Merge branch 'x86/amd-nb' into x86/urgent

Merge reason: This is one followup commit that was not in x86/mm - merge it via the urgent path

Signed-off-by: Ingo Molnar <mingo@elte.hu>

arch/x86/include/asm/amd_nb.h

@@ -9,15 +9,17 @@ struct amd_nb_bus_dev_range {
 	u8 dev_limit;
 };
 
-extern struct pci_device_id amd_nb_misc_ids[];
+extern const struct pci_device_id amd_nb_misc_ids[];
 extern const struct amd_nb_bus_dev_range amd_nb_bus_dev_ranges[];
 struct bootnode;
 
-extern int early_is_amd_nb(u32 value);
+extern bool early_is_amd_nb(u32 value);
 extern int amd_cache_northbridges(void);
 extern void amd_flush_garts(void);
 extern int amd_numa_init(unsigned long start_pfn, unsigned long end_pfn);
 extern int amd_scan_nodes(void);
+extern int amd_get_subcaches(int);
+extern int amd_set_subcaches(int, int);
 
 #ifdef CONFIG_NUMA_EMU
 extern void amd_fake_nodes(const struct bootnode *nodes, int nr_nodes);
@@ -26,6 +28,7 @@ extern void amd_get_nodes(struct bootnode *nodes);
 
 struct amd_northbridge {
 	struct pci_dev *misc;
+	struct pci_dev *link;
 };
 
 struct amd_northbridge_info {
@@ -35,17 +38,18 @@ struct amd_northbridge_info {
 };
 extern struct amd_northbridge_info amd_northbridges;
 
-#define AMD_NB_GART			0x1
-#define AMD_NB_L3_INDEX_DISABLE		0x2
+#define AMD_NB_GART			BIT(0)
+#define AMD_NB_L3_INDEX_DISABLE		BIT(1)
+#define AMD_NB_L3_PARTITIONING		BIT(2)
 
 #ifdef CONFIG_AMD_NB
 
-static inline int amd_nb_num(void)
+static inline u16 amd_nb_num(void)
 {
 	return amd_northbridges.num;
 }
 
-static inline int amd_nb_has_feature(int feature)
+static inline bool amd_nb_has_feature(unsigned feature)
 {
 	return ((amd_northbridges.flags & feature) == feature);
 }

arch/x86/kernel/amd_nb.c

@@ -12,14 +12,19 @@
 
 static u32 *flush_words;
 
-struct pci_device_id amd_nb_misc_ids[] = {
+const struct pci_device_id amd_nb_misc_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(amd_nb_misc_ids);
 
+static struct pci_device_id amd_nb_link_ids[] = {
+	{ PCI_DEVICE(PCI_VENDOR_ID_AMD, PCI_DEVICE_ID_AMD_15H_NB_LINK) },
+	{}
+};
+
 const struct amd_nb_bus_dev_range amd_nb_bus_dev_ranges[] __initconst = {
 	{ 0x00, 0x18, 0x20 },
 	{ 0xff, 0x00, 0x20 },
@@ -31,7 +36,7 @@ struct amd_northbridge_info amd_northbridges;
 EXPORT_SYMBOL(amd_northbridges);
 
 static struct pci_dev *next_northbridge(struct pci_dev *dev,
-					struct pci_device_id *ids)
+					const struct pci_device_id *ids)
 {
 	do {
 		dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, dev);
@@ -43,9 +48,9 @@ static struct pci_dev *next_northbridge(struct pci_dev *dev,
 
 int amd_cache_northbridges(void)
 {
-	int i = 0;
+	u16 i = 0;
 	struct amd_northbridge *nb;
-	struct pci_dev *misc;
+	struct pci_dev *misc, *link;
 
 	if (amd_nb_num())
 		return 0;
@@ -64,10 +69,12 @@ int amd_cache_northbridges(void)
 	amd_northbridges.nb = nb;
 	amd_northbridges.num = i;
 
-	misc = NULL;
+	link = misc = NULL;
 	for (i = 0; i != amd_nb_num(); i++) {
 		node_to_amd_nb(i)->misc = misc =
 			next_northbridge(misc, amd_nb_misc_ids);
+		node_to_amd_nb(i)->link = link =
+			next_northbridge(link, amd_nb_link_ids);
         }
 
 	/* some CPU families (e.g. family 0x11) do not support GART */
@@ -85,26 +92,95 @@ int amd_cache_northbridges(void)
 	     boot_cpu_data.x86_mask >= 0x1))
 		amd_northbridges.flags |= AMD_NB_L3_INDEX_DISABLE;
 
+	if (boot_cpu_data.x86 == 0x15)
+		amd_northbridges.flags |= AMD_NB_L3_INDEX_DISABLE;
+
+	/* L3 cache partitioning is supported on family 0x15 */
+	if (boot_cpu_data.x86 == 0x15)
+		amd_northbridges.flags |= AMD_NB_L3_PARTITIONING;
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(amd_cache_northbridges);
 
