no arch-specific mem_map init

PATCH New Dave Hansen 1 participant 1 message Aug 24, 2004

Subsystems (5)

Memory management Arm (32-bit) SuperH Page allocator +1 more subsystems

lore

Share & tools

lore

Dave Hansen

author

2004-08-24 18:03 UTC #0 lore raw

PATCH 1/1 no arch-specific mem_map init

So, this patch started out with me trying to keep from passing contiguous, node-specific mem_map into free_area_init_node() and cousins. Instead, I relied on some calls to pfn_to_page().

This works fine and dandy when all you need is the pgdat->node_mem_map to do pfn_to_page(). However, the non-NUMA/DISCONTIG architectures use the real, global mem_map[] instead of a node_mem_map in the pfn_to_page() calculation. So, I ended up effectively trying to initialize mem_map from itself, when it was NULL. That was bad, and caused some very pretty colors on someone's screen when he tested it.

So, I had to make sure to initialize the global mem_map[] before calling into free_area_init_node(). Then, I realized how many architectures do this on their own, and have comments like this:

        /* XXX: MRB-remove - this doesn't seem sane, should this be done somewhere else ?*/
        mem_map = NODE_DATA(0)->node_mem_map;

Also, ppc64 has some interesting hacks^Wcode in this area to make up for any empty-memory NUMA nodes, which I think can go away now.

The patch magically removes more code than it adds. It could be smaller, but I shamelessly added some comments.

Applies on top of 2.6.8.1-mm4. It touches enough architectures that it probably needs plenty of cooling-off time in -mm.

-- Dave

So, this patch started out with me trying to keep from passing contiguous, node-specific mem_map into free_area_init_node() and cousins. Instead, I relied on some calls to pfn_to_page().

So, I had to make sure to initialize the global mem_map[] before calling into free_area_init_node(). Then, I realized how many architectures do this on their own, and have comments like this:

        /* XXX: MRB-remove - this doesn't seem sane, should this be done somewhere else ?*/
        mem_map = NODE_DATA(0)->node_mem_map;

Also, ppc64 has some interesting hacks^Wcode in this area to make up for any empty-memory NUMA nodes, which I think can go away now.

The following patch does what my first one did (don't pass mem_map into the init functions), incorporates Jesse Barnes' ia64 fixes on top of that, and gets rid of all but one of the global mem_map initializations (parisc is weird). It also magically removes more code than it adds. It could be smaller, but I shamelessly added some comments.

Signed-off-by Dave Hansen <[email protected]>

---

arch/arm/mm/init.c +0 -4

diff -urp clean/arch/arm/mm/init.c btest/arch/arm/mm/init.c
@@ -499,10 +499,6 @@ void __init paging_init(struct meminfo *
 				bdata->node_boot_start >> PAGE_SHIFT, zhole_size);
 	}
 
-#ifndef CONFIG_DISCONTIGMEM
-	mem_map = contig_page_data.node_mem_map;
-#endif
-
 	/*
 	 * finish off the bad pages once
 	 * the mem_map is initialised

arch/arm26/mm/init.c +1 -3

diff -urp clean/arch/arm26/mm/init.c btest/arch/arm26/mm/init.c
@@ -306,11 +306,9 @@ void __init paging_init(struct meminfo *
 	if (!zone_size[0])
 		BUG();
 
-	free_area_init_node(0, pgdat, 0, zone_size,
+	free_area_init_node(0, pgdat, zone_size,
 			bdata->node_boot_start >> PAGE_SHIFT, zhole_size);
 
-	mem_map = NODE_DATA(0)->node_mem_map;
-
 	/*
 	 * finish off the bad pages once
 	 * the mem_map is initialised

arch/cris/arch-v10/mm/init.c +1 -2

diff -urp clean/arch/cris/arch-v10/mm/init.c btest/arch/cris/arch-v10/mm/init.c
@@ -183,8 +183,7 @@ paging_init(void)
 	 * mem_map page array.
 	 */
 
-	free_area_init_node(0, &contig_page_data, 0, zones_size, PAGE_OFFSET >> PAGE_SHIFT, 0);
-	mem_map = contig_page_data.node_mem_map;
+	free_area_init_node(0, &contig_page_data, zones_size, PAGE_OFFSET >> PAGE_SHIFT, 0);
 }
 
