[PATCH] ppc64: Convert NUMA to sparsemem (3)
Anton Blanchard [Fri, 11 Nov 2005 03:22:35 +0000 (14:22 +1100)]
Convert to sparsemem and remove all the discontigmem code in the
process. This has a few advantages:

- The old numa_memory_lookup_table can go away
- All the arch specific discontigmem magic can go away

We also remove the triple pass of memory properties and instead create a
list of per node extents that we iterate through. A final cleanup would
be to change our lmb code to store extents per node, then we can reuse
that information in the numa code.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

arch/powerpc/Kconfig
arch/powerpc/mm/numa.c
arch/ppc64/Kconfig
include/asm-powerpc/topology.h
include/asm-ppc64/mmzone.h
include/asm-ppc64/page.h

index ed31062..c523029 100644 (file)
@@ -581,17 +581,12 @@ config ARCH_FLATMEM_ENABLE
        def_bool y
        depends on PPC64 && !NUMA
 
-config ARCH_DISCONTIGMEM_ENABLE
-       def_bool y
-       depends on SMP && PPC_PSERIES
-
-config ARCH_DISCONTIGMEM_DEFAULT
+config ARCH_SPARSEMEM_ENABLE
        def_bool y
-       depends on ARCH_DISCONTIGMEM_ENABLE
 
-config ARCH_SPARSEMEM_ENABLE
+config ARCH_SPARSEMEM_DEFAULT
        def_bool y
-       depends on ARCH_DISCONTIGMEM_ENABLE
+       depends on SMP && PPC_PSERIES
 
 source "mm/Kconfig"
 
index c2d7fec..bd2cf13 100644 (file)
@@ -17,9 +17,8 @@
 #include <linux/nodemask.h>
 #include <linux/cpu.h>
 #include <linux/notifier.h>
+#include <asm/sparsemem.h>
 #include <asm/lmb.h>
-#include <asm/machdep.h>
-#include <asm/abs_addr.h>
 #include <asm/system.h>
 #include <asm/smp.h>
 
@@ -28,42 +27,113 @@ static int numa_enabled = 1;
 static int numa_debug;
 #define dbg(args...) if (numa_debug) { printk(KERN_INFO args); }
 
-#ifdef DEBUG_NUMA
-#define ARRAY_INITIALISER -1
-#else
-#define ARRAY_INITIALISER 0
-#endif
-
-int numa_cpu_lookup_table[NR_CPUS] = { [ 0 ... (NR_CPUS - 1)] =
-       ARRAY_INITIALISER};
-char *numa_memory_lookup_table;
+int numa_cpu_lookup_table[NR_CPUS];
 cpumask_t numa_cpumask_lookup_table[MAX_NUMNODES];
-
 struct pglist_data *node_data[MAX_NUMNODES];
-bootmem_data_t __initdata plat_node_bdata[MAX_NUMNODES];
+
+EXPORT_SYMBOL(numa_cpu_lookup_table);
+EXPORT_SYMBOL(numa_cpumask_lookup_table);
+EXPORT_SYMBOL(node_data);
+
+static bootmem_data_t __initdata plat_node_bdata[MAX_NUMNODES];
 static int min_common_depth;
 
 /*
- * We need somewhere to store start/span for each node until we have
+ * We need somewhere to store start/end/node for each region until we have
  * allocated the real node_data structures.
  */
+#define MAX_REGIONS    (MAX_LMB_REGIONS*2)
 static struct {
-       unsigned long node_start_pfn;
-       unsigned long node_end_pfn;
-       unsigned long node_present_pages;
-} init_node_data[MAX_NUMNODES] __initdata;
+       unsigned long start_pfn;
+       unsigned long end_pfn;
+       int nid;
+} init_node_data[MAX_REGIONS] __initdata;
 
