+		smp_num_siblings = ((ebx >> 8) & 3) + 1;
+		c->compute_unit_id = ebx & 0xff;
 	} else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) {
+		u64 value;
+
 		rdmsrl(MSR_FAM10H_NODE_ID, value);
 		nodes = ((value >> 3) & 7) + 1;
 		node_id = value & 7;
@@ -279,6 +287,8 @@ static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 
 	/* fixup multi-node processor information */
 	if (nodes > 1) {
+		u32 cores_per_node;
+
 		set_cpu_cap(c, X86_FEATURE_AMD_DCM);
 		cores_per_node = c->x86_max_cores / nodes;
 

[tip:x86/amd-nb] x86, amd: Use compute unit information to determine thread siblings

[tip-bot for Andreas Herrmann]

Commit-ID:  d4fbe4f03557e1fd4d9bbb3a1957aad560f39e96
Gitweb:     http://git.kernel.org/tip/d4fbe4f03557e1fd4d9bbb3a1957aad560f39e96
Author:     Andreas Herrmann <andreas.herrmann3@amd.com>
AuthorDate: Thu, 30 Sep 2010 14:41:56 +0200
Committer:  H. Peter Anvin <hpa@linux.intel.com>
CommitDate: Fri, 1 Oct 2010 16:18:32 -0700

x86, amd: Use compute unit information to determine thread siblings

This information is vital for different load balancing policies.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100930124156.GF20545@loge.amd.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
---
 arch/x86/kernel/smpboot.c |   28 ++++++++++++++++++++--------
 1 files changed, 20 insertions(+), 8 deletions(-)

diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index 8b3bfc4..bc2cc44 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -397,6 +397,19 @@ void __cpuinit smp_store_cpu_info(int id)
 		identify_secondary_cpu(c);
 }
 
+static void __cpuinit link_thread_siblings(int cpu1, int cpu2)
+{
+	struct cpuinfo_x86 *c1 = &cpu_data(cpu1);
+	struct cpuinfo_x86 *c2 = &cpu_data(cpu2);
+
+	cpumask_set_cpu(cpu1, cpu_sibling_mask(cpu2));
+	cpumask_set_cpu(cpu2, cpu_sibling_mask(cpu1));
+	cpumask_set_cpu(cpu1, cpu_core_mask(cpu2));
+	cpumask_set_cpu(cpu2, cpu_core_mask(cpu1));
+	cpumask_set_cpu(cpu1, c2->llc_shared_map);
+	cpumask_set_cpu(cpu2, c1->llc_shared_map);
+}
+
 
 void __cpuinit set_cpu_sibling_map(int cpu)
 {
@@ -409,14 +422,13 @@ void __cpuinit set_cpu_sibling_map(int cpu)
 		for_each_cpu(i, cpu_sibling_setup_mask) {
 			struct cpuinfo_x86 *o = &cpu_data(i);
 
-			if (c->phys_proc_id == o->phys_proc_id &&
-			    c->cpu_core_id == o->cpu_core_id) {
-				cpumask_set_cpu(i, cpu_sibling_mask(cpu));
-				cpumask_set_cpu(cpu, cpu_sibling_mask(i));
-				cpumask_set_cpu(i, cpu_core_mask(cpu));
-				cpumask_set_cpu(cpu, cpu_core_mask(i));
-				cpumask_set_cpu(i, c->llc_shared_map);
-				cpumask_set_cpu(cpu, o->llc_shared_map);
+			if (cpu_has(c, X86_FEATURE_TOPOEXT)) {
+				if (c->phys_proc_id == o->phys_proc_id &&
+				    c->compute_unit_id == o->compute_unit_id)
+					link_thread_siblings(cpu, i);
+			} else if (c->phys_proc_id == o->phys_proc_id &&
+				   c->cpu_core_id == o->cpu_core_id) {
+				link_thread_siblings(cpu, i);
 			}
 		}
 	} else {

[tip:x86/amd-nb] x86, amd_nb: Enable GART support for AMD family 0x15 CPUs

[tip-bot for Andreas Herrmann]

Commit-ID:  5c80cc78de46aef6cd5e714208da05c3f7f548f8
Gitweb:     http://git.kernel.org/tip/5c80cc78de46aef6cd5e714208da05c3f7f548f8
Author:     Andreas Herrmann <andreas.herrmann3@amd.com>
AuthorDate: Thu, 30 Sep 2010 14:43:16 +0200
Committer:  H. Peter Anvin <hpa@linux.intel.com>
CommitDate: Fri, 1 Oct 2010 16:18:32 -0700

x86, amd_nb: Enable GART support for AMD family 0x15 CPUs

AMD CPU family 0x15 still supports GART for compatibility reasons.

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100930124316.GG20545@loge.amd.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
---
 arch/x86/kernel/amd_nb.c |    4 +++-
 include/linux/pci_ids.h  |    1 +
 2 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/amd_nb.c b/arch/x86/kernel/amd_nb.c
index 4ffc38d..8f6463d 100644
--- a/arch/x86/kernel/amd_nb.c
+++ b/arch/x86/kernel/amd_nb.c
@@ -15,6 +15,7 @@ static u32 *flush_words;
 struct pci_device_id k8_nb_ids[] = {
 	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_K8_NB_MISC) },
 	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_10H_NB_MISC) },
+	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_15H_NB_MISC) },
 	{}
 };
 EXPORT_SYMBOL(k8_nb_ids);
@@ -45,7 +46,8 @@ int cache_k8_northbridges(void)
 		k8_northbridges.num++;
 
 	/* some CPU families (e.g. family 0x11) do not support GART */
-	if (boot_cpu_data.x86 == 0xf || boot_cpu_data.x86 == 0x10)
+	if (boot_cpu_data.x86 == 0xf || boot_cpu_data.x86 == 0x10 ||
+	    boot_cpu_data.x86 == 0x15)
 		k8_northbridges.gart_supported = 1;
 
 	k8_northbridges.nb_misc = kmalloc((k8_northbridges.num + 1) *
diff --git a/include/linux/pci_ids.h b/include/linux/pci_ids.h
index 10d3330..edc0279 100644
--- a/include/linux/pci_ids.h
+++ b/include/linux/pci_ids.h
@@ -514,6 +514,7 @@
 #define PCI_DEVICE_ID_AMD_11H_NB_DRAM	0x1302
 #define PCI_DEVICE_ID_AMD_11H_NB_MISC	0x1303
 #define PCI_DEVICE_ID_AMD_11H_NB_LINK	0x1304
+#define PCI_DEVICE_ID_AMD_15H_NB_MISC	0x1603
 #define PCI_DEVICE_ID_AMD_LANCE		0x2000
 #define PCI_DEVICE_ID_AMD_LANCE_HOME	0x2001
 #define PCI_DEVICE_ID_AMD_SCSI		0x2020