 /* Initialize remaps of some I/O-ports. It is important that this

arch/i386/mm/discontig.c +3 -6

diff -urp clean/arch/i386/mm/discontig.c btest/arch/i386/mm/discontig.c
@@ -417,18 +417,15 @@ 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),
-					zones_size, start, zholes_size);
-		else {
+		if (nid) {
 			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;
 			NODE_DATA(nid)->node_mem_map = (struct page *)lmem_map;
-			free_area_init_node(nid, NODE_DATA(nid), zones_size,
-				start, zholes_size);
 		}
+		free_area_init_node(nid, NODE_DATA(nid), zones_size,
+				    start, zholes_size);
 	}
 	return;
 }

arch/ia64/mm/contig.c +1 -3

diff -urp clean/arch/ia64/mm/contig.c btest/arch/ia64/mm/contig.c
@@ -269,7 +269,6 @@ paging_init (void)
 		vmem_map = (struct page *) 0;
 		free_area_init_node(0, &contig_page_data, zones_size, 0,
 				    zholes_size);
-		mem_map = contig_page_data.node_mem_map;
 	} else {
 		unsigned long map_size;

@@ -280,11 +279,10 @@ paging_init (void)
 		vmem_map = (struct page *) vmalloc_end;
 		efi_memmap_walk(create_mem_map_page_table, 0);
 
-		contig_page_data.node_mem_map = vmem_map;
+		NODE_DATA(0)->node_mem_map = vmem_map;
 		free_area_init_node(0, &contig_page_data, zones_size,
 				    0, zholes_size);
 
-		mem_map = contig_page_data.node_mem_map;
 		printk("Virtual mem_map starts at 0x%p\n", mem_map);
 	}
 #else /* !CONFIG_VIRTUAL_MEM_MAP */

arch/ia64/mm/discontig.c +1 -0

diff -urp clean/arch/ia64/mm/discontig.c btest/arch/ia64/mm/discontig.c
@@ -665,6 +665,7 @@ void paging_init(void)
 
 		pfn_offset = mem_data[node].min_pfn;
 
+		NODE_DATA(node)->node_mem_map = vmem_map + pfn_offset;
 		free_area_init_node(node, NODE_DATA(node), zones_size,
 				    pfn_offset, zholes_size);
 	}

arch/ia64/mm/init.c +1 -1

diff -urp clean/arch/ia64/mm/init.c btest/arch/ia64/mm/init.c
@@ -429,7 +429,7 @@ virtual_memmap_init (u64 start, u64 end,
 		    / sizeof(struct page));
 
 	if (map_start < map_end)
-		memmap_init_zone((unsigned long)(map_end - map_start),
+		memmap_init_zone(map_start, (unsigned long) (map_end - map_start),
 				 args->nid, args->zone, page_to_pfn(map_start));
 	return 0;
 }

arch/ppc64/mm/init.c +0 -1

diff -urp clean/arch/ppc64/mm/init.c btest/arch/ppc64/mm/init.c
@@ -615,7 +615,6 @@ void __init paging_init(void)
 
 	free_area_init_node(0, &contig_page_data, zones_size,
 			    __pa(PAGE_OFFSET) >> PAGE_SHIFT, zholes_size);
-	mem_map = contig_page_data.node_mem_map;
 }
 #endif /* CONFIG_DISCONTIGMEM */

arch/sh/mm/init.c +1 -3

diff -urp clean/arch/sh/mm/init.c btest/arch/sh/mm/init.c
@@ -215,9 +215,7 @@ void __init paging_init(void)
 	disable_mmu();
 #endif
 
-	free_area_init_node(0, NODE_DATA(0), 0, zones_size, __MEMORY_START >> PAGE_SHIFT, 0);
-	/* XXX: MRB-remove - this doesn't seem sane, should this be done somewhere else ?*/
-	mem_map = NODE_DATA(0)->node_mem_map;
+	free_area_init_node(0, NODE_DATA(0), zones_size, __MEMORY_START >> PAGE_SHIFT, 0);
 
 #ifdef CONFIG_DISCONTIGMEM
 	/*

arch/sh64/mm/init.c +1 -4

diff -urp clean/arch/sh64/mm/init.c btest/arch/sh64/mm/init.c
@@ -123,10 +123,7 @@ void __init paging_init(void)
          */
 	zones_size[ZONE_DMA] = MAX_LOW_PFN - START_PFN;
 