-EXPORT_SYMBOL(node_data);
-EXPORT_SYMBOL(numa_cpu_lookup_table);
-EXPORT_SYMBOL(numa_memory_lookup_table);
-EXPORT_SYMBOL(numa_cpumask_lookup_table);
+int __init early_pfn_to_nid(unsigned long pfn)
+{
+       unsigned int i;
+
+       for (i = 0; init_node_data[i].end_pfn; i++) {
+               unsigned long start_pfn = init_node_data[i].start_pfn;
+               unsigned long end_pfn = init_node_data[i].end_pfn;
+
+               if ((start_pfn <= pfn) && (pfn < end_pfn))
+                       return init_node_data[i].nid;
+       }
+
+       return -1;
+}
+
+void __init add_region(unsigned int nid, unsigned long start_pfn,
+                      unsigned long pages)
+{
+       unsigned int i;
+
+       dbg("add_region nid %d start_pfn 0x%lx pages 0x%lx\n",
+               nid, start_pfn, pages);
+
+       for (i = 0; init_node_data[i].end_pfn; i++) {
+               if (init_node_data[i].nid != nid)
+                       continue;
+               if (init_node_data[i].end_pfn == start_pfn) {
+                       init_node_data[i].end_pfn += pages;
+                       return;
+               }
+               if (init_node_data[i].start_pfn == (start_pfn + pages)) {
+                       init_node_data[i].start_pfn -= pages;
+                       return;
+               }
+       }
+
+       /*
+        * Leave last entry NULL so we dont iterate off the end (we use
+        * entry.end_pfn to terminate the walk).
+        */
+       if (i >= (MAX_REGIONS - 1)) {
+               printk(KERN_ERR "WARNING: too many memory regions in "
+                               "numa code, truncating\n");
+               return;
+       }
+
+       init_node_data[i].start_pfn = start_pfn;
+       init_node_data[i].end_pfn = start_pfn + pages;
+       init_node_data[i].nid = nid;
+}
+
+/* We assume init_node_data has no overlapping regions */
+void __init get_region(unsigned int nid, unsigned long *start_pfn,
+                      unsigned long *end_pfn, unsigned long *pages_present)
+{
+       unsigned int i;
+
+       *start_pfn = -1UL;
+       *end_pfn = *pages_present = 0;
+
+       for (i = 0; init_node_data[i].end_pfn; i++) {
+               if (init_node_data[i].nid != nid)
+                       continue;
+
+               *pages_present += init_node_data[i].end_pfn -
+                       init_node_data[i].start_pfn;
+
+               if (init_node_data[i].start_pfn < *start_pfn)
+                       *start_pfn = init_node_data[i].start_pfn;
+
+               if (init_node_data[i].end_pfn > *end_pfn)
+                       *end_pfn = init_node_data[i].end_pfn;
+       }
+
+       /* We didnt find a matching region, return start/end as 0 */
+       if (*start_pfn == -1UL)
+               start_pfn = 0;
+}
 
 static inline void map_cpu_to_node(int cpu, int node)
 {
        numa_cpu_lookup_table[cpu] = node;
-       if (!(cpu_isset(cpu, numa_cpumask_lookup_table[node]))) {
+
+       if (!(cpu_isset(cpu, numa_cpumask_lookup_table[node])))
                cpu_set(cpu, numa_cpumask_lookup_table[node]);
-       }
 }
 
 #ifdef CONFIG_HOTPLUG_CPU
@@ -82,7 +152,7 @@ static void unmap_cpu_from_node(unsigned long cpu)
 }
 #endif /* CONFIG_HOTPLUG_CPU */
 
-static struct device_node * __devinit find_cpu_node(unsigned int cpu)
+static struct device_node *find_cpu_node(unsigned int cpu)
 {
        unsigned int hw_cpuid = get_hard_smp_processor_id(cpu);
        struct device_node *cpu_node = NULL;
@@ -209,7 +279,7 @@ static int __init get_mem_size_cells(void)
        return rc;
 }
 
