ppc64: DISCONTIGMEM updates, rework to be like x86 version

arch/ppc64/mm/init.c

@@ -57,6 +57,7 @@
 #include <asm/naca.h>
 #include <asm/eeh.h>
 #include <asm/processor.h>
+#include <asm/mmzone.h>
 
 #include <asm/ppcdebug.h>
 
@@ -435,6 +436,7 @@ void __init mm_init_ppc64(void)
  * Initialize the bootmem system and give it all the memory we
  * have available.
  */
+#ifndef CONFIG_DISCONTIGMEM
 void __init do_init_bootmem(void)
 {
 	unsigned long i;
@@ -494,6 +496,7 @@ void __init paging_init(void)
 		zones_size[i] = 0;
 	free_area_init(zones_size);
 }
+#endif
 
 extern unsigned long prof_shift;
 extern unsigned long prof_len;
@@ -506,19 +509,39 @@ void initialize_paca_hardware_interrupt_stack(void);
 
 void __init mem_init(void)
 {
+#ifndef CONFIG_DISCONTIGMEM
 	extern char *sysmap; 
 	extern unsigned long sysmap_size;
 	unsigned long addr;
+#endif
 	int codepages = 0;
 	int datapages = 0;
 	int initpages = 0;
-	unsigned long va_rtas_base = (unsigned long)__va(rtas.base);
 
 	max_mapnr = max_low_pfn;
 	high_memory = (void *) __va(max_low_pfn * PAGE_SIZE);
 	num_physpages = max_mapnr;	/* RAM is assumed contiguous */
 	max_pfn = max_low_pfn;
 
+#ifdef CONFIG_DISCONTIGMEM
+{
+	int nid;
+
+        for (nid = 0; nid < MAX_NUMNODES; nid++) {
+		if (numa_node_exists[nid]) {
+			printk("freeing bootmem node %x\n", nid);
+			totalram_pages +=
+				free_all_bootmem_node(NODE_DATA(nid));
+		}
+	}
+
+	printk("Memory: %luk available (%dk kernel code, %dk data, %dk init) [%08lx,%08lx]\n",
+	       (unsigned long)nr_free_pages()<< (PAGE_SHIFT-10),
+	       codepages<< (PAGE_SHIFT-10), datapages<< (PAGE_SHIFT-10),
+	       initpages<< (PAGE_SHIFT-10),
+	       PAGE_OFFSET, (unsigned long)__va(lmb_end_of_DRAM()));
+}
+#else
 	totalram_pages += free_all_bootmem();
 
 	if ( sysmap_size )
@@ -546,6 +569,7 @@ void __init mem_init(void)
 	       codepages<< (PAGE_SHIFT-10), datapages<< (PAGE_SHIFT-10),
 	       initpages<< (PAGE_SHIFT-10),
 	       PAGE_OFFSET, (unsigned long)__va(lmb_end_of_DRAM()));
+#endif
 	mem_init_done = 1;
 
 	/* set the last page of each hardware interrupt stack to be protected */

include/asm-ppc64/io.h

@@ -155,7 +155,7 @@ static inline void * phys_to_virt(unsigned long address)
 /*
  * Change "struct page" to physical address.
  */
-#define page_to_phys(page)	((page - mem_map) << PAGE_SHIFT)
+#define page_to_phys(page)	(page_to_pfn(page) << PAGE_SHIFT)
 
 #if 0
 #define BIO_VMERGE_BOUNDARY	4096

include/asm-ppc64/mmzone.h

@@ -9,31 +9,67 @@
 
 #include <linux/config.h>
 
-typedef struct plat_pglist_data {
-	pg_data_t	gendata;
-} plat_pg_data_t;
+#ifdef CONFIG_DISCONTIGMEM
+
+extern struct pglist_data *node_data[];
 
 /*
- * Following are macros that are specific to this numa platform.
+ * Following are specific to this numa platform.
  */
 
