Merge

arch/i386/Kconfig

@@ -342,11 +342,6 @@ config X86_ALIGNMENT_16
 	depends on MWINCHIP3D || MWINCHIP2 || MWINCHIPC6 || MCYRIXIII || MELAN || MK6 || M586MMX || M586TSC || M586 || M486 || MVIAC3_2
 	default y
 
-config X86_TSC
-	bool
-	depends on MWINCHIP3D || MWINCHIP2 || MCRUSOE || MCYRIXIII || MK7 || MK6 || MPENTIUM4 || MPENTIUMIII || MPENTIUMII || M686 || M586MMX || M586TSC || MK8 || MVIAC3_2
-	default y
-
 config X86_GOOD_APIC
 	bool
 	depends on MK7 || MPENTIUM4 || MPENTIUMIII || MPENTIUMII || M686 || M586MMX || MK8
@@ -500,6 +495,11 @@ config HAVE_ARCH_BOOTMEM_NODE
 	depends on NUMA
 	default y
 
+config X86_TSC
+	bool
+	depends on (MWINCHIP3D || MWINCHIP2 || MCRUSOE || MCYRIXIII || MK7 || MK6 || MPENTIUM4 || MPENTIUMIII || MPENTIUMII || M686 || M586MMX || M586TSC || MK8 || MVIAC3_2) && !X86_NUMAQ
+	default y
+
 config X86_MCE
 	bool "Machine Check Exception"
 	---help---

arch/i386/kernel/i386_ksyms.c

@@ -68,7 +68,6 @@ EXPORT_SYMBOL(EISA_bus);
 EXPORT_SYMBOL(MCA_bus);
 #ifdef CONFIG_DISCONTIGMEM
 EXPORT_SYMBOL(node_data);
-EXPORT_SYMBOL(pfn_to_nid);
 #endif
 #ifdef CONFIG_X86_NUMAQ
 EXPORT_SYMBOL(xquad_portio);

arch/i386/kernel/mpparse.c

@@ -110,7 +110,7 @@ void __init MP_processor_info (struct mpc_config_processor *m)
 	if (!(m->mpc_cpuflag & CPU_ENABLED))
 		return;
 
-	apicid = mpc_apic_id(m, translation_table[mpc_record]->trans_quad);
+	apicid = mpc_apic_id(m, translation_table[mpc_record]);
 
 	if (m->mpc_featureflag&(1<<0))
 		Dprintk("    Floating point unit present.\n");

arch/i386/kernel/numaq.c

@@ -27,6 +27,7 @@
 #include <linux/mm.h>
 #include <linux/bootmem.h>
 #include <linux/mmzone.h>
+#include <linux/module.h>
 #include <asm/numaq.h>
 
 /* These are needed before the pgdat's are created */
@@ -82,19 +83,7 @@ static void __init smp_dump_qct(void)
  * physnode_map[8- ] = -1;
  */
 int physnode_map[MAX_ELEMENTS] = { [0 ... (MAX_ELEMENTS - 1)] = -1};
-
-#define PFN_TO_ELEMENT(pfn) (pfn / PAGES_PER_ELEMENT)
-#define PA_TO_ELEMENT(pa) (PFN_TO_ELEMENT(pa >> PAGE_SHIFT))
-
-int pfn_to_nid(unsigned long pfn)
-{
-	int nid = physnode_map[PFN_TO_ELEMENT(pfn)];
-
-	if (nid == -1)
-		BUG(); /* address is not present */
-
-	return nid;
-}
+EXPORT_SYMBOL(physnode_map);
 
 /*
  * for each node mark the regions

arch/i386/mm/discontig.c

@@ -48,6 +48,14 @@ extern unsigned long max_low_pfn;
 extern unsigned long totalram_pages;
 extern unsigned long totalhigh_pages;
 
+#define LARGE_PAGE_BYTES (PTRS_PER_PTE * PAGE_SIZE)
+
+unsigned long node_remap_start_pfn[MAX_NUMNODES];
+unsigned long node_remap_size[MAX_NUMNODES];
+unsigned long node_remap_offset[MAX_NUMNODES];
+void *node_remap_start_vaddr[MAX_NUMNODES];
+void set_pmd_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags);
+
 /*
  * Find the highest page frame number we have available for the node
  */
@@ -65,12 +73,13 @@ static void __init find_max_pfn_node(int nid)
  */
 static void __init allocate_pgdat(int nid)
 {
-	unsigned long node_datasz;
-
-	node_datasz = PFN_UP(sizeof(struct pglist_data));
+	if (nid)
+		NODE_DATA(nid) = (pg_data_t *)node_remap_start_vaddr[nid];
+	else {
 		NODE_DATA(nid) = (pg_data_t *)(__va(min_low_pfn << PAGE_SHIFT));
-	min_low_pfn += node_datasz;
-	memset(NODE_DATA(nid), 0, sizeof(struct pglist_data));
+		min_low_pfn += PFN_UP(sizeof(pg_data_t));
+		memset(NODE_DATA(nid), 0, sizeof(pg_data_t));
+	}
 }
 