-static unsigned long read_n_cells(int n, unsigned int **buf)
+static unsigned long __init read_n_cells(int n, unsigned int **buf)
 {
        unsigned long result = 0;
 
@@ -291,7 +361,8 @@ static int cpu_numa_callback(struct notifier_block *nfb,
  * or zero. If the returned value of size is 0 the region should be
  * discarded as it lies wholy above the memory limit.
  */
-static unsigned long __init numa_enforce_memory_limit(unsigned long start, unsigned long size)
+static unsigned long __init numa_enforce_memory_limit(unsigned long start,
+                                                     unsigned long size)
 {
        /*
         * We use lmb_end_of_DRAM() in here instead of memory_limit because
@@ -316,8 +387,7 @@ static int __init parse_numa_properties(void)
        struct device_node *cpu = NULL;
        struct device_node *memory = NULL;
        int addr_cells, size_cells;
-       int max_domain = 0;
-       long entries = lmb_end_of_DRAM() >> MEMORY_INCREMENT_SHIFT;
+       int max_domain;
        unsigned long i;
 
        if (numa_enabled == 0) {
@@ -325,13 +395,6 @@ static int __init parse_numa_properties(void)
                return -1;
        }
 
-       numa_memory_lookup_table =
-               (char *)abs_to_virt(lmb_alloc(entries * sizeof(char), 1));
-       memset(numa_memory_lookup_table, 0, entries * sizeof(char));
-
-       for (i = 0; i < entries ; i++)
-               numa_memory_lookup_table[i] = ARRAY_INITIALISER;
-
        min_common_depth = find_min_common_depth();
 
        dbg("NUMA associativity depth for CPU/Memory: %d\n", min_common_depth);
@@ -383,9 +446,6 @@ new_range:
                start = read_n_cells(addr_cells, &memcell_buf);
                size = read_n_cells(size_cells, &memcell_buf);
 
-               start = _ALIGN_DOWN(start, MEMORY_INCREMENT);
-               size = _ALIGN_UP(size, MEMORY_INCREMENT);
-
                numa_domain = of_node_numa_domain(memory);
 
                if (numa_domain >= MAX_NUMNODES) {
@@ -399,44 +459,15 @@ new_range:
                if (max_domain < numa_domain)
                        max_domain = numa_domain;
 
-               if (! (size = numa_enforce_memory_limit(start, size))) {
+               if (!(size = numa_enforce_memory_limit(start, size))) {
                        if (--ranges)
                                goto new_range;
                        else
                                continue;
                }
 
-               /*
-                * Initialize new node struct, or add to an existing one.
-                */
-               if (init_node_data[numa_domain].node_end_pfn) {
-                       if ((start / PAGE_SIZE) <
-                           init_node_data[numa_domain].node_start_pfn)
-                               init_node_data[numa_domain].node_start_pfn =
-                                       start / PAGE_SIZE;
-                       if (((start / PAGE_SIZE) + (size / PAGE_SIZE)) >
-                           init_node_data[numa_domain].node_end_pfn)
-                               init_node_data[numa_domain].node_end_pfn =
-                                       (start / PAGE_SIZE) +
-                                       (size / PAGE_SIZE);
-
-                       init_node_data[numa_domain].node_present_pages +=
-                               size / PAGE_SIZE;
-               } else {
-                       node_set_online(numa_domain);
-
-                       init_node_data[numa_domain].node_start_pfn =
-                               start / PAGE_SIZE;
-                       init_node_data[numa_domain].node_end_pfn =
-                               init_node_data[numa_domain].node_start_pfn +
-                               size / PAGE_SIZE;
-                       init_node_data[numa_domain].node_present_pages =
-                               size / PAGE_SIZE;
-               }
-
-               for (i = start ; i < (start+size); i += MEMORY_INCREMENT)
-                       numa_memory_lookup_table[i >> MEMORY_INCREMENT_SHIFT] =
-                               numa_domain;
+               add_region(numa_domain, start >> PAGE_SHIFT,
+                          size >> PAGE_SHIFT);
 