-extern plat_pg_data_t plat_node_data[];
+extern int numa_node_exists[];
+extern int numa_cpu_lookup_table[];
+extern int numa_memory_lookup_table[];
 
-#define MAX_NUMNODES 4
+#define MAX_NUMNODES 16
+#define MAX_MEMORY (1UL << 41)
+/* 256MB regions */
+#define MEMORY_INCREMENT_SHIFT 28
+#define MEMORY_INCREMENT (1UL << MEMORY_INCREMENT_SHIFT)
 
-/* XXX grab this from the device tree - Anton */
-#define MEMORY_ZONE_BITS	33
-#define CPU_SHIFT_BITS          1
+#undef DEBUG_NUMA
 
-#define PHYSADDR_TO_NID(pa)		((pa) >> MEMORY_ZONE_BITS)
-#define PLAT_NODE_DATA(n)		(&plat_node_data[(n)])
-#define PLAT_NODE_DATA_STARTNR(n)	\
-	(PLAT_NODE_DATA(n)->gendata.node_start_mapnr)
-#define PLAT_NODE_DATA_SIZE(n)		(PLAT_NODE_DATA(n)->gendata.node_size)
-#define PLAT_NODE_DATA_LOCALNR(p, n)	\
-	(((p) >> PAGE_SHIFT) - PLAT_NODE_DATA(n)->gendata.node_start_pfn)
+static inline int pa_to_nid(unsigned long pa)
+{
+        int nid;
 
-#ifdef CONFIG_DISCONTIGMEM
+        nid = numa_memory_lookup_table[pa >> MEMORY_INCREMENT_SHIFT];
+
+#ifdef DEBUG_NUMA
+        /* the physical address passed in is not in the map for the system */
+        if (nid == -1) {
+                printk("bad address: %lx\n", pa);
+                BUG();
+        }
+#endif
+
+        return nid;
+}
+
+#define pfn_to_nid(pfn)		pa_to_nid((pfn) << PAGE_SHIFT)
+
+#define node_startnr(nid)	(node_data[nid]->node_start_mapnr)
+#define node_size(nid)		(node_data[nid]->node_size)
+#define node_localnr(pfn, nid)	((pfn) - node_data[nid]->node_start_pfn)
+
+#ifdef CONFIG_NUMA
+
+static inline int __cpu_to_node(int cpu)
+{
+        int node;
+
+        node = numa_cpu_lookup_table[cpu];
+
+#ifdef DEBUG_NUMA
+        if (node == -1)
+                BUG();
+#endif
+
+        return node;
+}
+
+#define numa_node_id()	__cpu_to_node(smp_processor_id())
+#endif /* CONFIG_NUMA */
 
 /*
  * Following are macros that each numa implmentation must define.
@@ -42,55 +78,39 @@ extern plat_pg_data_t plat_node_data[];
 /*
  * Given a kernel address, find the home node of the underlying memory.
  */
-#define KVADDR_TO_NID(kaddr)	PHYSADDR_TO_NID(__pa(kaddr))
+#define kvaddr_to_nid(kaddr)	pa_to_nid(__pa(kaddr))
 
 /*
  * Return a pointer to the node data for node n.
  */
-#define NODE_DATA(n)	(&((PLAT_NODE_DATA(n))->gendata))
-
-/*
- * NODE_MEM_MAP gives the kaddr for the mem_map of the node.
- */
-#define NODE_MEM_MAP(nid)	(NODE_DATA(nid)->node_mem_map)
+#define NODE_DATA(nid)		(node_data[nid])
 
-/*
- * Given a kaddr, ADDR_TO_MAPBASE finds the owning node of the memory
- * and returns the mem_map of that node.
- */
-#define ADDR_TO_MAPBASE(kaddr) \
-			NODE_MEM_MAP(KVADDR_TO_NID((unsigned long)(kaddr)))
+#define node_mem_map(nid)	(NODE_DATA(nid)->node_mem_map)
+#define node_start_pfn(nid)	(NODE_DATA(nid)->node_start_pfn)
 