 /*
@@ -113,14 +122,6 @@ static void __init register_bootmem_low_pages(unsigned long system_max_low_pfn)
 	}
 }
 
-#define LARGE_PAGE_BYTES (PTRS_PER_PTE * PAGE_SIZE)
-
-unsigned long node_remap_start_pfn[MAX_NUMNODES];
-unsigned long node_remap_size[MAX_NUMNODES];
-unsigned long node_remap_offset[MAX_NUMNODES];
-void *node_remap_start_vaddr[MAX_NUMNODES];
-extern void set_pmd_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags);
-
 void __init remap_numa_kva(void)
 {
 	void *vaddr;
@@ -145,7 +146,7 @@ static unsigned long calculate_numa_remap_pages(void)
 	for (nid = 1; nid < numnodes; nid++) {
 		/* calculate the size of the mem_map needed in bytes */
 		size = (node_end_pfn[nid] - node_start_pfn[nid] + 1) 
-			* sizeof(struct page);
+			* sizeof(struct page) + sizeof(pg_data_t);
 		/* convert size to large (pmd size) pages, rounding up */
 		size = (size + LARGE_PAGE_BYTES - 1) / LARGE_PAGE_BYTES;
 		/* now the roundup is correct, convert to PAGE_SIZE pages */
@@ -195,9 +196,9 @@ unsigned long __init setup_memory(void)
 	printk("Low memory ends at vaddr %08lx\n",
 			(ulong) pfn_to_kaddr(max_low_pfn));
 	for (nid = 0; nid < numnodes; nid++) {
-		allocate_pgdat(nid);
 		node_remap_start_vaddr[nid] = pfn_to_kaddr(
 			highstart_pfn - node_remap_offset[nid]);
+		allocate_pgdat(nid);
 		printk ("node %d will remap to vaddr %08lx - %08lx\n", nid,
 			(ulong) node_remap_start_vaddr[nid],
 			(ulong) pfn_to_kaddr(highstart_pfn
@@ -251,13 +252,6 @@ unsigned long __init setup_memory(void)
 	 */
 	find_smp_config();
 
-	/*insert other nodes into pgdat_list*/
-	for (nid = 1; nid < numnodes; nid++){       
-		NODE_DATA(nid)->pgdat_next = pgdat_list;
-		pgdat_list = NODE_DATA(nid);
-	}
-       
-
 #ifdef CONFIG_BLK_DEV_INITRD
 	if (LOADER_TYPE && INITRD_START) {
 		if (INITRD_START + INITRD_SIZE <= (system_max_low_pfn << PAGE_SHIFT)) {
@@ -282,6 +276,18 @@ void __init zone_sizes_init(void)
 {
 	int nid;
 
+	/*
+	 * Insert nodes into pgdat_list backward so they appear in order.
+	 * Clobber node 0's links and NULL out pgdat_list before starting.
+	 */
+	pgdat_list = NULL;
+	for (nid = numnodes - 1; nid >= 0; nid--) {       
+		if (nid)
+			memset(NODE_DATA(nid), 0, sizeof(pg_data_t));
+		NODE_DATA(nid)->pgdat_next = pgdat_list;
+		pgdat_list = NODE_DATA(nid);
+	}
+
 	for (nid = 0; nid < numnodes; nid++) {
 		unsigned long zones_size[MAX_NR_ZONES] = {0, 0, 0};
 		unsigned long *zholes_size;
@@ -314,13 +320,18 @@ void __init zone_sizes_init(void)
 		 * normal bootmem allocator, but other nodes come from the
 		 * remapped KVA area - mbligh
 		 */
-		if (nid)
-			free_area_init_node(nid, NODE_DATA(nid), 
-				node_remap_start_vaddr[nid], zones_size, 
-				start, zholes_size);
-		else
+		if (!nid)
 			free_area_init_node(nid, NODE_DATA(nid), 0, 
 				zones_size, start, zholes_size);
+		else {
+			unsigned long lmem_map;
+			lmem_map = (unsigned long)node_remap_start_vaddr[nid];
+			lmem_map += sizeof(pg_data_t) + PAGE_SIZE - 1;
+			lmem_map &= PAGE_MASK;
+			free_area_init_node(nid, NODE_DATA(nid), 
+				(struct page *)lmem_map, zones_size, 
+				start, zholes_size);
+		}
 	}
 	return;
 }

arch/i386/pci/numa.c

@@ -17,7 +17,7 @@ static int __pci_conf1_mq_read (int seg, int bus, int dev, int fn, int reg, int
 {
 	unsigned long flags;
 