                if (--ranges)
                        goto new_range;
@@ -452,32 +483,15 @@ static void __init setup_nonnuma(void)
 {
        unsigned long top_of_ram = lmb_end_of_DRAM();
        unsigned long total_ram = lmb_phys_mem_size();
-       unsigned long i;
 
        printk(KERN_INFO "Top of RAM: 0x%lx, Total RAM: 0x%lx\n",
               top_of_ram, total_ram);
        printk(KERN_INFO "Memory hole size: %ldMB\n",
               (top_of_ram - total_ram) >> 20);
 
-       if (!numa_memory_lookup_table) {
-               long entries = top_of_ram >> MEMORY_INCREMENT_SHIFT;
-               numa_memory_lookup_table =
-                       (char *)abs_to_virt(lmb_alloc(entries * sizeof(char), 1));
-               memset(numa_memory_lookup_table, 0, entries * sizeof(char));
-               for (i = 0; i < entries ; i++)
-                       numa_memory_lookup_table[i] = ARRAY_INITIALISER;
-       }
-
        map_cpu_to_node(boot_cpuid, 0);
-
+       add_region(0, 0, lmb_end_of_DRAM() >> PAGE_SHIFT);
        node_set_online(0);
-
-       init_node_data[0].node_start_pfn = 0;
-       init_node_data[0].node_end_pfn = lmb_end_of_DRAM() / PAGE_SIZE;
-       init_node_data[0].node_present_pages = total_ram / PAGE_SIZE;
-
-       for (i = 0 ; i < top_of_ram; i += MEMORY_INCREMENT)
-               numa_memory_lookup_table[i >> MEMORY_INCREMENT_SHIFT] = 0;
 }
 
 static void __init dump_numa_topology(void)
@@ -495,8 +509,9 @@ static void __init dump_numa_topology(void)
 
                count = 0;
 
-               for (i = 0; i < lmb_end_of_DRAM(); i += MEMORY_INCREMENT) {
-                       if (numa_memory_lookup_table[i >> MEMORY_INCREMENT_SHIFT] == node) {
+               for (i = 0; i < lmb_end_of_DRAM();
+                    i += (1 << SECTION_SIZE_BITS)) {
+                       if (early_pfn_to_nid(i >> PAGE_SHIFT) == node) {
                                if (count == 0)
                                        printk(" 0x%lx", i);
                                ++count;
@@ -521,10 +536,12 @@ static void __init dump_numa_topology(void)
  *
  * Returns the physical address of the memory.
  */
-static unsigned long careful_allocation(int nid, unsigned long size,
-                                       unsigned long align, unsigned long end)
+static void __init *careful_allocation(int nid, unsigned long size,
+                                      unsigned long align,
+                                      unsigned long end_pfn)
 {
-       unsigned long ret = lmb_alloc_base(size, align, end);
+       int new_nid;
+       unsigned long ret = lmb_alloc_base(size, align, end_pfn << PAGE_SHIFT);
 
        /* retry over all memory */
        if (!ret)
@@ -538,28 +555,27 @@ static unsigned long careful_allocation(int nid, unsigned long size,
         * If the memory came from a previously allocated node, we must
         * retry with the bootmem allocator.
         */
-       if (pa_to_nid(ret) < nid) {
-               nid = pa_to_nid(ret);
-               ret = (unsigned long)__alloc_bootmem_node(NODE_DATA(nid),
+       new_nid = early_pfn_to_nid(ret >> PAGE_SHIFT);
+       if (new_nid < nid) {
+               ret = (unsigned long)__alloc_bootmem_node(NODE_DATA(new_nid),
                                size, align, 0);
 
                if (!ret)
                        panic("numa.c: cannot allocate %lu bytes on node %d",
-                             size, nid);
+                             size, new_nid);
 
