diff options
Diffstat (limited to 'arch/x86/mm')
| -rw-r--r-- | arch/x86/mm/Makefile | 9 | ||||
| -rw-r--r-- | arch/x86/mm/discontig_32.c | 285 | ||||
| -rw-r--r-- | arch/x86/mm/dump_pagetables.c | 2 | ||||
| -rw-r--r-- | arch/x86/mm/fault.c | 110 | ||||
| -rw-r--r-- | arch/x86/mm/init_32.c | 528 | ||||
| -rw-r--r-- | arch/x86/mm/init_64.c | 637 | ||||
| -rw-r--r-- | arch/x86/mm/ioremap.c | 60 | ||||
| -rw-r--r-- | arch/x86/mm/k8topology_64.c | 21 | ||||
| -rw-r--r-- | arch/x86/mm/kmmio.c | 510 | ||||
| -rw-r--r-- | arch/x86/mm/memtest.c | 123 | ||||
| -rw-r--r-- | arch/x86/mm/mmio-mod.c | 515 | ||||
| -rw-r--r-- | arch/x86/mm/numa_64.c | 93 | ||||
| -rw-r--r-- | arch/x86/mm/pageattr-test.c | 21 | ||||
| -rw-r--r-- | arch/x86/mm/pageattr.c | 67 | ||||
| -rw-r--r-- | arch/x86/mm/pat.c | 470 | ||||
| -rw-r--r-- | arch/x86/mm/pf_in.c | 489 | ||||
| -rw-r--r-- | arch/x86/mm/pf_in.h | 39 | ||||
| -rw-r--r-- | arch/x86/mm/pgtable.c | 190 | ||||
| -rw-r--r-- | arch/x86/mm/pgtable_32.c | 56 | ||||
| -rw-r--r-- | arch/x86/mm/srat_32.c | 279 | ||||
| -rw-r--r-- | arch/x86/mm/srat_64.c | 21 | ||||
| -rw-r--r-- | arch/x86/mm/testmmiotrace.c | 71 |
22 files changed, 3596 insertions, 1000 deletions
diff --git a/arch/x86/mm/Makefile b/arch/x86/mm/Makefile index b7b3e4c7cfc9..1fbb844c3d7a 100644 --- a/arch/x86/mm/Makefile +++ b/arch/x86/mm/Makefile @@ -8,10 +8,17 @@ obj-$(CONFIG_X86_PTDUMP) += dump_pagetables.o obj-$(CONFIG_HIGHMEM) += highmem_32.o +obj-$(CONFIG_MMIOTRACE_HOOKS) += kmmio.o +obj-$(CONFIG_MMIOTRACE) += mmiotrace.o +mmiotrace-y := pf_in.o mmio-mod.o +obj-$(CONFIG_MMIOTRACE_TEST) += testmmiotrace.o + ifeq ($(CONFIG_X86_32),y) obj-$(CONFIG_NUMA) += discontig_32.o else obj-$(CONFIG_NUMA) += numa_64.o obj-$(CONFIG_K8_NUMA) += k8topology_64.o -obj-$(CONFIG_ACPI_NUMA) += srat_64.o endif +obj-$(CONFIG_ACPI_NUMA) += srat_$(BITS).o + +obj-$(CONFIG_MEMTEST) += memtest.o diff --git a/arch/x86/mm/discontig_32.c b/arch/x86/mm/discontig_32.c index 914ccf983687..5dfef9fa061a 100644 --- a/arch/x86/mm/discontig_32.c +++ b/arch/x86/mm/discontig_32.c @@ -38,6 +38,7 @@ #include <asm/setup.h> #include <asm/mmzone.h> #include <asm/bios_ebda.h> +#include <asm/proto.h> struct pglist_data *node_data[MAX_NUMNODES] __read_mostly; EXPORT_SYMBOL(node_data); @@ -59,14 +60,14 @@ unsigned long node_end_pfn[MAX_NUMNODES] __read_mostly; /* * 4) physnode_map - the mapping between a pfn and owning node * physnode_map keeps track of the physical memory layout of a generic - * numa node on a 256Mb break (each element of the array will - * represent 256Mb of memory and will be marked by the node id. so, + * numa node on a 64Mb break (each element of the array will + * represent 64Mb of memory and will be marked by the node id. so, * if the first gig is on node 0, and the second gig is on node 1 * physnode_map will contain: * - * physnode_map[0-3] = 0; - * physnode_map[4-7] = 1; - * physnode_map[8- ] = -1; + * physnode_map[0-15] = 0; + * physnode_map[16-31] = 1; + * physnode_map[32- ] = -1; */ s8 physnode_map[MAX_ELEMENTS] __read_mostly = { [0 ... (MAX_ELEMENTS - 1)] = -1}; EXPORT_SYMBOL(physnode_map); @@ -75,15 +76,15 @@ void memory_present(int nid, unsigned long start, unsigned long end) { unsigned long pfn; - printk(KERN_INFO "Node: %d, start_pfn: %ld, end_pfn: %ld\n", + printk(KERN_INFO "Node: %d, start_pfn: %lx, end_pfn: %lx\n", nid, start, end); printk(KERN_DEBUG " Setting physnode_map array to node %d for pfns:\n", nid); printk(KERN_DEBUG " "); for (pfn = start; pfn < end; pfn += PAGES_PER_ELEMENT) { physnode_map[pfn / PAGES_PER_ELEMENT] = nid; - printk("%ld ", pfn); + printk(KERN_CONT "%lx ", pfn); } - printk("\n"); + printk(KERN_CONT "\n"); } unsigned long node_memmap_size_bytes(int nid, unsigned long start_pfn, @@ -99,7 +100,6 @@ unsigned long node_memmap_size_bytes(int nid, unsigned long start_pfn, #endif extern unsigned long find_max_low_pfn(void); -extern void add_one_highpage_init(struct page *, int, int); extern unsigned long highend_pfn, highstart_pfn; #define LARGE_PAGE_BYTES (PTRS_PER_PTE * PAGE_SIZE) @@ -117,13 +117,13 @@ static unsigned long kva_pages; */ int __init get_memcfg_numa_flat(void) { - printk("NUMA - single node, flat memory mode\n"); + printk(KERN_DEBUG "NUMA - single node, flat memory mode\n"); - /* Run the memory configuration and find the top of memory. */ - propagate_e820_map(); node_start_pfn[0] = 0; node_end_pfn[0] = max_pfn; + e820_register_active_regions(0, 0, max_pfn); memory_present(0, 0, max_pfn); + node_remap_size[0] = node_memmap_size_bytes(0, 0, max_pfn); /* Indicate there is one node available. */ nodes_clear(node_online_map); @@ -156,24 +156,32 @@ static void __init propagate_e820_map_node(int nid) */ static void __init allocate_pgdat(int nid) { - if (nid && node_has_online_mem(nid)) + char buf[16]; + + if (node_has_online_mem(nid) && node_remap_start_vaddr[nid]) NODE_DATA(nid) = (pg_data_t *)node_remap_start_vaddr[nid]; else { - NODE_DATA(nid) = (pg_data_t *)(pfn_to_kaddr(min_low_pfn)); - min_low_pfn += PFN_UP(sizeof(pg_data_t)); + unsigned long pgdat_phys; + pgdat_phys = find_e820_area(min_low_pfn<<PAGE_SHIFT, + max_pfn_mapped<<PAGE_SHIFT, + sizeof(pg_data_t), + PAGE_SIZE); + NODE_DATA(nid) = (pg_data_t *)(pfn_to_kaddr(pgdat_phys>>PAGE_SHIFT)); + memset(buf, 0, sizeof(buf)); + sprintf(buf, "NODE_DATA %d", nid); + reserve_early(pgdat_phys, pgdat_phys + sizeof(pg_data_t), buf); } + printk(KERN_DEBUG "allocate_pgdat: node %d NODE_DATA %08lx\n", + nid, (unsigned long)NODE_DATA(nid)); } -#ifdef CONFIG_DISCONTIGMEM /* - * In the discontig memory model, a portion of the kernel virtual area (KVA) - * is reserved and portions of nodes are mapped using it. This is to allow - * node-local memory to be allocated for structures that would normally require - * ZONE_NORMAL. The memory is allocated with alloc_remap() and callers - * should be prepared to allocate from the bootmem allocator instead. This KVA - * mechanism is incompatible with SPARSEMEM as it makes assumptions about the - * layout of memory that are broken if alloc_remap() succeeds for some of the - * map and fails for others + * In the DISCONTIGMEM and SPARSEMEM memory model, a portion of the kernel + * virtual address space (KVA) is reserved and portions of nodes are mapped + * using it. This is to allow node-local memory to be allocated for + * structures that would normally require ZONE_NORMAL. The memory is + * allocated with alloc_remap() and callers should be prepared to allocate + * from the bootmem allocator instead. */ static unsigned long node_remap_start_pfn[MAX_NUMNODES]; static void *node_remap_end_vaddr[MAX_NUMNODES]; @@ -195,15 +203,19 @@ void *alloc_remap(int nid, unsigned long size) return allocation; } -void __init remap_numa_kva(void) +static void __init remap_numa_kva(void) { void *vaddr; unsigned long pfn; int node; for_each_online_node(node) { + printk(KERN_DEBUG "remap_numa_kva: node %d\n", node); for (pfn=0; pfn < node_remap_size[node]; pfn += PTRS_PER_PTE) { vaddr = node_remap_start_vaddr[node]+(pfn<<PAGE_SHIFT); + printk(KERN_DEBUG "remap_numa_kva: %08lx to pfn %08lx\n", + (unsigned long)vaddr, + node_remap_start_pfn[node] + pfn); set_pmd_pfn((ulong) vaddr, node_remap_start_pfn[node] + pfn, PAGE_KERNEL_LARGE); @@ -215,17 +227,21 @@ static unsigned long calculate_numa_remap_pages(void) { int nid; unsigned long size, reserve_pages = 0; - unsigned long pfn; for_each_online_node(nid) { - unsigned old_end_pfn = node_end_pfn[nid]; + u64 node_kva_target; + u64 node_kva_final; /* * The acpi/srat node info can show hot-add memroy zones * where memory could be added but not currently present. */ + printk(KERN_DEBUG "node %d pfn: [%lx - %lx]\n", + nid, node_start_pfn[nid], node_end_pfn[nid]); if (node_start_pfn[nid] > max_pfn) continue; + if (!node_end_pfn[nid]) + continue; if (node_end_pfn[nid] > max_pfn) node_end_pfn[nid] = max_pfn; @@ -237,41 +253,48 @@ static unsigned long calculate_numa_remap_pages(void) /* now the roundup is correct, convert to PAGE_SIZE pages */ size = size * PTRS_PER_PTE; - /* - * Validate the region we are allocating only contains valid - * pages. - */ - for (pfn = node_end_pfn[nid] - size; - pfn < node_end_pfn[nid]; pfn++) - if (!page_is_ram(pfn)) - break; - - if (pfn != node_end_pfn[nid]) - size = 0; + node_kva_target = round_down(node_end_pfn[nid] - size, + PTRS_PER_PTE); + node_kva_target <<= PAGE_SHIFT; + do { + node_kva_final = find_e820_area(node_kva_target, + ((u64)node_end_pfn[nid])<<PAGE_SHIFT, + ((u64)size)<<PAGE_SHIFT, + LARGE_PAGE_BYTES); + node_kva_target -= LARGE_PAGE_BYTES; + } while (node_kva_final == -1ULL && + (node_kva_target>>PAGE_SHIFT) > (node_start_pfn[nid])); + + if (node_kva_final == -1ULL) + panic("Can not get kva ram\n"); - printk("Reserving %ld pages of KVA for lmem_map of node %d\n", - size, nid); node_remap_size[nid] = size; node_remap_offset[nid] = reserve_pages; reserve_pages += size; - printk("Shrinking node %d from %ld pages to %ld pages\n", - nid, node_end_pfn[nid], node_end_pfn[nid] - size); - - if (node_end_pfn[nid] & (PTRS_PER_PTE-1)) { - /* - * Align node_end_pfn[] and node_remap_start_pfn[] to - * pmd boundary. remap_numa_kva will barf otherwise. - */ - printk("Shrinking node %d further by %ld pages for proper alignment\n", - nid, node_end_pfn[nid] & (PTRS_PER_PTE-1)); - size += node_end_pfn[nid] & (PTRS_PER_PTE-1); - } + printk(KERN_DEBUG "Reserving %ld pages of KVA for lmem_map of" + " node %d at %llx\n", + size, nid, node_kva_final>>PAGE_SHIFT); + + /* + * prevent kva address below max_low_pfn want it on system + * with less memory later. + * layout will be: KVA address , KVA RAM + * + * we are supposed to only record the one less then max_low_pfn + * but we could have some hole in high memory, and it will only + * check page_is_ram(pfn) && !page_is_reserved_early(pfn) to decide + * to use it as free. + * So reserve_early here, hope we don't run out of that array + */ + reserve_early(node_kva_final, + node_kva_final+(((u64)size)<<PAGE_SHIFT), + "KVA RAM"); - node_end_pfn[nid] -= size; - node_remap_start_pfn[nid] = node_end_pfn[nid]; - shrink_active_range(nid, old_end_pfn, node_end_pfn[nid]); + node_remap_start_pfn[nid] = node_kva_final>>PAGE_SHIFT; + remove_active_range(nid, node_remap_start_pfn[nid], + node_remap_start_pfn[nid] + size); } - printk("Reserving total of %ld pages for numa KVA remap\n", + printk(KERN_INFO "Reserving total of %lx pages for numa KVA remap\n", reserve_pages); return reserve_pages; } @@ -285,37 +308,16 @@ static void init_remap_allocator(int nid) node_remap_alloc_vaddr[nid] = node_remap_start_vaddr[nid] + ALIGN(sizeof(pg_data_t), PAGE_SIZE); - printk ("node %d will remap to vaddr %08lx - %08lx\n", nid, + printk(KERN_DEBUG "node %d will remap to vaddr %08lx - %08lx\n", nid, (ulong) node_remap_start_vaddr[nid], - (ulong) pfn_to_kaddr(highstart_pfn - + node_remap_offset[nid] + node_remap_size[nid])); -} -#else -void *alloc_remap(int nid, unsigned long size) -{ - return NULL; -} - -static unsigned long calculate_numa_remap_pages(void) -{ - return 0; -} - -static void init_remap_allocator(int nid) -{ -} - -void __init remap_numa_kva(void) -{ + (ulong) node_remap_end_vaddr[nid]); } -#endif /* CONFIG_DISCONTIGMEM */ -extern void setup_bootmem_allocator(void); -unsigned long __init setup_memory(void) +void __init initmem_init(unsigned long start_pfn, + unsigned long end_pfn) { int nid; - unsigned long system_start_pfn, system_max_low_pfn; - unsigned long wasted_pages; + long kva_target_pfn; /* * When mapping a NUMA machine we allocate the node_mem_map arrays @@ -324,109 +326,77 @@ unsigned long __init setup_memory(void) * this space and use it to adjust the boundary between ZONE_NORMAL * and ZONE_HIGHMEM. */ - get_memcfg_numa(); - kva_pages = calculate_numa_remap_pages(); + get_memcfg_numa(); - /* partially used pages are not usable - thus round upwards */ - system_start_pfn = min_low_pfn = PFN_UP(init_pg_tables_end); + kva_pages = round_up(calculate_numa_remap_pages(), PTRS_PER_PTE); - kva_start_pfn = find_max_low_pfn() - kva_pages; + kva_target_pfn = round_down(max_low_pfn - kva_pages, PTRS_PER_PTE); + do { + kva_start_pfn = find_e820_area(kva_target_pfn<<PAGE_SHIFT, + max_low_pfn<<PAGE_SHIFT, + kva_pages<<PAGE_SHIFT, + PTRS_PER_PTE<<PAGE_SHIFT) >> PAGE_SHIFT; + kva_target_pfn -= PTRS_PER_PTE; + } while (kva_start_pfn == -1UL && kva_target_pfn > min_low_pfn); -#ifdef CONFIG_BLK_DEV_INITRD - /* Numa kva area is below the initrd */ - if (initrd_start) - kva_start_pfn = PFN_DOWN(initrd_start - PAGE_OFFSET) - - kva_pages; -#endif + if (kva_start_pfn == -1UL) + panic("Can not get kva space\n"); - /* - * We waste pages past at the end of the KVA for no good reason other - * than how it is located. This is bad. - */ - wasted_pages = kva_start_pfn & (PTRS_PER_PTE-1); - kva_start_pfn -= wasted_pages; - kva_pages += wasted_pages; - - system_max_low_pfn = max_low_pfn = find_max_low_pfn(); - printk("kva_start_pfn ~ %ld find_max_low_pfn() ~ %ld\n", + printk(KERN_INFO "kva_start_pfn ~ %lx max_low_pfn ~ %lx\n", kva_start_pfn, max_low_pfn); - printk("max_pfn = %ld\n", max_pfn); + printk(KERN_INFO "max_pfn = %lx\n", max_pfn); + + /* avoid clash with initrd */ + reserve_early(kva_start_pfn<<PAGE_SHIFT, + (kva_start_pfn + kva_pages)<<PAGE_SHIFT, + "KVA PG"); #ifdef CONFIG_HIGHMEM highstart_pfn = highend_pfn = max_pfn; - if (max_pfn > system_max_low_pfn) - highstart_pfn = system_max_low_pfn; + if (max_pfn > max_low_pfn) + highstart_pfn = max_low_pfn; printk(KERN_NOTICE "%ldMB HIGHMEM available.\n", pages_to_mb(highend_pfn - highstart_pfn)); num_physpages = highend_pfn; high_memory = (void *) __va(highstart_pfn * PAGE_SIZE - 1) + 1; #else - num_physpages = system_max_low_pfn; - high_memory = (void *) __va(system_max_low_pfn * PAGE_SIZE - 1) + 1; + num_physpages = max_low_pfn; + high_memory = (void *) __va(max_low_pfn * PAGE_SIZE - 1) + 1; #endif printk(KERN_NOTICE "%ldMB LOWMEM available.\n", - pages_to_mb(system_max_low_pfn)); - printk("min_low_pfn = %ld, max_low_pfn = %ld, highstart_pfn = %ld\n", - min_low_pfn, max_low_pfn, highstart_pfn); + pages_to_mb(max_low_pfn)); + printk(KERN_DEBUG "max_low_pfn = %lx, highstart_pfn = %lx\n", + max_low_pfn, highstart_pfn); - printk("Low memory ends at vaddr %08lx\n", + printk(KERN_DEBUG "Low memory ends at vaddr %08lx\n", (ulong) pfn_to_kaddr(max_low_pfn)); for_each_online_node(nid) { init_remap_allocator(nid); allocate_pgdat(nid); } - printk("High memory starts at vaddr %08lx\n", + remap_numa_kva(); + + printk(KERN_DEBUG "High memory starts at vaddr %08lx\n", (ulong) pfn_to_kaddr(highstart_pfn)); for_each_online_node(nid) propagate_e820_map_node(nid); - memset(NODE_DATA(0), 0, sizeof(struct pglist_data)); + for_each_online_node(nid) + memset(NODE_DATA(nid), 0, sizeof(struct pglist_data)); + NODE_DATA(0)->bdata = &node0_bdata; setup_bootmem_allocator(); - return max_low_pfn; -} - -void __init numa_kva_reserve(void) -{ - if (kva_pages) - reserve_bootmem(PFN_PHYS(kva_start_pfn), PFN_PHYS(kva_pages), - BOOTMEM_DEFAULT); } -void __init zone_sizes_init(void) -{ - int nid; - unsigned long max_zone_pfns[MAX_NR_ZONES]; - memset(max_zone_pfns, 0, sizeof(max_zone_pfns)); - max_zone_pfns[ZONE_DMA] = - virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT; - max_zone_pfns[ZONE_NORMAL] = max_low_pfn; -#ifdef CONFIG_HIGHMEM - max_zone_pfns[ZONE_HIGHMEM] = highend_pfn; -#endif - - /* If SRAT has not registered memory, register it now */ - if (find_max_pfn_with_active_regions() == 0) { - for_each_online_node(nid) { - if (node_has_online_mem(nid)) - add_active_range(nid, node_start_pfn[nid], - node_end_pfn[nid]); - } - } - - free_area_init_nodes(max_zone_pfns); - return; -} - -void __init set_highmem_pages_init(int bad_ppro) +void __init set_highmem_pages_init(void) { #ifdef CONFIG_HIGHMEM struct zone *zone; - struct page *page; + int nid; for_each_zone(zone) { - unsigned long node_pfn, zone_start_pfn, zone_end_pfn; + unsigned long zone_start_pfn, zone_end_pfn; if (!is_highmem(zone)) continue; @@ -434,16 +404,12 @@ void __init set_highmem_pages_init(int bad_ppro) zone_start_pfn = zone->zone_start_pfn; zone_end_pfn = zone_start_pfn + zone->spanned_pages; - printk("Initializing %s for node %d (%08lx:%08lx)\n", - zone->name, zone_to_nid(zone), - zone_start_pfn, zone_end_pfn); + nid = zone_to_nid(zone); + printk(KERN_INFO "Initializing %s for node %d (%08lx:%08lx)\n", + zone->name, nid, zone_start_pfn, zone_end_pfn); - for (node_pfn = zone_start_pfn; node_pfn < zone_end_pfn; node_pfn++) { - if (!pfn_valid(node_pfn)) - continue; - page = pfn_to_page(node_pfn); - add_one_highpage_init(page, node_pfn, bad_ppro); - } + add_highpages_with_active_regions(nid, zone_start_pfn, + zone_end_pfn); } totalram_pages += totalhigh_pages; #endif @@ -476,3 +442,4 @@ int memory_add_physaddr_to_nid(u64 addr) EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid); #endif + diff --git a/arch/x86/mm/dump_pagetables.c b/arch/x86/mm/dump_pagetables.c index 2c24bea92c66..0bb0caed8971 100644 --- a/arch/x86/mm/dump_pagetables.c +++ b/arch/x86/mm/dump_pagetables.c @@ -42,7 +42,7 @@ static struct addr_marker address_markers[] = { { 0, "User Space" }, #ifdef CONFIG_X86_64 { 0x8000000000000000UL, "Kernel Space" }, - { 0xffff810000000000UL, "Low Kernel Mapping" }, + { PAGE_OFFSET, "Low Kernel Mapping" }, { VMALLOC_START, "vmalloc() Area" }, { VMEMMAP_START, "Vmemmap" }, { __START_KERNEL_map, "High Kernel Mapping" }, diff --git a/arch/x86/mm/fault.c b/arch/x86/mm/fault.c index 8bcb6f40ccb6..455f3fe67b42 100644 --- a/arch/x86/mm/fault.c +++ b/arch/x86/mm/fault.c @@ -10,6 +10,7 @@ #include <linux/string.h> #include <linux/types.h> #include <linux/ptrace.h> +#include <linux/mmiotrace.h> #include <linux/mman.h> #include <linux/mm.h> #include <linux/smp.h> @@ -49,17 +50,23 @@ #define PF_RSVD (1<<3) #define PF_INSTR (1<<4) +static inline int kmmio_fault(struct pt_regs *regs, unsigned long addr) +{ +#ifdef CONFIG_MMIOTRACE_HOOKS + if (unlikely(is_kmmio_active())) + if (kmmio_handler(regs, addr) == 1) + return -1; +#endif + return 0; +} + static inline int notify_page_fault(struct pt_regs *regs) { #ifdef CONFIG_KPROBES int ret = 0; /* kprobe_running() needs smp_processor_id() */ -#ifdef CONFIG_X86_32 if (!user_mode_vm(regs)) { -#else - if (!user_mode(regs)) { -#endif preempt_disable(); if (kprobe_running() && kprobe_fault_handler(regs, 14)) ret = 1; @@ -396,11 +403,7 @@ static void show_fault_oops(struct pt_regs *regs, unsigned long error_code, printk(KERN_CONT "NULL pointer dereference"); else printk(KERN_CONT "paging request"); -#ifdef CONFIG_X86_32 - printk(KERN_CONT " at %08lx\n", address); -#else - printk(KERN_CONT " at %016lx\n", address); -#endif + printk(KERN_CONT " at %p\n", (void *) address); printk(KERN_ALERT "IP:"); printk_address(regs->ip, 1); dump_pagetable(address); @@ -606,6 +609,8 @@ void __kprobes do_page_fault(struct pt_regs *regs, unsigned long error_code) if (notify_page_fault(regs)) return; + if (unlikely(kmmio_fault(regs, address))) + return; /* * We fault-in kernel-space virtual memory on-demand. The @@ -800,14 +805,10 @@ bad_area_nosemaphore: if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) && printk_ratelimit()) { printk( -#ifdef CONFIG_X86_32 - "%s%s[%d]: segfault at %lx ip %08lx sp %08lx error %lx", -#else - "%s%s[%d]: segfault at %lx ip %lx sp %lx error %lx", -#endif + "%s%s[%d]: segfault at %lx ip %p sp %p error %lx", task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG, - tsk->comm, task_pid_nr(tsk), address, regs->ip, - regs->sp, error_code); + tsk->comm, task_pid_nr(tsk), address, + (void *) regs->ip, (void *) regs->sp, error_code); print_vma_addr(" in ", regs->ip); printk("\n"); } @@ -915,14 +916,7 @@ LIST_HEAD(pgd_list); void vmalloc_sync_all(void) { #ifdef CONFIG_X86_32 - /* - * Note that races in the updates of insync and start aren't - * problematic: insync can only get set bits added, and updates to - * start are only improving performance (without affecting correctness - * if undone). - */ - static DECLARE_BITMAP(insync, PTRS_PER_PGD); - static unsigned long start = TASK_SIZE; + unsigned long start = VMALLOC_START & PGDIR_MASK; unsigned long address; if (SHARED_KERNEL_PMD) @@ -930,56 +924,38 @@ void vmalloc_sync_all(void) BUILD_BUG_ON(TASK_SIZE & ~PGDIR_MASK); for (address = start; address >= TASK_SIZE; address += PGDIR_SIZE) { - if (!test_bit(pgd_index(address), insync)) { - unsigned long flags; - struct page *page; - - spin_lock_irqsave(&pgd_lock, flags); - list_for_each_entry(page, &pgd_list, lru) { - if (!vmalloc_sync_one(page_address(page), - address)) - break; - } - spin_unlock_irqrestore(&pgd_lock, flags); - if (!page) - set_bit(pgd_index(address), insync); + unsigned long flags; + struct page *page; + + spin_lock_irqsave(&pgd_lock, flags); + list_for_each_entry(page, &pgd_list, lru) { + if (!vmalloc_sync_one(page_address(page), + address)) + break; } - if (address == start && test_bit(pgd_index(address), insync)) - start = address + PGDIR_SIZE; + spin_unlock_irqrestore(&pgd_lock, flags); } #else /* CONFIG_X86_64 */ - /* - * Note that races in the updates of insync and start aren't - * problematic: insync can only get set bits added, and updates to - * start are only improving performance (without affecting correctness - * if undone). - */ - static DECLARE_BITMAP(insync, PTRS_PER_PGD); - static unsigned long start = VMALLOC_START & PGDIR_MASK; + unsigned long start = VMALLOC_START & PGDIR_MASK; unsigned long address; for (address = start; address <= VMALLOC_END; address += PGDIR_SIZE) { - if (!test_bit(pgd_index(address), insync)) { - const pgd_t *pgd_ref = pgd_offset_k(address); - unsigned long flags; - struct page *page; - - if (pgd_none(*pgd_ref)) - continue; - spin_lock_irqsave(&pgd_lock, flags); - list_for_each_entry(page, &pgd_list, lru) { - pgd_t *pgd; - pgd = (pgd_t *)page_address(page) + pgd_index(address); - if (pgd_none(*pgd)) - set_pgd(pgd, *pgd_ref); - else - BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref)); - } - spin_unlock_irqrestore(&pgd_lock, flags); - set_bit(pgd_index(address), insync); + const pgd_t *pgd_ref = pgd_offset_k(address); + unsigned long flags; + struct page *page; + + if (pgd_none(*pgd_ref)) + continue; + spin_lock_irqsave(&pgd_lock, flags); + list_for_each_entry(page, &pgd_list, lru) { + pgd_t *pgd; + pgd = (pgd_t *)page_address(page) + pgd_index(address); + if (pgd_none(*pgd)) + set_pgd(pgd, *pgd_ref); + else + BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref)); } - if (address == start) - start = address + PGDIR_SIZE; + spin_unlock_irqrestore(&pgd_lock, flags); } #endif } diff --git a/arch/x86/mm/init_32.c b/arch/x86/mm/init_32.c index ec30d10154b6..d37f29376b0c 100644 --- a/arch/x86/mm/init_32.c +++ b/arch/x86/mm/init_32.c @@ -50,6 +50,7 @@ unsigned int __VMALLOC_RESERVE = 128 << 20; +unsigned long max_low_pfn_mapped; unsigned long max_pfn_mapped; DEFINE_PER_CPU(struct mmu_gather, mmu_gathers); @@ -57,6 +58,27 @@ unsigned long highstart_pfn, highend_pfn; static noinline int do_test_wp_bit(void); + +static unsigned long __initdata table_start; +static unsigned long __meminitdata table_end; +static unsigned long __meminitdata table_top; + +static int __initdata after_init_bootmem; + +static __init void *alloc_low_page(unsigned long *phys) +{ + unsigned long pfn = table_end++; + void *adr; + + if (pfn >= table_top) + panic("alloc_low_page: ran out of memory"); + + adr = __va(pfn * PAGE_SIZE); + memset(adr, 0, PAGE_SIZE); + *phys = pfn * PAGE_SIZE; + return adr; +} + /* * Creates a middle page table and puts a pointer to it in the * given global directory entry. This only returns the gd entry @@ -68,9 +90,12 @@ static pmd_t * __init one_md_table_init(pgd_t *pgd) pmd_t *pmd_table; #ifdef CONFIG_X86_PAE + unsigned long phys; if (!(pgd_val(*pgd) & _PAGE_PRESENT)) { - pmd_table = (pmd_t *) alloc_bootmem_low_pages(PAGE_SIZE); - + if (after_init_bootmem) + pmd_table = (pmd_t *)alloc_bootmem_low_pages(PAGE_SIZE); + else + pmd_table = (pmd_t *)alloc_low_page(&phys); paravirt_alloc_pmd(&init_mm, __pa(pmd_table) >> PAGE_SHIFT); set_pgd(pgd, __pgd(__pa(pmd_table) | _PAGE_PRESENT)); pud = pud_offset(pgd, 0); @@ -92,12 +117,16 @@ static pte_t * __init one_page_table_init(pmd_t *pmd) if (!(pmd_val(*pmd) & _PAGE_PRESENT)) { pte_t *page_table = NULL; + if (after_init_bootmem) { #ifdef CONFIG_DEBUG_PAGEALLOC - page_table = (pte_t *) alloc_bootmem_pages(PAGE_SIZE); + page_table = (pte_t *) alloc_bootmem_pages(PAGE_SIZE); #endif - if (!page_table) { - page_table = + if (!page_table) + page_table = (pte_t *)alloc_bootmem_low_pages(PAGE_SIZE); + } else { + unsigned long phys; + page_table = (pte_t *)alloc_low_page(&phys); } paravirt_alloc_pte(&init_mm, __pa(page_table) >> PAGE_SHIFT); @@ -155,38 +184,44 @@ static inline int is_kernel_text(unsigned long addr) * of max_low_pfn pages, by creating page tables starting from address * PAGE_OFFSET: */ -static void __init kernel_physical_mapping_init(pgd_t *pgd_base) +static void __init kernel_physical_mapping_init(pgd_t *pgd_base, + unsigned long start_pfn, + unsigned long end_pfn, + int use_pse) { int pgd_idx, pmd_idx, pte_ofs; unsigned long pfn; pgd_t *pgd; pmd_t *pmd; pte_t *pte; + unsigned pages_2m = 0, pages_4k = 0; - pgd_idx = pgd_index(PAGE_OFFSET); - pgd = pgd_base + pgd_idx; - pfn = 0; + if (!cpu_has_pse) + use_pse = 0; + pfn = start_pfn; + pgd_idx = pgd_index((pfn<<PAGE_SHIFT) + PAGE_OFFSET); + pgd = pgd_base + pgd_idx; for (; pgd_idx < PTRS_PER_PGD; pgd++, pgd_idx++) { pmd = one_md_table_init(pgd); - if (pfn >= max_low_pfn) - continue; - for (pmd_idx = 0; - pmd_idx < PTRS_PER_PMD && pfn < max_low_pfn; + if (pfn >= end_pfn) + continue; +#ifdef CONFIG_X86_PAE + pmd_idx = pmd_index((pfn<<PAGE_SHIFT) + PAGE_OFFSET); + pmd += pmd_idx; +#else + pmd_idx = 0; +#endif + for (; pmd_idx < PTRS_PER_PMD && pfn < end_pfn; pmd++, pmd_idx++) { unsigned int addr = pfn * PAGE_SIZE + PAGE_OFFSET; /* * Map with big pages if possible, otherwise * create normal page tables: - * - * Don't use a large page for the first 2/4MB of memory - * because there are often fixed size MTRRs in there - * and overlapping MTRRs into large pages can cause - * slowdowns. */ - if (cpu_has_pse && !(pgd_idx == 0 && pmd_idx == 0)) { + if (use_pse) { unsigned int addr2; pgprot_t prot = PAGE_KERNEL_LARGE; @@ -197,34 +232,30 @@ static void __init kernel_physical_mapping_init(pgd_t *pgd_base) is_kernel_text(addr2)) prot = PAGE_KERNEL_LARGE_EXEC; + pages_2m++; set_pmd(pmd, pfn_pmd(pfn, prot)); pfn += PTRS_PER_PTE; - max_pfn_mapped = pfn; continue; } pte = one_page_table_init(pmd); - for (pte_ofs = 0; - pte_ofs < PTRS_PER_PTE && pfn < max_low_pfn; + pte_ofs = pte_index((pfn<<PAGE_SHIFT) + PAGE_OFFSET); + pte += pte_ofs; + for (; pte_ofs < PTRS_PER_PTE && pfn < end_pfn; pte++, pfn++, pte_ofs++, addr += PAGE_SIZE) { pgprot_t prot = PAGE_KERNEL; if (is_kernel_text(addr)) prot = PAGE_KERNEL_EXEC; + pages_4k++; set_pte(pte, pfn_pte(pfn, prot)); } - max_pfn_mapped = pfn; } } -} - -static inline int page_kills_ppro(unsigned long pagenr) -{ - if (pagenr >= 0x70000 && pagenr <= 0x7003F) - return 1; - return 0; + update_page_count(PG_LEVEL_2M, pages_2m); + update_page_count(PG_LEVEL_4K, pages_4k); } /* @@ -287,29 +318,62 @@ static void __init permanent_kmaps_init(pgd_t *pgd_base) pkmap_page_table = pte; } -void __init add_one_highpage_init(struct page *page, int pfn, int bad_ppro) +static void __init add_one_highpage_init(struct page *page, int pfn) { - if (page_is_ram(pfn) && !(bad_ppro && page_kills_ppro(pfn))) { - ClearPageReserved(page); - init_page_count(page); - __free_page(page); - totalhigh_pages++; - } else - SetPageReserved(page); + ClearPageReserved(page); + init_page_count(page); + __free_page(page); + totalhigh_pages++; } -#ifndef CONFIG_NUMA -static void __init set_highmem_pages_init(int bad_ppro) +struct add_highpages_data { + unsigned long start_pfn; + unsigned long end_pfn; +}; + +static int __init add_highpages_work_fn(unsigned long start_pfn, + unsigned long end_pfn, void *datax) { - int pfn; + int node_pfn; + struct page *page; + unsigned long final_start_pfn, final_end_pfn; + struct add_highpages_data *data; - for (pfn = highstart_pfn; pfn < highend_pfn; pfn++) { - /* - * Holes under sparsemem might not have no mem_map[]: - */ - if (pfn_valid(pfn)) - add_one_highpage_init(pfn_to_page(pfn), pfn, bad_ppro); + data = (struct add_highpages_data *)datax; + + final_start_pfn = max(start_pfn, data->start_pfn); + final_end_pfn = min(end_pfn, data->end_pfn); + if (final_start_pfn >= final_end_pfn) + return 0; + + for (node_pfn = final_start_pfn; node_pfn < final_end_pfn; + node_pfn++) { + if (!pfn_valid(node_pfn)) + continue; + page = pfn_to_page(node_pfn); + add_one_highpage_init(page, node_pfn); } + + return 0; + +} + +void __init add_highpages_with_active_regions(int nid, unsigned long start_pfn, + unsigned long end_pfn) +{ + struct add_highpages_data data; + + data.start_pfn = start_pfn; + data.end_pfn = end_pfn; + + work_with_active_regions(nid, add_highpages_work_fn, &data); +} + +#ifndef CONFIG_NUMA +static void __init set_highmem_pages_init(void) +{ + add_highpages_with_active_regions(0, highstart_pfn, highend_pfn); + totalram_pages += totalhigh_pages; } #endif /* !CONFIG_NUMA */ @@ -317,14 +381,9 @@ static void __init set_highmem_pages_init(int bad_ppro) #else # define kmap_init() do { } while (0) # define permanent_kmaps_init(pgd_base) do { } while (0) -# define set_highmem_pages_init(bad_ppro) do { } while (0) +# define set_highmem_pages_init() do { } while (0) #endif /* CONFIG_HIGHMEM */ -pteval_t __PAGE_KERNEL = _PAGE_KERNEL; -EXPORT_SYMBOL(__PAGE_KERNEL); - -pteval_t __PAGE_KERNEL_EXEC = _PAGE_KERNEL_EXEC; - void __init native_pagetable_setup_start(pgd_t *base) { unsigned long pfn, va; @@ -380,27 +439,10 @@ void __init native_pagetable_setup_done(pgd_t *base) * be partially populated, and so it avoids stomping on any existing * mappings. */ -static void __init pagetable_init(void) +static void __init early_ioremap_page_table_range_init(pgd_t *pgd_base) { - pgd_t *pgd_base = swapper_pg_dir; unsigned long vaddr, end; - paravirt_pagetable_setup_start(pgd_base); - - /* Enable PSE if available */ - if (cpu_has_pse) - set_in_cr4(X86_CR4_PSE); - - /* Enable PGE if available */ - if (cpu_has_pge) { - set_in_cr4(X86_CR4_PGE); - __PAGE_KERNEL |= _PAGE_GLOBAL; - __PAGE_KERNEL_EXEC |= _PAGE_GLOBAL; - } - - kernel_physical_mapping_init(pgd_base); - remap_numa_kva(); - /* * Fixed mappings, only the page table structure has to be * created - mappings will be set by set_fixmap(): @@ -410,6 +452,13 @@ static void __init pagetable_init(void) end = (FIXADDR_TOP + PMD_SIZE - 1) & PMD_MASK; page_table_range_init(vaddr, end, pgd_base); early_ioremap_reset(); +} + +static void __init pagetable_init(void) +{ + pgd_t *pgd_base = swapper_pg_dir; + + paravirt_pagetable_setup_start(pgd_base); permanent_kmaps_init(pgd_base); @@ -456,7 +505,7 @@ void zap_low_mappings(void) int nx_enabled; -pteval_t __supported_pte_mask __read_mostly = ~_PAGE_NX; +pteval_t __supported_pte_mask __read_mostly = ~(_PAGE_NX | _PAGE_GLOBAL); EXPORT_SYMBOL_GPL(__supported_pte_mask); #ifdef CONFIG_X86_PAE @@ -509,27 +558,319 @@ static void __init set_nx(void) } #endif +/* user-defined highmem size */ +static unsigned int highmem_pages = -1; + /* - * paging_init() sets up the page tables - note that the first 8MB are - * already mapped by head.S. - * - * This routines also unmaps the page at virtual kernel address 0, so - * that we can trap those pesky NULL-reference errors in the kernel. + * highmem=size forces highmem to be exactly 'size' bytes. + * This works even on boxes that have no highmem otherwise. + * This also works to reduce highmem size on bigger boxes. */ -void __init paging_init(void) +static int __init parse_highmem(char *arg) +{ + if (!arg) + return -EINVAL; + + highmem_pages = memparse(arg, &arg) >> PAGE_SHIFT; + return 0; +} +early_param("highmem", parse_highmem); + +/* + * Determine low and high memory ranges: + */ +void __init find_low_pfn_range(void) +{ + /* it could update max_pfn */ + + /* max_low_pfn is 0, we already have early_res support */ + + max_low_pfn = max_pfn; + if (max_low_pfn > MAXMEM_PFN) { + if (highmem_pages == -1) + highmem_pages = max_pfn - MAXMEM_PFN; + if (highmem_pages + MAXMEM_PFN < max_pfn) + max_pfn = MAXMEM_PFN + highmem_pages; + if (highmem_pages + MAXMEM_PFN > max_pfn) { + printk(KERN_WARNING "only %luMB highmem pages " + "available, ignoring highmem size of %uMB.\n", + pages_to_mb(max_pfn - MAXMEM_PFN), + pages_to_mb(highmem_pages)); + highmem_pages = 0; + } + max_low_pfn = MAXMEM_PFN; +#ifndef CONFIG_HIGHMEM + /* Maximum memory usable is what is directly addressable */ + printk(KERN_WARNING "Warning only %ldMB will be used.\n", + MAXMEM>>20); + if (max_pfn > MAX_NONPAE_PFN) + printk(KERN_WARNING + "Use a HIGHMEM64G enabled kernel.\n"); + else + printk(KERN_WARNING "Use a HIGHMEM enabled kernel.\n"); + max_pfn = MAXMEM_PFN; +#else /* !CONFIG_HIGHMEM */ +#ifndef CONFIG_HIGHMEM64G + if (max_pfn > MAX_NONPAE_PFN) { + max_pfn = MAX_NONPAE_PFN; + printk(KERN_WARNING "Warning only 4GB will be used." + "Use a HIGHMEM64G enabled kernel.\n"); + } +#endif /* !CONFIG_HIGHMEM64G */ +#endif /* !CONFIG_HIGHMEM */ + } else { + if (highmem_pages == -1) + highmem_pages = 0; +#ifdef CONFIG_HIGHMEM + if (highmem_pages >= max_pfn) { + printk(KERN_ERR "highmem size specified (%uMB) is " + "bigger than pages available (%luMB)!.\n", + pages_to_mb(highmem_pages), + pages_to_mb(max_pfn)); + highmem_pages = 0; + } + if (highmem_pages) { + if (max_low_pfn - highmem_pages < + 64*1024*1024/PAGE_SIZE){ + printk(KERN_ERR "highmem size %uMB results in " + "smaller than 64MB lowmem, ignoring it.\n" + , pages_to_mb(highmem_pages)); + highmem_pages = 0; + } + max_low_pfn -= highmem_pages; + } +#else + if (highmem_pages) + printk(KERN_ERR "ignoring highmem size on non-highmem" + " kernel!\n"); +#endif + } +} + +#ifndef CONFIG_NEED_MULTIPLE_NODES +void __init initmem_init(unsigned long start_pfn, + unsigned long end_pfn) { +#ifdef CONFIG_HIGHMEM + highstart_pfn = highend_pfn = max_pfn; + if (max_pfn > max_low_pfn) + highstart_pfn = max_low_pfn; + memory_present(0, 0, highend_pfn); + e820_register_active_regions(0, 0, highend_pfn); + printk(KERN_NOTICE "%ldMB HIGHMEM available.\n", + pages_to_mb(highend_pfn - highstart_pfn)); + num_physpages = highend_pfn; + high_memory = (void *) __va(highstart_pfn * PAGE_SIZE - 1) + 1; +#else + memory_present(0, 0, max_low_pfn); + e820_register_active_regions(0, 0, max_low_pfn); + num_physpages = max_low_pfn; + high_memory = (void *) __va(max_low_pfn * PAGE_SIZE - 1) + 1; +#endif +#ifdef CONFIG_FLATMEM + max_mapnr = num_physpages; +#endif + printk(KERN_NOTICE "%ldMB LOWMEM available.