-	if (!value || (bus > 255) || (dev > 31) || (fn > 7) || (reg > 255))
+	if (!value || (bus > MAX_MP_BUSSES) || (dev > 31) || (fn > 7) || (reg > 255))
 		return -EINVAL;
 
 	spin_lock_irqsave(&pci_config_lock, flags);
@@ -45,7 +45,7 @@ static int __pci_conf1_mq_write (int seg, int bus, int dev, int fn, int reg, int
 {
 	unsigned long flags;
 
-	if ((bus > 255) || (dev > 31) || (fn > 7) || (reg > 255)) 
+	if ((bus > MAX_MP_BUSSES) || (dev > 31) || (fn > 7) || (reg > 255)) 
 		return -EINVAL;
 
 	spin_lock_irqsave(&pci_config_lock, flags);

include/asm-generic/topology.h

@@ -47,6 +47,9 @@
 #ifndef node_to_memblk
 #define node_to_memblk(node)	(0)
 #endif
+#ifndef pcibus_to_cpumask
+#define pcibus_to_cpumask(bus)	(cpu_online_map)
+#endif
 
 /* Cross-node load balancing interval. */
 #ifndef NODE_BALANCE_RATE

include/asm-i386/mach-bigsmp/mach_apic.h

@@ -87,7 +87,8 @@ static inline int cpu_to_logical_apicid(int cpu)
        return (int)cpu_2_logical_apicid[cpu];
  }
 
-static inline int mpc_apic_id(struct mpc_config_processor *m, int quad)
+static inline int mpc_apic_id(struct mpc_config_processor *m,
+			struct mpc_config_translation *translation_record)
 {
 	printk("Processor #%d %ld:%ld APIC version %d\n",
 	        m->mpc_apicid,

include/asm-i386/mach-default/mach_apic.h

@@ -79,7 +79,8 @@ static inline unsigned long apicid_to_cpu_present(int phys_apicid)
 	return (1ul << phys_apicid);
 }
 
-static inline int mpc_apic_id(struct mpc_config_processor *m, int quad)
+static inline int mpc_apic_id(struct mpc_config_processor *m, 
+			struct mpc_config_translation *translation_record)
 {
 	printk("Processor #%d %ld:%ld APIC version %d\n",
 			m->mpc_apicid,

include/asm-i386/mach-numaq/mach_apic.h

@@ -73,8 +73,10 @@ static inline unsigned long apicid_to_cpu_present(int logical_apicid)
 	return ( (logical_apicid&0xf) << (4*apicid_to_node(logical_apicid)) );
 }
 
-static inline int mpc_apic_id(struct mpc_config_processor *m, int quad)
+static inline int mpc_apic_id(struct mpc_config_processor *m, 
+			struct mpc_config_translation *translation_record)
 {
+	int quad = translation_record->trans_quad;
 	int logical_apicid = generate_logical_apicid(quad, m->mpc_apicid);
 
 	printk("Processor #%d %ld:%ld APIC version %d (quad %d, apic %d)\n",

include/asm-i386/mach-summit/mach_apic.h

@@ -90,7 +90,8 @@ static inline unsigned long apicid_to_cpu_present(int apicid)
 		return (1ul << apicid);
 }
 
-static inline int mpc_apic_id(struct mpc_config_processor *m, int quad)
+static inline int mpc_apic_id(struct mpc_config_processor *m, 
+			struct mpc_config_translation *translation_record)
 {
 	printk("Processor #%d %ld:%ld APIC version %d\n",
 			m->mpc_apicid,

include/asm-i386/numaq.h

@@ -36,10 +36,11 @@
 #define MAX_ELEMENTS 256
 #define PAGES_PER_ELEMENT (16777216/256)
 
+extern int physnode_map[];
+#define pfn_to_nid(pfn)	({ physnode_map[(pfn) / PAGES_PER_ELEMENT]; })
 #define pfn_to_pgdat(pfn) NODE_DATA(pfn_to_nid(pfn))
 #define PHYSADDR_TO_NID(pa) pfn_to_nid(pa >> PAGE_SHIFT)
 #define MAX_NUMNODES		8
-extern int pfn_to_nid(unsigned long);
 extern void get_memcfg_numaq(void);
 #define get_memcfg_numa() get_memcfg_numaq()
 

include/asm-i386/topology.h

@@ -29,6 +29,8 @@
 
 #ifdef CONFIG_NUMA
 
+#include <asm/mpspec.h>
+
 /* Mappings between logical cpu number and node number */
 extern volatile unsigned long node_2_cpu_mask[];
 extern volatile int cpu_2_node[];
@@ -61,6 +63,12 @@ static inline int node_to_first_cpu(int node)
 /* Returns the number of the first MemBlk on Node 'node' */
 #define node_to_memblk(node) (node)
 
+/* Returns the number of the node containing PCI bus 'bus' */
+static inline unsigned long pcibus_to_cpumask(int bus)
+{
+	return node_to_cpumask(mp_bus_id_to_node[bus]);
+}
+
 /* Cross-node load balancing interval. */
 #define NODE_BALANCE_RATE 100