-/* Ignores subdevice/subvendor but as far as I can figure out
-   they're useless anyways */
-int __init early_is_amd_nb(u32 device)
+/*
+ * Ignores subdevice/subvendor but as far as I can figure out
+ * they're useless anyways
+ */
+bool __init early_is_amd_nb(u32 device)
 {
-	struct pci_device_id *id;
+	const struct pci_device_id *id;
 	u32 vendor = device & 0xffff;
+
 	device >>= 16;
 	for (id = amd_nb_misc_ids; id->vendor; id++)
 		if (vendor == id->vendor && device == id->device)
-			return 1;
+			return true;
+	return false;
+}
+
+int amd_get_subcaches(int cpu)
+{
+	struct pci_dev *link = node_to_amd_nb(amd_get_nb_id(cpu))->link;
+	unsigned int mask;
+	int cuid = 0;
+
+	if (!amd_nb_has_feature(AMD_NB_L3_PARTITIONING))
+		return 0;
+
+	pci_read_config_dword(link, 0x1d4, &mask);
+
+#ifdef CONFIG_SMP
+	cuid = cpu_data(cpu).compute_unit_id;
+#endif
+	return (mask >> (4 * cuid)) & 0xf;
+}
+
+int amd_set_subcaches(int cpu, int mask)
+{
+	static unsigned int reset, ban;
+	struct amd_northbridge *nb = node_to_amd_nb(amd_get_nb_id(cpu));
+	unsigned int reg;
+	int cuid = 0;
+
+	if (!amd_nb_has_feature(AMD_NB_L3_PARTITIONING) || mask > 0xf)
+		return -EINVAL;
+
+	/* if necessary, collect reset state of L3 partitioning and BAN mode */
+	if (reset == 0) {
+		pci_read_config_dword(nb->link, 0x1d4, &reset);
+		pci_read_config_dword(nb->misc, 0x1b8, &ban);
+		ban &= 0x180000;
+	}
+
+	/* deactivate BAN mode if any subcaches are to be disabled */
+	if (mask != 0xf) {
+		pci_read_config_dword(nb->misc, 0x1b8, &reg);
+		pci_write_config_dword(nb->misc, 0x1b8, reg & ~0x180000);
+	}
+
+#ifdef CONFIG_SMP
+	cuid = cpu_data(cpu).compute_unit_id;
+#endif
+	mask <<= 4 * cuid;
+	mask |= (0xf ^ (1 << cuid)) << 26;
+
+	pci_write_config_dword(nb->link, 0x1d4, mask);
+
+	/* reset BAN mode if L3 partitioning returned to reset state */
+	pci_read_config_dword(nb->link, 0x1d4, &reg);
+	if (reg == reset) {
+		pci_read_config_dword(nb->misc, 0x1b8, &reg);
+		reg &= ~0x180000;
+		pci_write_config_dword(nb->misc, 0x1b8, reg | ban);
+	}
+
 	return 0;
 }
 
-int amd_cache_gart(void)
+static int amd_cache_gart(void)
 {
-       int i;
+	u16 i;
 
        if (!amd_nb_has_feature(AMD_NB_GART))
                return 0;

arch/x86/kernel/cpu/amd.c

@@ -261,7 +261,7 @@ static int __cpuinit nearby_node(int apicid)
 #ifdef CONFIG_X86_HT
 static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 {
-	u32 nodes;
+	u32 nodes, cores_per_cu = 1;
 	u8 node_id;
 	int cpu = smp_processor_id();
 
@@ -276,6 +276,7 @@ static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 		/* get compute unit information */
 		smp_num_siblings = ((ebx >> 8) & 3) + 1;
 		c->compute_unit_id = ebx & 0xff;
+		cores_per_cu += ((ebx >> 8) & 3);
 	} else if (cpu_has(c, X86_FEATURE_NODEID_MSR)) {
 		u64 value;
 
@@ -288,15 +289,18 @@ static void __cpuinit amd_get_topology(struct cpuinfo_x86 *c)
 	/* fixup multi-node processor information */
 	if (nodes > 1) {
 		u32 cores_per_node;
+		u32 cus_per_node;
 
 		set_cpu_cap(c, X86_FEATURE_AMD_DCM);
 		cores_per_node = c->x86_max_cores / nodes;
+		cus_per_node = cores_per_node / cores_per_cu;
 
 		/* store NodeID, use llc_shared_map to store sibling info */
 		per_cpu(cpu_llc_id, cpu) = node_id;
 
-		/* 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 has to be in the [0 .. cores_per_node - 1] range */
+		c->cpu_core_id %= cores_per_node;
+		c->compute_unit_id %= cus_per_node;
 	}
 }
 #endif