\n", + pages_to_mb(max_low_pfn)); + + setup_bootmem_allocator(); +} +#endif /* !CONFIG_NEED_MULTIPLE_NODES */ + +static void __init zone_sizes_init(void) +{ + unsigned long max_zone_pfns[MAX_NR_ZONES]; + memset(max_zone_pfns, 0, sizeof(max_zone_pfns)); + max_zone_pfns[ZONE_DMA] = + virt_to_phys((char *)MAX_DMA_ADDRESS) >> PAGE_SHIFT; + max_zone_pfns[ZONE_NORMAL] = max_low_pfn; +#ifdef CONFIG_HIGHMEM + max_zone_pfns[ZONE_HIGHMEM] = highend_pfn; +#endif + + free_area_init_nodes(max_zone_pfns); +} + +void __init setup_bootmem_allocator(void) +{ + int i; + unsigned long bootmap_size, bootmap; + /* + * Initialize the boot-time allocator (with low memory only): + */ + bootmap_size = bootmem_bootmap_pages(max_low_pfn)<<PAGE_SHIFT; + bootmap = find_e820_area(min_low_pfn<<PAGE_SHIFT, + max_pfn_mapped<<PAGE_SHIFT, bootmap_size, + PAGE_SIZE); + if (bootmap == -1L) + panic("Cannot find bootmem map of size %ld\n", bootmap_size); + reserve_early(bootmap, bootmap + bootmap_size, "BOOTMAP"); + + /* don't touch min_low_pfn */ + bootmap_size = init_bootmem_node(NODE_DATA(0), bootmap >> PAGE_SHIFT, + min_low_pfn, max_low_pfn); + printk(KERN_INFO " mapped low ram: 0 - %08lx\n", + max_pfn_mapped<<PAGE_SHIFT); + printk(KERN_INFO " low ram: %08lx - %08lx\n", + min_low_pfn<<PAGE_SHIFT, max_low_pfn<<PAGE_SHIFT); + printk(KERN_INFO " bootmap %08lx - %08lx\n", + bootmap, bootmap + bootmap_size); + for_each_online_node(i) + free_bootmem_with_active_regions(i, max_low_pfn); + early_res_to_bootmem(0, max_low_pfn<<PAGE_SHIFT); + + after_init_bootmem = 1; +} + +static void __init find_early_table_space(unsigned long end) +{ + unsigned long puds, pmds, ptes, tables, start; + + puds = (end + PUD_SIZE - 1) >> PUD_SHIFT; + tables = PAGE_ALIGN(puds * sizeof(pud_t)); + + pmds = (end + PMD_SIZE - 1) >> PMD_SHIFT; + tables += PAGE_ALIGN(pmds * sizeof(pmd_t)); + + if (cpu_has_pse) { + unsigned long extra; + + extra = end - ((end>>PMD_SHIFT) << PMD_SHIFT); + extra += PMD_SIZE; + ptes = (extra + PAGE_SIZE - 1) >> PAGE_SHIFT; + } else + ptes = (end + PAGE_SIZE - 1) >> PAGE_SHIFT; + + tables += PAGE_ALIGN(ptes * sizeof(pte_t)); + + /* for fixmap */ + tables += PAGE_SIZE * 2; + + /* + * RED-PEN putting page tables only on node 0 could + * cause a hotspot and fill up ZONE_DMA. The page tables + * need roughly 0.5KB per GB. + */ + start = 0x7000; + table_start = find_e820_area(start, max_pfn_mapped<<PAGE_SHIFT, + tables, PAGE_SIZE); + if (table_start == -1UL) + panic("Cannot find space for the kernel page tables"); + + table_start >>= PAGE_SHIFT; + table_end = table_start; + table_top = table_start + (tables>>PAGE_SHIFT); + + printk(KERN_DEBUG "kernel direct mapping tables up to %lx @ %lx-%lx\n", + end, table_start << PAGE_SHIFT, + (table_start << PAGE_SHIFT) + tables); +} + +unsigned long __init_refok init_memory_mapping(unsigned long start, + unsigned long end) +{ + pgd_t *pgd_base = swapper_pg_dir; + unsigned long start_pfn, end_pfn; + unsigned long big_page_start; + + /* + * Find space for the kernel direct mapping tables. + */ + if (!after_init_bootmem) + find_early_table_space(end); + #ifdef CONFIG_X86_PAE set_nx(); if (nx_enabled) printk(KERN_INFO "NX (Execute Disable) protection: active\n"); #endif - pagetable_init(); + + /* Enable PSE if available */ + if (cpu_has_pse) + set_in_cr4(X86_CR4_PSE); + + /* Enable PGE if available */ + if (cpu_has_pge) { + set_in_cr4(X86_CR4_PGE); + __supported_pte_mask |= _PAGE_GLOBAL; + } + + /* + * Don't use a large page for the first 2/4MB of memory + * because there are often fixed size MTRRs in there + * and overlapping MTRRs into large pages can cause + * slowdowns. + */ + big_page_start = PMD_SIZE; + + if (start < big_page_start) { + start_pfn = start >> PAGE_SHIFT; + end_pfn = min(big_page_start>>PAGE_SHIFT, end>>PAGE_SHIFT); + } else { + /* head is not big page alignment ? */ + start_pfn = start >> PAGE_SHIFT; + end_pfn = ((start + (PMD_SIZE - 1))>>PMD_SHIFT) + << (PMD_SHIFT - PAGE_SHIFT); + } + if (start_pfn < end_pfn) + kernel_physical_mapping_init(pgd_base, start_pfn, end_pfn, 0); + + /* big page range */ + start_pfn = ((start + (PMD_SIZE - 1))>>PMD_SHIFT) + << (PMD_SHIFT - PAGE_SHIFT); + if (start_pfn < (big_page_start >> PAGE_SHIFT)) + start_pfn = big_page_start >> PAGE_SHIFT; + end_pfn = (end>>PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT); + if (start_pfn < end_pfn) + kernel_physical_mapping_init(pgd_base, start_pfn, end_pfn, + cpu_has_pse); + + /* tail is not big page alignment ? */ + start_pfn = end_pfn; + if (start_pfn > (big_page_start>>PAGE_SHIFT)) { + end_pfn = end >> PAGE_SHIFT; + if (start_pfn < end_pfn) + kernel_physical_mapping_init(pgd_base, start_pfn, + end_pfn, 0); + } + + early_ioremap_page_table_range_init(pgd_base); load_cr3(swapper_pg_dir); __flush_tlb_all(); + if (!after_init_bootmem) + reserve_early(table_start << PAGE_SHIFT, + table_end << PAGE_SHIFT, "PGTABLE"); + + if (!after_init_bootmem) + early_memtest(start, end); + + return end >> PAGE_SHIFT; +} + + +/* + * paging_init() sets up the page tables - note that the first 8MB are + * already mapped by head.S. + * + * This routines also unmaps the page at virtual kernel address 0, so + * that we can trap those pesky NULL-reference errors in the kernel. + */ +void __init paging_init(void) +{ + pagetable_init(); + + __flush_tlb_all(); + kmap_init(); + + /* + * NOTE: at this point the bootmem allocator is fully available. + */ + sparse_init(); + zone_sizes_init(); } /* @@ -564,24 +905,11 @@ static struct kcore_list kcore_mem, kcore_vmalloc; void __init mem_init(void) { int codesize, reservedpages, datasize, initsize; - int tmp, bad_ppro; + int tmp; #ifdef CONFIG_FLATMEM BUG_ON(!mem_map); #endif - bad_ppro = ppro_with_ram_bug(); - -#ifdef CONFIG_HIGHMEM - /* check that fixmap and pkmap do not overlap */ - if (PKMAP_BASE + LAST_PKMAP*PAGE_SIZE >= FIXADDR_START) { - printk(KERN_ERR - "fixmap and kmap areas overlap - this will crash\n"); - printk(KERN_ERR "pkstart: %lxh pkend: %lxh fixstart %lxh\n", - PKMAP_BASE, PKMAP_BASE + LAST_PKMAP*PAGE_SIZE, - FIXADDR_START); - BUG(); - } -#endif /* this will put all low memory onto the freelists */ totalram_pages += free_all_bootmem(); @@ -593,7 +921,7 @@ void __init mem_init(void) if (page_is_ram(tmp) && PageReserved(pfn_to_page(tmp))) reservedpages++; - set_highmem_pages_init(bad_ppro); + set_highmem_pages_init(); codesize = (unsigned long) &_etext - (unsigned long) &_text; datasize = (unsigned long) &_edata - (unsigned long) &_etext; @@ -614,7 +942,6 @@ void __init mem_init(void) (unsigned long) (totalhigh_pages << (PAGE_SHIFT-10)) ); -#if 1 /* double-sanity-check paranoia */ printk(KERN_INFO "virtual kernel memory layout:\n" " fixmap : 0x%08lx - 0x%08lx (%4ld kB)\n" #ifdef CONFIG_HIGHMEM @@ -655,7 +982,6 @@ void __init mem_init(void) #endif BUG_ON(VMALLOC_START > VMALLOC_END); BUG_ON((unsigned long)high_memory > VMALLOC_START); -#endif /* double-sanity-check paranoia */ if (boot_cpu_data.wp_works_ok < 0) test_wp_bit(); @@ -710,6 +1036,8 @@ void mark_rodata_ro(void) unsigned long start = PFN_ALIGN(_text); unsigned long size = PFN_ALIGN(_etext) - start; +#ifndef CONFIG_DYNAMIC_FTRACE + /* Dynamic tracing modifies the kernel text section */ set_pages_ro(virt_to_page(start), size >> PAGE_SHIFT); printk(KERN_INFO "Write protecting the kernel text: %luk\n", size >> 10); @@ -722,6 +1050,8 @@ void mark_rodata_ro(void) printk(KERN_INFO "Testing CPA: write protecting again\n"); set_pages_ro(virt_to_page(start), size>>PAGE_SHIFT); #endif +#endif /* CONFIG_DYNAMIC_FTRACE */ + start += size; size = (unsigned long)__end_rodata - start; set_pages_ro(virt_to_page(start), size >> PAGE_SHIFT); @@ -784,3 +1114,9 @@ void free_initrd_mem(unsigned long start, unsigned long end) free_init_pages("initrd memory", start, end); } #endif + +int __init reserve_bootmem_generic(unsigned long phys, unsigned long len, + int flags) +{ + return reserve_bootmem(phys, len, flags); +} diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c index 156e6d7b0e32..ec37121f6709 100644 --- a/arch/x86/mm/init_64.c +++ b/arch/x86/mm/init_64.c @@ -18,6 +18,7 @@ #include <linux/swap.h> #include <linux/smp.h> #include <linux/init.h> +#include <linux/initrd.h> #include <linux/pagemap.h> #include <linux/bootmem.h> #include <linux/proc_fs.h> @@ -47,6 +48,14 @@ #include <asm/numa.h> #include <asm/cacheflush.h> +/* + * end_pfn only includes RAM, while max_pfn_mapped includes all e820 entries. + * The direct mapping extends to max_pfn_mapped, so that we can directly access + * apertures, ACPI and other tables without having to play with fixmaps. + */ +unsigned long max_low_pfn_mapped; +unsigned long max_pfn_mapped; + static unsigned long dma_reserve __initdata; DEFINE_PER_CPU(struct mmu_gather, mmu_gathers); @@ -135,26 +144,17 @@ static __init void *spp_getpage(void) return ptr; } -static void -set_pte_phys(unsigned long vaddr, unsigned long phys, pgprot_t prot) +void +set_pte_vaddr_pud(pud_t *pud_page, unsigned long vaddr, pte_t new_pte) { - pgd_t *pgd; pud_t *pud; pmd_t *pmd; - pte_t *pte, new_pte; - - pr_debug("set_pte_phys %lx to %lx\n", vaddr, phys); + pte_t *pte; - pgd = pgd_offset_k(vaddr); - if (pgd_none(*pgd)) { - printk(KERN_ERR - "PGD FIXMAP MISSING, it should be setup in head.S!\n"); - return; - } - pud = pud_offset(pgd, vaddr); + pud = pud_page + pud_index(vaddr); if (pud_none(*pud)) { pmd = (pmd_t *) spp_getpage(); - set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE | _PAGE_USER)); + pud_populate(&init_mm, pud, pmd); if (pmd != pmd_offset(pud, 0)) { printk(KERN_ERR "PAGETABLE BUG #01! %p <-> %p\n", pmd, pmd_offset(pud, 0)); @@ -164,13 +164,12 @@ set_pte_phys(unsigned long vaddr, unsigned long phys, pgprot_t prot) pmd = pmd_offset(pud, vaddr); if (pmd_none(*pmd)) { pte = (pte_t *) spp_getpage(); - set_pmd(pmd, __pmd(__pa(pte) | _KERNPG_TABLE | _PAGE_USER)); + pmd_populate_kernel(&init_mm, pmd, pte); if (pte != pte_offset_kernel(pmd, 0)) { printk(KERN_ERR "PAGETABLE BUG #02!\n"); return; } } - new_pte = pfn_pte(phys >> PAGE_SHIFT, prot); pte = pte_offset_kernel(pmd, vaddr); if (!pte_none(*pte) && pte_val(new_pte) && @@ -185,6 +184,64 @@ set_pte_phys(unsigned long vaddr, unsigned long phys, pgprot_t prot) __flush_tlb_one(vaddr); } +void +set_pte_vaddr(unsigned long vaddr, pte_t pteval) +{ + pgd_t *pgd; + pud_t *pud_page; + + pr_debug("set_pte_vaddr %lx to %lx\n", vaddr, native_pte_val(pteval)); + + pgd = pgd_offset_k(vaddr); + if (pgd_none(*pgd)) { + printk(KERN_ERR + "PGD FIXMAP MISSING, it should be setup in head.S!\n"); + return; + } + pud_page = (pud_t*)pgd_page_vaddr(*pgd); + set_pte_vaddr_pud(pud_page, vaddr, pteval); +} + +/* + * Create large page table mappings for a range of physical addresses. + */ +static void __init __init_extra_mapping(unsigned long phys, unsigned long size, + pgprot_t prot) +{ + pgd_t *pgd; + pud_t *pud; + pmd_t *pmd; + + BUG_ON((phys & ~PMD_MASK) || (size & ~PMD_MASK)); + for (; size; phys += PMD_SIZE, size -= PMD_SIZE) { + pgd = pgd_offset_k((unsigned long)__va(phys)); + if (pgd_none(*pgd)) { + pud = (pud_t *) spp_getpage(); + set_pgd(pgd, __pgd(__pa(pud) | _KERNPG_TABLE | + _PAGE_USER)); + } + pud = pud_offset(pgd, (unsigned long)__va(phys)); + if (pud_none(*pud)) { + pmd = (pmd_t *) spp_getpage(); + set_pud(pud, __pud(__pa(pmd) | _KERNPG_TABLE | + _PAGE_USER)); + } + pmd = pmd_offset(pud, phys); + BUG_ON(!pmd_none(*pmd)); + set_pmd(pmd, __pmd(phys | pgprot_val(prot))); + } +} + +void __init init_extra_mapping_wb(unsigned long phys, unsigned long size) +{ + __init_extra_mapping(phys, size, PAGE_KERNEL_LARGE); +} + +void __init init_extra_mapping_uc(unsigned long phys, unsigned long size) +{ + __init_extra_mapping(phys, size, PAGE_KERNEL_LARGE_NOCACHE); +} + /* * The head.S code sets up the kernel high mapping: * @@ -213,20 +270,9 @@ void __init cleanup_highmap(void) } } -/* NOTE: this is meant to be run only at boot */ -void __set_fixmap(enum fixed_addresses idx, unsigned long phys, pgprot_t prot) -{ - unsigned long address = __fix_to_virt(idx); - - if (idx >= __end_of_fixed_addresses) { - printk(KERN_ERR "Invalid __set_fixmap\n"); - return; - } - set_pte_phys(address, phys, prot); -} - static unsigned long __initdata table_start; static unsigned long __meminitdata table_end; +static unsigned long __meminitdata table_top; static __meminit void *alloc_low_page(unsigned long *phys) { @@ -240,7 +286,7 @@ static __meminit void *alloc_low_page(unsigned long *phys) return adr; } - if (pfn >= end_pfn) + if (pfn >= table_top) panic("alloc_low_page: ran out of memory"); adr = early_ioremap(pfn * PAGE_SIZE, PAGE_SIZE); @@ -257,65 +303,61 @@ static __meminit void unmap_low_page(void *adr) early_iounmap(adr, PAGE_SIZE); } -/* Must run before zap_low_mappings */ -__meminit void *early_ioremap(unsigned long addr, unsigned long size) +static unsigned long __meminit +phys_pte_init(pte_t *pte_page, unsigned long addr, unsigned long end) { - pmd_t *pmd, *last_pmd; - unsigned long vaddr; - int i, pmds; + unsigned pages = 0; + unsigned long last_map_addr = end; + int i; + + pte_t *pte = pte_page + pte_index(addr); - pmds = ((addr & ~PMD_MASK) + size + ~PMD_MASK) / PMD_SIZE; - vaddr = __START_KERNEL_map; - pmd = level2_kernel_pgt; - last_pmd = level2_kernel_pgt + PTRS_PER_PMD - 1; + for(i = pte_index(addr); i < PTRS_PER_PTE; i++, addr += PAGE_SIZE, pte++) { - for (; pmd <= last_pmd; pmd++, vaddr += PMD_SIZE) { - for (i = 0; i < pmds; i++) { - if (pmd_present(pmd[i])) - goto continue_outer_loop; + if (addr >= end) { + if (!after_bootmem) { + for(; i < PTRS_PER_PTE; i++, pte++) + set_pte(pte, __pte(0)); + } + break; } - vaddr += addr & ~PMD_MASK; - addr &= PMD_MASK; - for (i = 0; i < pmds; i++, addr += PMD_SIZE) - set_pmd(pmd+i, __pmd(addr | __PAGE_KERNEL_LARGE_EXEC)); - __flush_tlb_all(); + if (pte_val(*pte)) + continue; - return (void *)vaddr; -continue_outer_loop: - ; + if (0) + printk(" pte=%p addr=%lx pte=%016lx\n", + pte, addr, pfn_pte(addr >> PAGE_SHIFT, PAGE_KERNEL).pte); + set_pte(pte, pfn_pte(addr >> PAGE_SHIFT, PAGE_KERNEL)); + last_map_addr = (addr & PAGE_MASK) + PAGE_SIZE; + pages++; } - printk(KERN_ERR "early_ioremap(0x%lx, %lu) failed\n", addr, size); + update_page_count(PG_LEVEL_4K, pages); - return NULL; + return last_map_addr; } -/* - * To avoid virtual aliases later: - */ -__meminit void early_iounmap(void *addr, unsigned long size) +static unsigned long __meminit +phys_pte_update(pmd_t *pmd, unsigned long address, unsigned long end) { - unsigned long vaddr; - pmd_t *pmd; - int i, pmds; - - vaddr = (unsigned long)addr; - pmds = ((vaddr & ~PMD_MASK) + size + ~PMD_MASK) / PMD_SIZE; - pmd = level2_kernel_pgt + pmd_index(vaddr); - - for (i = 0; i < pmds; i++) - pmd_clear(pmd + i); + pte_t *pte = (pte_t *)pmd_page_vaddr(*pmd); - __flush_tlb_all(); + return phys_pte_init(pte, address, end); } static unsigned long __meminit -phys_pmd_init(pmd_t *pmd_page, unsigned long address, unsigned long end) +phys_pmd_init(pmd_t *pmd_page, unsigned long address, unsigned long end, + unsigned long page_size_mask) { + unsigned long pages = 0; + unsigned long last_map_addr = end; + int i = pmd_index(address); for (; i < PTRS_PER_PMD; i++, address += PMD_SIZE) { + unsigned long pte_phys; pmd_t *pmd = pmd_page + pmd_index(address); + pte_t *pte; if (address >= end) { if (!after_bootmem) { @@ -325,31 +367,50 @@ phys_pmd_init(pmd_t *pmd_page, unsigned long address, unsigned long end) break; } - if (pmd_val(*pmd)) + if (pmd_val(*pmd)) { + if (!pmd_large(*pmd)) + last_map_addr = phys_pte_update(pmd, address, + end); + continue; + } + + if (page_size_mask & (1<<PG_LEVEL_2M)) { + pages++; + set_pte((pte_t *)pmd, + pfn_pte(address >> PAGE_SHIFT, PAGE_KERNEL_LARGE)); + last_map_addr = (address & PMD_MASK) + PMD_SIZE; continue; + } + + pte = alloc_low_page(&pte_phys); + last_map_addr = phys_pte_init(pte, address, end); + unmap_low_page(pte); - set_pte((pte_t *)pmd, - pfn_pte(address >> PAGE_SHIFT, PAGE_KERNEL_LARGE)); + pmd_populate_kernel(&init_mm, pmd, __va(pte_phys)); } - return address; + update_page_count(PG_LEVEL_2M, pages); + return last_map_addr; } static unsigned long __meminit -phys_pmd_update(pud_t *pud, unsigned long address, unsigned long end) +phys_pmd_update(pud_t *pud, unsigned long address, unsigned long end, + unsigned long page_size_mask) { pmd_t *pmd = pmd_offset(pud, 0); unsigned long last_map_addr; spin_lock(&init_mm.page_table_lock); - last_map_addr = phys_pmd_init(pmd, address, end); + last_map_addr = phys_pmd_init(pmd, address, end, page_size_mask); spin_unlock(&init_mm.page_table_lock); __flush_tlb_all(); return last_map_addr; } static unsigned long __meminit -phys_pud_init(pud_t *pud_page, unsigned long addr, unsigned long end) +phys_pud_init(pud_t *pud_page, unsigned long addr, unsigned long end, + unsigned long page_size_mask) { + unsigned long pages = 0; unsigned long last_map_addr = end; int i = pud_index(addr); @@ -369,11 +430,13 @@ phys_pud_init(pud_t *pud_page, unsigned long addr, unsigned long end) if (pud_val(*pud)) { if (!pud_large(*pud)) - last_map_addr = phys_pmd_update(pud, addr, end); + last_map_addr = phys_pmd_update(pud, addr, end, + page_size_mask); continue; } - if (direct_gbpages) { + if (page_size_mask & (1<<PG_LEVEL_1G)) { + pages++; set_pte((pte_t *)pud, pfn_pte(addr >> PAGE_SHIFT, PAGE_KERNEL_LARGE)); last_map_addr = (addr & PUD_MASK) + PUD_SIZE; @@ -383,27 +446,50 @@ phys_pud_init(pud_t *pud_page, unsigned long addr, unsigned long end) pmd = alloc_low_page(&pmd_phys); spin_lock(&init_mm.page_table_lock); - set_pud(pud, __pud(pmd_phys | _KERNPG_TABLE)); - last_map_addr = phys_pmd_init(pmd, addr, end); + last_map_addr = phys_pmd_init(pmd, addr, end, page_size_mask); + unmap_low_page(pmd); + pud_populate(&init_mm, pud, __va(pmd_phys)); spin_unlock(&init_mm.page_table_lock); - unmap_low_page(pmd); } __flush_tlb_all(); + update_page_count(PG_LEVEL_1G, pages); - return last_map_addr >> PAGE_SHIFT; + return last_map_addr; +} + +static unsigned long __meminit +phys_pud_update(pgd_t *pgd, unsigned long addr, unsigned long end, + unsigned long page_size_mask) +{ + pud_t *pud; + + pud = (pud_t *)pgd_page_vaddr(*pgd); + + return phys_pud_init(pud, addr, end, page_size_mask); } static void __init find_early_table_space(unsigned long end) { - unsigned long puds, pmds, tables, start; + unsigned long puds, pmds, ptes, tables, start; puds = (end + PUD_SIZE - 1) >> PUD_SHIFT; tables = round_up(puds * sizeof(pud_t), PAGE_SIZE); - if (!direct_gbpages) { + if (direct_gbpages) { + unsigned long extra; + extra = end - ((end>>PUD_SHIFT) << PUD_SHIFT); + pmds = (extra + PMD_SIZE - 1) >> PMD_SHIFT; + } else pmds = (end + PMD_SIZE - 1) >> PMD_SHIFT; - tables += round_up(pmds * sizeof(pmd_t), PAGE_SIZE); - } + tables += round_up(pmds * sizeof(pmd_t), PAGE_SIZE); + + if (cpu_has_pse) { + unsigned long extra; + extra = end - ((end>>PMD_SHIFT) << PMD_SHIFT); + ptes = (extra + PAGE_SIZE - 1) >> PAGE_SHIFT; + } else + ptes = (end + PAGE_SIZE - 1) >> PAGE_SHIFT; + tables += round_up(ptes * sizeof(pte_t), PAGE_SIZE); /* * RED-PEN putting page tables only on node 0 could @@ -417,10 +503,10 @@ static void __init find_early_table_space(unsigned long end) table_start >>= PAGE_SHIFT; table_end = table_start; + table_top = table_start + (tables >> PAGE_SHIFT); - early_printk("kernel direct mapping tables up to %lx @ %lx-%lx\n", - end, table_start << PAGE_SHIFT, - (table_start << PAGE_SHIFT) + tables); + printk(KERN_DEBUG "kernel direct mapping tables up to %lx @ %lx-%lx\n", + end, table_start << PAGE_SHIFT, table_top << PAGE_SHIFT); } static void __init init_gbpages(void) @@ -431,125 +517,85 @@ static void __init init_gbpages(void) direct_gbpages = 0; } -#ifdef CONFIG_MEMTEST_BOOTPARAM - -static void __init memtest(unsigned long start_phys, unsigned long size, - unsigned pattern) -{ - unsigned long i; - unsigned long *start; - unsigned long start_bad; - unsigned long last_bad; - unsigned long