-	free_area_init_node(0, NODE_DATA(0), 0, zones_size, __MEMORY_START >> PAGE_SHIFT, 0);
-
-	/* XXX: MRB-remove - this doesn't seem sane, should this be done somewhere else ?*/
-	mem_map = NODE_DATA(0)->node_mem_map;
+	free_area_init_node(0, NODE_DATA(0), zones_size, __MEMORY_START >> PAGE_SHIFT, 0);
 }
 
 void __init mem_init(void)

arch/sparc/mm/srmmu.c +0 -1

diff -urp clean/arch/sparc/mm/srmmu.c btest/arch/sparc/mm/srmmu.c
@@ -1343,7 +1343,6 @@ void __init srmmu_paging_init(void)
 
 		free_area_init_node(0, &contig_page_data, zones_size,
 				    pfn_base, zholes_size);
-		mem_map = contig_page_data.node_mem_map;
 	}
 }

arch/sparc/mm/sun4c.c +0 -1

diff -urp clean/arch/sparc/mm/sun4c.c btest/arch/sparc/mm/sun4c.c
@@ -2116,7 +2116,6 @@ void __init sun4c_paging_init(void)
 
 		free_area_init_node(0, &contig_page_data, zones_size,
 				    pfn_base, zholes_size);
-		mem_map = contig_page_data.node_mem_map;
 	}
 
 	cnt = 0;

arch/sparc64/mm/init.c +0 -1

diff -urp clean/arch/sparc64/mm/init.c btest/arch/sparc64/mm/init.c
@@ -1504,7 +1504,6 @@ void __init paging_init(void)
 
 		free_area_init_node(0, &contig_page_data, zones_size,
 				    phys_base >> PAGE_SHIFT, zholes_size);
-		mem_map = contig_page_data.node_mem_map;
 	}
 
 	device_scan();

arch/v850/kernel/setup.c +1 -2

diff -urp clean/arch/v850/kernel/setup.c btest/arch/v850/kernel/setup.c
@@ -281,7 +281,6 @@ init_mem_alloc (unsigned long ram_start,
 #error MAX_ORDER is too large for given PAGE_OFFSET (use CONFIG_FORCE_MAX_ZONEORDER to change it)
 #endif
 
-	free_area_init_node (0, NODE_DATA(0), 0, zones_size,
+	free_area_init_node (0, NODE_DATA(0), zones_size,
 			     ADDR_TO_PAGE (PAGE_OFFSET), 0);
-	mem_map = NODE_DATA(0)->node_mem_map;
 }

mm/page_alloc.c +14 -4

diff -urp clean/mm/page_alloc.c btest/mm/page_alloc.c
@@ -1690,14 +1690,25 @@ static void __init free_area_init_core(s
 	}
 }
 
-void __init node_alloc_mem_map(struct pglist_data *pgdat)
+static void __init alloc_node_mem_map(struct pglist_data *pgdat)
 {
 	unsigned long size;
 
+	/*
+	 * Make sure that the architecture hasn't already allocated
+	 * a node_mem_map, and that the node contains memory.
+	 */
+	if (pgdat->node_mem_map || !pgdat->node_spanned_pages)
+		return;
+
 	size = (pgdat->node_spanned_pages + 1) * sizeof(struct page);
 	pgdat->node_mem_map = alloc_bootmem_node(pgdat, size);
 #ifndef CONFIG_DISCONTIGMEM
-	mem_map = contig_page_data.node_mem_map;
+	/*
+	 * With no DISCONTIG, the global mem_map is just set as node 0's
+	 */
+	if (pgdat == NODE_DATA(0))
+		mem_map = NODE_DATA(0)->node_mem_map;
 #endif
 }

@@ -1709,8 +1720,7 @@ void __init free_area_init_node(int nid,
 	pgdat->node_start_pfn = node_start_pfn;
 	calculate_zone_totalpages(pgdat, zones_size, zholes_size);
 
-	if (!pfn_to_page(node_start_pfn))
-		node_alloc_mem_map(pgdat);
+	alloc_node_mem_map(pgdat);
 
 	free_area_init_core(pgdat, zones_size, zholes_size);
 }