-/*
- * Given a kaddr, LOCAL_BASE_ADDR finds the owning node of the memory
- * and returns the kaddr corresponding to first physical page in the
- * node's mem_map.
- */
-#define LOCAL_BASE_ADDR(kaddr) \
-	((unsigned long)__va(NODE_DATA(KVADDR_TO_NID(kaddr))->node_start_pfn << PAGE_SHIFT))
-
-#define LOCAL_MAP_NR(kvaddr) \
-	(((unsigned long)(kvaddr)-LOCAL_BASE_ADDR(kvaddr)) >> PAGE_SHIFT)
+#define local_mapnr(kvaddr) \
+	( (__pa(kvaddr) >> PAGE_SHIFT) - node_start_pfn(kvaddr_to_nid(kvaddr)) 
 
 #if 0
 /* XXX fix - Anton */
-#define kern_addr_valid(kaddr)	test_bit(LOCAL_MAP_NR(kaddr), \
-					 NODE_DATA(KVADDR_TO_NID(kaddr))->valid_addr_bitmap)
+#define kern_addr_valid(kaddr)	test_bit(local_mapnr(kaddr), \
+		 NODE_DATA(kvaddr_to_nid(kaddr))->valid_addr_bitmap)
 #endif
 
+/* Written this way to avoid evaluating arguments twice */
 #define discontigmem_pfn_to_page(pfn) \
 ({ \
-	unsigned long kaddr = (unsigned long)__va(pfn << PAGE_SHIFT); \
-	(ADDR_TO_MAPBASE(kaddr) + LOCAL_MAP_NR(kaddr)); \
+	unsigned long __tmp = pfn; \
+	(node_mem_map(pfn_to_nid(__tmp)) + \
+	 node_localnr(__tmp, pfn_to_nid(__tmp))); \
 })
 
-#ifdef CONFIG_NUMA
-
-/* XXX grab this from the device tree - Anton */
-#define cputonode(cpu)	((cpu) >> CPU_SHIFT_BITS)
-
-#define numa_node_id()	cputonode(smp_processor_id())
+#define discontigmem_page_to_pfn(p)	\
+({ \
+	struct page *__tmp = p; \
+	(((__tmp) - page_zone(__tmp)->zone_mem_map) + \
+	 page_zone(__tmp)->zone_start_pfn); \
+})
 
-#endif /* CONFIG_NUMA */
 #endif /* CONFIG_DISCONTIGMEM */
 #endif /* _ASM_MMZONE_H_ */

include/asm-ppc64/page.h

@@ -205,9 +205,7 @@ static inline int get_order(unsigned long size)
 #define __a2v(x) ((void *) __va(absolute_to_phys(x)))
 
 #ifdef CONFIG_DISCONTIGMEM
-#define page_to_pfn(page) \
-		((page) - page_zone(page)->zone_mem_map + \
-		(page_zone(page)->zone_start_pfn))
+#define page_to_pfn(page)	discontigmem_page_to_pfn(page)
 #define pfn_to_page(pfn)	discontigmem_pfn_to_page(pfn)
 #else
 #define pfn_to_page(pfn)	(mem_map + (pfn))

include/asm-ppc64/pgalloc.h

 #ifndef _PPC64_PGALLOC_H
 #define _PPC64_PGALLOC_H
 
-#include <linux/threads.h>
+#include <linux/mm.h>
 #include <asm/processor.h>
 
 /*
@@ -78,8 +78,16 @@ pte_alloc_one_kernel(struct mm_struct *mm, unsigned long addr)
 	return pte;
 }
 
-#define pte_alloc_one(mm, address) \
-	virt_to_page(pte_alloc_one_kernel((mm), (address)))
+static inline struct page *
+pte_alloc_one(struct mm_struct *mm, unsigned long address)
+{
+	pte_t *pte = pte_alloc_one_kernel(mm, address);
+
+	if (pte)
+		return virt_to_page(pte);
+
+	return NULL;
+}
 
 static inline void
 pte_free_kernel(pte_t *pte)