val; - unsigned long start_phys_aligned; - unsigned long count; - unsigned long incr; - - switch (pattern) { - case 0: - val = 0UL; - break; - case 1: - val = -1UL; - break; - case 2: - val = 0x5555555555555555UL; - break; - case 3: - val = 0xaaaaaaaaaaaaaaaaUL; - break; - default: - return; - } +static unsigned long __init kernel_physical_mapping_init(unsigned long start, + unsigned long end, + unsigned long page_size_mask) +{ - incr = sizeof(unsigned long); - start_phys_aligned = ALIGN(start_phys, incr); - count = (size - (start_phys_aligned - start_phys))/incr; - start = __va(start_phys_aligned); - start_bad = 0; - last_bad = 0; - - for (i = 0; i < count; i++) - start[i] = val; - for (i = 0; i < count; i++, start++, start_phys_aligned += incr) { - if (*start != val) { - if (start_phys_aligned == last_bad + incr) { - last_bad += incr; - } else { - if (start_bad) { - printk(KERN_CONT "\n %016lx bad mem addr %016lx - %016lx reserved", - val, start_bad, last_bad + incr); - reserve_early(start_bad, last_bad - start_bad, "BAD RAM"); - } - start_bad = last_bad = start_phys_aligned; - } - } - } - if (start_bad) { - printk(KERN_CONT "\n %016lx bad mem addr %016lx - %016lx reserved", - val, start_bad, last_bad + incr); - reserve_early(start_bad, last_bad - start_bad, "BAD RAM"); - } + unsigned long next, last_map_addr = end; -} + start = (unsigned long)__va(start); + end = (unsigned long)__va(end); -static int memtest_pattern __initdata = CONFIG_MEMTEST_BOOTPARAM_VALUE; + for (; start < end; start = next) { + pgd_t *pgd = pgd_offset_k(start); + unsigned long pud_phys; + pud_t *pud; -static int __init parse_memtest(char *arg) -{ - if (arg) - memtest_pattern = simple_strtoul(arg, NULL, 0); - return 0; -} + next = (start + PGDIR_SIZE) & PGDIR_MASK; + if (next > end) + next = end; -early_param("memtest", parse_memtest); + if (pgd_val(*pgd)) { + last_map_addr = phys_pud_update(pgd, __pa(start), + __pa(end), page_size_mask); + continue; + } -static void __init early_memtest(unsigned long start, unsigned long end) -{ - u64 t_start, t_size; - unsigned pattern; + if (after_bootmem) + pud = pud_offset(pgd, start & PGDIR_MASK); + else + pud = alloc_low_page(&pud_phys); - if (!memtest_pattern) - return; + last_map_addr = phys_pud_init(pud, __pa(start), __pa(next), + page_size_mask); + unmap_low_page(pud); + pgd_populate(&init_mm, pgd_offset_k(start), + __va(pud_phys)); + } - printk(KERN_INFO "early_memtest: pattern num %d", memtest_pattern); - for (pattern = 0; pattern < memtest_pattern; pattern++) { - t_start = start; - t_size = 0; - while (t_start < end) { - t_start = find_e820_area_size(t_start, &t_size, 1); + return last_map_addr; +} - /* done ? */ - if (t_start >= end) - break; - if (t_start + t_size > end) - t_size = end - t_start; +struct map_range { + unsigned long start; + unsigned long end; + unsigned page_size_mask; +}; - printk(KERN_CONT "\n %016llx - %016llx pattern %d", - t_start, t_start + t_size, pattern); +#define NR_RANGE_MR 5 - memtest(t_start, t_size, pattern); +static int save_mr(struct map_range *mr, int nr_range, + unsigned long start_pfn, unsigned long end_pfn, + unsigned long page_size_mask) +{ - t_start += t_size; - } + if (start_pfn < end_pfn) { + if (nr_range >= NR_RANGE_MR) + panic("run out of range for init_memory_mapping\n"); + mr[nr_range].start = start_pfn<<PAGE_SHIFT; + mr[nr_range].end = end_pfn<<PAGE_SHIFT; + mr[nr_range].page_size_mask = page_size_mask; + nr_range++; } - printk(KERN_CONT "\n"); -} -#else -static void __init early_memtest(unsigned long start, unsigned long end) -{ + + return nr_range; } -#endif /* * Setup the direct mapping of the physical memory at PAGE_OFFSET. * This runs before bootmem is initialized and gets pages directly from * the physical memory. To access them they are temporarily mapped. */ -unsigned long __init_refok init_memory_mapping(unsigned long start, unsigned long end) +unsigned long __init_refok init_memory_mapping(unsigned long start, + unsigned long end) { - unsigned long next, last_map_addr = end; - unsigned long start_phys = start, end_phys = end; + unsigned long last_map_addr = 0; + unsigned long page_size_mask = 0; + unsigned long start_pfn, end_pfn; + + struct map_range mr[NR_RANGE_MR]; + int nr_range, i; printk(KERN_INFO "init_memory_mapping\n"); @@ -560,48 +606,115 @@ unsigned long __init_refok init_memory_mapping(unsigned long start, unsigned lon * memory mapped. Unfortunately this is done currently before the * nodes are discovered. */ - if (!after_bootmem) { + if (!after_bootmem) init_gbpages(); - find_early_table_space(end); + + if (direct_gbpages) + page_size_mask |= 1 << PG_LEVEL_1G; + if (cpu_has_pse) + page_size_mask |= 1 << PG_LEVEL_2M; + + memset(mr, 0, sizeof(mr)); + nr_range = 0; + + /* head if not big page alignment ?*/ + start_pfn = start >> PAGE_SHIFT; + end_pfn = ((start + (PMD_SIZE - 1)) >> PMD_SHIFT) + << (PMD_SHIFT - PAGE_SHIFT); + nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0); + + /* big page (2M) range*/ + start_pfn = ((start + (PMD_SIZE - 1))>>PMD_SHIFT) + << (PMD_SHIFT - PAGE_SHIFT); + end_pfn = ((start + (PUD_SIZE - 1))>>PUD_SHIFT) + << (PUD_SHIFT - PAGE_SHIFT); + if (end_pfn > ((end>>PUD_SHIFT)<<(PUD_SHIFT - PAGE_SHIFT))) + end_pfn = ((end>>PUD_SHIFT)<<(PUD_SHIFT - PAGE_SHIFT)); + nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, + page_size_mask & (1<<PG_LEVEL_2M)); + + /* big page (1G) range */ + start_pfn = end_pfn; + end_pfn = (end>>PUD_SHIFT) << (PUD_SHIFT - PAGE_SHIFT); + nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, + page_size_mask & + ((1<<PG_LEVEL_2M)|(1<<PG_LEVEL_1G))); + + /* tail is not big page (1G) alignment */ + start_pfn = end_pfn; + end_pfn = (end>>PMD_SHIFT) << (PMD_SHIFT - PAGE_SHIFT); + nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, + page_size_mask & (1<<PG_LEVEL_2M)); + + /* tail is not big page (2M) alignment */ + start_pfn = end_pfn; + end_pfn = end>>PAGE_SHIFT; + nr_range = save_mr(mr, nr_range, start_pfn, end_pfn, 0); + + /* try to merge same page size and continuous */ + for (i = 0; nr_range > 1 && i < nr_range - 1; i++) { + unsigned long old_start; + if (mr[i].end != mr[i+1].start || + mr[i].page_size_mask != mr[i+1].page_size_mask) + continue; + /* move it */ + old_start = mr[i].start; + memmove(&mr[i], &mr[i+1], + (nr_range - 1 - i) * sizeof (struct map_range)); + mr[i].start = old_start; + nr_range--; } - start = (unsigned long)__va(start); - end = (unsigned long)__va(end); + for (i = 0; i < nr_range; i++) + printk(KERN_DEBUG " %010lx - %010lx page %s\n", + mr[i].start, mr[i].end, + (mr[i].page_size_mask & (1<<PG_LEVEL_1G))?"1G":( + (mr[i].page_size_mask & (1<<PG_LEVEL_2M))?"2M":"4k")); - for (; start < end; start = next) { - pgd_t *pgd = pgd_offset_k(start); - unsigned long pud_phys; - pud_t *pud; - - if (after_bootmem) - pud = pud_offset(pgd, start & PGDIR_MASK); - else - pud = alloc_low_page(&pud_phys); + if (!after_bootmem) + find_early_table_space(end); - next = start + PGDIR_SIZE; - if (next > end) - next = end; - last_map_addr = phys_pud_init(pud, __pa(start), __pa(next)); - if (!after_bootmem) - set_pgd(pgd_offset_k(start), mk_kernel_pgd(pud_phys)); - unmap_low_page(pud); - } + for (i = 0; i < nr_range; i++) + last_map_addr = kernel_physical_mapping_init( + mr[i].start, mr[i].end, + mr[i].page_size_mask); if (!after_bootmem) mmu_cr4_features = read_cr4(); __flush_tlb_all(); - if (!after_bootmem) + if (!after_bootmem && table_end > table_start) reserve_early(table_start << PAGE_SHIFT, table_end << PAGE_SHIFT, "PGTABLE"); + printk(KERN_INFO "last_map_addr: %lx end: %lx\n", + last_map_addr, end); + if (!after_bootmem) - early_memtest(start_phys, end_phys); + early_memtest(start, end); - return last_map_addr; + return last_map_addr >> PAGE_SHIFT; } #ifndef CONFIG_NUMA +void __init initmem_init(unsigned long start_pfn, unsigned long end_pfn) +{ + unsigned long bootmap_size, bootmap; + + bootmap_size = bootmem_bootmap_pages(end_pfn)<<PAGE_SHIFT; + bootmap = find_e820_area(0, end_pfn<<PAGE_SHIFT, bootmap_size, + PAGE_SIZE); + if (bootmap == -1L) + panic("Cannot find bootmem map of size %ld\n", bootmap_size); + /* don't touch min_low_pfn */ + bootmap_size = init_bootmem_node(NODE_DATA(0), bootmap >> 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<<PAGE_SHIFT); + reserve_bootmem(bootmap, bootmap_size, BOOTMEM_DEFAULT); +} + void __init paging_init(void) { unsigned long max_zone_pfns[MAX_NR_ZONES]; @@ -609,9 +722,9 @@ void __init paging_init(void) memset(max_zone_pfns, 0, sizeof(max_zone_pfns)); max_zone_pfns[ZONE_DMA] = MAX_DMA_PFN; max_zone_pfns[ZONE_DMA32] = MAX_DMA32_PFN; - max_zone_pfns[ZONE_NORMAL] = end_pfn; + max_zone_pfns[ZONE_NORMAL] = max_pfn; - memory_present(0, 0, end_pfn); + memory_present(0, 0, max_pfn); sparse_init(); free_area_init_nodes(max_zone_pfns); } @@ -693,8 +806,8 @@ void __init mem_init(void) #else totalram_pages = free_all_bootmem(); #endif - reservedpages = end_pfn - totalram_pages - - absent_pages_in_range(0, end_pfn); + reservedpages = max_pfn - totalram_pages - + absent_pages_in_range(0, max_pfn); after_bootmem = 1; codesize = (unsigned long) &_etext - (unsigned long) &_text; @@ -713,7 +826,7 @@ void __init mem_init(void) printk(KERN_INFO "Memory: %luk/%luk available (%ldk kernel code, " "%ldk reserved, %ldk data, %ldk init)\n", (unsigned long) nr_free_pages() << (PAGE_SHIFT-10), - end_pfn << (PAGE_SHIFT-10), + max_pfn << (PAGE_SHIFT-10), codesize >> 10, reservedpages << (PAGE_SHIFT-10), datasize >> 10, @@ -766,6 +879,13 @@ EXPORT_SYMBOL_GPL(rodata_test_data); void mark_rodata_ro(void) { unsigned long start = PFN_ALIGN(_stext), end = PFN_ALIGN(__end_rodata); + unsigned long rodata_start = + ((unsigned long)__start_rodata + PAGE_SIZE - 1) & PAGE_MASK; + +#ifdef CONFIG_DYNAMIC_FTRACE + /* Dynamic tracing modifies the kernel text section */ + start = rodata_start; +#endif printk(KERN_INFO "Write protecting the kernel read-only data: %luk\n", (end - start) >> 10); @@ -775,8 +895,7 @@ void mark_rodata_ro(void) * The rodata section (but not the kernel text!) should also be * not-executable. */ - start = ((unsigned long)__start_rodata + PAGE_SIZE - 1) & PAGE_MASK; - set_memory_nx(start, (end - start) >> PAGE_SHIFT); + set_memory_nx(rodata_start, (end - rodata_start) >> PAGE_SHIFT); rodata_test(); @@ -798,24 +917,26 @@ void free_initrd_mem(unsigned long start, unsigned long end) } #endif -void __init reserve_bootmem_generic(unsigned long phys, unsigned len) +int __init reserve_bootmem_generic(unsigned long phys, unsigned long len, + int flags) { #ifdef CONFIG_NUMA int nid, next_nid; + int ret; #endif unsigned long pfn = phys >> PAGE_SHIFT; - if (pfn >= end_pfn) { + if (pfn >= max_pfn) { /* * This can happen with kdump kernels when accessing * firmware tables: */ if (pfn < max_pfn_mapped) - return; + return -EFAULT; - printk(KERN_ERR "reserve_bootmem: illegal reserve %lx %u\n", + printk(KERN_ERR "reserve_bootmem: illegal reserve %lx %lu\n", phys, len); - return; + return -EFAULT; } /* Should check here against the e820 map to avoid double free */ @@ -823,9 +944,13 @@ void __init reserve_bootmem_generic(unsigned long phys, unsigned len) nid = phys_to_nid(phys); next_nid = phys_to_nid(phys + len - 1); if (nid == next_nid) - reserve_bootmem_node(NODE_DATA(nid), phys, len, BOOTMEM_DEFAULT); + ret = reserve_bootmem_node(NODE_DATA(nid), phys, len, flags); else - reserve_bootmem(phys, len, BOOTMEM_DEFAULT); + ret = reserve_bootmem(phys, len, flags); + + if (ret != 0) + return ret; + #else reserve_bootmem(phys, len, BOOTMEM_DEFAULT); #endif @@ -834,6 +959,8 @@ void __init reserve_bootmem_generic(unsigned long phys, unsigned len) dma_reserve += len / PAGE_SIZE; set_dma_reserve(dma_reserve); } + + return 0; } int kern_addr_valid(unsigned long addr) @@ -938,7 +1065,7 @@ vmemmap_populate(struct page *start_page, unsigned long size, int node) pmd_t *pmd; for (; addr < end; addr = next) { - next = pmd_addr_end(addr, end); + void *p = NULL; pgd = vmemmap_pgd_populate(addr, node); if (!pgd) @@ -948,33 +1075,51 @@ vmemmap_populate(struct page *start_page, unsigned long size, int node) if (!pud) return -ENOMEM; - pmd = pmd_offset(pud, addr); - if (pmd_none(*pmd)) { - pte_t entry; - void *p; + if (!cpu_has_pse) { + next = (addr + PAGE_SIZE) & PAGE_MASK; + pmd = vmemmap_pmd_populate(pud, addr, node); + + if (!pmd) + return -ENOMEM; + + p = vmemmap_pte_populate(pmd, addr, node); - p = vmemmap_alloc_block(PMD_SIZE, node); if (!p) return -ENOMEM; - entry = pfn_pte(__pa(p) >> PAGE_SHIFT, - PAGE_KERNEL_LARGE); - set_pmd(pmd, __pmd(pte_val(entry))); - - /* check to see if we have contiguous blocks */ - if (p_end != p || node_start != node) { - if (p_start) - printk(KERN_DEBUG " [%lx-%lx] PMD -> [%p-%p] on node %d\n", - addr_start, addr_end-1, p_start, p_end-1, node_start); - addr_start = addr; - node_start = node; - p_start = p; - } - addr_end = addr + PMD_SIZE; - p_end = p + PMD_SIZE; + addr_end = addr + PAGE_SIZE; + p_end = p + PAGE_SIZE; } else { - vmemmap_verify((pte_t *)pmd, node, addr, next); + next = pmd_addr_end(addr, end); + + pmd = pmd_offset(pud, addr); + if (pmd_none(*pmd)) { + pte_t entry; + + p = vmemmap_alloc_block(PMD_SIZE, node); + if (!p) + return -ENOMEM; + + entry = pfn_pte(__pa(p) >> PAGE_SHIFT, + PAGE_KERNEL_LARGE); + set_pmd(pmd, __pmd(pte_val(entry))); + + /* check to see if we have contiguous blocks */ + if (p_end != p || node_start != node) { + if (p_start) + printk(KERN_DEBUG " [%lx-%lx] PMD -> [%p-%p] on node %d\n", + addr_start, addr_end-1, p_start, p_end-1, node_start); + addr_start = addr; + node_start = node; + p_start = p; + } + + addr_end = addr + PMD_SIZE; + p_end = p + PMD_SIZE; + } else + vmemmap_verify((pte_t *)pmd, node, addr, next); } + } return 0; } diff --git a/arch/x86/mm/ioremap.c b/arch/x86/mm/ioremap.c index 2b2bb3f9b683..24c1d3c30186 100644 --- a/arch/x86/mm/ioremap.c +++ b/arch/x86/mm/ioremap.c @@ -12,6 +12,7 @@ #include <linux/module.h> #include <linux/slab.h> #include <linux/vmalloc.h> +#include <linux/mmiotrace.h> #include <asm/cacheflush.h> #include <asm/e820.h> @@ -122,10 +123,13 @@ static void __iomem *__ioremap_caller(resource_size_t phys_addr, { unsigned long pfn, offset, vaddr; resource_size_t last_addr; + const resource_size_t unaligned_phys_addr = phys_addr; + const unsigned long unaligned_size = size; struct vm_struct *area; unsigned long new_prot_val; pgprot_t prot; int retval; + void __iomem *ret_addr; /* Don't allow wraparound or zero size */ last_addr = phys_addr + size - 1; @@ -142,7 +146,7 @@ static void __iomem *__ioremap_caller(resource_size_t phys_addr, /* * Don't remap the low PCI/ISA area, it's always mapped.. */ - if (phys_addr >= ISA_START_ADDRESS && last_addr < ISA_END_ADDRESS) + if (is_ISA_range(phys_addr, last_addr)) return (__force void __iomem *)phys_to_virt(phys_addr); /* @@ -233,7 +237,10 @@ static void __iomem *__ioremap_caller(resource_size_t phys_addr, return NULL; } - return (void __iomem *) (vaddr + offset); + ret_addr = (void __iomem *) (vaddr + offset); + mmiotrace_ioremap(unaligned_phys_addr, unaligned_size, ret_addr); + + return ret_addr; } /** @@ -261,7 +268,7 @@ void __iomem *ioremap_nocache(resource_size_t phys_addr, unsigned long size) { /* * Ideally, this should be: - * pat_wc_enabled ? _PAGE_CACHE_UC : _PAGE_CACHE_UC_MINUS; + * pat_enabled ? _PAGE_CACHE_UC : _PAGE_CACHE_UC_MINUS; * * Till we fix all X drivers to use ioremap_wc(), we will use * UC MINUS. @@ -285,7 +292,7 @@ EXPORT_SYMBOL(ioremap_nocache); */ void __iomem *ioremap_wc(unsigned long phys_addr, unsigned long size) { - if (pat_wc_enabled) + if (pat_enabled) return __ioremap_caller(phys_addr, size, _PAGE_CACHE_WC, __builtin_return_address(0)); else @@ -300,6 +307,29 @@ void __iomem *ioremap_cache(resource_size_t phys_addr, unsigned long size) } EXPORT_SYMBOL(ioremap_cache); +static void __iomem *ioremap_default(resource_size_t phys_addr, + unsigned long size) +{ + unsigned long flags; + void *ret; + int err; + + /* + * - WB for WB-able memory and no other conflicting mappings + * - UC_MINUS for non-WB-able memory with no other conflicting mappings + * - Inherit from confliting mappings otherwise + */ + err = reserve_memtype(phys_addr, phys_addr + size, -1, &flags); + if (err < 0) + return NULL; + + ret = (void *) __ioremap_caller(phys_addr, size, flags, + __builtin_return_address(0)); + + free_memtype(phys_addr, phys_addr + size); + return (void __iomem *)ret; +} + /** * iounmap - Free a IO remapping * @addr: virtual address from ioremap_* @@ -318,13 +348,15 @@ void iounmap(volatile void __iomem *addr) * vm_area and by simply returning an address into the kernel mapping * of ISA space. So handle that here. */ - if (addr >= phys_to_virt(ISA_START_ADDRESS) && - addr < phys_to_virt(ISA_END_ADDRESS)) + if ((void __force *)addr >= phys_to_virt(ISA_START_ADDRESS) && + (void __force *)addr < phys_to_virt(ISA_END_ADDRESS)) return; addr = (volatile void __iomem *) (PAGE_MASK & (unsigned long __force)addr); + mmiotrace_iounmap(addr); + /* Use the vm area unlocked, assuming the caller ensures there isn't another iounmap for the same address in parallel. Reuse of the virtual address is prevented by @@ -332,7 +364,7 @@ void iounmap(volatile void __iomem *addr) cpa takes care of the direct mappings. */ read_lock(&vmlist_lock); for (p = vmlist; p; p = p->next) { - if (p->addr == addr) + if (p->addr == (void __force *)addr) break; } read_unlock(&vmlist_lock); @@ -346,7 +378,7 @@ void iounmap(volatile void __iomem *addr) free_memtype(p->phys_addr, p->phys_addr + get_vm_area_size(p)); /* Finally remove it */ - o = remove_vm_area((void *)addr); + o = remove_vm_area((void __force *)addr); BUG_ON(p != o || o == NULL); kfree(p); } @@ -365,7 +397,7 @@ void *xlate_dev_mem_ptr(unsigned long phys) if (page_is_ram(start >> PAGE_SHIFT)) return __va(phys); - addr = (void *)ioremap(start, PAGE_SIZE); + addr = (void __force *)ioremap_default(start, PAGE_SIZE); if (addr) addr = (void *)((unsigned long)addr | (phys & ~PAGE_MASK)); @@ -381,8 +413,6 @@ void unxlate_dev_mem_ptr(unsigned long phys, void *addr) return; } -#ifdef CONFIG_X86_32 - int __initdata early_ioremap_debug; static int __init early_ioremap_debug_setup(char *str) @@ -394,8 +424,7 @@ static int __init