-               ret = virt_to_abs(ret);
+               ret = __pa(ret);
 
                dbg("alloc_bootmem %lx %lx\n", ret, size);
        }
 
-       return ret;
+       return (void *)ret;
 }
 
 void __init do_init_bootmem(void)
 {
        int nid;
-       int addr_cells, size_cells;
-       struct device_node *memory = NULL;
+       unsigned int i;
        static struct notifier_block ppc64_numa_nb = {
                .notifier_call = cpu_numa_callback,
                .priority = 1 /* Must run before sched domains notifier. */
@@ -577,99 +593,66 @@ void __init do_init_bootmem(void)
        register_cpu_notifier(&ppc64_numa_nb);
 
        for_each_online_node(nid) {
-               unsigned long start_paddr, end_paddr;
-               int i;
+               unsigned long start_pfn, end_pfn, pages_present;
                unsigned long bootmem_paddr;
                unsigned long bootmap_pages;
 
-               start_paddr = init_node_data[nid].node_start_pfn * PAGE_SIZE;
-               end_paddr = init_node_data[nid].node_end_pfn * PAGE_SIZE;
+               get_region(nid, &start_pfn, &end_pfn, &pages_present);
 
                /* Allocate the node structure node local if possible */
-               NODE_DATA(nid) = (struct pglist_data *)careful_allocation(nid,
+               NODE_DATA(nid) = careful_allocation(nid,
                                        sizeof(struct pglist_data),
-                                       SMP_CACHE_BYTES, end_paddr);
-               NODE_DATA(nid) = abs_to_virt(NODE_DATA(nid));
+                                       SMP_CACHE_BYTES, end_pfn);
+               NODE_DATA(nid) = __va(NODE_DATA(nid));
                memset(NODE_DATA(nid), 0, sizeof(struct pglist_data));
 
                dbg("node %d\n", nid);
                dbg("NODE_DATA() = %p\n", NODE_DATA(nid));
 
                NODE_DATA(nid)->bdata = &plat_node_bdata[nid];
-               NODE_DATA(nid)->node_start_pfn =
-                       init_node_data[nid].node_start_pfn;
-               NODE_DATA(nid)->node_spanned_pages =
-                       end_paddr - start_paddr;
+               NODE_DATA(nid)->node_start_pfn = start_pfn;
+               NODE_DATA(nid)->node_spanned_pages = end_pfn - start_pfn;
 
                if (NODE_DATA(nid)->node_spanned_pages == 0)
                        continue;
 
-               dbg("start_paddr = %lx\n", start_paddr);
-               dbg("end_paddr = %lx\n", end_paddr);
+               dbg("start_paddr = %lx\n", start_pfn << PAGE_SHIFT);
+               dbg("end_paddr = %lx\n", end_pfn << PAGE_SHIFT);
 
-               bootmap_pages = bootmem_bootmap_pages((end_paddr - start_paddr) >> PAGE_SHIFT);
+               bootmap_pages = bootmem_bootmap_pages(end_pfn - start_pfn);
+               bootmem_paddr = (unsigned long)careful_allocation(nid,
+                                       bootmap_pages << PAGE_SHIFT,
+                                       PAGE_SIZE, end_pfn);
+               memset(__va(bootmem_paddr), 0, bootmap_pages << PAGE_SHIFT);
 
-               bootmem_paddr = careful_allocation(nid,
-                               bootmap_pages << PAGE_SHIFT,
-                               PAGE_SIZE, end_paddr);
-               memset(abs_to_virt(bootmem_paddr), 0,
-                      bootmap_pages << PAGE_SHIFT);
                dbg("bootmap_paddr = %lx\n", bootmem_paddr);
 
                init_bootmem_node(NODE_DATA(nid), bootmem_paddr >> PAGE_SHIFT,
-                                 start_paddr >> PAGE_SHIFT,
-                                 end_paddr >> PAGE_SHIFT);
+                                 start_pfn, end_pfn);
 
