From 499a5f1efa0b0ac56ec5d060412aed84ae68e63e Mon Sep 17 00:00:00 2001 From: Jan Beulich Date: Fri, 18 Dec 2009 16:05:51 +0000 Subject: x86: Lift restriction on the location of FIX_BTMAP_* The early ioremap fixmap entries cover half (or for 32-bit non-PAE, a quarter) of a page table, yet they got uncondtitionally aligned so far to a 256-entry boundary. This is not necessary if the range of page table entries anyway falls into a single page table. This buys back, for (theoretically) 50% of all configurations (25% of all non-PAE ones), at least some of the lowmem necessarily lost with commit e621bd18958ef5dbace3129ebe17a0a475e127d9. Signed-off-by: Jan Beulich Cc: Linus Torvalds Cc: Andrew Morton LKML-Reference: <4B2BB66F0200007800026AD6@vpn.id2.novell.com> Signed-off-by: Ingo Molnar --- arch/x86/mm/ioremap.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'arch/x86/mm') diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c index c246d259822..03c75ffd5c2 100644 --- a/arch/x86/mm/ioremap.c +++ b/arch/x86/mm/ioremap.c @@ -422,6 +422,10 @@ void __init early_ioremap_init(void) * The boot-ioremap range spans multiple pmds, for which * we are not prepared: */ +#define __FIXADDR_TOP (-PAGE_SIZE) + BUILD_BUG_ON((__fix_to_virt(FIX_BTMAP_BEGIN) >> PMD_SHIFT) + != (__fix_to_virt(FIX_BTMAP_END) >> PMD_SHIFT)); +#undef __FIXADDR_TOP if (pmd != early_ioremap_pmd(fix_to_virt(FIX_BTMAP_END))) { WARN_ON(1); printk(KERN_WARNING "pmd %p != %p\n", -- cgit v1.2.3-70-g09d2 From 2854e72b58dad82f5248b30a45dda0df49e5fe05 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Wed, 27 Jan 2010 17:32:22 +0100 Subject: x86: Use helpers for rlimits Make sure compiler won't do weird things with limits. Fetching them twice may return 2 different values after writable limits are implemented. We can either use rlimit helpers added in 3e10e716abf3c71bdb5d86b8f507f9e72236c9cd or ACCESS_ONCE if not applicable; this patch uses the helpers. Signed-off-by: Jiri Slaby LKML-Reference: <1264609942-24621-1-git-send-email-jslaby@suse.cz> Cc: Ingo Molnar Cc: Thomas Gleixner Signed-off-by: H. Peter Anvin --- arch/x86/ia32/ia32_aout.c | 2 +- arch/x86/mm/mmap.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/ia32/ia32_aout.c b/arch/x86/ia32/ia32_aout.c index 2a4d073d2cf..06474788723 100644 --- a/arch/x86/ia32/ia32_aout.c +++ b/arch/x86/ia32/ia32_aout.c @@ -297,7 +297,7 @@ static int load_aout_binary(struct linux_binprm *bprm, struct pt_regs *regs) * size limits imposed on them by creating programs with large * arrays in the data or bss. */ - rlim = current->signal->rlim[RLIMIT_DATA].rlim_cur; + rlim = rlimit(RLIMIT_DATA); if (rlim >= RLIM_INFINITY) rlim = ~0; if (ex.a_data + ex.a_bss > rlim) diff --git a/arch/x86/mm/mmap.c b/arch/x86/mm/mmap.c index c8191defc38..1dab5194fd9 100644 --- a/arch/x86/mm/mmap.c +++ b/arch/x86/mm/mmap.c @@ -71,7 +71,7 @@ static int mmap_is_legacy(void) if (current->personality & ADDR_COMPAT_LAYOUT) return 1; - if (current->signal->rlim[RLIMIT_STACK].rlim_cur == RLIM_INFINITY) + if (rlimit(RLIMIT_STACK) == RLIM_INFINITY) return 1; return sysctl_legacy_va_layout; @@ -96,7 +96,7 @@ static unsigned long mmap_rnd(void) static unsigned long mmap_base(void) { - unsigned long gap = current->signal->rlim[RLIMIT_STACK].rlim_cur; + unsigned long gap = rlimit(RLIMIT_STACK); if (gap < MIN_GAP) gap = MIN_GAP; -- cgit v1.2.3-70-g09d2 From 1b5576e69a5fe168c08a159685ac366316ac9bbc Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Fri, 22 Jan 2010 11:21:04 +0800 Subject: x86: Remove BIOS data range from e820 In preparation for moving to the generic page_is_ram(), make explicit what we expect to be reserved and not reserved. Tested-by: Wu Fengguang Signed-off-by: Yinghai Lu LKML-Reference: <20100122033004.335813103@intel.com> Cc: Andrew Morton Signed-off-by: H. Peter Anvin --- arch/x86/kernel/e820.c | 8 ++++++++ arch/x86/kernel/setup.c | 19 ++++++++++++++++++- arch/x86/mm/ioremap.c | 16 ---------------- 3 files changed, 26 insertions(+), 17 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index d17d482a04f..230687ba5ba 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c @@ -517,11 +517,19 @@ u64 __init e820_remove_range(u64 start, u64 size, unsigned old_type, int checktype) { int i; + u64 end; u64 real_removed_size = 0; if (size > (ULLONG_MAX - start)) size = ULLONG_MAX - start; + end = start + size; + printk(KERN_DEBUG "e820 remove range: %016Lx - %016Lx ", + (unsigned long long) start, + (unsigned long long) end); + e820_print_type(old_type); + printk(KERN_CONT "\n"); + for (i = 0; i < e820.nr_map; i++) { struct e820entry *ei = &e820.map[i]; u64 final_start, final_end; diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index cdb6a8a506d..f9b1f4e5ab7 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -650,6 +650,23 @@ static struct dmi_system_id __initdata bad_bios_dmi_table[] = { {} }; +static void __init trim_bios_range(void) +{ + /* + * A special case is the first 4Kb of memory; + * This is a BIOS owned area, not kernel ram, but generally + * not listed as such in the E820 table. + */ + e820_update_range(0, PAGE_SIZE, E820_RAM, E820_RESERVED); + /* + * special case: Some BIOSen report the PC BIOS + * area (640->1Mb) as ram even though it is not. + * take them out. + */ + e820_remove_range(BIOS_BEGIN, BIOS_END - BIOS_BEGIN, E820_RAM, 1); + sanitize_e820_map(e820.map, ARRAY_SIZE(e820.map), &e820.nr_map); +} + /* * Determine if we were loaded by an EFI loader. If so, then we have also been * passed the efi memmap, systab, etc., so we should use these data structures @@ -813,7 +830,7 @@ void __init setup_arch(char **cmdline_p) insert_resource(&iomem_resource, &data_resource); insert_resource(&iomem_resource, &bss_resource); - + trim_bios_range(); #ifdef CONFIG_X86_32 if (ppro_with_ram_bug()) { e820_update_range(0x70000000ULL, 0x40000ULL, E820_RAM, diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c index 334e63ca7b2..30e068d6462 100644 --- a/arch/x86/mm/ioremap.c +++ b/arch/x86/mm/ioremap.c @@ -29,22 +29,6 @@ int page_is_ram(unsigned long pagenr) resource_size_t addr, end; int i; - /* - * A special case is the first 4Kb of memory; - * This is a BIOS owned area, not kernel ram, but generally - * not listed as such in the E820 table. - */ - if (pagenr == 0) - return 0; - - /* - * Second special case: Some BIOSen report the PC BIOS - * area (640->1Mb) as ram even though it is not. - */ - if (pagenr >= (BIOS_BEGIN >> PAGE_SHIFT) && - pagenr < (BIOS_END >> PAGE_SHIFT)) - return 0; - for (i = 0; i < e820.nr_map; i++) { /* * Not usable memory: -- cgit v1.2.3-70-g09d2 From 13ca0fcaa33f6b1984c4111b6ec5df42689fea6f Mon Sep 17 00:00:00 2001 From: Wu Fengguang Date: Fri, 22 Jan 2010 11:21:05 +0800 Subject: x86: Use the generic page_is_ram() The generic resource based page_is_ram() works better with memory hotplug/hotremove. So switch the x86 e820map based code to it. CC: Andi Kleen CC: KAMEZAWA Hiroyuki CC: Yinghai Lu Signed-off-by: Wu Fengguang LKML-Reference: <20100122033004.470767217@intel.com> Cc: Andrew Morton Signed-off-by: H. Peter Anvin --- arch/x86/include/asm/page_types.h | 1 - arch/x86/mm/ioremap.c | 21 --------------------- 2 files changed, 22 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/include/asm/page_types.h b/arch/x86/include/asm/page_types.h index 642fe34b36a..a667f24c725 100644 --- a/arch/x86/include/asm/page_types.h +++ b/arch/x86/include/asm/page_types.h @@ -40,7 +40,6 @@ #ifndef __ASSEMBLY__ -extern int page_is_ram(unsigned long pagenr); extern int devmem_is_allowed(unsigned long pagenr); extern unsigned long max_low_pfn_mapped; diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c index 30e068d6462..1bf9e08ed73 100644 --- a/arch/x86/mm/ioremap.c +++ b/arch/x86/mm/ioremap.c @@ -24,27 +24,6 @@ #include "physaddr.h" -int page_is_ram(unsigned long pagenr) -{ - resource_size_t addr, end; - int i; - - for (i = 0; i < e820.nr_map; i++) { - /* - * Not usable memory: - */ - if (e820.map[i].type != E820_RAM) - continue; - addr = (e820.map[i].addr + PAGE_SIZE-1) >> PAGE_SHIFT; - end = (e820.map[i].addr + e820.map[i].size) >> PAGE_SHIFT; - - - if ((pagenr >= addr) && (pagenr < end)) - return 1; - } - return 0; -} - /* * Fix up the linear direct mapping of the kernel to avoid cache attribute * conflicts. -- cgit v1.2.3-70-g09d2 From 1842f90cc98625d4d9bf8f8b927f17705ceb4e9c Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 10 Feb 2010 01:20:15 -0800 Subject: x86: Call early_res_to_bootmem one time Simplify setup_node_mem: don't use bootmem from other node, instead just find_e820_area in early_node_mem. This keeps the boundary between early_res and boot mem more clear, and lets us only call early_res_to_bootmem() one time instead of for all nodes. Signed-off-by: Yinghai Lu LKML-Reference: <1265793639-15071-12-git-send-email-yinghai@kernel.org> Signed-off-by: H. Peter Anvin --- arch/x86/kernel/setup.c | 1 + arch/x86/mm/init_32.c | 1 - arch/x86/mm/init_64.c | 3 +-- arch/x86/mm/numa_64.c | 62 ++++++++++++++++--------------------------------- 4 files changed, 22 insertions(+), 45 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index 3499b4fabc9..48cadbb1d28 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -967,6 +967,7 @@ void __init setup_arch(char **cmdline_p) #endif initmem_init(0, max_pfn, acpi, k8); + early_res_to_bootmem(0, max_low_pfn<> PAGE_SHIFT, 0, end_pfn); e820_register_active_regions(0, start_pfn, end_pfn); free_bootmem_with_active_regions(0, end_pfn); - early_res_to_bootmem(0, end_pfn<bdata = &bootmem_node_data[nodeid]; @@ -227,11 +234,7 @@ setup_node_bootmem(int nodeid, unsigned long start, unsigned long end) * of alloc_bootmem, that could clash with reserved range */ bootmap_pages = bootmem_bootmap_pages(last_pfn - start_pfn); - nid = phys_to_nid(nodedata_phys); - if (nid == nodeid) - bootmap_start = roundup(nodedata_phys + pgdat_size, PAGE_SIZE); - else - bootmap_start = roundup(start, PAGE_SIZE); + bootmap_start = roundup(nodedata_phys + pgdat_size, PAGE_SIZE); /* * SMP_CACHE_BYTES could be enough, but init_bootmem_node like * to use that to align to PAGE_SIZE @@ -239,18 +242,13 @@ setup_node_bootmem(int