early_ioremap_debug_setup(char *str) early_param("early_ioremap_debug", early_ioremap_debug_setup); static __initdata int after_paging_init; -static pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)] - __section(.bss.page_aligned); +static pte_t bm_pte[PAGE_SIZE/sizeof(pte_t)] __page_aligned_bss; static inline pmd_t * __init early_ioremap_pmd(unsigned long addr) { @@ -484,10 +513,11 @@ static void __init __early_set_fixmap(enum fixed_addresses idx, return; } pte = early_ioremap_pte(addr); + if (pgprot_val(flags)) set_pte(pte, pfn_pte(phys >> PAGE_SHIFT, flags)); else - pte_clear(NULL, addr, pte); + pte_clear(&init_mm, addr, pte); __flush_tlb_one(addr); } @@ -625,5 +655,3 @@ void __this_fixmap_does_not_exist(void) { WARN_ON(1); } - -#endif /* CONFIG_X86_32 */ diff --git a/arch/x86/mm/k8topology_64.c b/arch/x86/mm/k8topology_64.c index 1f476e477844..41f1b5c00a1d 100644 --- a/arch/x86/mm/k8topology_64.c +++ b/arch/x86/mm/k8topology_64.c @@ -22,6 +22,7 @@ #include <asm/numa.h> #include <asm/mpspec.h> #include <asm/apic.h> +#include <asm/k8.h> static __init int find_northbridge(void) { @@ -56,34 +57,33 @@ static __init void early_get_boot_cpu_id(void) /* * Find possible boot-time SMP configuration: */ +#ifdef CONFIG_X86_MPPARSE early_find_smp_config(); +#endif #ifdef CONFIG_ACPI /* * Read APIC information from ACPI tables. */ early_acpi_boot_init(); #endif +#ifdef CONFIG_X86_MPPARSE /* * get boot-time SMP configuration: */ if (smp_found_config) early_get_smp_config(); +#endif early_init_lapic_mapping(); } int __init k8_scan_nodes(unsigned long start, unsigned long end) { + unsigned numnodes, cores, bits, apicid_base; unsigned long prevbase; struct bootnode nodes[8]; - int nodeid, i, nb; unsigned char nodeids[8]; - int found = 0; - u32 reg; - unsigned numnodes; - unsigned cores; - unsigned bits; - int j; - unsigned apicid_base; + int i, j, nb, found = 0; + u32 nodeid, reg; if (!early_pci_allowed()) return -1; @@ -105,7 +105,6 @@ int __init k8_scan_nodes(unsigned long start, unsigned long end) prevbase = 0; for (i = 0; i < 8; i++) { unsigned long base, limit; - u32 nodeid; base = read_pci_config(0, nb, 1, 0x40 + i*8); limit = read_pci_config(0, nb, 1, 0x44 + i*8); @@ -144,8 +143,8 @@ int __init k8_scan_nodes(unsigned long start, unsigned long end) limit |= (1<<24)-1; limit++; - if (limit > end_pfn << PAGE_SHIFT) - limit = end_pfn << PAGE_SHIFT; + if (limit > max_pfn << PAGE_SHIFT) + limit = max_pfn << PAGE_SHIFT; if (limit <= base) continue; diff --git a/arch/x86/mm/kmmio.c b/arch/x86/mm/kmmio.c new file mode 100644 index 000000000000..93d82038af4b --- /dev/null +++ b/arch/x86/mm/kmmio.c @@ -0,0 +1,510 @@ +/* Support for MMIO probes. + * Benfit many code from kprobes + * (C) 2002 Louis Zhuang <louis.zhuang@intel.com>. + * 2007 Alexander Eichner + * 2008 Pekka Paalanen <pq@iki.fi> + */ + +#include <linux/list.h> +#include <linux/rculist.h> +#include <linux/spinlock.h> +#include <linux/hash.h> +#include <linux/init.h> +#include <linux/module.h> +#include <linux/kernel.h> +#include <linux/uaccess.h> +#include <linux/ptrace.h> +#include <linux/preempt.h> +#include <linux/percpu.h> +#include <linux/kdebug.h> +#include <linux/mutex.h> +#include <linux/io.h> +#include <asm/cacheflush.h> +#include <asm/tlbflush.h> +#include <linux/errno.h> +#include <asm/debugreg.h> +#include <linux/mmiotrace.h> + +#define KMMIO_PAGE_HASH_BITS 4 +#define KMMIO_PAGE_TABLE_SIZE (1 << KMMIO_PAGE_HASH_BITS) + +struct kmmio_fault_page { + struct list_head list; + struct kmmio_fault_page *release_next; + unsigned long page; /* location of the fault page */ + + /* + * Number of times this page has been registered as a part + * of a probe. If zero, page is disarmed and this may be freed. + * Used only by writers (RCU). + */ + int count; +}; + +struct kmmio_delayed_release { + struct rcu_head rcu; + struct kmmio_fault_page *release_list; +}; + +struct kmmio_context { + struct kmmio_fault_page *fpage; + struct kmmio_probe *probe; + unsigned long saved_flags; + unsigned long addr; + int active; +}; + +static DEFINE_SPINLOCK(kmmio_lock); + +/* Protected by kmmio_lock */ +unsigned int kmmio_count; + +/* Read-protected by RCU, write-protected by kmmio_lock. */ +static struct list_head kmmio_page_table[KMMIO_PAGE_TABLE_SIZE]; +static LIST_HEAD(kmmio_probes); + +static struct list_head *kmmio_page_list(unsigned long page) +{ + return &kmmio_page_table[hash_long(page, KMMIO_PAGE_HASH_BITS)]; +} + +/* Accessed per-cpu */ +static DEFINE_PER_CPU(struct kmmio_context, kmmio_ctx); + +/* + * this is basically a dynamic stabbing problem: + * Could use the existing prio tree code or + * Possible better implementations: + * The Interval Skip List: A Data Structure for Finding All Intervals That + * Overlap a Point (might be simple) + * Space Efficient Dynamic Stabbing with Fast Queries - Mikkel Thorup + */ +/* Get the kmmio at this addr (if any). You must be holding RCU read lock. */ +static struct kmmio_probe *get_kmmio_probe(unsigned long addr) +{ + struct kmmio_probe *p; + list_for_each_entry_rcu(p, &kmmio_probes, list) { + if (addr >= p->addr && addr <= (p->addr + p->len)) + return p; + } + return NULL; +} + +/* You must be holding RCU read lock. */ +static struct kmmio_fault_page *get_kmmio_fault_page(unsigned long page) +{ + struct list_head *head; + struct kmmio_fault_page *p; + + page &= PAGE_MASK; + head = kmmio_page_list(page); + list_for_each_entry_rcu(p, head, list) { + if (p->page == page) + return p; + } + return NULL; +} + +static void set_page_present(unsigned long addr, bool present, + unsigned int *pglevel) +{ + pteval_t pteval; + pmdval_t pmdval; + unsigned int level; + pmd_t *pmd; + pte_t *pte = lookup_address(addr, &level); + + if (!pte) { + pr_err("kmmio: no pte for page 0x%08lx\n", addr); + return; + } + + if (pglevel) + *pglevel = level; + + switch (level) { + case PG_LEVEL_2M: + pmd = (pmd_t *)pte; + pmdval = pmd_val(*pmd) & ~_PAGE_PRESENT; + if (present) + pmdval |= _PAGE_PRESENT; + set_pmd(pmd, __pmd(pmdval)); + break; + + case PG_LEVEL_4K: + pteval = pte_val(*pte) & ~_PAGE_PRESENT; + if (present) + pteval |= _PAGE_PRESENT; + set_pte_atomic(pte, __pte(pteval)); + break; + + default: + pr_err("kmmio: unexpected page level 0x%x.\n", level); + return; + } + + __flush_tlb_one(addr); +} + +/** Mark the given page as not present. Access to it will trigger a fault. */ +static void arm_kmmio_fault_page(unsigned long page, unsigned int *pglevel) +{ + set_page_present(page & PAGE_MASK, false, pglevel); +} + +/** Mark the given page as present. */ +static void disarm_kmmio_fault_page(unsigned long page, unsigned int *pglevel) +{ + set_page_present(page & PAGE_MASK, true, pglevel); +} + +/* + * This is being called from do_page_fault(). + * + * We may be in an interrupt or a critical section. Also prefecthing may + * trigger a page fault. We may be in the middle of process switch. + * We cannot take any locks, because we could be executing especially + * within a kmmio critical section. + * + * Local interrupts are disabled, so preemption cannot happen. + * Do not enable interrupts, do not sleep, and watch out for other CPUs. + */ +/* + * Interrupts are disabled on entry as trap3 is an interrupt gate + * and they remain disabled thorough out this function. + */ +int kmmio_handler(struct pt_regs *regs, unsigned long addr) +{ + struct kmmio_context *ctx; + struct kmmio_fault_page *faultpage; + int ret = 0; /* default to fault not handled */ + + /* + * Preemption is now disabled to prevent process switch during + * single stepping. We can only handle one active kmmio trace + * per cpu, so ensure that we finish it before something else + * gets to run. We also hold the RCU read lock over single + * stepping to avoid looking up the probe and kmmio_fault_page + * again. + */ + preempt_disable(); + rcu_read_lock(); + + faultpage = get_kmmio_fault_page(addr); + if (!faultpage) { + /* + * Either this page fault is not caused by kmmio, or + * another CPU just pulled the kmmio probe from under + * our feet. The latter case should not be possible. + */ + goto no_kmmio; + } + + ctx = &get_cpu_var(kmmio_ctx); + if (ctx->active) { + disarm_kmmio_fault_page(faultpage->page, NULL); + if (addr == ctx->addr) { + /* + * On SMP we sometimes get recursive probe hits on the + * same address. Context is already saved, fall out. + */ + pr_debug("kmmio: duplicate probe hit on CPU %d, for " + "address 0x%08lx.\n", + smp_processor_id(), addr); + ret = 1; + goto no_kmmio_ctx; + } + /* + * Prevent overwriting already in-flight context. + * This should not happen, let's hope disarming at least + * prevents a panic. + */ + pr_emerg("kmmio: recursive probe hit on CPU %d, " + "for address 0x%08lx. Ignoring.\n", + smp_processor_id(), addr); + pr_emerg("kmmio: previous hit was at 0x%08lx.\n", + ctx->addr); + goto no_kmmio_ctx; + } + ctx->active++; + + ctx->fpage = faultpage; + ctx->probe = get_kmmio_probe(addr); + ctx->saved_flags = (regs->flags & (X86_EFLAGS_TF | X86_EFLAGS_IF)); + ctx->addr = addr; + + if (ctx->probe && ctx->probe->pre_handler) + ctx->probe->pre_handler(ctx->probe, regs, addr); + + /* + * Enable single-stepping and disable interrupts for the faulting + * context. Local interrupts must not get enabled during stepping. + */ + regs->flags |= X86_EFLAGS_TF; + regs->flags &= ~X86_EFLAGS_IF; + + /* Now we set present bit in PTE and single step. */ + disarm_kmmio_fault_page(ctx->fpage->page, NULL); + + /* + * If another cpu accesses the same page while we are stepping, + * the access will not be caught. It will simply succeed and the + * only downside is we lose the event. If this becomes a problem, + * the user should drop to single cpu before tracing. + */ + + put_cpu_var(kmmio_ctx); + return 1; /* fault handled */ + +no_kmmio_ctx: + put_cpu_var(kmmio_ctx); +no_kmmio: + rcu_read_unlock(); + preempt_enable_no_resched(); + return ret; +} + +/* + * Interrupts are disabled on entry as trap1 is an interrupt gate + * and they remain disabled thorough out this function. + * This must always get called as the pair to kmmio_handler(). + */ +static int post_kmmio_handler(unsigned long condition, struct pt_regs *regs) +{ + int ret = 0; + struct kmmio_context *ctx = &get_cpu_var(kmmio_ctx); + + if (!ctx->active) { + pr_debug("kmmio: spurious debug trap on CPU %d.\n", + smp_processor_id()); + goto out; + } + + if (ctx->probe && ctx->probe->post_handler) + ctx->probe->post_handler(ctx->probe, condition, regs); + + arm_kmmio_fault_page(ctx->fpage->page, NULL); + + regs->flags &= ~X86_EFLAGS_TF; + regs->flags |= ctx->saved_flags; + + /* These were acquired in kmmio_handler(). */ + ctx->active--; + BUG_ON(ctx->active); + rcu_read_unlock(); + preempt_enable_no_resched(); + + /* + * if somebody else is singlestepping across a probe point, flags + * will have TF set, in which case, continue the remaining processing + * of do_debug, as if this is not a probe hit. + */ + if (!(regs->flags & X86_EFLAGS_TF)) + ret = 1; +out: + put_cpu_var(kmmio_ctx); + return ret; +} + +/* You must be holding kmmio_lock. */ +static int add_kmmio_fault_page(unsigned long page) +{ + struct kmmio_fault_page *f; + + page &= PAGE_MASK; + f = get_kmmio_fault_page(page); + if (f) { + if (!f->count) + arm_kmmio_fault_page(f->page, NULL); + f->count++; + return 0; + } + + f = kmalloc(sizeof(*f), GFP_ATOMIC); + if (!f) + return -1; + + f->count = 1; + f->page = page; + list_add_rcu(&f->list, kmmio_page_list(f->page)); + + arm_kmmio_fault_page(f->page, NULL); + + return 0; +} + +/* You must be holding kmmio_lock. */ +static void release_kmmio_fault_page(unsigned long page, + struct kmmio_fault_page **release_list) +{ + struct kmmio_fault_page *f; + + page &= PAGE_MASK; + f = get_kmmio_fault_page(page); + if (!f) + return; + + f->count--; + BUG_ON(f->count < 0); + if (!f->count) { + disarm_kmmio_fault_page(f->page, NULL); + f->release_next = *release_list; + *release_list = f; + } +} + +/* + * With page-unaligned ioremaps, one or two armed pages may contain + * addresses from outside the intended mapping. Events for these addresses + * are currently silently dropped. The events may result only from programming + * mistakes by accessing addresses before the beginning or past the end of a + * mapping. + */ +int register_kmmio_probe(struct kmmio_probe *p) +{ + unsigned long flags; + int ret = 0; + unsigned long size = 0; + const unsigned long size_lim = p->len + (p->addr & ~PAGE_MASK); + + spin_lock_irqsave(&kmmio_lock, flags); + if (get_kmmio_probe(p->addr)) { + ret = -EEXIST; + goto out; + } + kmmio_count++; + list_add_rcu(&p->list, &kmmio_probes); + while (size < size_lim) { + if (add_kmmio_fault_page(p->addr + size)) + pr_err("kmmio: Unable to set page fault.\n"); + size += PAGE_SIZE; + } +out: + spin_unlock_irqrestore(&kmmio_lock, flags); + /* + * XXX: What should I do here? + * Here was a call to global_flush_tlb(), but it does not exist + * anymore. It seems it's not needed after all. + */ + return ret; +} +EXPORT_SYMBOL(register_kmmio_probe); + +static void rcu_free_kmmio_fault_pages(struct rcu_head *head) +{ + struct kmmio_delayed_release *dr = container_of( + head, + struct kmmio_delayed_release, + rcu); + struct kmmio_fault_page *p = dr->release_list; + while (p) { + struct kmmio_fault_page *next = p->release_next; + BUG_ON(p->count); + kfree(p); + p = next; + } + kfree(dr); +} + +static void remove_kmmio_fault_pages(struct rcu_head *head) +{ + struct kmmio_delayed_release *dr = container_of( + head, + struct kmmio_delayed_release, + rcu); + struct kmmio_fault_page *p = dr->release_list; + struct kmmio_fault_page **prevp = &dr->release_list; + unsigned long flags; + spin_lock_irqsave(&kmmio_lock, flags); + while (p) { + if (!p->count) + list_del_rcu(&p->list); + else + *prevp = p->release_next; + prevp = &p->release_next; + p = p->release_next; + } + spin_unlock_irqrestore(&kmmio_lock, flags); + /* This is the real RCU destroy call. */ + call_rcu(&dr->rcu, rcu_free_kmmio_fault_pages); +} + +/* + * Remove a kmmio probe. You have to synchronize_rcu() before you can be + * sure that the callbacks will not be called anymore. Only after that + * you may actually release your struct kmmio_probe. + * + * Unregistering a kmmio fault page has three steps: + * 1. release_kmmio_fault_page() + * Disarm the page, wait a grace period to let all faults finish. + * 2. remove_kmmio_fault_pages() + * Remove the pages from kmmio_page_table. + * 3. rcu_free_kmmio_fault_pages() + * Actally free the kmmio_fault_page structs as with RCU. + */ +void unregister_kmmio_probe(struct kmmio_probe *p) +{ + unsigned long flags; + unsigned long size = 0; + const unsigned long size_lim = p->len + (p->addr & ~PAGE_MASK); + struct kmmio_fault_page *release_list = NULL; + struct kmmio_delayed_release *drelease; + + spin_lock_irqsave(&kmmio_lock, flags); + while (size < size_lim) { + release_kmmio_fault_page(p->addr + size, &release_list); + size += PAGE_SIZE; + } + list_del_rcu(&p->list); + kmmio_count--; + spin_unlock_irqrestore(&kmmio_lock, flags); + + drelease = kmalloc(sizeof(*drelease), GFP_ATOMIC); + if (!drelease) { + pr_crit("kmmio: leaking kmmio_fault_page objects.\n"); + return; + } + drelease->release_list = release_list; + + /* + * This is not really RCU here. We have just disarmed a set of + * pages so that they cannot trigger page faults anymore. However, + * we cannot remove the pages from kmmio_page_table, + * because a probe hit might be in flight on another CPU. The + * pages are collected into a list, and they will be removed from + * kmmio_page_table when it is certain that no probe hit related to + * these pages can be in flight. RCU grace period sounds like a + * good choice. + * + * If we removed the pages too early, kmmio page fault handler might + * not find the respective kmmio_fault_page and determine it's not + * a kmmio fault, when it actually is. This would lead to madness. + */ + call_rcu(&drelease->rcu, remove_kmmio_fault_pages); +} +EXPORT_SYMBOL(unregister_kmmio_probe); + +static int kmmio_die_notifier(struct notifier_block *nb, unsigned long val, + void *args) +{ + struct die_args *arg = args; + + if (val == DIE_DEBUG && (arg->err & DR_STEP)) + if (post_kmmio_handler(arg->err, arg->regs) == 1) + return NOTIFY_STOP; + + return NOTIFY_DONE; +} + +static struct notifier_block nb_die = { + .notifier_call = kmmio_die_notifier +}; + +static int __init init_kmmio(void) +{ + int i; + for (i = 0; i < KMMIO_PAGE_TABLE_SIZE; i++) + INIT_LIST_HEAD(&kmmio_page_table[i]); + return register_die_notifier(&nb_die); +} +fs_initcall(init_kmmio); /* should be before device_initcall() */ diff --git a/arch/x86/mm/memtest.c b/arch/x86/mm/memtest.c new file mode 100644 index 000000000000..672e17f8262a --- /dev/null +++ b/arch/x86/mm/memtest.c @@ -0,0 +1,123 @@ +#include <linux/kernel.h> +#include <linux/errno.h> +#include <linux/string.h> +#include <linux/types.h> +#include <linux/mm.h> +#include <linux/smp.h> +#include <linux/init.h> +#include <linux/pfn.h> + +#include <asm/e820.h> + +static void __init memtest(unsigned long start_phys, unsigned long size, + unsigned pattern) +{ + unsigned long i; + unsigned long *start; + unsigned long start_bad; + unsigned long last_bad; + unsigned long val; + unsigned long start_phys_aligned; + unsigned long count; + unsigned long incr; + + switch (pattern) { + case 0: + val = 0UL; + break; + case 1: + val = -1UL; + break; + case 2: +#ifdef CONFIG_X86_64 + val = 0x5555555555555555UL; +#else + val = 0x55555555UL; +#endif + break; + case 3: +#ifdef CONFIG_X86_64 + val = 0xaaaaaaaaaaaaaaaaUL; +#else + val = 0xaaaaaaaaUL; +#endif + break; + default: + return; + } + + incr = sizeof(unsigned long); + start_phys_aligned = ALIGN(start_phys, incr); + count = (size - (start_phys_aligned - start_phys))/incr; + start = __va(start_phys_aligned); + start_bad = 0; + last_bad = 0; + + for (i = 0; i < count; i++) + start[i] = val; + for (i = 0; i < count; i++, start++, start_phys_aligned += incr) { + if (*start != val) { + if (start_phys_aligned == last_bad + incr) { + last_bad += incr; + } else { + if (start_bad) { + printk(KERN_CONT "\n %010lx bad mem addr %010lx - %010lx reserved", + val, start_bad, last_bad + incr); + reserve_early(start_bad, last_bad - start_bad, "BAD RAM"); + } + start_bad = last_bad = start_phys_aligned; + } + } + } + if (start_bad) { + printk(KERN_CONT "\n %016lx bad mem addr %010lx - %010lx reserved", + val, start_bad, last_bad + incr); + reserve_early(start_bad, last_bad - start_bad, "BAD RAM"); + } + +} + +/* default is disabled */ +static int memtest_pattern __initdata; + +static int __init parse_memtest(char *arg) +{ + if (arg) + memtest_pattern = simple_strtoul(arg, NULL, 0); + return 0; +} + +early_param("memtest", parse_memtest); + +void __init early_memtest(unsigned long start, unsigned long end) +{ + u64 t_start, t_size; + unsigned pattern; + + if (!memtest_pattern) + return; + + printk(KERN_INFO "early_memtest: pattern num %d", memtest_pattern); + for (pattern = 0; pattern < memtest_pattern; pattern++) { + t_start = start; + t_size = 0; + while (t_start < end) { + t_start = find_e820_area_size(t_start, &t_size, 1); + + /* done ? */ + if (t_start >= end) + break; + if (t_start + t_size > end) + t_size = end - t_start; + + printk(KERN_CONT "\n %010llx - %010llx pattern %d", + (unsigned long long)t_start, + (unsigned long long)t_start + t_size, pattern); + + memtest(t_start, t_size, pattern); + + t_start += t_size; + } + } + printk(KERN_CONT "\n"); +} diff --git a/arch/x86/mm/mmio-mod.c b/arch/x86/mm/mmio-mod.c new file mode 100644 index 000000000000..e7397e108beb --- /dev/null +++ b/arch/x86/mm/mmio-mod.c @@ -0,0 +1,515 @@ +/* + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * Copyright (C) IBM Corporation, 2005 + * Jeff Muizelaar, 2006, 2007 + * Pekka Paalanen, 2008 <pq@iki.fi> + * + * Derived from the read-mod example from relay-examples by Tom Zanussi. + */ +#define DEBUG 1 + +#include <linux/module.h> +#include <linux/debugfs.h> +#include <linux/uaccess.h> +#include <linux/io.h> +#include <linux/version.h> +#include <linux/kallsyms.h> +#include <asm/pgtable.h> +#include <linux/mmiotrace.h> +#include <asm/e820.h> /* for ISA_START_ADDRESS */ +#include <asm/atomic.h> +#include <linux/percpu.h> +#include <linux/cpu.h> + +#include "pf_in.h" + +#define NAME "mmiotrace: " + +struct trap_reason { + unsigned long addr; + unsigned long ip; + enum reason_type type; + int active_traces; +}; + +struct remap_trace { + struct list_head list; + struct kmmio_probe probe; + resource_size_t phys; + unsigned long id; +}; + +/* Accessed per-cpu. */ +static DEFINE_PER_CPU(struct trap_reason, pf_reason); +static DEFINE_PER_CPU(struct mmiotrace_rw, cpu_trace); + +#if 0 /* XXX: no way gather this info anymore */ +/* Access to this is not per-cpu. */ +static DEFINE_PER_CPU(atomic_t, dropped); +#endif + +static struct dentry *marker_file; + +static DEFINE_MUTEX(mmiotrace_mutex); +static DEFINE_SPINLOCK(trace_lock); +static atomic_t mmiotrace_enabled; +static LIST_HEAD(trace_list); /* struct remap_trace */ + +/* + * Locking in this file: + * - mmiotrace_mutex enforces enable/disable_mmiotrace() critical sections. + * - mmiotrace_enabled may be modified only when holding mmiotrace_mutex + * and trace_lock. + * - Routines depending on is_enabled() must take trace_lock. + * - trace_list users must hold trace_lock. + * - is_enabled() guarantees that mmio_trace_record is allowed. + * - pre/post callbacks assume the effect of is_enabled() being true. + */ + +/* module parameters */ +static unsigned long filter_offset; +static int nommiotrace; +static int trace_pc; + +module_param(filter_offset, ulong, 0); +module_param(nommiotrace, bool, 0); +module_param(trace_pc, bool, 0); + +MODULE_PARM_DESC(filter_offset, "Start address of traced mappings."); +MODULE_PARM_DESC(nommiotrace, "Disable actual MMIO tracing."); +MODULE_PARM_DESC(trace_pc, "Record address of faulting instructions."); + +static bool is_enabled(void) +{ + return atomic_read(&mmiotrace_enabled); +} + +#if 0 /* XXX: needs rewrite */ +/* + * Write callback for the debugfs entry: + * Read a marker and write it to the mmio trace log + */ +static ssize_t write_marker(struct file *file, const char __user *buffer, + size_t count, loff_t *ppos) +{ + char *event = NULL; + struct mm_io_header *headp; + ssize_t len = (count > 65535) ? 65535 : count; + + event = kzalloc(sizeof(*headp) + len, GFP_KERNEL); + if (!event) + return -ENOMEM; + + headp = (struct mm_io_header *)event; + headp->type = MMIO_MAGIC | (MMIO_MARKER << MMIO_OPCODE_SHIFT); + headp->data_len = len; + + if (copy_from_user(event + sizeof(*headp), buffer, len)) { + kfree(event); + return -EFAULT; + } + + spin_lock_irq(&trace_lock); +#if 0 /* XXX: convert this to use tracing */ + if (is_enabled()) + relay_write(chan, event, sizeof(*headp) + len); + else +#endif + len = -EINVAL; + spin_unlock_irq(&trace_lock); + kfree(event); + return len; +} +#endif + +static void print_pte(unsigned long address) +{ + unsigned int level; + pte_t *pte = lookup_address(address, &level); + + if (!pte) { + pr_err(NAME "Error in %s: no pte for page 0x%08lx\n", + __func__, address); + return; + } + + if (level == PG_LEVEL_2M) { + pr_emerg(NAME "4MB pages are not currently supported: " + "0x%08lx\n", address); + BUG(); + } + pr_info(NAME "pte for 0x%lx: 0x%llx 0x%llx\n", address, + (unsigned long long)pte_val(*pte), + (unsigned long long)pte_val(*pte) & _PAGE_PRESENT); +} + +/* + * For some reason the pre/post pairs have been called in an + * unmatched order. Report and die. + */ +static void die_kmmio_nesting_error(struct pt_regs *regs, unsigned long addr) +{ + const struct trap_reason *my_reason = &get_cpu_var(pf_reason); + pr_emerg(NAME "unexpected fault for address: 0x%08lx, " + "last fault for address: 0x%08lx\n", + addr, my_reason->addr); + print_pte(addr); + print_symbol(KERN_EMERG "faulting IP is at %s\n", regs->ip); + print_symbol(KERN_EMERG "last faulting IP was at %s\n", my_reason->ip); +#ifdef __i386__ + pr_emerg("eax: %08lx ebx: %08lx ecx: %08lx edx: %08lx\n", + regs->ax, regs->bx, regs->cx, regs->dx); + pr_emerg("esi: %08lx edi: %08lx ebp: %08lx esp: %08lx\n", + regs->si, regs->di, regs->bp, regs->sp); +#else + pr_emerg("rax: %016lx rcx: %016lx rdx: %016lx\n", + regs->ax, regs->cx, regs->dx); + pr_emerg("rsi: %016lx rdi: %016lx rbp: %016lx rsp: %016lx\n", + regs->si, regs->di, regs->bp, regs->sp); +#endif + put_cpu_var(pf_reason); + BUG(); +} + +static void pre(struct kmmio_probe *p, struct pt_regs *regs, + unsigned long addr) +{ + struct trap_reason *my_reason = &get_cpu_var(pf_reason); + struct mmiotrace_rw *my_trace = &get_cpu_var(cpu_trace); + const unsigned long instptr = instruction_pointer(regs); + const enum reason_type type = get_ins_type(instptr); + struct remap_trace *trace = p->private; + + /* it doesn't make sense to have more than one active trace per cpu */ + if (my_reason->active_traces) + die_kmmio_nesting_error(regs, addr); + else + my_reason->active_traces++; + + my_reason->type = type; + my_reason->addr = addr; + my_reason->ip = instptr; + + my_trace->phys = addr - trace->probe.addr + trace->phys; + my_trace->map_id = trace->id; + + /* + * Only record the program counter when requested. + * It may taint clean-room reverse engineering. + */ + if (trace_pc) + my_trace->pc = instptr; + else + my_trace->pc = 0; + + /* + * XXX: the timestamp recorded will be *after* the tracing has been + * done, not at the time we hit the instruction. SMP implications + * on event ordering? + */ + + switch (type) { + case REG_READ: + my_trace->opcode = MMIO_READ; + my_trace->width = get_ins_mem_width(instptr); + break; + case REG_WRITE: + my_trace->opcode = MMIO_WRITE; + my_trace->width = get_ins_mem_width(instptr); + my_trace->value = get_ins_reg_val(instptr, regs); + break; + case IMM_WRITE: + my_trace->opcode = MMIO_WRITE; + my_trace->width = get_ins_mem_width(instptr); + my_trace->value = get_ins_imm_val(instptr); + break; + default: + { + unsigned char *ip = (unsigned char *)instptr; + my_trace->opcode = MMIO_UNKNOWN_OP; + my_trace->width = 0; + my_trace->value = (*ip) << 16 | *(ip + 1) << 8 | + *(ip + 2); + } + } + put_cpu_var(cpu_trace); + put_cpu_var(pf_reason); +} + +static void post(struct kmmio_probe *p, unsigned long condition, + struct pt_regs *regs) +{ + struct trap_reason *my_reason = &get_cpu_var(pf_reason); + struct mmiotrace_rw *my_trace = &get_cpu_var(cpu_trace); + + /* this should always return the active_trace count to 0 */ + my_reason->active_traces--; + if (my_reason->active_traces) { + pr_emerg(NAME "unexpected post handler"); + BUG(); + } + + switch (my_reason->type) { + case REG_READ: + my_trace->value = get_ins_reg_val(my_reason->ip, regs); + break; + default: + break; + } + + mmio_trace_rw(my_trace); + put_cpu_var(cpu_trace); + put_cpu_var(pf_reason); +} + +static void ioremap_trace_core(resource_size_t offset, unsigned long size, + void __iomem *addr) +{ + static atomic_t next_id; + struct remap_trace *trace = kmalloc(sizeof(*trace), GFP_KERNEL); + /* These are page-unaligned. */ + struct mmiotrace_map map = { + .phys = offset, + .virt = (unsigned long)addr, + .len = size, + .opcode = MMIO_PROBE + }; + + if (!trace) { + pr_err(NAME "kmalloc failed in ioremap\n"); + return; + } + + *trace = (struct remap_trace) { + .probe = { + .addr = (unsigned long)addr, + .len = size, + .pre_handler = pre, + .post_handler = post, + .private = trace + }, + .phys = offset, + .id = atomic_inc_return(&next_id) + }; + map.map_id = trace->id; + + spin_lock_irq(&trace_lock); + if (!is_enabled()) + goto not_enabled; + + mmio_trace_mapping(&map); + list_add_tail(&trace->list, &trace_list); + if (!nommiotrace) + register_kmmio_probe(&trace->probe); + +not_enabled: + spin_unlock_irq(&trace_lock); +} + +void mmiotrace_ioremap(resource_size_t offset, unsigned long size, + void __iomem *addr) +{ + if (!is_enabled()) /* recheck and proper locking in *_core() */ + return; + + pr_debug(NAME "ioremap_*(0x%llx, 0x%lx) = %p\n", + (unsigned long long)offset, size, addr); + if ((filter_offset) && (offset != filter_offset)) + return; + ioremap_trace_core(offset, size, addr); +} + +static void iounmap_trace_core(volatile void __iomem *addr) +{ + struct mmiotrace_map map = { + .phys = 0, + .virt = (unsigned long)addr, + .len = 0, + .opcode = MMIO_UNPROBE + }; + struct remap_trace *trace; + struct remap_trace *tmp; + struct remap_trace *found_trace = NULL; + + pr_debug(NAME "Unmapping %p.\n", addr); + + spin_lock_irq(&trace_lock); + if (!is_enabled()) + goto not_enabled; + + list_for_each_entry_safe(trace, tmp, &trace_list, list) { + if ((unsigned long)addr == trace->probe.addr) { + if (!nommiotrace) + unregister_kmmio_probe(&trace->probe); + list_del(&trace->list); + found_trace = trace; + break; + } + } + map.map_id = (found_trace) ? found_trace->id : -1; + mmio_trace_mapping(&map); + +not_enabled: + spin_unlock_irq(&trace_lock); + if (found_trace) { + synchronize_rcu(); /* unregister_kmmio_probe() requirement */ + kfree(found_trace); + } +} + +void mmiotrace_iounmap(volatile void __iomem *addr) +{ + might_sleep(); + if (is_enabled()) /* recheck and proper locking in *_core() */ + iounmap_trace_core(addr); +} + +static void clear_trace_list(void) +{ + struct remap_trace *trace; + struct remap_trace *tmp; + + /* + * No locking required, because the caller ensures we are in a + * critical section via mutex, and is_enabled() is false, + * i.e. nothing can traverse or modify this list. + * Caller also ensures is_enabled() cannot change. + */ + list_for_each_entry(trace, &trace_list, list) { + pr_notice(NAME "purging non-iounmapped " + "trace @0x%08lx, size 0x%lx.\n", + trace->probe.addr, trace->probe.len); + if (!nommiotrace) + unregister_kmmio_probe(&trace->probe); + } + synchronize_rcu(); /* unregister_kmmio_probe() requirement */ + + list_for_each_entry_safe(trace, tmp, &trace_list, list) { + list_del(&trace->list); + kfree(trace); + } +} + +#ifdef CONFIG_HOTPLUG_CPU +static cpumask_t downed_cpus; + +static void enter_uniprocessor(void) +{ + int cpu; + int err; + + get_online_cpus(); + downed_cpus = cpu_online_map; + cpu_clear(first_cpu(cpu_online_map), downed_cpus); + if (num_online_cpus() > 1) + pr_notice(NAME "Disabling non-boot CPUs...\n"); + put_online_cpus(); + + for_each_cpu_mask(cpu, downed_cpus) { + err = cpu_down(cpu); + if (!err) + pr_info(NAME "CPU%d is down.\n", cpu); + else + pr_err(NAME "Error taking CPU%d down: %d\n", cpu, err); + } + if (num_online_cpus() > 1) + pr_warning(NAME "multiple CPUs still online, " + "may miss events.\n"); +} + +static void leave_uniprocessor(void) +{ + int cpu; + int err; + + if (cpus_weight(downed_cpus) == 0) + return; + pr_notice(NAME "Re-enabling CPUs...\n"); + for_each_cpu_mask(cpu, downed_cpus) { + err = cpu_up(cpu); + if (!err) + pr_info(NAME "enabled CPU%d.\n", cpu); + else + pr_err(NAME "cannot re-enable CPU%d: %d\n", cpu, err); + } +} + +#else /* !CONFIG_HOTPLUG_CPU */ +static void enter_uniprocessor(void) +{ + if (num_online_cpus() > 1) + pr_warning(NAME "multiple CPUs are online, may miss events. " + "Suggest booting with maxcpus=1 kernel argument.\n"); +} + +static void leave_uniprocessor(void) +{ +} +#endif + +#if 0 /* XXX: out of order */ +static struct file_operations fops_marker = { + .owner = THIS_MODULE, + .write = write_marker +}; +#endif + +void enable_mmiotrace(void) +{ + mutex_lock(&mmiotrace_mutex); + if (is_enabled()) + goto out; + +#if 0 /* XXX: tracing does not support text entries */ + marker_file = debugfs_create_file("marker", 0660, dir, NULL, + &fops_marker); + if (!marker_file) + pr_err(NAME "marker file creation failed.\n"); +#endif + + if (nommiotrace) + pr_info(NAME "MMIO tracing disabled.\n"); + enter_uniprocessor(); + spin_lock_irq(&trace_lock); + atomic_inc(&mmiotrace_enabled); + spin_unlock_irq(&trace_lock); + pr_info(NAME "enabled.\n"); +out: + mutex_unlock(&mmiotrace_mutex); +} + +void disable_mmiotrace(void) +{ + mutex_lock(&mmiotrace_mutex); + if (!is_enabled()) + goto out; + + spin_lock_irq(&trace_lock); + atomic_dec(&mmiotrace_enabled); + BUG_ON(is_enabled()); + spin_unlock_irq(&trace_lock); + + clear_trace_list(); /* guarantees: no more kmmio callbacks */ + leave_uniprocessor(); + if (marker_file) { + debugfs_remove(marker_file); + marker_file = NULL; + } + + pr_info(NAME "disabled.\n"); +out: + mutex_unlock(&mmiotrace_mutex); +} diff --git a/arch/x86/mm/numa_64.c b/arch/x86/mm/numa_64.c index c5066d519e5d..b432d5781773 100644 --- a/arch/x86/mm/numa_64.c +++ b/arch/x86/mm/numa_64.c @@ -27,30 +27,17 @@ struct pglist_data *node_data[MAX_NUMNODES] __read_mostly; EXPORT_SYMBOL(node_data); -bootmem_data_t plat_node_bdata[MAX_NUMNODES]; +static bootmem_data_t plat_node_bdata[MAX_NUMNODES]; struct memnode memnode; -#ifdef CONFIG_SMP -int x86_cpu_to_node_map_init[NR_CPUS] = { - [0 ... NR_CPUS-1] = NUMA_NO_NODE -}; -void *x86_cpu_to_node_map_early_ptr; -EXPORT_SYMBOL(x86_cpu_to_node_map_early_ptr); -#endif -DEFINE_PER_CPU(int, x86_cpu_to_node_map) = NUMA_NO_NODE; -EXPORT_PER_CPU_SYMBOL(x86_cpu_to_node_map); - s16 apicid_to_node[MAX_LOCAL_APIC] __cpuinitdata = { [0 ... MAX_LOCAL_APIC-1] = NUMA_NO_NODE }; -cpumask_t node_to_cpumask_map[MAX_NUMNODES] __read_mostly; -EXPORT_SYMBOL(node_to_cpumask_map); - int numa_off __initdata; -unsigned long __initdata nodemap_addr; -unsigned long __initdata nodemap_size; +static unsigned long __initdata nodemap_addr; +static unsigned long __initdata nodemap_size; /* * Given a shift value, try to populate memnodemap[] @@ -99,7 +86,7 @@ static int __init allocate_cachealigned_memnodemap(void) addr = 0x8000; nodemap_size = round_up(sizeof(s16) * memnodemapsize, L1_CACHE_BYTES); - nodemap_addr = find_e820_area(addr, end_pfn<<PAGE_SHIFT, + nodemap_addr = find_e820_area(addr, max_pfn<<PAGE_SHIFT, nodemap_size, L1_CACHE_BYTES); if (nodemap_addr == -1UL) { printk(KERN_ERR @@ -192,7 +179,7 @@ static void * __init early_node_mem(int nodeid, unsigned long start, void __init setup_node_bootmem(int nodeid, unsigned long start, unsigned long end) { - unsigned long start_pfn, end_pfn, bootmap_pages, bootmap_size; + unsigned long start_pfn, last_pfn, bootmap_pages, bootmap_size; unsigned long bootmap_start, nodedata_phys; void *bootmap; const int pgdat_size = round_up(sizeof(pg_data_t), PAGE_SIZE); @@ -204,7 +191,7 @@ void __init setup_node_bootmem(int nodeid, unsigned long start, start, end); start_pfn = start >> PAGE_SHIFT; - end_pfn = end >> PAGE_SHIFT; + last_pfn = end >> PAGE_SHIFT; node_data[nodeid] = early_node_mem(nodeid, start, end, pgdat_size, SMP_CACHE_BYTES); @@ -217,7 +204,7 @@ void __init setup_node_bootmem(int nodeid, unsigned long start, memset(NODE_DATA(nodeid), 0, sizeof(pg_data_t)); NODE_DATA(nodeid)->bdata = &plat_node_bdata[nodeid]; NODE_DATA(nodeid)->node_start_pfn = start_pfn; - NODE_DATA(nodeid)->node_spanned_pages = end_pfn - start_pfn; + NODE_DATA(nodeid)->node_spanned_pages = last_pfn - start_pfn; /* * Find a place for the bootmem map @@ -226,14 +213,14 @@ void __init setup_node_bootmem(int nodeid, unsigned long start, * early_node_mem will get that with find_e820_area instead * of alloc_bootmem, that could clash with reserved range */ - bootmap_pages = bootmem_bootmap_pages(end_pfn - start_pfn); + bootmap_pages = bootmem_bootmap_pages(last_pfn - start_pfn); nid = phys_to_nid(nodedata_phys); if (nid == nodeid) bootmap_start = round_up(nodedata_phys + pgdat_size, PAGE_SIZE); else bootmap_start = round_up(start, PAGE_SIZE); /* - * SMP_CAHCE_BYTES could be enough, but init_bootmem_node like + * SMP_CACHE_BYTES could be enough, but init_bootmem_node like * to use that to align to PAGE_SIZE */ bootmap = early_node_mem(nodeid, bootmap_start, end, @@ -248,7 +235,7 @@ void __init setup_node_bootmem(int nodeid, unsigned long start, bootmap_size = init_bootmem_node(NODE_DATA(nodeid), bootmap_start >> PAGE_SHIFT, - start_pfn, end_pfn); + start_pfn, last_pfn); printk(KERN_INFO " bootmap [%016lx - %016lx] pages %lx\n", bootmap_start, bootmap_start + bootmap_size - 1, @@ -309,7 +296,7 @@ void __init numa_init_array(void) #ifdef CONFIG_NUMA_EMU /* Numa emulation */ -char *cmdline __initdata; +static char *cmdline __initdata; /* * Setups up nid to range from addr to addr + size. If the end @@ -413,15 +400,15 @@ static int __init split_nodes_by_size(struct bootnode *nodes, u64 *addr, } /* - * Sets up the system RAM area from start_pfn to end_pfn according to the + * Sets up the