-               /*
-                * We need to do another scan of all memory sections to
-                * associate memory with the correct node.
-                */
-               addr_cells = get_mem_addr_cells();
-               size_cells = get_mem_size_cells();
-               memory = NULL;
-               while ((memory = of_find_node_by_type(memory, "memory")) != NULL) {
-                       unsigned long mem_start, mem_size;
-                       int numa_domain, ranges;
-                       unsigned int *memcell_buf;
-                       unsigned int len;
-
-                       memcell_buf = (unsigned int *)get_property(memory, "reg", &len);
-                       if (!memcell_buf || len <= 0)
-                               continue;
+               /* Add free regions on this node */
+               for (i = 0; init_node_data[i].end_pfn; i++) {
+                       unsigned long start, end;
 
-                       ranges = memory->n_addrs;       /* ranges in cell */
-new_range:
-                       mem_start = read_n_cells(addr_cells, &memcell_buf);
-                       mem_size = read_n_cells(size_cells, &memcell_buf);
-                       if (numa_enabled) {
-                               numa_domain = of_node_numa_domain(memory);
-                               if (numa_domain  >= MAX_NUMNODES)
-                                       numa_domain = 0;
-                       } else
-                               numa_domain =  0;
-
-                       if (numa_domain != nid)
+                       if (init_node_data[i].nid != nid)
                                continue;
 
-                       mem_size = numa_enforce_memory_limit(mem_start, mem_size);
-                       if (mem_size) {
-                               dbg("free_bootmem %lx %lx\n", mem_start, mem_size);
-                               free_bootmem_node(NODE_DATA(nid), mem_start, mem_size);
-                       }
+                       start = init_node_data[i].start_pfn << PAGE_SHIFT;
+                       end = init_node_data[i].end_pfn << PAGE_SHIFT;
 
-                       if (--ranges)           /* process all ranges in cell */
-                               goto new_range;
+                       dbg("free_bootmem %lx %lx\n", start, end - start);
+                       free_bootmem_node(NODE_DATA(nid), start, end - start);
                }
 
-               /*
-                * Mark reserved regions on this node
-                */
+               /* Mark reserved regions on this node */
                for (i = 0; i < lmb.reserved.cnt; i++) {
                        unsigned long physbase = lmb.reserved.region[i].base;
                        unsigned long size = lmb.reserved.region[i].size;
+                       unsigned long start_paddr = start_pfn << PAGE_SHIFT;
+                       unsigned long end_paddr = end_pfn << PAGE_SHIFT;
 
-                       if (pa_to_nid(physbase) != nid &&
-                           pa_to_nid(physbase+size-1) != nid)
+                       if (early_pfn_to_nid(physbase >> PAGE_SHIFT) != nid &&
+                           early_pfn_to_nid((physbase+size-1) >> PAGE_SHIFT) != nid)
                                continue;
 
                        if (physbase < end_paddr &&
@@ -689,46 +672,19 @@ new_range:
                                                     size);
                        }
                }
-               /*
-                * This loop may look famaliar, but we have to do it again
-                * after marking our reserved memory to mark memory present
-                * for sparsemem.
-                */
-               addr_cells = get_mem_addr_cells();
-               size_cells = get_mem_size_cells();
-               memory = NULL;
-               while ((memory = of_find_node_by_type(memory, "memory")) != NULL) {
-                       unsigned long mem_start, mem_size;
-                       int numa_domain, ranges;
-                       unsigned int *memcell_buf;
-                       unsigned int len;
-
-                       memcell_buf = (unsigned int *)get_property(memory, "reg", &len);
-                       if (!memcell_buf || len <= 0)
-                               continue;
 