nodeid, unsigned long start, unsigned long end) bootmap = early_node_mem(nodeid, bootmap_start, end, bootmap_pages<= end) { - /* - * only need to free it if it is from other node - * bootmem - */ - if (nid != nodeid) - free_bootmem(nodedata_phys, pgdat_size); - } + free_early(nodedata_phys, nodedata_phys + pgdat_size); node_data[nodeid] = NULL; return; } bootmap_start = __pa(bootmap); + reserve_early(bootmap_start, bootmap_start+(bootmap_pages<> PAGE_SHIFT, @@ -259,31 +257,11 @@ setup_node_bootmem(int nodeid, unsigned long start, unsigned long end) printk(KERN_INFO " bootmap [%016lx - %016lx] pages %lx\n", bootmap_start, bootmap_start + bootmap_size - 1, bootmap_pages); - - free_bootmem_with_active_regions(nodeid, end); - - /* - * convert early reserve to bootmem reserve earlier - * otherwise early_node_mem could use early reserved mem - * on previous node - */ - early_res_to_bootmem(start, end); - - /* - * in some case early_node_mem could use alloc_bootmem - * to get range on other node, don't reserve that again - */ - if (nid != nodeid) - printk(KERN_INFO " NODE_DATA(%d) on node %d\n", nodeid, nid); - else - reserve_bootmem_node(NODE_DATA(nodeid), nodedata_phys, - pgdat_size, BOOTMEM_DEFAULT); nid = phys_to_nid(bootmap_start); if (nid != nodeid) printk(KERN_INFO " bootmap(%d) on node %d\n", nodeid, nid); - else - reserve_bootmem_node(NODE_DATA(nodeid), bootmap_start, - bootmap_pages< Date: Wed, 10 Feb 2010 01:20:18 -0800 Subject: x86: Make early_node_mem get mem > 4 GB if possible So we could put pgdata for the node high, and later sparse vmmap will get the section nr that need. With this patch will make <4 GB ram not use a sparse vmmap. before this patch, will get, before swiotlb try get bootmem [ 0.000000] nid=1 start=0 end=2080000 aligned=1 [ 0.000000] free [10 - 96] [ 0.000000] free [b12 - 1000] [ 0.000000] free [359f - 38a3] [ 0.000000] free [38b5 - 3a00] [ 0.000000] free [41e01 - 42000] [ 0.000000] free [73dde - 73e00] [ 0.000000] free [73fdd - 74000] [ 0.000000] free [741dd - 74200] [ 0.000000] free [743dd - 74400] [ 0.000000] free [745dd - 74600] [ 0.000000] free [747dd - 74800] [ 0.000000] free [749dd - 74a00] [ 0.000000] free [74bdd - 74c00] [ 0.000000] free [74ddd - 74e00] [ 0.000000] free [74fdd - 75000] [ 0.000000] free [751dd - 75200] [ 0.000000] free [753dd - 75400] [ 0.000000] free [755dd - 75600] [ 0.000000] free [757dd - 75800] [ 0.000000] free [759dd - 75a00] [ 0.000000] free [75bdd - 7bf5f] [ 0.000000] free [7f730 - 7f750] [ 0.000000] free [100000 - 2080000] [ 0.000000] total free 1f87170 [ 93.301474] Placing 64MB software IO TLB between ffff880075bdd000 - ffff880079bdd000 [ 93.311814] software IO TLB at phys 0x75bdd000 - 0x79bdd000 with this patch will get: before swiotlb try get bootmem [ 0.000000] nid=1 start=0 end=2080000 aligned=1 [ 0.000000] free [a - 96] [ 0.000000] free [702 - 1000] [ 0.000000] free [359f - 3600] [ 0.000000] free [37de - 3800] [ 0.000000] free [39dd - 3a00] [ 0.000000] free [3bdd - 3c00] [ 0.000000] free [3ddd - 3e00] [ 0.000000] free [3fdd - 4000] [ 0.000000] free [41dd - 4200] [ 0.000000] free [43dd - 4400] [ 0.000000] free [45dd - 4600] [ 0.000000] free [47dd - 4800] [ 0.000000] free [49dd - 4a00] [ 0.000000] free [4bdd - 4c00] [ 0.000000] free [4ddd - 4e00] [ 0.000000] free [4fdd - 5000] [ 0.000000] free [51dd - 5200] [ 0.000000] free [53dd - 5400] [ 0.000000] free [55dd - 7bf5f] [ 0.000000] free [7f730 - 7f750] [ 0.000000] free [100428 - 100600] [ 0.000000] free [13ea01 - 13ec00] [ 0.000000] free [170800 - 2080000] [ 0.000000] total free 1f87170 [ 92.689485] PCI-DMA: Using software bounce buffering for IO (SWIOTLB) [ 92.699799] Placing 64MB software IO TLB between ffff8800055dd000 - ffff8800095dd000 [ 92.710916] software IO TLB at phys 0x55dd000 - 0x95dd000 so will get enough space below 4G, aka pfn 0x100000 Signed-off-by: Yinghai Lu LKML-Reference: <1265793639-15071-15-git-send-email-yinghai@kernel.org> Signed-off-by: H. Peter Anvin --- arch/x86/mm/numa_64.c | 21 +++++++++++++++++---- 1 file changed, 17 insertions(+), 4 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c index 3232148756c..02f13cb99bc 100644 --- a/arch/x86/mm/numa_64.c +++ b/arch/x86/mm/numa_64.c @@ -163,14 +163,27 @@ static void * __init early_node_mem(int nodeid, unsigned long start, unsigned long end, unsigned long size, unsigned long align) { - unsigned long mem = find_e820_area(start, end, size, align); + unsigned long mem; + /* + * put it on high as possible + * something will go with NODE_DATA + */ + if (start < (MAX_DMA_PFN< (MAX_DMA32_PFN< (MAX_DMA32_PFN< Date: Wed, 10 Feb 2010 01:20:20 -0800 Subject: x86: Make 64 bit use early_res instead of bootmem before slab Finally we can use early_res to replace bootmem for x86_64 now. Still can use CONFIG_NO_BOOTMEM to enable it or not. -v2: fix 32bit compiling about MAX_DMA32_PFN -v3: folded bug fix from LKML message below Signed-off-by: Yinghai Lu LKML-Reference: <4B747239.4070907@kernel.org> Signed-off-by: H. Peter Anvin --- arch/x86/Kconfig | 13 +++ arch/x86/include/asm/e820.h | 6 ++ arch/x86/kernel/e820.c | 159 +++++++++++++++++++++++++++++++++--- arch/x86/kernel/setup.c | 2 + arch/x86/mm/init_64.c | 4 + arch/x86/mm/numa_64.c | 20 +++-- include/linux/bootmem.h | 7 ++ include/linux/mm.h | 5 ++ include/linux/mmzone.h | 2 + mm/bootmem.c | 195 +++++++++++++++++++++++++++++++++++++++++++- mm/page_alloc.c | 59 +++++++++++++- mm/percpu.c | 3 + mm/sparse-vmemmap.c | 2 +- 13 files changed, 454 insertions(+), 23 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index eb4092568f9..95439843ceb 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -568,6 +568,19 @@ config PARAVIRT_DEBUG Enable to debug paravirt_ops internals. Specifically, BUG if a paravirt_op is missing when it is called. +config NO_BOOTMEM + default y + bool "Disable Bootmem code" + depends on X86_64 + ---help--- + Use early_res directly instead of bootmem before slab is ready. + - allocator (buddy) [generic] + - early allocator (bootmem) [generic] + - very early allocator (reserve_early*()) [x86] + - very very early allocator (early brk model) [x86] + So reduce one layer between early allocator to final allocator + + config MEMTEST bool "Memtest" ---help--- diff --git a/arch/x86/include/asm/e820.h b/arch/x86/include/asm/e820.h index 761249e396f..7d72e5fb700 100644 --- a/arch/x86/include/asm/e820.h +++ b/arch/x86/include/asm/e820.h @@ -117,6 +117,12 @@ extern void free_early(u64 start, u64 end); extern void early_res_to_bootmem(u64 start, u64 end); extern u64 early_reserve_e820(u64 startt, u64 sizet, u64 align); +void reserve_early_without_check(u64 start, u64 end, char *name); +u64 find_early_area(u64 ei_start, u64 ei_last, u64 start, u64 end, + u64 size, u64 align); +#include +int get_free_all_memory_range(struct range **rangep, int nodeid); + extern unsigned long e820_end_of_ram_pfn(void); extern unsigned long e820_end_of_low_ram_pfn(void); extern int e820_find_active_region(const struct e820entry *ei, diff --git a/arch/x86/kernel/e820.c b/arch/x86/kernel/e820.c index e09c18c8f3c..90a85295f33 100644 --- a/arch/x86/kernel/e820.c +++ b/arch/x86/kernel/e820.c @@ -977,6 +977,25 @@ void __init reserve_early(u64 start, u64 end, char *name) __reserve_early(start, end, name, 0); } +void __init reserve_early_without_check(u64 start, u64 end, char *name) +{ + struct early_res *r; + + if (start >= end) + return; + + __check_and_double_early_res(end); + + r = &early_res[early_res_count]; + + r->start = start; + r->end = end; + r->overlap_ok = 0; + if (name) + strncpy(r->name, name, sizeof(r->name) - 1); + early_res_count++; +} + void __init free_early(u64 start, u64 end) { struct early_res *r; @@ -991,6 +1010,94 @@ void __init free_early(u64 start, u64 end) drop_range(i); } +#ifdef CONFIG_NO_BOOTMEM +static void __init subtract_early_res(struct range *range, int az) +{ + int i, count; + u64 final_start, final_end; + int idx = 0; + + count = 0; + for (i = 0; i < max_early_res && early_res[i].end; i++) + count++; + + /* need to skip first one ?*/ + if (early_res != early_res_x) + idx = 1; + +#if 1 + printk(KERN_INFO "Subtract (%d early reservations)\n", count); +#endif + for (i = idx; i < count; i++) { + struct early_res *r = &early_res[i]; +#if 0 + printk(KERN_INFO " #%d [%010llx - %010llx] %15s", i, + r->start, r->end, r->name); +#endif + final_start = PFN_DOWN(r->start); + final_end = PFN_UP(r->end); + if (final_start >= final_end) { +#if 0 + printk(KERN_CONT "\n"); +#endif + continue; + } +#if 0 + printk(KERN_CONT " subtract pfn [%010llx - %010llx]\n", + final_start, final_end); +#endif + subtract_range(range, az, final_start, final_end); + } + +} + +int __init get_free_all_memory_range(struct range **rangep, int nodeid) +{ + int i, count; + u64 start = 0, end; + u64 size; + u64 mem; + struct range *range; + int nr_range; + + count = 0; + for (i = 0; i < max_early_res && early_res[i].end; i++) + count++; + + count *= 2; + + size = sizeof(struct range) * count; +#ifdef MAX_DMA32_PFN + if (max_pfn_mapped > MAX_DMA32_PFN) + start = MAX_DMA32_PFN << PAGE_SHIFT; +#endif + end = max_pfn_mapped << PAGE_SHIFT; + mem = find_e820_area(start, end, size, sizeof(struct range)); + if (mem == -1ULL) + panic("can not find more space for range free"); + + range = __va(mem); + /* use early_node_map[] and early_res to get range array at first */ + memset(range, 0, size); + nr_range = 0; + + /* need to go over early_node_map to find out good range for node */ + nr_range = add_from_early_node_map(range, count, nr_range, nodeid); + subtract_early_res(range, count); + nr_range = clean_sort_range(range, count); + + /* need to clear it ? */ + if (nodeid == MAX_NUMNODES) { + memset(&early_res[0], 0, + sizeof(struct early_res) * max_early_res); + early_res = NULL; + max_early_res = 0; + } + + *rangep = range; + return nr_range; +} +#else void __init early_res_to_bootmem(u64 start, u64 end) { int i, count; @@ -1028,6 +1135,7 @@ void __init early_res_to_bootmem(u64 start, u64 end) max_early_res = 0; early_res_count = 0; } +#endif /* Check for already reserved areas */ static inline int __init bad_addr(u64 *addrp, u64 size, u64 align) @@ -1081,6 +1189,35 @@ again: return changed; } +/* + * Find a free area with specified alignment in a specific range. + * only with the area.between start to end is active range from early_node_map + * so they are good as RAM + */ +u64 __init find_early_area(u64 ei_start, u64 ei_last, u64 start, u64 end, + u64 size, u64 align) +{ + u64 addr, last; + + addr = round_up(ei_start, align); + if (addr < start) + addr = round_up(start, align); + if (addr >= ei_last) + goto out; + while (bad_addr(&addr, size, align) && addr+size <= ei_last) + ; + last = addr + size; + if (last > ei_last) + goto out; + if (last > end) + goto out; + + return addr; + +out: + return -1ULL; +} + /* * Find a free area with specified alignment in a specific range. */ @@ -1090,24 +1227,20 @@ u64 __init find_e820_area(u64 start, u64 end, u64 size, u64 align) for (i = 0; i < e820.nr_map; i++) { struct e820entry *ei = &e820.map[i]; - u64 addr, last; - u64 ei_last; + u64 addr; + u64 ei_start, ei_last; if (ei->type != E820_RAM) continue; - addr = round_up(ei->addr, align); + ei_last = ei->addr + ei->size; - if (addr < start) - addr = round_up(start, align); - if (addr >= ei_last) - continue; - while (bad_addr(&addr, size, align) && addr+size <= ei_last) - ; - last = addr + size; - if (last > ei_last) - continue; - if (last > end) + ei_start = ei->addr; + addr = find_early_area(ei_start, ei_last, start, end, + size, align); + + if (addr == -1ULL) continue; + return addr; } return -1ULL; diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c index ea4141b4851..d49e168bda8 100644 --- a/arch/x86/kernel/setup.c +++ b/arch/x86/kernel/setup.c @@ -967,7 +967,9 @@ void __init setup_arch(char **cmdline_p) #endif initmem_init(0, max_pfn, acpi, k8); +#ifndef CONFIG_NO_BOOTMEM early_res_to_bootmem(0, max_low_pfn<> PAGE_SHIFT; @@ -235,10 +237,13 @@ setup_node_bootmem(int nodeid, unsigned long start, unsigned long end) printk(KERN_INFO " NODE_DATA(%d) on node %d\n", nodeid, nid); memset(NODE_DATA(nodeid), 0, sizeof(pg_data_t)); - NODE_DATA(nodeid)->bdata = &bootmem_node_data[nodeid]; + NODE_DATA(nodeid)->node_id = nodeid; NODE_DATA(nodeid)->node_start_pfn = start_pfn; NODE_DATA(nodeid)->node_spanned_pages = last_pfn - start_pfn; +#ifndef CONFIG_NO_BOOTMEM + NODE_DATA(nodeid)->bdata = &bootmem_node_data[nodeid]; + /* * Find a place for the bootmem map * nodedata_phys could be on other nodes by alloc_bootmem, @@ -275,6 +280,7 @@ setup_node_bootmem(int nodeid, unsigned long start, unsigned long end) printk(KERN_INFO " bootmap(%d) on node %d\n", nodeid, nid); free_bootmem_with_active_regions(nodeid, end); +#endif node_set_online(nodeid); } @@ -733,6 +739,10 @@ unsigned long __init numa_free_all_bootmem(void) for_each_online_node(i) pages += free_all_bootmem_node(NODE_DATA(i)); +#ifdef CONFIG_NO_BOOTMEM + pages += free_all_memory_core_early(MAX_NUMNODES); +#endif + return pages; } diff --git a/include/linux/bootmem.h b/include/linux/bootmem.h index b10ec49ee2d..266ab929123 100644 --- a/include/linux/bootmem.h +++ b/include/linux/bootmem.h @@ -23,6 +23,7 @@ extern unsigned long max_pfn; extern unsigned long saved_max_pfn; #endif +#ifndef CONFIG_NO_BOOTMEM /* * node_bootmem_map is a map pointer - the bits represent all physical * memory pages (including holes) on the node. @@ -37,6 +38,7 @@ typedef struct bootmem_data { } bootmem_data_t; extern bootmem_data_t bootmem_node_data[]; +#endif extern unsigned long bootmem_bootmap_pages(unsigned long); @@ -46,6 +48,7 @@ extern unsigned long init_bootmem_node(pg_data_t *pgdat, unsigned long endpfn); extern unsigned long init_bootmem(unsigned long addr, unsigned long memend); +unsigned long free_all_memory_core_early(int nodeid); extern unsigned long free_all_bootmem_node(pg_data_t *pgdat); extern unsigned long free_all_bootmem(void); @@ -84,6 +87,10 @@ extern void *__alloc_bootmem_node(pg_data_t *pgdat, unsigned long size, unsigned long align, unsigned long goal); +void *__alloc_bootmem_node_high(pg_data_t *pgdat, + unsigned long size, + unsigned long align, + unsigned long goal); extern void *__alloc_bootmem_node_nopanic(pg_data_t *pgdat, unsigned long size, unsigned long align, diff --git a/include/linux/mm.h b/include/linux/mm.h index 8b2fa8593c6..f2c5b3cee8a 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -12,6 +12,7 @@ #include #include #include +#include struct mempolicy; struct anon_vma; @@ -1049,6 +1050,10 @@ extern void get_pfn_range_for_nid(unsigned int nid, extern unsigned long find_min_pfn_with_active_regions(void); extern void free_bootmem_with_active_regions(int nid, unsigned long max_low_pfn); +int add_from_early_node_map(struct range *range, int az, + int nr_range, int nid); +void *__alloc_memory_core_early(int nodeid, u64 size, u64 align, + u64 goal, u64 limit); typedef int (*work_fn_t)(unsigned long, unsigned long, void *); extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data); extern void sparse_memory_present_with_active_regions(int nid); diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h index 30fe668c254..eae8387b600 100644 --- a/include/linux/mmzone.h +++ b/include/linux/mmzone.h @@ -620,7 +620,9 @@ typedef struct pglist_data { struct page_cgroup *node_page_cgroup; #endif #endif +#ifndef CONFIG_NO_BOOTMEM struct bootmem_data *bdata; +#endif #ifdef CONFIG_MEMORY_HOTPLUG /* * Must be held any time you expect node_start_pfn, node_present_pages diff --git a/mm/bootmem.c b/mm/bootmem.c index 7d1486875e1..d7c791ef003 100644 --- a/mm/bootmem.c +++ b/mm/bootmem.c @@ -13,6 +13,7 @@ #include #include #include +#include #include #include @@ -32,6 +33,7 @@ unsigned long max_pfn; unsigned long saved_max_pfn; #endif +#ifndef CONFIG_NO_BOOTMEM bootmem_data_t bootmem_node_data[MAX_NUMNODES] __initdata; static struct list_head bdata_list __initdata = LIST_HEAD_INIT(bdata_list); @@ -142,7 +144,7 @@ unsigned long __init init_bootmem(unsigned long start, unsigned long pages) min_low_pfn = start; return init_bootmem_core(NODE_DATA(0)->bdata, start, 0, pages); } - +#endif /* * free_bootmem_late - free bootmem pages directly to page allocator * @addr: starting address of the range @@ -167,6 +169,60 @@ void __init free_bootmem_late(unsigned long addr, unsigned long size) } } +#ifdef CONFIG_NO_BOOTMEM +static void __init __free_pages_memory(unsigned long start, unsigned long end) +{ + int i; + unsigned long start_aligned, end_aligned; + int order = ilog2(BITS_PER_LONG); + + start_aligned = (start + (BITS_PER_LONG - 1)) & ~(BITS_PER_LONG - 1); + end_aligned = end & ~(BITS_PER_LONG - 1); + + if (end_aligned <= start_aligned) { +#if 1 + printk(KERN_DEBUG " %lx - %lx\n", start, end); +#endif + for (i = start; i < end; i++) + __free_pages_bootmem(pfn_to_page(i), 0); + + return; + } + +#if 1 + printk(KERN_DEBUG " %lx %lx - %lx %lx\n", + start, start_aligned, end_aligned, end); +#endif + for (i = start; i < start_aligned; i++) + __free_pages_bootmem(pfn_to_page(i), 0); + + for (i = start_aligned; i < end_aligned; i += BITS_PER_LONG) + __free_pages_bootmem(pfn_to_page(i), order); + + for (i = end_aligned; i < end; i++) + __free_pages_bootmem(pfn_to_page(i), 0); +} + +unsigned long __init free_all_memory_core_early(int nodeid) +{ + int i; + u64 start, end; + unsigned long count = 0; + struct range *range = NULL; + int nr_range; + + nr_range = get_free_all_memory_range(&range, nodeid); + + for (i = 0; i < nr_range; i++) { + start = range[i].start; + end = range[i].end; + count += end - start; + __free_pages_memory(start, end); + } + + return count; +} +#else static unsigned long __init free_all_bootmem_core(bootmem_data_t *bdata) { int aligned; @@ -227,6 +283,7 @@ static unsigned long __init free_all_bootmem_core(bootmem_data_t *bdata) return count; } +#endif /** * free_all_bootmem_node - release a node's free pages to the buddy allocator @@ -237,7 +294,12 @@ static unsigned long __init free_all_bootmem_core(bootmem_data_t *bdata) unsigned long __init free_all_bootmem_node(pg_data_t *pgdat) { register_page_bootmem_info_node(pgdat); +#ifdef CONFIG_NO_BOOTMEM + /* free_all_memory_core_early(MAX_NUMNODES) will be called later */ + return 0; +#else return free_all_bootmem_core(pgdat->bdata); +#endif } /** @@ -247,9 +309,14 @@ unsigned long __init free_all_bootmem_node(pg_data_t *pgdat) */ unsigned long __init free_all_bootmem(void) { +#ifdef CONFIG_NO_BOOTMEM + return free_all_memory_core_early(NODE_DATA(0)->node_id); +#else return free_all_bootmem_core(NODE_DATA(0)->bdata); +#endif } +#ifndef CONFIG_NO_BOOTMEM static void __init __free(bootmem_data_t *bdata, unsigned long sidx, unsigned long eidx) { @@ -344,6 +411,7 @@ static int __init mark_bootmem(unsigned long start, unsigned long end, } BUG(); } +#endif /** * free_bootmem_node - mark a page range as usable @@ -358,6 +426,12 @@ static int __init mark_bootmem(unsigned long start, unsigned long end, void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr, unsigned long size) { +#ifdef CONFIG_NO_BOOTMEM + free_early(physaddr, physaddr + size); +#if 0 + printk(KERN_DEBUG "free %lx %lx\n", physaddr, size); +#endif +#else unsigned long start, end; kmemleak_free_part(__va(physaddr), size); @@ -366,6 +440,7 @@ void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr, end = PFN_DOWN(physaddr + size); mark_bootmem_node(pgdat->bdata, start, end, 0, 0); +#endif } /** @@ -379,6 +454,12 @@ void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr, */ void __init free_bootmem(unsigned long addr, unsigned long size) { +#ifdef CONFIG_NO_BOOTMEM + free_early(addr, addr + size); +#if 0 + printk(KERN_DEBUG "free %lx %lx\n", addr, size); +#endif +#else unsigned long start, end; kmemleak_free_part(__va(addr), size); @@ -387,6 +468,7 @@ void __init free_bootmem(unsigned long addr, unsigned long size) end = PFN_DOWN(addr + size); mark_bootmem(start, end, 0, 0); +#endif } /** @@ -403,12 +485,17 @@ void __init free_bootmem(unsigned long addr, unsigned long size) int __init reserve_bootmem_node(pg_data_t *pgdat, unsigned long physaddr, unsigned long size, int flags) { +#ifdef CONFIG_NO_BOOTMEM + panic("no bootmem"); + return 0; +#else unsigned long start, end; start = PFN_DOWN(physaddr); end = PFN_UP(physaddr + size); return mark_bootmem_node(pgdat->bdata, start, end, 1, flags); +#endif } /** @@ -424,14 +511,20 @@ int __init reserve_bootmem_node(pg_data_t *pgdat, unsigned long physaddr, int __init reserve_bootmem(unsigned long addr, unsigned long size, int flags) { +#ifdef CONFIG_NO_BOOTMEM + panic("no bootmem"); + return 0; +#else unsigned long start, end; start = PFN_DOWN(addr); end = PFN_UP(addr + size); return mark_bootmem(start, end, 1, flags); +#endif } +#ifndef CONFIG_NO_BOOTMEM static unsigned long __init align_idx(struct bootmem_data *bdata, unsigned long idx, unsigned long step) { @@ -582,12 +675,33 @@ static void * __init alloc_arch_preferred_bootmem(bootmem_data_t *bdata, #endif return NULL; } +#endif static void * __init ___alloc_bootmem_nopanic(unsigned long size, unsigned long align, unsigned long goal, unsigned long limit) { +#ifdef CONFIG_NO_BOOTMEM + void *ptr; + + if (WARN_ON_ONCE(slab_is_available())) + return kzalloc(size, GFP_NOWAIT); + +restart: + + ptr = __alloc_memory_core_early(MAX_NUMNODES, size, align, goal, limit); + + if (ptr) + return ptr; + + if (goal != 0) { + goal = 0; + goto restart; + } + + return NULL; +#else bootmem_data_t *bdata; void *region; @@ -613,6 +727,7 @@ restart: } return NULL; +#endif } /** @@ -631,7 +746,13 @@ restart: void * __init __alloc_bootmem_nopanic(unsigned long size, unsigned long align, unsigned long goal) { - return ___alloc_bootmem_nopanic(size, align, goal, 0); + unsigned long limit = 0; + +#ifdef CONFIG_NO_BOOTMEM + limit = -1UL; +#endif + + return ___alloc_bootmem_nopanic(size, align, goal, limit); } static void * __init ___alloc_bootmem(unsigned long size, unsigned long align, @@ -665,9 +786,16 @@ static void * __init ___alloc_bootmem(unsigned long size, unsigned long align, void * __init __alloc_bootmem(unsigned long size, unsigned long align, unsigned long goal) { - return ___alloc_bootmem(size, align, goal, 0); + unsigned long limit = 0; + +#ifdef CONFIG_NO_BOOTMEM + limit = -1UL; +#endif + + return ___alloc_bootmem(size, align, goal, limit); } +#ifndef CONFIG_NO_BOOTMEM static void * __init ___alloc_bootmem_node(bootmem_data_t *bdata, unsigned long size, unsigned long align, unsigned long goal, unsigned long limit) @@ -684,6 +812,7 @@ static void * __init ___alloc_bootmem_node(bootmem_data_t *bdata, return ___alloc_bootmem(size, align, goal, limit); } +#endif /** * __alloc_bootmem_node - allocate boot memory from a specific node @@ -706,7 +835,46 @@ void * __init __alloc_bootmem_node(pg_data_t *pgdat, unsigned long size, if (WARN_ON_ONCE(slab_is_available())) return kzalloc_node(size, GFP_NOWAIT, pgdat->node_id); +#ifdef CONFIG_NO_BOOTMEM + return __alloc_memory_core_early(pgdat->node_id, size, align, + goal, -1ULL); +#else return ___alloc_bootmem_node(pgdat->bdata, size, align, goal, 0); +#endif +} + +void * __init __alloc_bootmem_node_high(pg_data_t *pgdat, unsigned long size, + unsigned long align, unsigned long goal) +{ +#ifdef MAX_DMA32_PFN + unsigned long end_pfn; + + if (WARN_ON_ONCE(slab_is_available())) + return kzalloc_node(size, GFP_NOWAIT, pgdat->node_id); + + /* update goal according ...MAX_DMA32_PFN */ + end_pfn = pgdat->node_start_pfn + pgdat->node_spanned_pages; + + if (end_pfn > MAX_DMA32_PFN + (128 >> (20 - PAGE_SHIFT)) && + (goal >> PAGE_SHIFT) < MAX_DMA32_PFN) { + void *ptr; + unsigned long new_goal; + + new_goal = MAX_DMA32_PFN << PAGE_SHIFT; +#ifdef CONFIG_NO_BOOTMEM + ptr = __alloc_memory_core_early(pgdat->node_id, size, align, + new_goal, -1ULL); +#else + ptr = alloc_bootmem_core(pgdat->bdata, size, align, + new_goal, 0); +#endif + if (ptr) + return ptr; + } +#endif + + return __alloc_bootmem_node(pgdat, size, align, goal); + } #ifdef CONFIG_SPARSEMEM @@ -720,6 +888,16 @@ void * __init __alloc_bootmem_node(pg_data_t *pgdat, unsigned long size, void * __init alloc_bootmem_section(unsigned long size, unsigned long section_nr) { +#ifdef CONFIG_NO_BOOTMEM + unsigned long pfn, goal, limit; + + pfn = section_nr_to_pfn(section_nr); + goal = pfn << PAGE_SHIFT; + limit = section_nr_to_pfn(section_nr + 1) << PAGE_SHIFT; + + return __alloc_memory_core_early(early_pfn_to_nid(pfn), size, + SMP_CACHE_BYTES, goal, limit); +#else bootmem_data_t *bdata; unsigned long pfn, goal, limit; @@ -729,6 +907,7 @@ void * __init alloc_bootmem_section(unsigned long size, bdata = &bootmem_node_data[early_pfn_to_nid(pfn)]; return alloc_bootmem_core(bdata, size, SMP_CACHE_BYTES, goal, limit); +#endif } #endif @@ -740,11 +919,16 @@ void * __init __alloc_bootmem_node_nopanic(pg_data_t *pgdat, unsigned long size, if (WARN_ON_ONCE(slab_is_available())) return kzalloc_node(size, GFP_NOWAIT, pgdat->node_id); +#ifdef CONFIG_NO_BOOTMEM + ptr = __alloc_memory_core_early(pgdat->node_id, size, align, + goal, -1ULL); +#else ptr = alloc_arch_preferred_bootmem(pgdat->bdata, size, align, goal, 0); if (ptr) return ptr; ptr = alloc_bootmem_core(pgdat->bdata, size, align, goal, 0); +#endif if (ptr) return ptr; @@ -795,6 +979,11 @@ void * __init __alloc_bootmem_low_node(pg_data_t *pgdat, unsigned long size, if (WARN_ON_ONCE(slab_is_available())) return kzalloc_node(size, GFP_NOWAIT, pgdat->node_id); +#ifdef CONFIG_NO_BOOTMEM + return __alloc_memory_core_early(pgdat->node_id, size, align, + goal, ARCH_LOW_ADDRESS_LIMIT); +#else return ___alloc_bootmem_node(pgdat->bdata, size, align, goal, ARCH_LOW_ADDRESS_LIMIT); +#endif } diff --git a/mm/page_alloc.c b/mm/page_alloc.c index 8deb9d0fd5b..78821a28e39 100644 --- a/mm/page_alloc.c +++ b/mm/page_alloc.c @@ -3435,6 +3435,59 @@ void __init free_bootmem_with_active_regions(int nid, } } +int __init add_from_early_node_map(struct range *range, int az, + int nr_range, int nid) +{ + int i; + u64 start, end; + + /* need to go over early_node_map to find out good range for node */ + for_each_active_range_index_in_nid(i, nid) { + start = early_node_map[i].start_pfn; + end = early_node_map[i].end_pfn; + nr_range = add_range(range, az, nr_range, start, end); + } + return nr_range; +} + +void * __init __alloc_memory_core_early(int nid, u64 size, u64 align, + u64 goal, u64 limit) +{ + int i; + void *ptr; + + /* need to go over early_node_map to find out good range for node */ + for_each_active_range_index_in_nid(i, nid) { + u64 addr; + u64 ei_start, ei_last; + + ei_last = early_node_map[i].end_pfn; + ei_last <<= PAGE_SHIFT; + ei_start = early_node_map[i].start_pfn; + ei_start <<= PAGE_SHIFT; + addr = find_early_area(ei_start, ei_last, + goal, limit, size, align); + + if (addr == -1ULL) + continue; + +#if 0 + printk(KERN_DEBUG "alloc (nid=%d %llx - %llx) (%llx - %llx) %llx %llx => %llx\n", + nid, + ei_start, ei_last, goal, limit, size, + align, addr); +#endif + + ptr = phys_to_virt(addr); + memset(ptr, 0, size); + reserve_early_without_check(addr, addr + size, "BOOTMEM"); + return ptr; + } + + return NULL; +} + + void __init work_with_active_regions(int nid, work_fn_t work_fn, void *data) { int i; @@ -4467,7 +4520,11 @@ void __init set_dma_reserve(unsigned long new_dma_reserve) } #ifndef CONFIG_NEED_MULTIPLE_NODES -struct pglist_data __refdata contig_page_data = { .bdata = &bootmem_node_data[0] }; +struct pglist_data __refdata contig_page_data = { +#ifndef CONFIG_NO_BOOTMEM + .bdata = &bootmem_node_data[0] +#endif + }; EXPORT_SYMBOL(contig_page_data); #endif diff --git a/mm/percpu.c b/mm/percpu.c index 083e7c91e5f..841defeeef8 100644 --- a/mm/percpu.c +++ b/mm/percpu.c @@ -1929,7 +1929,10 @@ int __init pcpu_embed_first_chunk(size_t reserved_size, ssize_t dyn_size, } /* copy and return the unused part */ memcpy(ptr, __per_cpu_load, ai->static_size); +#ifndef CONFIG_NO_BOOTMEM + /* fix partial free ! */ free_fn(ptr + size_sum, ai->unit_size - size_sum); +#endif } } diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c index d9714bdcb4a..9506c39942f 100644 --- a/mm/sparse-vmemmap.c +++ b/mm/sparse-vmemmap.c @@ -40,7 +40,7 @@ static void * __init_refok __earlyonly_bootmem_alloc(int node, unsigned long align, unsigned long goal) { - return __alloc_bootmem_node(NODE_DATA(node), size, align, goal); + return __alloc_bootmem_node_high(NODE_DATA(node), size, align, goal); } -- cgit v1.2.3-70-g09d2 From 9bdac914240759457175ac0d6529a37d2820bc4d Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 10 Feb 2010 01:20:22 -0800 Subject: sparsemem: Put mem map for one node together. Add vmemmap_alloc_block_buf for mem map only. It will fallback to the old way if it cannot get a block that big. Before this patch, when a node have 128g ram installed, memmap are split into two parts or more. [ 0.000000] [ffffea0000000000-ffffea003fffffff] PMD -> [ffff880100600000-ffff88013e9fffff] on node 1 [ 0.000000] [ffffea0040000000-ffffea006fffffff] PMD -> [ffff88013ec00000-ffff88016ebfffff] on node 1 [ 0.000000] [ffffea0070000000-ffffea007fffffff] PMD -> [ffff882000600000-ffff8820105fffff] on node 0 [ 0.000000] [ffffea0080000000-ffffea00bfffffff] PMD -> [ffff882010800000-ffff8820507fffff] on node 0 [ 0.000000] [ffffea00c0000000-ffffea00dfffffff] PMD -> [ffff882050a00000-ffff8820709fffff] on node 0 [ 0.000000] [ffffea00e0000000-ffffea00ffffffff] PMD -> [ffff884000600000-ffff8840205fffff] on node 2 [ 0.000000] [ffffea0100000000-ffffea013fffffff] PMD -> [ffff884020800000-ffff8840607fffff] on node 2 [ 0.000000] [ffffea0140000000-ffffea014fffffff] PMD -> [ffff884060a00000-ffff8840709fffff] on node 2 [ 0.000000] [ffffea0150000000-ffffea017fffffff] PMD -> [ffff886000600000-ffff8860305fffff] on node 3 [ 0.000000] [ffffea0180000000-ffffea01bfffffff] PMD -> [ffff886030800000-ffff8860707fffff] on node 3 [ 0.000000] [ffffea01c0000000-ffffea01ffffffff] PMD -> [ffff888000600000-ffff8880405fffff] on node 4 [ 0.000000] [ffffea0200000000-ffffea022fffffff] PMD -> [ffff888040800000-ffff8880707fffff] on node 4 [ 0.000000] [ffffea0230000000-ffffea023fffffff] PMD -> [ffff88a000600000-ffff88a0105fffff] on node 5 [ 0.000000] [ffffea0240000000-ffffea027fffffff] PMD -> [ffff88a010800000-ffff88a0507fffff] on node 5 [ 0.000000] [ffffea0280000000-ffffea029fffffff] PMD -> [ffff88a050a00000-ffff88a0709fffff] on node 5 [ 0.000000] [ffffea02a0000000-ffffea02bfffffff] PMD -> [ffff88c000600000-ffff88c0205fffff] on node 6 [ 0.000000] [ffffea02c0000000-ffffea02ffffffff] PMD -> [ffff88c020800000-ffff88c0607fffff] on node 6 [ 0.000000] [ffffea0300000000-ffffea030fffffff] PMD -> [ffff88c060a00000-ffff88c0709fffff] on node 6 [ 0.000000] [ffffea0310000000-ffffea033fffffff] PMD -> [ffff88e000600000-ffff88e0305fffff] on node 7 [ 0.000000] [ffffea0340000000-ffffea037fffffff] PMD -> [ffff88e030800000-ffff88e0707fffff] on node 7 after patch will get [ 0.000000] [ffffea0000000000-ffffea006fffffff] PMD -> [ffff880100200000-ffff88016e5fffff] on node 0 [ 0.000000] [ffffea0070000000-ffffea00dfffffff] PMD -> [ffff882000200000-ffff8820701fffff] on node 1 [ 0.000000] [ffffea00e0000000-ffffea014fffffff] PMD -> [ffff884000200000-ffff8840701fffff] on node 2 [ 0.000000] [ffffea0150000000-ffffea01bfffffff] PMD -> [ffff886000200000-ffff8860701fffff] on node 3 [ 0.000000] [ffffea01c0000000-ffffea022fffffff] PMD -> [ffff888000200000-ffff8880701fffff] on node 4 [ 0.000000] [ffffea0230000000-ffffea029fffffff] PMD -> [ffff88a000200000-ffff88a0701fffff] on node 5 [ 0.000000] [ffffea02a0000000-ffffea030fffffff] PMD -> [ffff88c000200000-ffff88c0701fffff] on node 6 [ 0.000000] [ffffea0310000000-ffffea037fffffff] PMD -> [ffff88e000200000-ffff88e0701fffff] on node 7 -v2: change buf to vmemmap_buf instead according to Ingo also add CONFIG_SPARSEMEM_ALLOC_MEM_MAP_TOGETHER according to Ingo -v3: according to Andrew, use sizeof(name) instead of hard coded 15 Signed-off-by: Yinghai Lu LKML-Reference: <1265793639-15071-19-git-send-email-yinghai@kernel.org> Cc: Christoph Lameter Acked-by: Christoph Lameter Signed-off-by: H. Peter Anvin --- arch/x86/mm/init_64.c | 2 +- include/linux/mm.h | 7 ++++ mm/Kconfig | 4 ++ mm/sparse-vmemmap.c | 74 ++++++++++++++++++++++++++++++++- mm/sparse.c | 111 +++++++++++++++++++++++++++++++++++++++++++++++++- 5 files changed, 195 insertions(+), 3 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c index 53158b7e5d4..e9b040e1cde 100644 --- a/arch/x86/mm/init_64.c +++ b/arch/x86/mm/init_64.c @@ -977,7 +977,7 @@ vmemmap_populate(struct page *start_page, unsigned long size, int node) if (pmd_none(*pmd)) { pte_t entry; - p = vmemmap_alloc_block(PMD_SIZE, node); + p = vmemmap_alloc_block_buf(PMD_SIZE, node); if (!p) return -ENOMEM; diff --git a/include/linux/mm.h b/include/linux/mm.h index f2c5b3cee8a..f6002e5dc18 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -1326,12 +1326,19 @@ extern int randomize_va_space; const char * arch_vma_name(struct vm_area_struct *vma); void print_vma_addr(char *prefix, unsigned long rip); +void sparse_mem_maps_populate_node(struct page **map_map, + unsigned long pnum_begin, + unsigned long pnum_end, + unsigned long map_count, + int nodeid); + struct page *sparse_mem_map_populate(unsigned long pnum, int nid); pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); void *vmemmap_alloc_block(unsigned long size, int node); +void *vmemmap_alloc_block_buf(unsigned long size, int node); void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); int vmemmap_populate_basepages(struct page *start_page, unsigned long pages, int node); diff --git a/mm/Kconfig b/mm/Kconfig index 17b8947aa7d..e4a33b9479b 100644 --- a/mm/Kconfig +++ b/mm/Kconfig @@ -115,6 +115,10 @@ config SPARSEMEM_EXTREME config SPARSEMEM_VMEMMAP_ENABLE bool +config SPARSEMEM_ALLOC_MEM_MAP_TOGETHER + def_bool y + depends on SPARSEMEM && X86_64 + config SPARSEMEM_VMEMMAP bool "Sparse Memory virtual memmap" depends on SPARSEMEM && SPARSEMEM_VMEMMAP_ENABLE diff --git a/mm/sparse-vmemmap.c b/mm/sparse-vmemmap.c index 9506c39942f..392b9bb5bc0 100644 --- a/mm/sparse-vmemmap.c +++ b/mm/sparse-vmemmap.c @@ -43,6 +43,8 @@ static void * __init_refok __earlyonly_bootmem_alloc(int node, return __alloc_bootmem_node_high(NODE_DATA(node), size, align, goal); } +static void *vmemmap_buf; +static void *vmemmap_buf_end; void * __meminit vmemmap_alloc_block(unsigned long size, int node) { @@ -64,6 +66,24 @@ void * __meminit vmemmap_alloc_block(unsigned long size, int node) __pa(MAX_DMA_ADDRESS)); } +/* need to make sure size is all the same during early stage */ +void * __meminit vmemmap_alloc_block_buf(unsigned long size, int node) +{ + void *ptr; + + if (!vmemmap_buf) + return vmemmap_alloc_block(size, node); + + /* take the from buf */ + ptr = (void *)ALIGN((unsigned long)vmemmap_buf, size); + if (ptr + size > vmemmap_buf_end) + return vmemmap_alloc_block(size, node); + + vmemmap_buf = ptr + size; + + return ptr; +} + void __meminit vmemmap_verify(pte_t *pte, int node, unsigned long start, unsigned long end) { @@ -80,7 +100,7 @@ pte_t * __meminit vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node) pte_t *pte = pte_offset_kernel(pmd, addr); if (pte_none(*pte)) { pte_t entry; - void *p = vmemmap_alloc_block(PAGE_SIZE, node); + void *p = vmemmap_alloc_block_buf(PAGE_SIZE, node); if (!p) return NULL; entry = pfn_pte(__pa(p) >> PAGE_SHIFT, PAGE_KERNEL); @@ -163,3 +183,55 @@ struct page * __meminit sparse_mem_map_populate(unsigned long pnum, int nid) return map; } + +void __init sparse_mem_maps_populate_node(struct page **map_map, + unsigned long pnum_begin, + unsigned long pnum_end, + unsigned long map_count, int nodeid) +{ + unsigned long pnum; + unsigned long size = sizeof(struct page) * PAGES_PER_SECTION; + void *vmemmap_buf_start; + + size = ALIGN(size, PMD_SIZE); + vmemmap_buf_start = __earlyonly_bootmem_alloc(nodeid, size * map_count, + PMD_SIZE, __pa(MAX_DMA_ADDRESS)); + + if (vmemmap_buf_start) { + vmemmap_buf = vmemmap_buf_start; + vmemmap_buf_end = vmemmap_buf_start + size * map_count; + } + + for (pnum = pnum_begin; pnum < pnum_end; pnum++) { + struct mem_section *ms; + + if (!present_section_nr(pnum)) + continue; + + map_map[pnum] = sparse_mem_map_populate(pnum, nodeid); + if (map_map[pnum]) + continue; + ms = __nr_to_section(pnum); + printk(KERN_ERR "%s: sparsemem memory map backing failed " + "some memory will not be available.\n", __func__); + ms->section_mem_map = 0; + } + + if (vmemmap_buf_start) { + /* need to free left buf */ +#ifdef CONFIG_NO_BOOTMEM + free_early(__pa(vmemmap_buf_start), __pa(vmemmap_buf_end)); + if (vmemmap_buf_start < vmemmap_buf) { + char name[15]; + + snprintf(name, sizeof(name), "MEMMAP %d", nodeid); + reserve_early_without_check(__pa(vmemmap_buf_start), + __pa(vmemmap_buf), name); + } +#else + free_bootmem(__pa(vmemmap_buf), vmemmap_buf_end - vmemmap_buf); +#endif + vmemmap_buf = NULL; + vmemmap_buf_end = NULL; + } +} diff --git a/mm/sparse.c b/mm/sparse.c index 0cdaf0b5845..9b6b93a4d78 100644 --- a/mm/sparse.c +++ b/mm/sparse.c @@ -390,8 +390,65 @@ struct page __init *sparse_mem_map_populate(unsigned long pnum, int nid) PAGE_ALIGN(sizeof(struct page) * PAGES_PER_SECTION)); return map; } +void __init sparse_mem_maps_populate_node(struct page **map_map, + unsigned long pnum_begin, + unsigned long pnum_end, + unsigned long map_count, int nodeid) +{ + void *map; + unsigned long pnum; + unsigned long size = sizeof(struct page) * PAGES_PER_SECTION; + + map = alloc_remap(nodeid, size * map_count); + if (map) { + for (pnum = pnum_begin; pnum < pnum_end; pnum++) { + if (!present_section_nr(pnum)) + continue; + map_map[pnum] = map; + map += size; + } + return; + } + + size = PAGE_ALIGN(size); + map = alloc_bootmem_pages_node(NODE_DATA(nodeid), size * map_count); + if (map) { + for (pnum = pnum_begin; pnum < pnum_end; pnum++) { + if (!present_section_nr(pnum)) + continue; + map_map[pnum] = map; + map += size; + } + return; + } + + /* fallback */ + for (pnum = pnum_begin; pnum < pnum_end; pnum++) { + struct mem_section *ms; + + if (!present_section_nr(pnum)) + continue; + map_map[pnum] = sparse_mem_map_populate(pnum, nodeid); + if (map_map[pnum]) + continue; + ms = __nr_to_section(pnum); + printk(KERN_ERR "%s: sparsemem memory map backing failed " + "some memory will not be available.\n", __func__); + ms->section_mem_map = 0; + } +} #endif /* !CONFIG_SPARSEMEM_VMEMMAP */ +static void __init sparse_early_mem_maps_alloc_node(struct page **map_map, + unsigned long pnum_begin, + unsigned long pnum_end, + unsigned long map_count, int nodeid) +{ + sparse_mem_maps_populate_node(map_map, pnum_begin, pnum_end, + map_count, nodeid); +} + +#ifndef CONFIG_SPARSEMEM_ALLOC_MEM_MAP_TOGETHER static struct page __init *sparse_early_mem_map_alloc(unsigned long pnum) { struct page *map; @@ -407,6 +464,7 @@ static struct page __init *sparse_early_mem_map_alloc(unsigned long pnum) ms->section_mem_map = 0; return NULL; } +#endif void __attribute__((weak)) __meminit vmemmap_populate_print_last(void) { @@ -420,12 +478,14 @@ void __init sparse_init(void) { unsigned long pnum; struct page *map; + struct page **map_map; unsigned long *usemap; unsigned long **usemap_map; - int size; + int size, size2; int nodeid_begin = 0; unsigned long pnum_begin = 0; unsigned long usemap_count; + unsigned long map_count; /* * map is using big page (aka 2M in x86 64 bit) @@ -478,6 +538,48 @@ void __init sparse_init(void) sparse_early_usemaps_alloc_node(usemap_map, pnum_begin, NR_MEM_SECTIONS, usemap_count, nodeid_begin); +#ifdef CONFIG_SPARSEMEM_ALLOC_MEM_MAP_TOGETHER + size2 = sizeof(struct page *) * NR_MEM_SECTIONS; + map_map = alloc_bootmem(size2); + if (!map_map) + panic("can not allocate map_map\n"); + + for (pnum = 0; pnum < NR_MEM_SECTIONS; pnum++) { + struct mem_section *ms; + + if (!present_section_nr(pnum)) + continue; + ms = __nr_to_section(pnum); + nodeid_begin = sparse_early_nid(ms); + pnum_begin = pnum; + break; + } + map_count = 1; + for (pnum = pnum_begin + 1; pnum < NR_MEM_SECTIONS; pnum++) { + struct mem_section *ms; + int nodeid; + + if (!present_section_nr(pnum)) + continue; + ms = __nr_to_section(pnum); + nodeid = sparse_early_nid(ms); + if (nodeid == nodeid_begin) { + map_count++; + continue; + } + /* ok, we need to take cake of from pnum_begin to pnum - 1*/ + sparse_early_mem_maps_alloc_node(map_map, pnum_begin, pnum, + map_count, nodeid_begin); + /* new start, update count etc*/ + nodeid_begin = nodeid; + pnum_begin = pnum; + map_count = 1; + } + /* ok, last chunk */ + sparse_early_mem_maps_alloc_node(map_map, pnum_begin, NR_MEM_SECTIONS, + map_count, nodeid_begin); +#endif + for (pnum = 0; pnum < NR_MEM_SECTIONS; pnum++) { if (!present_section_nr(pnum)) continue; @@ -486,7 +588,11 @@ void __init sparse_init(void) if (!usemap) continue; +#ifdef CONFIG_SPARSEMEM_ALLOC_MEM_MAP_TOGETHER + map = map_map[pnum]; +#else map = sparse_early_mem_map_alloc(pnum); +#endif if (!map) continue; @@ -496,6 +602,9 @@ void __init sparse_init(void) vmemmap_populate_print_last(); +#ifdef CONFIG_SPARSEMEM_ALLOC_MEM_MAP_TOGETHER + free_bootmem(__pa(map_map), size2); +#endif free_bootmem(__pa(usemap_map), size); } -- cgit v1.2.3-70-g09d2 From 59be5a8e8ce765cf739ec7f07176219972de7481 Mon Sep 17 00:00:00 2001 From: Yinghai Lu Date: Wed, 10 Feb 2010 01:20:28 -0800 Subject: x86: Make 32bit support NO_BOOTMEM Let's make 32bit consistent with 64bit. -v2: Andrew pointed out for 32bit that we should use -1ULL Signed-off-by: Yinghai Lu LKML-Reference: <1265793639-15071-25-git-send-email-yinghai@kernel.org> Signed-off-by: H. Peter Anvin --- arch/x86/Kconfig | 1 - arch/x86/kernel/early_res.c | 3 +++ arch/x86/mm/init_32.c | 6 ++++++ arch/x86/mm/numa_32.c | 3 +++ 4 files changed, 12 insertions(+), 1 deletion(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig index 95439843ceb..29f9efb74fc 100644 --- a/arch/x86/Kconfig +++ b/arch/x86/Kconfig @@ -571,7 +571,6 @@ config PARAVIRT_DEBUG config NO_BOOTMEM default y bool "Disable Bootmem code" - depends on X86_64 ---help--- Use early_res directly instead of bootmem before slab is ready. - allocator (buddy) [generic] diff --git a/arch/x86/kernel/early_res.c b/arch/x86/kernel/early_res.c index dbf08bd0125..656cdf86a2f 100644 --- a/arch/x86/kernel/early_res.c +++ b/arch/x86/kernel/early_res.c @@ -354,6 +354,9 @@ int __init get_free_all_memory_range(struct range **rangep, int nodeid) /* need to go over early_node_map to find out good range for node */ nr_range = add_from_early_node_map(range, count, nr_range, nodeid); +#ifdef CONFIG_X86_32 + subtract_range(range, count, max_low_pfn, -1ULL); +#endif subtract_early_res(range, count); nr_range = clean_sort_range(range, count); diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c index 2dccde06d22..262867a7d43 100644 --- a/arch/x86/mm/init_32.c +++ b/arch/x86/mm/init_32.c @@ -748,6 +748,7 @@ static void __init zone_sizes_init(void) free_area_init_nodes(max_zone_pfns); } +#ifndef CONFIG_NO_BOOTMEM static unsigned long __init setup_node_bootmem(int nodeid, unsigned long start_pfn, unsigned long end_pfn, @@ -767,9 +768,11 @@ static unsigned long __init setup_node_bootmem(int nodeid, return bootmap + bootmap_size; } +#endif void __init setup_bootmem_allocator(void) { +#ifndef CONFIG_NO_BOOTMEM int nodeid; unsigned long bootmap_size, bootmap; /* @@ -781,11 +784,13 @@ void __init setup_bootmem_allocator(void) if (bootmap == -1L) panic("Cannot find bootmem map of size %ld\n", bootmap_size); reserve_early(bootmap, bootmap + bootmap_size, "BOOTMAP"); +#endif printk(KERN_INFO " mapped low ram: 0 - %08lx\n", max_pfn_mapped<node_id = nid; +#ifndef CONFIG_NO_BOOTMEM NODE_DATA(nid)->bdata = &bootmem_node_data[nid]; +#endif } setup_bootmem_allocator(); -- cgit v1.2.3-70-g09d2 From 68fd111e02b979876359c7b471a8bcbca0628b75 Mon Sep 17 00:00:00 2001 From: David Rientjes Date: Mon, 15 Feb 2010 13:43:25 -0800 Subject: x86, numa: Fix numa emulation calculation of big nodes numa=fake=N uses split_nodes_interleave() to partition the system into N fake nodes. Each node size must have be a multiple of FAKE_NODE_MIN_SIZE, otherwise it is possible to get strange alignments. Because of this, the remaining memory from each node when rounded to FAKE_NODE_MIN_SIZE is consolidated into a number of "big nodes" that are bigger than the rest. The calculation of the number of big nodes is incorrect since it is using a logical AND operator when it should be multiplying the rounded-off portion of each node with N. Signed-off-by: David Rientjes LKML-Reference: Signed-off-by: H. Peter Anvin --- arch/x86/mm/numa_64.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c index 83bbc70d11b..2ecbe0ca0df 100644 --- a/arch/x86/mm/numa_64.c +++ b/arch/x86/mm/numa_64.c @@ -427,7 +427,7 @@ static int __init split_nodes_interleave(u64 addr, u64 max_addr, * Calculate the number of big nodes that can be allocated as a result * of consolidating the remainder. */ - big = ((size & ~FAKE_NODE_MIN_HASH_MASK) & nr_nodes) / + big = ((size & ~FAKE_NODE_MIN_HASH_MASK) * nr_nodes) / FAKE_NODE_MIN_SIZE; size &= FAKE_NODE_MIN_HASH_MASK; -- cgit v1.2.3-70-g09d2 From 8df5bb34defd685fe86f60746bbf3d47d1c6f033 Mon Sep 17 00:00:00 2001 From: David Rientjes Date: Mon, 15 Feb 2010 13:43:30 -0800 Subject: x86, numa: Add fixed node size option for numa emulation numa=fake=N specifies the number of fake nodes, N, to partition the system into and then allocates them by interleaving over physical nodes. This requires knowledge of the system capacity when attempting to allocate nodes of a certain size: either very large nodes to benchmark scalability of code that operates on individual nodes, or very small nodes to find bugs in the VM. This patch introduces numa=fake=[MG] so it is possible to specify the size of each node to allocate. When used, nodes of the size specified will be allocated and interleaved over the set of physical nodes. FAKE_NODE_MIN_SIZE was also moved to the more-appropriate include/asm/numa_64.h. Signed-off-by: David Rientjes LKML-Reference: Signed-off-by: H. Peter Anvin --- Documentation/x86/x86_64/boot-options.txt | 4 + arch/x86/include/asm/mmzone_64.h | 6 -- arch/x86/include/asm/numa_64.h | 5 ++ arch/x86/mm/numa_64.c | 117 ++++++++++++++++++++++++++++-- 4 files changed, 118 insertions(+), 14 deletions(-) (limited to 'arch/x86/mm') diff --git a/Documentation/x86/x86_64/boot-options.txt b/Documentation/x86/x86_64/boot-options.txt index 29a6ff8bc7d..01150c64aa7 100644 --- a/Documentation/x86/x86_64/boot-options.txt +++ b/Documentation/x86/x86_64/boot-options.txt @@ -166,6 +166,10 @@ NUMA numa=noacpi Don't parse the SRAT table for NUMA setup + numa=fake=[MG] + If given as a memory unit, fills all system RAM with nodes of + size interleaved over physical nodes. + numa=fake=CMDLINE If a number, fakes CMDLINE nodes and ignores NUMA setup of the actual machine. Otherwise, system memory is configured diff --git a/arch/x86/include/asm/mmzone_64.h b/arch/x86/include/asm/mmzone_64.h index a29f48c2a32..288b96f815a 100644 --- a/arch/x86/include/asm/mmzone_64.h +++ b/arch/x86/include/asm/mmzone_64.h @@ -39,11 +39,5 @@ static inline __attribute__((pure)) int phys_to_nid(unsigned long addr) #define node_start_pfn(nid) (NODE_DATA(nid)->node_start_pfn) #define node_end_pfn(nid) (NODE_DATA(nid)->node_start_pfn + \ NODE_DATA(nid)->node_spanned_pages) - -#ifdef CONFIG_NUMA_EMU -#define FAKE_NODE_MIN_SIZE (64 * 1024 * 1024) -#define FAKE_NODE_MIN_HASH_MASK (~(FAKE_NODE_MIN_SIZE - 1UL)) -#endif - #endif #endif /* _ASM_X86_MMZONE_64_H */ diff --git a/arch/x86/include/asm/numa_64.h b/arch/x86/include/asm/numa_64.h index c4ae822e415..823e070e7c2 100644 --- a/arch/x86/include/asm/numa_64.h +++ b/arch/x86/include/asm/numa_64.h @@ -36,6 +36,11 @@ extern void __cpuinit numa_set_node(int cpu, int node); extern void __cpuinit numa_clear_node(int cpu); extern void __cpuinit numa_add_cpu(int cpu); extern void __cpuinit numa_remove_cpu(int cpu); + +#ifdef CONFIG_NUMA_EMU +#define FAKE_NODE_MIN_SIZE ((u64)64 << 20) +#define FAKE_NODE_MIN_HASH_MASK (~(FAKE_NODE_MIN_SIZE - 1UL)) +#endif /* CONFIG_NUMA_EMU */ #else static inline void init_cpu_to_node(void) { } static inline void numa_set_node(int cpu, int node) { } diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c index 2ecbe0ca0df..c47c78ba3ac 100644 --- a/arch/x86/mm/numa_64.c +++ b/arch/x86/mm/numa_64.c @@ -501,6 +501,102 @@ static int __init split_nodes_interleave(u64 addr, u64 max_addr, return ret; } +/* + * Returns the end address of a node so that there is at least `size' amount of + * non-reserved memory or `max_addr' is reached. + */ +static u64 __init find_end_of_node(u64 start, u64 max_addr, u64 size) +{ + u64 end = start + size; + + while (end - start - e820_hole_size(start, end) < size) { + end += FAKE_NODE_MIN_SIZE; + if (end > max_addr) { + end = max_addr; + break; + } + } + return end; +} + +/* + * Sets up fake nodes of `size' interleaved over physical nodes ranging from + * `addr' to `max_addr'. The return value is the number of nodes allocated. + */ +static int __init split_nodes_size_interleave(u64 addr, u64 max_addr, u64 size) +{ + nodemask_t physnode_mask = NODE_MASK_NONE; + u64 min_size; + int ret = 0; + int i; + + if (!size) + return -1; + /* + * The limit on emulated nodes is MAX_NUMNODES, so the size per node is + * increased accordingly if the requested size is too small. This + * creates a uniform distribution of node sizes across the entire + * machine (but not necessarily over physical nodes). + */ + min_size = (max_addr - addr - e820_hole_size(addr, max_addr)) / + MAX_NUMNODES; + min_size = max(min_size, FAKE_NODE_MIN_SIZE); + if ((min_size & FAKE_NODE_MIN_HASH_MASK) < min_size) + min_size = (min_size + FAKE_NODE_MIN_SIZE) & + FAKE_NODE_MIN_HASH_MASK; + if (size < min_size) { + pr_err("Fake node size %LuMB too small, increasing to %LuMB\n", + size >> 20, min_size >> 20); + size = min_size; + } + size &= FAKE_NODE_MIN_HASH_MASK; + + for (i = 0; i < MAX_NUMNODES; i++) + if (physnodes[i].start != physnodes[i].end) + node_set(i, physnode_mask); + /* + * Fill physical nodes with fake nodes of size until there is no memory + * left on any of them. + */ + while (nodes_weight(physnode_mask)) { + for_each_node_mask(i, physnode_mask) { + u64 dma32_end = MAX_DMA32_PFN << PAGE_SHIFT; + u64 end; + + end = find_end_of_node(physnodes[i].start, + physnodes[i].end, size); + /* + * If there won't be at least FAKE_NODE_MIN_SIZE of + * non-reserved memory in ZONE_DMA32 for the next node, + * this one must extend to the boundary. + */ + if (end < dma32_end && dma32_end - end - + e820_hole_size(end, dma32_end) < FAKE_NODE_MIN_SIZE) + end = dma32_end; + + /* + * If there won't be enough non-reserved memory for the + * next node, this one must extend to the end of the + * physical node. + */ + if (physnodes[i].end - end - + e820_hole_size(end, physnodes[i].end) < size) + end = physnodes[i].end; + + /* + * Setup the fake node that will be allocated as bootmem + * later. If setup_node_range() returns non-zero, there + * is no more memory available on this physical node. + */ + if (setup_node_range(ret++, &physnodes[i].start, + end - physnodes[i].start, + physnodes[i].end) < 0) + node_clear(i, physnode_mask); + } + } + return ret; +} + /* * Splits num_nodes nodes up equally starting at node_start. The return value * is the number of nodes split up and addr is adjusted to be at the end of the @@ -546,14 +642,7 @@ static int __init split_nodes_equally(u64 *addr, u64 max_addr, int node_start, if (i == num_nodes + node_start - 1) end = max_addr; else - while (end - *addr - e820_hole_size(*addr, end) < - size) { - end += FAKE_NODE_MIN_SIZE; - if (end > max_addr) { - end = max_addr; - break; - } - } + end = find_end_of_node(*addr, max_addr, size); if (setup_node_range(i, addr, end - *addr, max_addr) < 0) break; } @@ -588,6 +677,18 @@ static int __init numa_emulation(unsigned long start_pfn, int num_phys_nodes; num_phys_nodes = setup_physnodes(addr, max_addr, acpi, k8); + /* + * If the numa=fake command-line contains a 'M' or 'G', it represents + * the fixed node size. + */ + if (strchr(cmdline, 'M') || strchr(cmdline, 'G')) { + size = memparse(cmdline, &cmdline); + num_nodes = split_nodes_size_interleave(addr, max_addr, size); + if (num_nodes < 0) + return num_nodes; + goto out; + } + /* * If the numa=fake command-line is just a single number N, split the * system RAM into N fake nodes. -- cgit v1.2.3-70-g09d2 From ca2107c9d6cf44fb915402d6f12b9d9ff3925cd7 Mon Sep 17 00:00:00 2001 From: David Rientjes Date: Mon, 15 Feb 2010 13:43:33 -0800 Subject: x86, numa: Remove configurable node size support for numa emulation Now that numa=fake=[MG] is implemented, it is possible to remove configurable node size support. The command-line parsing was already broken (numa=fake=*128, for example, would not work) and since fake nodes are now interleaved over physical nodes, this support is no longer required. Signed-off-by: David Rientjes LKML-Reference: Signed-off-by: H. Peter Anvin --- Documentation/x86/x86_64/boot-options.txt | 16 +-- arch/x86/mm/numa_64.c | 160 +++--------------------------- 2 files changed, 16 insertions(+), 160 deletions(-) (limited to 'arch/x86/mm') diff --git a/Documentation/x86/x86_64/boot-options.txt b/Documentation/x86/x86_64/boot-options.txt index 01150c64aa7..7fbbaf85f5b 100644 --- a/Documentation/x86/x86_64/boot-options.txt +++ b/Documentation/x86/x86_64/boot-options.txt @@ -170,19 +170,9 @@ NUMA If given as a memory unit, fills all system RAM with nodes of size interleaved over physical nodes. - numa=fake=CMDLINE - If a number, fakes CMDLINE nodes and ignores NUMA setup of the - actual machine. Otherwise, system memory is configured - depending on the sizes and coefficients listed. For example: - numa=fake=2*512,1024,4*256,*128 - gives two 512M nodes, a 1024M node, four 256M nodes, and the - rest split into 128M chunks. If the last character of CMDLINE - is a *, the remaining memory is divided up equally among its - coefficient: - numa=fake=2*512,2* - gives two 512M nodes and the rest split into two nodes. - Otherwise, the remaining system RAM is allocated to an - additional node. + numa=fake= + If given as an integer, fills all system RAM with N fake nodes + interleaved over physical nodes. ACPI diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c index c47c78ba3ac..3307ea8bd43 100644 --- a/arch/x86/mm/numa_64.c +++ b/arch/x86/mm/numa_64.c @@ -597,73 +597,6 @@ static int __init split_nodes_size_interleave(u64 addr, u64 max_addr, u64 size) return ret; } -/* - * Splits num_nodes nodes up equally starting at node_start. The return value - * is the number of nodes split up and addr is adjusted to be at the end of the - * last node allocated. - */ -static int __init split_nodes_equally(u64 *addr, u64 max_addr, int node_start, - int num_nodes) -{ - unsigned int big; - u64 size; - int i; - - if (num_nodes <= 0) - return -1; - if (num_nodes > MAX_NUMNODES) - num_nodes = MAX_NUMNODES; - size = (max_addr - *addr - e820_hole_size(*addr, max_addr)) / - num_nodes; - /* - * Calculate the number of big nodes that can be allocated as a result - * of consolidating the leftovers. - */ - big = ((size & ~FAKE_NODE_MIN_HASH_MASK) * num_nodes) / - FAKE_NODE_MIN_SIZE; - - /* Round down to nearest FAKE_NODE_MIN_SIZE. */ - size &= FAKE_NODE_MIN_HASH_MASK; - if (!size) { - printk(KERN_ERR "Not enough memory for each node. " - "NUMA emulation disabled.\n"); - return -1; - } - - for (i = node_start; i < num_nodes + node_start; i++) { - u64 end = *addr + size; - - if (i < big) - end += FAKE_NODE_MIN_SIZE; - /* - * The final node can have the remaining system RAM. Other - * nodes receive roughly the same amount of available pages. - */ - if (i == num_nodes + node_start - 1) - end = max_addr; - else - end = find_end_of_node(*addr, max_addr, size); - if (setup_node_range(i, addr, end - *addr, max_addr) < 0) - break; - } - return i - node_start + 1; -} - -/* - * Splits the remaining system RAM into chunks of size. The remaining memory is - * always assigned to a final node and can be asymmetric. Returns the number of - * nodes split. - */ -static int __init split_nodes_by_size(u64 *addr, u64 max_addr, int node_start, - u64 size) -{ - int i = node_start; - size = (size << 20) & FAKE_NODE_MIN_HASH_MASK; - while (!setup_node_range(i++, addr, size, max_addr)) - ; - return i - node_start; -} - /* * Sets up the system RAM area from start_pfn to last_pfn according to the * numa=fake command-line option. @@ -671,99 +604,32 @@ static int __init split_nodes_by_size(u64 *addr, u64 max_addr, int node_start, static int __init numa_emulation(unsigned long start_pfn, unsigned long last_pfn, int acpi, int k8) { - u64 size, addr = start_pfn << PAGE_SHIFT; + u64 addr = start_pfn << PAGE_SHIFT; u64 max_addr = last_pfn << PAGE_SHIFT; - int num_nodes = 0, num = 0, coeff_flag, coeff = -1, i; int num_phys_nodes; + int num_nodes; + int i; num_phys_nodes = setup_physnodes(addr, max_addr, acpi, k8); /* * If the numa=fake command-line contains a 'M' or 'G', it represents - * the fixed node size. + * the fixed node size. Otherwise, if it is just a single number N, + * split the system RAM into N fake nodes. */ if (strchr(cmdline, 'M') || strchr(cmdline, 'G')) { + u64 size; + size = memparse(cmdline, &cmdline); num_nodes = split_nodes_size_interleave(addr, max_addr, size); - if (num_nodes < 0) - return num_nodes; - goto out; - } + } else { + unsigned long n; - /* - * If the numa=fake command-line is just a single number N, split the - * system RAM into N fake nodes. - */ - if (!strchr(cmdline, '*') && !strchr(cmdline, ',')) { - long n = simple_strtol(cmdline, NULL, 0); - - num_nodes = split_nodes_interleave(addr, max_addr, - num_phys_nodes, n); - if (num_nodes < 0) - return num_nodes; - goto out; + n = simple_strtoul(cmdline, NULL, 0); + num_nodes = split_nodes_interleave(addr, max_addr, num_phys_nodes, n); } - /* Parse the command line. */ - for (coeff_flag = 0; ; cmdline++) { - if (*cmdline && isdigit(*cmdline)) { - num = num * 10 + *cmdline - '0'; - continue; - } - if (*cmdline == '*') { - if (num > 0) - coeff = num; - coeff_flag = 1; - } - if (!*cmdline || *cmdline == ',') { - if (!coeff_flag) - coeff = 1; - /* - * Round down to the nearest FAKE_NODE_MIN_SIZE. - * Command-line coefficients are in megabytes. - */ - size = ((u64)num << 20) & FAKE_NODE_MIN_HASH_MASK; - if (size) - for (i = 0; i < coeff; i++, num_nodes++) - if (setup_node_range(num_nodes, &addr, - size, max_addr) < 0) - goto done; - if (!*cmdline) - break; - coeff_flag = 0; - coeff = -1; - } - num = 0; - } -done: - if (!num_nodes) - return -1; - /* Fill remainder of system RAM, if appropriate. */ - if (addr < max_addr) { - if (coeff_flag && coeff < 0) { - /* Split remaining nodes into num-sized chunks */ - num_nodes += split_nodes_by_size(&addr, max_addr, - num_nodes, num); - goto out; - } - switch (*(cmdline - 1)) { - case '*': - /* Split remaining nodes into coeff chunks */ - if (coeff <= 0) - break; - num_nodes += split_nodes_equally(&addr, max_addr, - num_nodes, coeff); - break; - case ',': - /* Do not allocate remaining system RAM */ - break; - default: - /* Give one final node */ - setup_node_range(num_nodes, &addr, max_addr - addr, - max_addr); - num_nodes++; - } - } -out: + if (num_nodes < 0) + return num_nodes; memnode_shift = compute_hash_shift(nodes, num_nodes, NULL); if (memnode_shift < 0) { memnode_shift = 0; -- cgit v1.2.3-70-g09d2 From 39c662f60c556908faf861ef0430549b1731b891 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sat, 25 Jul 2009 19:15:48 +0200 Subject: x86: Convert tlbstate_lock to raw_spinlock Signed-off-by: Thomas Gleixner --- arch/x86/mm/tlb.