-                       ranges = memory->n_addrs;       /* ranges in cell */
-new_range2:
-                       mem_start = read_n_cells(addr_cells, &memcell_buf);
-                       mem_size = read_n_cells(size_cells, &memcell_buf);
-                       if (numa_enabled) {
-                               numa_domain = of_node_numa_domain(memory);
-                               if (numa_domain  >= MAX_NUMNODES)
-                                       numa_domain = 0;
-                       } else
-                               numa_domain =  0;
-
-                       if (numa_domain != nid)
+               /* Add regions into sparsemem */
+               for (i = 0; init_node_data[i].end_pfn; i++) {
+                       unsigned long start, end;
+
+                       if (init_node_data[i].nid != nid)
                                continue;
 
-                       mem_size = numa_enforce_memory_limit(mem_start, mem_size);
-                       memory_present(numa_domain, mem_start >> PAGE_SHIFT,
-                                      (mem_start + mem_size) >> PAGE_SHIFT);
+                       start = init_node_data[i].start_pfn;
+                       end = init_node_data[i].end_pfn;
 
-                       if (--ranges)           /* process all ranges in cell */
-                               goto new_range2;
+                       memory_present(nid, start, end);
                }
-
        }
 }
 
@@ -742,21 +698,18 @@ void __init paging_init(void)
        memset(zholes_size, 0, sizeof(zholes_size));
 
        for_each_online_node(nid) {
-               unsigned long start_pfn;
-               unsigned long end_pfn;
+               unsigned long start_pfn, end_pfn, pages_present;
 
-               start_pfn = init_node_data[nid].node_start_pfn;
-               end_pfn = init_node_data[nid].node_end_pfn;
+               get_region(nid, &start_pfn, &end_pfn, &pages_present);
 
                zones_size[ZONE_DMA] = end_pfn - start_pfn;
-               zholes_size[ZONE_DMA] = zones_size[ZONE_DMA] -
-                       init_node_data[nid].node_present_pages;
+               zholes_size[ZONE_DMA] = zones_size[ZONE_DMA] - pages_present;
 
                dbg("free_area_init node %d %lx %lx (hole: %lx)\n", nid,
                    zones_size[ZONE_DMA], start_pfn, zholes_size[ZONE_DMA]);
 
-               free_area_init_node(nid, NODE_DATA(nid), zones_size,
-                                                       start_pfn, zholes_size);
+               free_area_init_node(nid, NODE_DATA(nid), zones_size, start_pfn,
+                                   zholes_size);
        }
 }
 
index c9d32db..9d10c12 100644 (file)
@@ -279,17 +279,12 @@ config ARCH_FLATMEM_ENABLE
        def_bool y
        depends on !NUMA
 
-config ARCH_DISCONTIGMEM_ENABLE
-       def_bool y
-       depends on SMP && PPC_PSERIES
-
-config ARCH_DISCONTIGMEM_DEFAULT
+config ARCH_SPARSEMEM_ENABLE
        def_bool y
-       depends on ARCH_DISCONTIGMEM_ENABLE
 
-config ARCH_SPARSEMEM_ENABLE
+config ARCH_SPARSEMEM_DEFAULT
        def_bool y
-       depends on ARCH_DISCONTIGMEM_ENABLE
+       depends on NUMA
 
 source "mm/Kconfig"
 
index 388cc9f..015d287 100644 (file)
@@ -9,15 +9,7 @@
 
 static inline int cpu_to_node(int cpu)
 {
-       int node;
-
-       node = numa_cpu_lookup_table[cpu];
-
-#ifdef DEBUG_NUMA
-       BUG_ON(node == -1);
-#endif
-
-       return node;
+       return numa_cpu_lookup_table[cpu];
 }
 
 #define parent_node(node)      (node)
index 7f06afb..54958d6 100644 (file)
@@ -8,15 +8,14 @@
 #define _ASM_MMZONE_H_
 