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/mm/tlb.c b/arch/x86/mm/tlb.c index 65b58e4b0b8..426f3a1a64d 100644 --- a/arch/x86/mm/tlb.c +++ b/arch/x86/mm/tlb.c @@ -41,7 +41,7 @@ union smp_flush_state { struct { struct mm_struct *flush_mm; unsigned long flush_va; - spinlock_t tlbstate_lock; + raw_spinlock_t tlbstate_lock; DECLARE_BITMAP(flush_cpumask, NR_CPUS); }; char pad[INTERNODE_CACHE_BYTES]; @@ -181,7 +181,7 @@ static void flush_tlb_others_ipi(const struct cpumask *cpumask, * num_online_cpus() <= NUM_INVALIDATE_TLB_VECTORS, but it is * probably not worth checking this for a cache-hot lock. */ - spin_lock(&f->tlbstate_lock); + raw_spin_lock(&f->tlbstate_lock); f->flush_mm = mm; f->flush_va = va; @@ -199,7 +199,7 @@ static void flush_tlb_others_ipi(const struct cpumask *cpumask, f->flush_mm = NULL; f->flush_va = 0; - spin_unlock(&f->tlbstate_lock); + raw_spin_unlock(&f->tlbstate_lock); } void native_flush_tlb_others(const struct cpumask *cpumask, @@ -223,7 +223,7 @@ static int __cpuinit init_smp_flush(void) int i; for (i = 0; i < ARRAY_SIZE(flush_state); i++) - spin_lock_init(&flush_state[i].tlbstate_lock); + raw_spin_lock_init(&flush_state[i].tlbstate_lock); return 0; } -- cgit v1.2.3-70-g09d2 From 81fc03909a80bead8f553287a2b749a1d29dca64 Mon Sep 17 00:00:00 2001 From: Catalin Marinas Date: Mon, 8 Feb 2010 11:16:24 +0000 Subject: kmemcheck: Test the full object in kmemcheck_is_obj_initialized() This is a fix for bug #14845 (bugzilla.kernel.org). The update_checksum() function in mm/kmemleak.c calls kmemcheck_is_obj_initialised() before scanning an object. When KMEMCHECK_PARTIAL_OK is enabled, this function returns true. However, the crc32_le() reads smaller intervals (32-bit) for which kmemleak_is_obj_initialised() may be false leading to a kmemcheck warning. Note that kmemcheck_is_obj_initialized() is currently only used by kmemleak before scanning a memory location. Signed-off-by: Catalin Marinas Cc: Andrew Morton Cc: Christian Casteyde Cc: Vegard Nossum Signed-off-by: Pekka Enberg --- arch/x86/mm/kmemcheck/kmemcheck.c | 2 +- arch/x86/mm/kmemcheck/shadow.c | 16 ++++++++++++++-- arch/x86/mm/kmemcheck/shadow.h | 2 ++ 3 files changed, 17 insertions(+), 3 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/mm/kmemcheck/kmemcheck.c b/arch/x86/mm/kmemcheck/kmemcheck.c index 8cc18334414..b3b531a4f8e 100644 --- a/arch/x86/mm/kmemcheck/kmemcheck.c +++ b/arch/x86/mm/kmemcheck/kmemcheck.c @@ -337,7 +337,7 @@ bool kmemcheck_is_obj_initialized(unsigned long addr, size_t size) if (!shadow) return true; - status = kmemcheck_shadow_test(shadow, size); + status = kmemcheck_shadow_test_all(shadow, size); return status == KMEMCHECK_SHADOW_INITIALIZED; } diff --git a/arch/x86/mm/kmemcheck/shadow.c b/arch/x86/mm/kmemcheck/shadow.c index 3f66b82076a..aec124214d9 100644 --- a/arch/x86/mm/kmemcheck/shadow.c +++ b/arch/x86/mm/kmemcheck/shadow.c @@ -125,12 +125,12 @@ void kmemcheck_mark_initialized_pages(struct page *p, unsigned int n) enum kmemcheck_shadow kmemcheck_shadow_test(void *shadow, unsigned int size) { +#ifdef CONFIG_KMEMCHECK_PARTIAL_OK uint8_t *x; unsigned int i; x = shadow; -#ifdef CONFIG_KMEMCHECK_PARTIAL_OK /* * Make sure _some_ bytes are initialized. Gcc frequently generates * code to access neighboring bytes. @@ -139,13 +139,25 @@ enum kmemcheck_shadow kmemcheck_shadow_test(void *shadow, unsigned int size) if (x[i] == KMEMCHECK_SHADOW_INITIALIZED) return x[i]; } + + return x[0]; #else + return kmemcheck_shadow_test_all(shadow, size); +#endif +} + +enum kmemcheck_shadow kmemcheck_shadow_test_all(void *shadow, unsigned int size) +{ + uint8_t *x; + unsigned int i; + + x = shadow; + /* All bytes must be initialized. */ for (i = 0; i < size; ++i) { if (x[i] != KMEMCHECK_SHADOW_INITIALIZED) return x[i]; } -#endif return x[0]; } diff --git a/arch/x86/mm/kmemcheck/shadow.h b/arch/x86/mm/kmemcheck/shadow.h index af46d9ab9d8..ff0b2f70fbc 100644 --- a/arch/x86/mm/kmemcheck/shadow.h +++ b/arch/x86/mm/kmemcheck/shadow.h @@ -11,6 +11,8 @@ enum kmemcheck_shadow { void *kmemcheck_shadow_lookup(unsigned long address); enum kmemcheck_shadow kmemcheck_shadow_test(void *shadow, unsigned int size); +enum kmemcheck_shadow kmemcheck_shadow_test_all(void *shadow, + unsigned int size); void kmemcheck_shadow_set(void *shadow, unsigned int size); #endif -- cgit v1.2.3-70-g09d2 From 14315592009c17035cac81f4954d5a1f4d71e489 Mon Sep 17 00:00:00 2001 From: Ian Campbell Date: Wed, 17 Feb 2010 10:38:10 +0000 Subject: x86, mm: Allow highmem user page tables to be disabled at boot time Distros generally (I looked at Debian, RHEL5 and SLES11) seem to enable CONFIG_HIGHPTE for any x86 configuration which has highmem enabled. This means that the overhead applies even to machines which have a fairly modest amount of high memory and which therefore do not really benefit from allocating PTEs in high memory but still pay the price of the additional mapping operations. Running kernbench on a 4G box I found that with CONFIG_HIGHPTE=y but no actual highptes being allocated there was a reduction in system time used from 59.737s to 55.9s. With CONFIG_HIGHPTE=y and highmem PTEs being allocated: Average Optimal load -j 4 Run (std deviation): Elapsed Time 175.396 (0.238914) User Time 515.983 (5.85019) System Time 59.737 (1.26727) Percent CPU 263.8 (71.6796) Context Switches 39989.7 (4672.64) Sleeps 42617.7 (246.307) With CONFIG_HIGHPTE=y but with no highmem PTEs being allocated: Average Optimal load -j 4 Run (std deviation): Elapsed Time 174.278 (0.831968) User Time 515.659 (6.07012) System Time 55.9 (1.07799) Percent CPU 263.8 (71.266) Context Switches 39929.6 (4485.13) Sleeps 42583.7 (373.039) This patch allows the user to control the allocation of PTEs in highmem from the command line ("userpte=nohigh") but retains the status-quo as the default. It is possible that some simple heuristic could be developed which allows auto-tuning of this option however I don't have a sufficiently large machine available to me to perform any particularly meaningful experiments. We could probably handwave up an argument for a threshold at 16G of total RAM. Assuming 768M of lowmem we have 196608 potential lowmem PTE pages. Each page can map 2M of RAM in a PAE-enabled configuration, meaning a maximum of 384G of RAM could potentially be mapped using lowmem PTEs. Even allowing generous factor of 10 to account for other required lowmem allocations, generous slop to account for page sharing (which reduces the total amount of RAM mappable by a given number of PT pages) and other innacuracies in the estimations it would seem that even a 32G machine would not have a particularly pressing need for highmem PTEs. I think 32G could be considered to be at the upper bound of what might be sensible on a 32 bit machine (although I think in practice 64G is still supported). It's seems questionable if HIGHPTE is even a win for any amount of RAM you would sensibly run a 32 bit kernel on rather than going 64 bit. Signed-off-by: Ian Campbell LKML-Reference: <1266403090-20162-1-git-send-email-ian.campbell@citrix.com> Signed-off-by: H. Peter Anvin --- Documentation/kernel-parameters.txt | 7 +++++++ arch/x86/include/asm/pgalloc.h | 5 +++++ arch/x86/mm/pgtable.c | 31 ++++++++++++++++++++++++++----- 3 files changed, 38 insertions(+), 5 deletions(-) (limited to 'arch/x86/mm') diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt index 736d4560288..67c69ffe7b7 100644 --- a/Documentation/kernel-parameters.txt +++ b/Documentation/kernel-parameters.txt @@ -2694,6 +2694,13 @@ and is between 256 and 4096 characters. It is defined in the file medium is write-protected). Example: quirks=0419:aaf5:rl,0421:0433:rc + userpte= + [X86] Flags controlling user PTE allocations. + + nohigh = do not allocate PTE pages in + HIGHMEM regardless of setting + of CONFIG_HIGHPTE. + vdso= [X86,SH] vdso=2: enable compat VDSO (default with COMPAT_VDSO) vdso=1: enable VDSO (default) diff --git a/arch/x86/include/asm/pgalloc.h b/arch/x86/include/asm/pgalloc.h index 0e8c2a0fd92..271de94c381 100644 --- a/arch/x86/include/asm/pgalloc.h +++ b/arch/x86/include/asm/pgalloc.h @@ -22,6 +22,11 @@ static inline void paravirt_release_pmd(unsigned long pfn) {} static inline void paravirt_release_pud(unsigned long pfn) {} #endif +/* + * Flags to use when allocating a user page table page. + */ +extern gfp_t __userpte_alloc_gfp; + /* * Allocate and free page tables. */ diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c index ed34f5e3599..c9ba9deafe8 100644 --- a/arch/x86/mm/pgtable.c +++ b/arch/x86/mm/pgtable.c @@ -6,6 +6,14 @@ #define PGALLOC_GFP GFP_KERNEL | __GFP_NOTRACK | __GFP_REPEAT | __GFP_ZERO +#ifdef CONFIG_HIGHPTE +#define PGALLOC_USER_GFP __GFP_HIGHMEM +#else +#define PGALLOC_USER_GFP 0 +#endif + +gfp_t __userpte_alloc_gfp = PGALLOC_GFP | PGALLOC_USER_GFP; + pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address) { return (pte_t *)__get_free_page(PGALLOC_GFP); @@ -15,16 +23,29 @@ pgtable_t pte_alloc_one(struct mm_struct *mm, unsigned long address) { struct page *pte; -#ifdef CONFIG_HIGHPTE - pte = alloc_pages(PGALLOC_GFP | __GFP_HIGHMEM, 0); -#else - pte = alloc_pages(PGALLOC_GFP, 0); -#endif + pte = alloc_pages(__userpte_alloc_gfp, 0); if (pte) pgtable_page_ctor(pte); return pte; } +static int __init setup_userpte(char *arg) +{ + if (!arg) + return -EINVAL; + + /* + * "userpte=nohigh" disables allocation of user pagetables in + * high memory. + */ + if (strcmp(arg, "nohigh") == 0) + __userpte_alloc_gfp &= ~__GFP_HIGHMEM; + else + return -EINVAL; + return 0; +} +early_param("userpte", setup_userpte); + void ___pte_free_tlb(struct mmu_gather *tlb, struct page *pte) { pgtable_page_dtor(pte); -- cgit v1.2.3-70-g09d2 From c1fd1b43831fa20c91cdd461342af8edf2e87c2f Mon Sep 17 00:00:00 2001 From: Pekka Enberg Date: Wed, 24 Feb 2010 17:04:47 +0200 Subject: x86, mm: Unify kernel_physical_mapping_init() API This patch changes the 32-bit version of kernel_physical_mapping_init() to return the last mapped address like the 64-bit one so that we can unify the call-site in init_memory_mapping(). Cc: Yinghai Lu Cc: KAMEZAWA Hiroyuki Signed-off-by: Pekka Enberg LKML-Reference: Signed-off-by: H. Peter Anvin --- arch/x86/mm/init.c | 7 ------- arch/x86/mm/init_32.c | 8 +++++--- 2 files changed, 5 insertions(+), 10 deletions(-) (limited to 'arch/x86/mm') diff --git a/arch/x86/mm/init.c b/arch/x86/mm/init.c index d406c523901..e71c5cbc8f3 100644 --- a/arch/x86/mm/init.c +++ b/arch/x86/mm/init.c @@ -266,16 +266,9 @@ unsigned long __init_refok init_memory_mapping(unsigned long start, if (!after_bootmem) find_early_table_space(end, use_pse, use_gbpages); -#ifdef CONFIG_X86_32 - for (i = 0; i < nr_range; i++) - kernel_physical_mapping_init(mr[i].start, mr[i].end, - mr[i].page_size_mask); - ret = end; -#else /* CONFIG_X86_64 */ for (i = 0; i < nr_range; i++) ret = kernel_physical_mapping_init(mr[i].start, mr[i].end, mr[i].page_size_mask); -#endif #ifdef CONFIG_X86_32 early_ioremap_page_table_range_init(); diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c index 9a0c258a86b..2226f2c70ea 100644 --- a/arch/x86/mm/init_32.c +++ b/arch/x86/mm/init_32.c @@ -241,6 +241,7 @@ kernel_physical_mapping_init(unsigned long start, unsigned long page_size_mask) { int use_pse = page_size_mask == (1<