 #include <linux/config.h>
-#include <asm/smp.h>
 
-/* generic non-linear memory support:
+/*
+ * generic non-linear memory support:
  *
  * 1) we will not split memory into more chunks than will fit into the
  *    flags field of the struct page
  */
 
-
 #ifdef CONFIG_NEED_MULTIPLE_NODES
 
 extern struct pglist_data *node_data[];
@@ -30,41 +29,11 @@ extern struct pglist_data *node_data[];
  */
 
 extern int numa_cpu_lookup_table[];
-extern char *numa_memory_lookup_table;
 extern cpumask_t numa_cpumask_lookup_table[];
 #ifdef CONFIG_MEMORY_HOTPLUG
 extern unsigned long max_pfn;
 #endif
 
-/* 16MB regions */
-#define MEMORY_INCREMENT_SHIFT 24
-#define MEMORY_INCREMENT (1UL << MEMORY_INCREMENT_SHIFT)
-
-/* NUMA debugging, will not work on a DLPAR machine */
-#undef DEBUG_NUMA
-
-static inline int pa_to_nid(unsigned long pa)
-{
-       int nid;
-
-#ifdef CONFIG_MEMORY_HOTPLUG
-       /* kludge hot added sections default to node 0 */
-       if (pa >= (max_pfn << PAGE_SHIFT))
-               return 0;
-#endif
-       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;
-}
-
 /*
  * Following are macros that each numa implmentation must define.
  */
@@ -72,36 +41,10 @@ static inline int pa_to_nid(unsigned long pa)
 #define node_start_pfn(nid)    (NODE_DATA(nid)->node_start_pfn)
 #define node_end_pfn(nid)      (NODE_DATA(nid)->node_end_pfn)
 
-#ifdef CONFIG_DISCONTIGMEM
-
-#define node_localnr(pfn, nid) ((pfn) - NODE_DATA(nid)->node_start_pfn)
-
-#define pfn_to_nid(pfn)                pa_to_nid((unsigned long)(pfn) << PAGE_SHIFT)
-
-/* Written this way to avoid evaluating arguments twice */
-#define discontigmem_pfn_to_page(pfn) \
-({ \
-       unsigned long __tmp = pfn; \
-       (NODE_DATA(pfn_to_nid(__tmp))->node_mem_map + \
-        node_localnr(__tmp, pfn_to_nid(__tmp))); \
-})
-
-#define discontigmem_page_to_pfn(p) \
-({ \
-       struct page *__tmp = p; \
-       (((__tmp) - page_zone(__tmp)->zone_mem_map) + \
-        page_zone(__tmp)->zone_start_pfn); \
-})
-
-/* XXX fix for discontiguous physical memory */
-#define discontigmem_pfn_valid(pfn)            ((pfn) < num_physpages)
-
-#endif /* CONFIG_DISCONTIGMEM */
-
 #endif /* CONFIG_NEED_MULTIPLE_NODES */
 
 #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
-#define early_pfn_to_nid(pfn)  pa_to_nid(((unsigned long)pfn) << PAGE_SHIFT)
+extern int __init early_pfn_to_nid(unsigned long pfn);
 #endif
 
 #endif /* _ASM_MMZONE_H_ */
index e32f118..3efc328 100644 (file)
@@ -279,11 +279,6 @@ extern u64 ppc64_pft_size;         /* Log 2 of page table size */
 
 #define __va(x) ((void *)((unsigned long)(x) + KERNELBASE))
 
-#ifdef CONFIG_DISCONTIGMEM
-#define page_to_pfn(page)      discontigmem_page_to_pfn(page)
-#define pfn_to_page(pfn)       discontigmem_pfn_to_page(pfn)
-#define pfn_valid(pfn)         discontigmem_pfn_valid(pfn)
-#endif
 #ifdef CONFIG_FLATMEM
 #define pfn_to_page(pfn)       (mem_map + (pfn))
 #define page_to_pfn(page)      ((unsigned long)((page) - mem_map))