system RAM area from start_pfn to last_pfn according to the * numa=fake command-line option. */ static struct bootnode nodes[MAX_NUMNODES] __initdata; -static int __init numa_emulation(unsigned long start_pfn, unsigned long end_pfn) +static int __init numa_emulation(unsigned long start_pfn, unsigned long last_pfn) { u64 size, addr = start_pfn << PAGE_SHIFT; - u64 max_addr = end_pfn << PAGE_SHIFT; + u64 max_addr = last_pfn << PAGE_SHIFT; int num_nodes = 0, num = 0, coeff_flag, coeff = -1, i; memset(&nodes, 0, sizeof(nodes)); @@ -527,7 +514,7 @@ out: } #endif /* CONFIG_NUMA_EMU */ -void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn) +void __init initmem_init(unsigned long start_pfn, unsigned long last_pfn) { int i; @@ -535,7 +522,7 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn) nodes_clear(node_online_map); #ifdef CONFIG_NUMA_EMU - if (cmdline && !numa_emulation(start_pfn, end_pfn)) + if (cmdline && !numa_emulation(start_pfn, last_pfn)) return; nodes_clear(node_possible_map); nodes_clear(node_online_map); @@ -543,7 +530,7 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn) #ifdef CONFIG_ACPI_NUMA if (!numa_off && !acpi_scan_nodes(start_pfn << PAGE_SHIFT, - end_pfn << PAGE_SHIFT)) + last_pfn << PAGE_SHIFT)) return; nodes_clear(node_possible_map); nodes_clear(node_online_map); @@ -551,7 +538,7 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn) #ifdef CONFIG_K8_NUMA if (!numa_off && !k8_scan_nodes(start_pfn<<PAGE_SHIFT, - end_pfn<<PAGE_SHIFT)) + last_pfn<<PAGE_SHIFT)) return; nodes_clear(node_possible_map); nodes_clear(node_online_map); @@ -561,7 +548,7 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn) printk(KERN_INFO "Faking a node at %016lx-%016lx\n", start_pfn << PAGE_SHIFT, - end_pfn << PAGE_SHIFT); + last_pfn << PAGE_SHIFT); /* setup dummy node covering all memory */ memnode_shift = 63; memnodemap = memnode.embedded_map; @@ -570,29 +557,8 @@ void __init numa_initmem_init(unsigned long start_pfn, unsigned long end_pfn) node_set(0, node_possible_map); for (i = 0; i < NR_CPUS; i++) numa_set_node(i, 0); - /* cpumask_of_cpu() may not be available during early startup */ - memset(&node_to_cpumask_map[0], 0, sizeof(node_to_cpumask_map[0])); - cpu_set(0, node_to_cpumask_map[0]); - e820_register_active_regions(0, start_pfn, end_pfn); - setup_node_bootmem(0, start_pfn << PAGE_SHIFT, end_pfn << PAGE_SHIFT); -} - -__cpuinit void numa_add_cpu(int cpu) -{ - set_bit(cpu, - (unsigned long *)&node_to_cpumask_map[early_cpu_to_node(cpu)]); -} - -void __cpuinit numa_set_node(int cpu, int node) -{ - int *cpu_to_node_map = x86_cpu_to_node_map_early_ptr; - - if(cpu_to_node_map) - cpu_to_node_map[cpu] = node; - else if(per_cpu_offset(cpu)) - per_cpu(x86_cpu_to_node_map, cpu) = node; - else - Dprintk(KERN_INFO "Setting node for non-present cpu %d\n", cpu); + e820_register_active_regions(0, start_pfn, last_pfn); + setup_node_bootmem(0, start_pfn << PAGE_SHIFT, last_pfn << PAGE_SHIFT); } unsigned long __init numa_free_all_bootmem(void) @@ -613,7 +579,7 @@ void __init paging_init(void) memset(max_zone_pfns, 0, sizeof(max_zone_pfns)); max_zone_pfns[ZONE_DMA] = MAX_DMA_PFN; max_zone_pfns[ZONE_DMA32] = MAX_DMA32_PFN; - max_zone_pfns[ZONE_NORMAL] = end_pfn; + max_zone_pfns[ZONE_NORMAL] = max_pfn; sparse_memory_present_with_active_regions(MAX_NUMNODES); sparse_init(); @@ -641,6 +607,7 @@ static __init int numa_setup(char *opt) } early_param("numa", numa_setup); +#ifdef CONFIG_NUMA /* * Setup early cpu_to_node. * @@ -652,14 +619,19 @@ early_param("numa", numa_setup); * is already initialized in a round robin manner at numa_init_array, * prior to this call, and this initialization is good enough * for the fake NUMA cases. + * + * Called before the per_cpu areas are setup. */ void __init init_cpu_to_node(void) { - int i; + int cpu; + u16 *cpu_to_apicid = early_per_cpu_ptr(x86_cpu_to_apicid); - for (i = 0; i < NR_CPUS; i++) { + BUG_ON(cpu_to_apicid == NULL); + + for_each_possible_cpu(cpu) { int node; - u16 apicid = x86_cpu_to_apicid_init[i]; + u16 apicid = cpu_to_apicid[cpu]; if (apicid == BAD_APICID) continue; @@ -668,8 +640,9 @@ void __init init_cpu_to_node(void) continue; if (!node_online(node)) continue; - numa_set_node(i, node); + numa_set_node(cpu, node); } } +#endif diff --git a/arch/x86/mm/pageattr-test.c b/arch/x86/mm/pageattr-test.c index 75f1b109aae8..0dcd42eb94e6 100644 --- a/arch/x86/mm/pageattr-test.c +++ b/arch/x86/mm/pageattr-test.c @@ -1,8 +1,8 @@ /* * self test for change_page_attr. * - * Clears the global bit on random pages in the direct mapping, then reverts - * and compares page tables forwards and afterwards. + * Clears the a test pte bit on random pages in the direct mapping, + * then reverts and compares page tables forwards and afterwards. */ #include <linux/bootmem.h> #include <linux/kthread.h> @@ -32,6 +32,13 @@ enum { GPS = (1<<30) }; +#define PAGE_TESTBIT __pgprot(_PAGE_UNUSED1) + +static int pte_testbit(pte_t pte) +{ + return pte_flags(pte) & _PAGE_UNUSED1; +} + struct split_state { long lpg, gpg, spg, exec; long min_exec, max_exec; @@ -165,15 +172,14 @@ static int pageattr_test(void) continue; } - err = change_page_attr_clear(addr[i], len[i], - __pgprot(_PAGE_GLOBAL)); + err = change_page_attr_set(addr[i], len[i], PAGE_TESTBIT); if (err < 0) { printk(KERN_ERR "CPA %d failed %d\n", i, err); failed++; } pte = lookup_address(addr[i], &level); - if (!pte || pte_global(*pte) || pte_huge(*pte)) { + if (!pte || !pte_testbit(*pte) || pte_huge(*pte)) { printk(KERN_ERR "CPA %lx: bad pte %Lx\n", addr[i], pte ? (u64)pte_val(*pte) : 0ULL); failed++; @@ -198,14 +204,13 @@ static int pageattr_test(void) failed++; continue; } - err = change_page_attr_set(addr[i], len[i], - __pgprot(_PAGE_GLOBAL)); + err = change_page_attr_clear(addr[i], len[i], PAGE_TESTBIT); if (err < 0) { printk(KERN_ERR "CPA reverting failed: %d\n", err); failed++; } pte = lookup_address(addr[i], &level); - if (!pte || !pte_global(*pte)) { + if (!pte || pte_testbit(*pte)) { printk(KERN_ERR "CPA %lx: bad pte after revert %Lx\n", addr[i], pte ? (u64)pte_val(*pte) : 0ULL); failed++; diff --git a/arch/x86/mm/pageattr.c b/arch/x86/mm/pageattr.c index 60bcb5b6a37e..65c6e46bf059 100644 --- a/arch/x86/mm/pageattr.c +++ b/arch/x86/mm/pageattr.c @@ -34,6 +34,41 @@ struct cpa_data { unsigned force_split : 1; }; +#ifdef CONFIG_PROC_FS +static unsigned long direct_pages_count[PG_LEVEL_NUM]; + +void update_page_count(int level, unsigned long pages) +{ + unsigned long flags; + + /* Protect against CPA */ + spin_lock_irqsave(&pgd_lock, flags); + direct_pages_count[level] += pages; + spin_unlock_irqrestore(&pgd_lock, flags); +} + +static void split_page_count(int level) +{ + direct_pages_count[level]--; + direct_pages_count[level - 1] += PTRS_PER_PTE; +} + +int arch_report_meminfo(char *page) +{ + int n = sprintf(page, "DirectMap4k: %8lu\n" + "DirectMap2M: %8lu\n", + direct_pages_count[PG_LEVEL_4K], + direct_pages_count[PG_LEVEL_2M]); +#ifdef CONFIG_X86_64 + n += sprintf(page + n, "DirectMap1G: %8lu\n", + direct_pages_count[PG_LEVEL_1G]); +#endif + return n; +} +#else +static inline void split_page_count(int level) { } +#endif + #ifdef CONFIG_X86_64 static inline unsigned long highmap_start_pfn(void) @@ -106,7 +141,7 @@ static void cpa_flush_all(unsigned long cache) { BUG_ON(irqs_disabled()); - on_each_cpu(__cpa_flush_all, (void *) cache, 1, 1); + on_each_cpu(__cpa_flush_all, (void *) cache, 1); } static void __cpa_flush_range(void *arg) @@ -127,7 +162,7 @@ static void cpa_flush_range(unsigned long start, int numpages, int cache) BUG_ON(irqs_disabled()); WARN_ON(PAGE_ALIGN(start) != start); - on_each_cpu(__cpa_flush_range, NULL, 1, 1); + on_each_cpu(__cpa_flush_range, NULL, 1); if (!cache) return; @@ -227,6 +262,7 @@ pte_t *lookup_address(unsigned long address, unsigned int *level) return pte_offset_kernel(pmd, address); } +EXPORT_SYMBOL_GPL(lookup_address); /* * Set the new pmd in all the pgds we know about: @@ -500,6 +536,16 @@ static int split_large_page(pte_t *kpte, unsigned long address) for (i = 0; i < PTRS_PER_PTE; i++, pfn += pfninc) set_pte(&pbase[i], pfn_pte(pfn, ref_prot)); + if (address >= (unsigned long)__va(0) && + address < (unsigned long)__va(max_low_pfn_mapped << PAGE_SHIFT)) + split_page_count(level); + +#ifdef CONFIG_X86_64 + if (address >= (unsigned long)__va(1UL<<32) && + address < (unsigned long)__va(max_pfn_mapped << PAGE_SHIFT)) + split_page_count(level); +#endif + /* * Install the new, split up pagetable. Important details here: * @@ -613,15 +659,24 @@ static int cpa_process_alias(struct cpa_data *cpa) struct cpa_data alias_cpa; int ret = 0; - if (cpa->pfn > max_pfn_mapped) + if (cpa->pfn >= max_pfn_mapped) return 0; +#ifdef CONFIG_X86_64 + if (cpa->pfn >= max_low_pfn_mapped && cpa->pfn < (1UL<<(32-PAGE_SHIFT))) + return 0; +#endif /* * No need to redo, when the primary call touched the direct * mapping already: */ - if (!within(cpa->vaddr, PAGE_OFFSET, - PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT))) { + if (!(within(cpa->vaddr, PAGE_OFFSET, + PAGE_OFFSET + (max_low_pfn_mapped << PAGE_SHIFT)) +#ifdef CONFIG_X86_64 + || within(cpa->vaddr, PAGE_OFFSET + (1UL<<32), + PAGE_OFFSET + (max_pfn_mapped << PAGE_SHIFT)) +#endif + )) { alias_cpa = *cpa; alias_cpa.vaddr = (unsigned long) __va(cpa->pfn << PAGE_SHIFT); @@ -805,7 +860,7 @@ int _set_memory_wc(unsigned long addr, int numpages) int set_memory_wc(unsigned long addr, int numpages) { - if (!pat_wc_enabled) + if (!pat_enabled) return set_memory_uc(addr, numpages); if (reserve_memtype(addr, addr + numpages * PAGE_SIZE, diff --git a/arch/x86/mm/pat.c b/arch/x86/mm/pat.c index 06b7a1c90fb8..2fe30916d4b6 100644 --- a/arch/x86/mm/pat.c +++ b/arch/x86/mm/pat.c @@ -12,6 +12,8 @@ #include <linux/gfp.h> #include <linux/fs.h> #include <linux/bootmem.h> +#include <linux/debugfs.h> +#include <linux/seq_file.h> #include <asm/msr.h> #include <asm/tlbflush.h> @@ -26,11 +28,11 @@ #include <asm/io.h> #ifdef CONFIG_X86_PAT -int __read_mostly pat_wc_enabled = 1; +int __read_mostly pat_enabled = 1; void __cpuinit pat_disable(char *reason) { - pat_wc_enabled = 0; + pat_enabled = 0; printk(KERN_INFO "%s\n", reason); } @@ -42,6 +44,19 @@ static int __init nopat(char *str) early_param("nopat", nopat); #endif + +static int debug_enable; +static int __init pat_debug_setup(char *str) +{ + debug_enable = 1; + return 0; +} +__setup("debugpat", pat_debug_setup); + +#define dprintk(fmt, arg...) \ + do { if (debug_enable) printk(KERN_INFO fmt, ##arg); } while (0) + + static u64 __read_mostly boot_pat_state; enum { @@ -53,24 +68,25 @@ enum { PAT_UC_MINUS = 7, /* UC, but can be overriden by MTRR */ }; -#define PAT(x,y) ((u64)PAT_ ## y << ((x)*8)) +#define PAT(x, y) ((u64)PAT_ ## y << ((x)*8)) void pat_init(void) { u64 pat; - if (!pat_wc_enabled) + if (!pat_enabled) return; /* Paranoia check. */ - if (!cpu_has_pat) { - printk(KERN_ERR "PAT enabled, but CPU feature cleared\n"); + if (!cpu_has_pat && boot_pat_state) { /* - * Panic if this happens on the secondary CPU, and we + * If this happens we are on a secondary CPU, but * switched to PAT on the boot CPU. We have no way to * undo PAT. - */ - BUG_ON(boot_pat_state); + */ + printk(KERN_ERR "PAT enabled, " + "but not supported by secondary CPU\n"); + BUG(); } /* Set PWT to Write-Combining. All other bits stay the same */ @@ -86,8 +102,8 @@ void pat_init(void) * 011 UC _PAGE_CACHE_UC * PAT bit unused */ - pat = PAT(0,WB) | PAT(1,WC) | PAT(2,UC_MINUS) | PAT(3,UC) | - PAT(4,WB) | PAT(5,WC) | PAT(6,UC_MINUS) | PAT(7,UC); + pat = PAT(0, WB) | PAT(1, WC) | PAT(2, UC_MINUS) | PAT(3, UC) | + PAT(4, WB) | PAT(5, WC) | PAT(6, UC_MINUS) | PAT(7, UC); /* Boot CPU check */ if (!boot_pat_state) @@ -103,11 +119,11 @@ void pat_init(void) static char *cattr_name(unsigned long flags) { switch (flags & _PAGE_CACHE_MASK) { - case _PAGE_CACHE_UC: return "uncached"; - case _PAGE_CACHE_UC_MINUS: return "uncached-minus"; - case _PAGE_CACHE_WB: return "write-back"; - case _PAGE_CACHE_WC: return "write-combining"; - default: return "broken"; + case _PAGE_CACHE_UC: return "uncached"; + case _PAGE_CACHE_UC_MINUS: return "uncached-minus"; + case _PAGE_CACHE_WB: return "write-back"; + case _PAGE_CACHE_WC: return "write-combining"; + default: return "broken"; } } @@ -145,47 +161,50 @@ static DEFINE_SPINLOCK(memtype_lock); /* protects memtype list */ * The intersection is based on "Effective Memory Type" tables in IA-32 * SDM vol 3a */ -static int pat_x_mtrr_type(u64 start, u64 end, unsigned long prot, - unsigned long *ret_prot) +static unsigned long pat_x_mtrr_type(u64 start, u64 end, unsigned long req_type) { - unsigned long pat_type; - u8 mtrr_type; - - pat_type = prot & _PAGE_CACHE_MASK; - prot &= (~_PAGE_CACHE_MASK); - - /* - * We return the PAT request directly for types where PAT takes - * precedence with respect to MTRR and for UC_MINUS. - * Consistency checks with other PAT requests is done later - * while going through memtype list. - */ - if (pat_type == _PAGE_CACHE_WC) { - *ret_prot = prot | _PAGE_CACHE_WC; - return 0; - } else if (pat_type == _PAGE_CACHE_UC_MINUS) { - *ret_prot = prot | _PAGE_CACHE_UC_MINUS; - return 0; - } else if (pat_type == _PAGE_CACHE_UC) { - *ret_prot = prot | _PAGE_CACHE_UC; - return 0; - } - /* * Look for MTRR hint to get the effective type in case where PAT * request is for WB. */ - mtrr_type = mtrr_type_lookup(start, end); + if (req_type == _PAGE_CACHE_WB) { + u8 mtrr_type; + + mtrr_type = mtrr_type_lookup(start, end); + if (mtrr_type == MTRR_TYPE_UNCACHABLE) + return _PAGE_CACHE_UC; + if (mtrr_type == MTRR_TYPE_WRCOMB) + return _PAGE_CACHE_WC; + } - if (mtrr_type == MTRR_TYPE_UNCACHABLE) { - *ret_prot = prot | _PAGE_CACHE_UC; - } else if (mtrr_type == MTRR_TYPE_WRCOMB) { - *ret_prot = prot | _PAGE_CACHE_WC; - } else { - *ret_prot = prot | _PAGE_CACHE_WB; + return req_type; +} + +static int chk_conflict(struct memtype *new, struct memtype *entry, + unsigned long *type) +{ + if (new->type != entry->type) { + if (type) { + new->type = entry->type; + *type = entry->type; + } else + goto conflict; } + /* check overlaps with more than one entry in the list */ + list_for_each_entry_continue(entry, &memtype_list, nd) { + if (new->end <= entry->start) + break; + else if (new->type != entry->type) + goto conflict; + } return 0; + + conflict: + printk(KERN_INFO "%s:%d conflicting memory types " + "%Lx-%Lx %s<->%s\n", current->comm, current->pid, new->start, + new->end, cattr_name(new->type), cattr_name(entry->type)); + return -EBUSY; } /* @@ -198,37 +217,36 @@ static int pat_x_mtrr_type(u64 start, u64 end, unsigned long prot, * req_type will have a special case value '-1', when requester want to inherit * the memory type from mtrr (if WB), existing PAT, defaulting to UC_MINUS. * - * If ret_type is NULL, function will return an error if it cannot reserve the - * region with req_type. If ret_type is non-null, function will return - * available type in ret_type in case of no error. In case of any error + * If new_type is NULL, function will return an error if it cannot reserve the + * region with req_type. If new_type is non-NULL, function will return + * available type in new_type in case of no error. In case of any error * it will return a negative return value. */ int reserve_memtype(u64 start, u64 end, unsigned long req_type, - unsigned long *ret_type) + unsigned long *new_type) { - struct memtype *new_entry = NULL; - struct memtype *parse; + struct memtype *new, *entry; unsigned long actual_type; + struct list_head *where; int err = 0; - /* Only track when pat_wc_enabled */ - if (!pat_wc_enabled) { + BUG_ON(start >= end); /* end is exclusive */ + + if (!pat_enabled) { /* This is identical to page table setting without PAT */ - if (ret_type) { - if (req_type == -1) { - *ret_type = _PAGE_CACHE_WB; - } else { - *ret_type = req_type; - } + if (new_type) { + if (req_type == -1) + *new_type = _PAGE_CACHE_WB; + else + *new_type = req_type & _PAGE_CACHE_MASK; } return 0; } /* Low ISA region is always mapped WB in page table. No need to track */ - if (start >= ISA_START_ADDRESS && (end - 1) <= ISA_END_ADDRESS) { - if (ret_type) - *ret_type = _PAGE_CACHE_WB; - + if (is_ISA_range(start, end - 1)) { + if (new_type) + *new_type = _PAGE_CACHE_WB; return 0; } @@ -241,206 +259,92 @@ int reserve_memtype(u64 start, u64 end, unsigned long req_type, */ u8 mtrr_type = mtrr_type_lookup(start, end); - if (mtrr_type == MTRR_TYPE_WRBACK) { - req_type = _PAGE_CACHE_WB; + if (mtrr_type == MTRR_TYPE_WRBACK) actual_type = _PAGE_CACHE_WB; - } else { - req_type = _PAGE_CACHE_UC_MINUS; + else actual_type = _PAGE_CACHE_UC_MINUS; - } - } else { - req_type &= _PAGE_CACHE_MASK; - err = pat_x_mtrr_type(start, end, req_type, &actual_type); - } - - if (err) { - if (ret_type) - *ret_type = actual_type; + } else + actual_type = pat_x_mtrr_type(start, end, + req_type & _PAGE_CACHE_MASK); - return -EINVAL; - } - - new_entry = kmalloc(sizeof(struct memtype), GFP_KERNEL); - if (!new_entry) + new = kmalloc(sizeof(struct memtype), GFP_KERNEL); + if (!new) return -ENOMEM; - new_entry->start = start; - new_entry->end = end; - new_entry->type = actual_type; + new->start = start; + new->end = end; + new->type = actual_type; - if (ret_type) - *ret_type = actual_type; + if (new_type) + *new_type = actual_type; spin_lock(&memtype_lock); /* Search for existing mapping that overlaps the current range */ - list_for_each_entry(parse, &memtype_list, nd) { - struct memtype *saved_ptr; - - if (parse->start >= end) { - pr_debug("New Entry\n"); - list_add(&new_entry->nd, parse->nd.prev); - new_entry = NULL; + where = NULL; + list_for_each_entry(entry, &memtype_list, nd) { + if (end <= entry->start) { + where = entry->nd.prev; break; - } - - if (start <= parse->start && end >= parse->start) { - if (actual_type != parse->type && ret_type) { - actual_type = parse->type; - *ret_type = actual_type; - new_entry->type = actual_type; + } else if (start <= entry->start) { /* end > entry->start */ + err = chk_conflict(new, entry, new_type); + if (!err) { + dprintk("Overlap at 0x%Lx-0x%Lx\n", + entry->start, entry->end); + where = entry->nd.prev; } - - if (actual_type != parse->type) { - printk( - KERN_INFO "%s:%d conflicting memory types %Lx-%Lx %s<->%s\n", - current->comm, current->pid, - start, end, - cattr_name(actual_type), - cattr_name(parse->type)); - err = -EBUSY; - break; - } - - saved_ptr = parse; - /* - * Check to see whether the request overlaps more - * than one entry in the list - */ - list_for_each_entry_continue(parse, &memtype_list, nd) { - if (end <= parse->start) { - break; - } - - if (actual_type != parse->type) { - printk( - KERN_INFO "%s:%d conflicting memory types %Lx-%Lx %s<->%s\n", - current->comm, current->pid, - start, end, - cattr_name(actual_type), - cattr_name(parse->type)); - err = -EBUSY; - break; - } - } - - if (err) { - break; - } - - pr_debug("Overlap at 0x%Lx-0x%Lx\n", - saved_ptr->start, saved_ptr->end); - /* No conflict. Go ahead and add this new entry */ - list_add(&new_entry->nd, saved_ptr->nd.prev); - new_entry = NULL; break; - } - - if (start < parse->end) { - if (actual_type != parse->type && ret_type) { - actual_type = parse->type; - *ret_type = actual_type; - new_entry->type = actual_type; - } - - if (actual_type != parse->type) { - printk( - KERN_INFO "%s:%d conflicting memory types %Lx-%Lx %s<->%s\n", - current->comm, current->pid, - start, end, - cattr_name(actual_type), - cattr_name(parse->type)); - err = -EBUSY; - break; - } - - saved_ptr = parse; - /* - * Check to see whether the request overlaps more - * than one entry in the list - */ - list_for_each_entry_continue(parse, &memtype_list, nd) { - if (end <= parse->start) { - break; - } - - if (actual_type != parse->type) { - printk( - KERN_INFO "%s:%d conflicting memory types %Lx-%Lx %s<->%s\n", - current->comm, current->pid, - start, end, - cattr_name(actual_type), - cattr_name(parse->type)); - err = -EBUSY; - break; - } + } else if (start < entry->end) { /* start > entry->start */ + err = chk_conflict(new, entry, new_type); + if (!err) { + dprintk("Overlap at 0x%Lx-0x%Lx\n", + entry->start, entry->end); + where = &entry->nd; } - - if (err) { - break; - } - - pr_debug(KERN_INFO "Overlap at 0x%Lx-0x%Lx\n", - saved_ptr->start, saved_ptr->end); - /* No conflict. Go ahead and add this new entry */ - list_add(&new_entry->nd, &saved_ptr->nd); - new_entry = NULL; break; } } if (err) { - printk(KERN_INFO - "reserve_memtype failed 0x%Lx-0x%Lx, track %s, req %s\n", - start, end, cattr_name(new_entry->type), - cattr_name(req_type)); - kfree(new_entry); + printk(KERN_INFO "reserve_memtype failed 0x%Lx-0x%Lx, " + "track %s, req %s\n", + start, end, cattr_name(new->type), cattr_name(req_type)); + kfree(new); spin_unlock(&memtype_lock); return err; } - if (new_entry) { - /* No conflict. Not yet added to the list. Add to the tail */ - list_add_tail(&new_entry->nd, &memtype_list); - pr_debug("New Entry\n"); - } - - if (ret_type) { - pr_debug( - "reserve_memtype added 0x%Lx-0x%Lx, track %s, req %s, ret %s\n", - start, end, cattr_name(actual_type), - cattr_name(req_type), cattr_name(*ret_type)); - } else { - pr_debug( - "reserve_memtype added 0x%Lx-0x%Lx, track %s, req %s\n", - start, end, cattr_name(actual_type), - cattr_name(req_type)); - } + if (where) + list_add(&new->nd, where); + else + list_add_tail(&new->nd, &memtype_list); spin_unlock(&memtype_lock); + + dprintk("reserve_memtype added 0x%Lx-0x%Lx, track %s, req %s, ret %s\n", + start, end, cattr_name(new->type), cattr_name(req_type), + new_type ? cattr_name(*new_type) : "-"); + return err; } int free_memtype(u64 start, u64 end) { - struct memtype *ml; + struct memtype *entry; int err = -EINVAL; - /* Only track when pat_wc_enabled */ - if (!pat_wc_enabled) { + if (!pat_enabled) return 0; - } /* Low ISA region is always mapped WB. No need to track */ - if (start >= ISA_START_ADDRESS && end <= ISA_END_ADDRESS) { + if (is_ISA_range(start, end - 1)) return 0; - } spin_lock(&memtype_lock); - list_for_each_entry(ml, &memtype_list, nd) { - if (ml->start == start && ml->end == end) { - list_del(&ml->nd); - kfree(ml); + list_for_each_entry(entry, &memtype_list, nd) { + if (entry->start == start && entry->end == end) { + list_del(&entry->nd); + kfree(entry); err = 0; break; } @@ -452,7 +356,7 @@ int free_memtype(u64 start, u64 end) current->comm, current->pid, start, end); } - pr_debug("free_memtype request 0x%Lx-0x%Lx\n", start, end); + dprintk("free_memtype request 0x%Lx-0x%Lx\n", start, end); return err; } @@ -471,8 +375,8 @@ pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn, return vma_prot; } -#ifdef CONFIG_NONPROMISC_DEVMEM -/* This check is done in drivers/char/mem.c in case of NONPROMISC_DEVMEM*/ +#ifdef CONFIG_STRICT_DEVMEM +/* This check is done in drivers/char/mem.c in case of STRICT_DEVMEM*/ static inline int range_is_allowed(unsigned long pfn, unsigned long size) { return 1; @@ -496,7 +400,7 @@ static inline int range_is_allowed(unsigned long pfn, unsigned long size) } return 1; } -#endif /* CONFIG_NONPROMISC_DEVMEM */ +#endif /* CONFIG_STRICT_DEVMEM */ int phys_mem_access_prot_allowed(struct file *file, unsigned long pfn, unsigned long size, pgprot_t *vma_prot) @@ -521,12 +425,12 @@ int phys_mem_access_prot_allowed(struct file *file, unsigned long pfn, * caching for the high addresses through the KEN pin, but * we maintain the tradition of paranoia in this code. */ - if (!pat_wc_enabled && - ! ( test_bit(X86_FEATURE_MTRR, boot_cpu_data.x86_capability) || - test_bit(X86_FEATURE_K6_MTRR, boot_cpu_data.x86_capability) || - test_bit(X86_FEATURE_CYRIX_ARR, boot_cpu_data.x86_capability) || - test_bit(X86_FEATURE_CENTAUR_MCR, boot_cpu_data.x86_capability)) && - (pfn << PAGE_SHIFT) >= __pa(high_memory)) { + if (!pat_enabled && + !(boot_cpu_has(X86_FEATURE_MTRR) || + boot_cpu_has(X86_FEATURE_K6_MTRR) || + boot_cpu_has(X86_FEATURE_CYRIX_ARR) || + boot_cpu_has(X86_FEATURE_CENTAUR_MCR)) && + (pfn << PAGE_SHIFT) >= __pa(high_memory)) { flags = _PAGE_CACHE_UC; } #endif @@ -547,8 +451,9 @@ int phys_mem_access_prot_allowed(struct file *file, unsigned long pfn, if (retval < 0) return 0; - if (pfn <= max_pfn_mapped && - ioremap_change_attr((unsigned long)__va(offset), size, flags) < 0) { + if (((pfn < max_low_pfn_mapped) || + (pfn >= (1UL<<(32 - PAGE_SHIFT)) && pfn < max_pfn_mapped)) && + ioremap_change_attr((unsigned long)__va(offset), size, flags) < 0) { free_memtype(offset, offset + size); printk(KERN_INFO "%s:%d /dev/mem ioremap_change_attr failed %s for %Lx-%Lx\n", @@ -587,3 +492,88 @@ void unmap_devmem(unsigned long pfn, unsigned long size, pgprot_t vma_prot) free_memtype(addr, addr + size); } +#if defined(CONFIG_DEBUG_FS) + +/* get Nth element of the linked list */ +static struct memtype *memtype_get_idx(loff_t pos) +{ + struct memtype *list_node, *print_entry; + int i = 1; + + print_entry = kmalloc(sizeof(struct memtype), GFP_KERNEL); + if (!print_entry) + return NULL; + + spin_lock(&memtype_lock); + list_for_each_entry(list_node, &memtype_list, nd) { + if (pos == i) { + *print_entry = *list_node; + spin_unlock(&memtype_lock); + return print_entry; + } + ++i; + } + spin_unlock(&memtype_lock); + kfree(print_entry); + return NULL; +} + +static void *memtype_seq_start(struct seq_file *seq, loff_t *pos) +{ + if (*pos == 0) { + ++*pos; + seq_printf(seq, "PAT memtype list:\n"); + } + + return memtype_get_idx(*pos); +} + +static void *memtype_seq_next(struct seq_file *seq, void *v, loff_t *pos) +{ + ++*pos; + return memtype_get_idx(*pos); +} + +static void memtype_seq_stop(struct seq_file *seq, void *v) +{ +} + +static int memtype_seq_show(struct seq_file *seq, void *v) +{ + struct memtype *print_entry = (struct memtype *)v; + + seq_printf(seq, "%s @ 0x%Lx-0x%Lx\n", cattr_name(print_entry->type), + print_entry->start, print_entry->end); + kfree(print_entry); + return 0; +} + +static struct seq_operations memtype_seq_ops = { + .start = memtype_seq_start, + .next = memtype_seq_next, + .stop = memtype_seq_stop, + .show = memtype_seq_show, +}; + +static int memtype_seq_open(struct inode *inode, struct file *file) +{ + return seq_open(file, &memtype_seq_ops); +} + +static const struct file_operations memtype_fops = { + .open = memtype_seq_open, + .read = seq_read, + .llseek = seq_lseek, + .release = seq_release, +}; + +static int __init pat_memtype_list_init(void) +{ + debugfs_create_file("pat_memtype_list", S_IRUSR, arch_debugfs_dir, + NULL, &memtype_fops); + return 0; +} + +late_initcall(pat_memtype_list_init); + +#endif /* CONFIG_DEBUG_FS */ diff --git a/arch/x86/mm/pf_in.c b/arch/x86/mm/pf_in.c new file mode 100644 index 000000000000..efa1911e20ca --- /dev/null +++ b/arch/x86/mm/pf_in.c @@ -0,0 +1,489 @@ +/* + * Fault Injection Test harness (FI) + * Copyright (C) Intel Crop. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, + * USA. + * + */ + +/* Id: pf_in.c,v 1.1.1.1 2002/11/12 05:56:32 brlock Exp + * Copyright by Intel Crop., 2002 + * Louis Zhuang (louis.zhuang@intel.com) + * + * Bjorn Steinbrink (B.Steinbrink@gmx.de), 2007 + */ + +#include <linux/module.h> +#include <linux/ptrace.h> /* struct pt_regs */ +#include "pf_in.h" + +#ifdef __i386__ +/* IA32 Manual 3, 2-1 */ +static unsigned char prefix_codes[] = { + 0xF0, 0xF2, 0xF3, 0x2E, 0x36, 0x3E, 0x26, 0x64, + 0x65, 0x2E, 0x3E, 0x66, 0x67 +}; +/* IA32 Manual 3, 3-432*/ +static unsigned int reg_rop[] = { + 0x8A, 0x8B, 0xB60F, 0xB70F, 0xBE0F, 0xBF0F +}; +static unsigned int reg_wop[] = { 0x88, 0x89 }; +static unsigned int imm_wop[] = { 0xC6, 0xC7 }; +/* IA32 Manual 3, 3-432*/ +static unsigned int rw8[] = { 0x88, 0x8A, 0xC6 }; +static unsigned int rw32[] = { + 0x89, 0x8B, 0xC7, 0xB60F, 0xB70F, 0xBE0F, 0xBF0F +}; +static unsigned int mw8[] = { 0x88, 0x8A, 0xC6, 0xB60F, 0xBE0F }; +static unsigned int mw16[] = { 0xB70F, 0xBF0F }; +static unsigned int mw32[] = { 0x89, 0x8B, 0xC7 }; +static unsigned int mw64[] = {}; +#else /* not __i386__ */ +static unsigned char prefix_codes[] = { + 0x66, 0x67, 0x2E, 0x3E, 0x26, 0x64, 0x65, 0x36, + 0xF0, 0xF3, 0xF2, + /* REX Prefixes */ + 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, + 0x48, 0x49, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, 0x4f +}; +/* AMD64 Manual 3, Appendix A*/ +static unsigned int reg_rop[] = { + 0x8A, 0x8B, 0xB60F, 0xB70F, 0xBE0F, 0xBF0F +}; +static unsigned int reg_wop[] = { 0x88, 0x89 }; +static unsigned int imm_wop[] = { 0xC6, 0xC7 }; +static unsigned int rw8[] = { 0xC6, 0x88, 0x8A }; +static unsigned int rw32[] = { + 0xC7, 0x89, 0x8B, 0xB60F, 0xB70F, 0xBE0F, 0xBF0F +}; +/* 8 bit only */ +static unsigned int mw8[] = { 0xC6, 0x88, 0x8A, 0xB60F, 0xBE0F }; +/* 16 bit only */ +static unsigned int mw16[] = { 0xB70F, 0xBF0F }; +/* 16 or 32 bit */ +static unsigned int mw32[] = { 0xC7 }; +/* 16, 32 or 64 bit */ +static unsigned int mw64[] = { 0x89, 0x8B }; +#endif /* not __i386__ */ + +static int skip_prefix(unsigned char *addr, int *shorted, int *enlarged, + int *rexr) +{ + int i; + unsigned char *p = addr; + *shorted = 0; + *enlarged = 0; + *rexr = 0; + +restart: + for (i = 0; i < ARRAY_SIZE(prefix_codes); i++) { + if (*p == prefix_codes[i]) { + if (*p == 0x66) + *shorted = 1; +#ifdef __amd64__ + if ((*p & 0xf8) == 0x48) + *enlarged = 1; + if ((*p & 0xf4) == 0x44) + *rexr = 1; +#endif + p++; + goto restart; + } + } + + return (p - addr); +} + +static int get_opcode(unsigned char *addr, unsigned int *opcode) +{ + int len; + + if (*addr == 0x0F) { + /* 0x0F is extension instruction */ + *opcode = *(unsigned short *)addr; + len = 2; + } else { + *opcode = *addr; + len = 1; + } + + return len; +} + +#define CHECK_OP_TYPE(opcode, array, type) \ + for (i = 0; i < ARRAY_SIZE(array); i++) { \ + if (array[i] == opcode) { \ + rv = type; \ + goto exit; \ + } \ + } + +enum reason_type get_ins_type(unsigned long ins_addr) +{ + unsigned int opcode; + unsigned char *p; + int shorted, enlarged, rexr; + int i; + enum reason_type rv = OTHERS; + + p = (unsigned char *)ins_addr; + p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += get_opcode(p, &opcode); + + CHECK_OP_TYPE(opcode, reg_rop, REG_READ); + CHECK_OP_TYPE(opcode, reg_wop, REG_WRITE); + CHECK_OP_TYPE(opcode, imm_wop, IMM_WRITE); + +exit: + return rv; +} +#undef CHECK_OP_TYPE + +static unsigned int get_ins_reg_width(unsigned long ins_addr) +{ + unsigned int opcode; + unsigned char *p; + int i, shorted, enlarged, rexr; + + p = (unsigned char *)ins_addr; + p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += get_opcode(p, &opcode); + + for (i = 0; i < ARRAY_SIZE(rw8); i++) + if (rw8[i] == opcode) + return 1; + + for (i = 0; i < ARRAY_SIZE(rw32); i++) + if (rw32[i] == opcode) + return (shorted ? 2 : (enlarged ? 8 : 4)); + + printk(KERN_ERR "mmiotrace: Unknown opcode 0x%02x\n", opcode); + return 0; +} + +unsigned int get_ins_mem_width(unsigned long ins_addr) +{ + unsigned int opcode; + unsigned char *p; + int i, shorted, enlarged, rexr; + + p = (unsigned char *)ins_addr; + p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += get_opcode(p, &opcode); + + for (i = 0; i < ARRAY_SIZE(mw8); i++) + if (mw8[i] == opcode) + return 1; + + for (i = 0; i < ARRAY_SIZE(mw16); i++) + if (mw16[i] == opcode) + return 2; + + for (i = 0; i < ARRAY_SIZE(mw32); i++) + if (mw32[i] == opcode) + return shorted ? 2 : 4; + + for (i = 0; i < ARRAY_SIZE(mw64); i++) + if (mw64[i] == opcode) + return shorted ? 2 : (enlarged ? 8 : 4); + + printk(KERN_ERR "mmiotrace: Unknown opcode 0x%02x\n", opcode); + return 0; +} + +/* + * Define register ident in mod/rm byte. + * Note: these are NOT the same as in ptrace-abi.h. + */ +enum { + arg_AL = 0, + arg_CL = 1, + arg_DL = 2, + arg_BL = 3, + arg_AH = 4, + arg_CH = 5, + arg_DH = 6, + arg_BH = 7, + + arg_AX = 0, + arg_CX = 1, + arg_DX = 2, + arg_BX = 3, + arg_SP = 4, + arg_BP = 5, + arg_SI = 6, + arg_DI = 7, +#ifdef __amd64__ + arg_R8 = 8, + arg_R9 = 9, + arg_R10 = 10, + arg_R11 = 11, + arg_R12 = 12, + arg_R13 = 13, + arg_R14 = 14, + arg_R15 = 15 +#endif +}; + +static unsigned char *get_reg_w8(int no, struct pt_regs *regs) +{ + unsigned char *rv = NULL; + + switch (no) { + case arg_AL: + rv = (unsigned char *)®s->ax; + break; + case arg_BL: + rv = (unsigned char *)®s->bx; + break; + case arg_CL: + rv = (unsigned char *)®s->cx; + break; + case arg_DL: + rv = (unsigned char *)®s->dx; + break; + case arg_AH: + rv = 1 + (unsigned char *)®s->ax; + break; + case arg_BH: + rv = 1 + (unsigned char *)®s->bx; + break; + case arg_CH: + rv = 1 + (unsigned char *)®s->cx; + break; + case arg_DH: + rv = 1 + (unsigned char *)®s->dx; + break; +#ifdef __amd64__ + case arg_R8: + rv = (unsigned char *)®s->r8; + break; + case arg_R9: + rv = (unsigned char *)®s->r9; + break; + case arg_R10: + rv = (unsigned char *)®s->r10; + break; + case arg_R11: + rv = (unsigned char *)®s->r11; + break; + case arg_R12: + rv = (unsigned char *)®s->r12; + break; + case arg_R13: + rv = (unsigned char *)®s->r13; + break; + case arg_R14: + rv = (unsigned char *)®s->r14; + break; + case arg_R15: + rv = (unsigned char *)®s->r15; + break; +#endif + default: + printk(KERN_ERR "mmiotrace: Error reg no# %d\n", no); + break; + } + return rv; +} + +static unsigned long *get_reg_w32(int no, struct pt_regs *regs) +{ + unsigned long *rv = NULL; + + switch (no) { + case arg_AX: + rv = ®s->ax; + break; + case arg_BX: + rv = ®s->bx; + break; + case arg_CX: + rv = ®s->cx; + break; + case arg_DX: + rv = ®s->dx; + break; + case arg_SP: + rv = ®s->sp; + break; + case arg_BP: + rv = ®s->bp; + break; + case arg_SI: + rv = ®s->si; + break; + case arg_DI: + rv = ®s->di; + break; +#ifdef __amd64__ + case arg_R8: + rv = ®s->r8; + break; + case arg_R9: + rv = ®s->r9; + break; + case arg_R10: + rv = ®s->r10; + break; + case arg_R11: + rv = ®s->r11; + break; + case arg_R12: + rv = ®s->r12; + break; + case arg_R13: + rv = ®s->r13; + break; + case arg_R14: + rv = ®s->r14; + break; + case arg_R15: + rv = ®s->r15; + break; +#endif + default: + printk(KERN_ERR "mmiotrace: Error reg no# %d\n", no); + } + + return rv; +} + +unsigned long get_ins_reg_val(unsigned long ins_addr, struct pt_regs *regs) +{ + unsigned int opcode; + unsigned char mod_rm; + int reg; + unsigned char *p; + int i, shorted, enlarged, rexr; + unsigned long rv; + + p = (unsigned char *)ins_addr; + p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += get_opcode(p, &opcode); + for (i = 0; i < ARRAY_SIZE(reg_rop); i++) + if (reg_rop[i] == opcode) { + rv = REG_READ; + goto do_work; + } + + for (i = 0; i < ARRAY_SIZE(reg_wop); i++) + if (reg_wop[i] == opcode) { + rv = REG_WRITE; + goto do_work; + } + + printk(KERN_ERR "mmiotrace: Not a register instruction, opcode " + "0x%02x\n", opcode); + goto err; + +do_work: + mod_rm = *p; + reg = ((mod_rm >> 3) & 0x7) | (rexr << 3); + switch (get_ins_reg_width(ins_addr)) { + case 1: + return *get_reg_w8(reg, regs); + + case 2: + return *(unsigned short *)get_reg_w32(reg, regs); + + case 4: + return *(unsigned int *)get_reg_w32(reg, regs); + +#ifdef __amd64__ + case 8: + return *(unsigned long *)get_reg_w32(reg, regs); +#endif + + default: + printk(KERN_ERR "mmiotrace: Error width# %d\n", reg); + } + +err: + return 0; +} + +unsigned long get_ins_imm_val(unsigned long ins_addr) +{ + unsigned int opcode; + unsigned char mod_rm; + unsigned char mod; + unsigned char *p; + int i, shorted, enlarged, rexr; + unsigned long rv; + + p = (unsigned char *)ins_addr; + p += skip_prefix(p, &shorted, &enlarged, &rexr); + p += get_opcode(p, &opcode); + for (i = 0; i < ARRAY_SIZE(imm_wop); i++) + if (imm_wop[i] == opcode) { + rv = IMM_WRITE; + goto do_work; + } + + printk(KERN_ERR "mmiotrace: Not an immediate instruction, opcode " + "0x%02x\n", opcode); + goto err; + +do_work: + mod_rm = *p; + mod = mod_rm >> 6; + p++; + switch (mod) { + case 0: + /* if r/m is 5 we have a 32 disp (IA32 Manual 3, Table 2-2) */ + /* AMD64: XXX Check for address size prefix? */ + if ((mod_rm & 0x7) == 0x5) + p += 4; + break; + + case 1: + p += 1; + break; + + case 2: + p += 4; + break; + + case 3: + default: + printk(KERN_ERR "mmiotrace: not a memory access instruction " + "at 0x%lx, rm_mod=0x%02x\n", + ins_addr, mod_rm); + } + + switch (get_ins_reg_width(ins_addr)) { + case 1: + return *(unsigned char *)p; + + case 2: + return *(unsigned short *)p; + + case 4: + return *(unsigned int *)p; + +#ifdef __amd64__ + case 8: + return *(unsigned long *)p; +#endif + + default: + printk(KERN_ERR "mmiotrace: Error: width.\n"); + } + +err: + return 0; +} diff --git a/arch/x86/mm/pf_in.h b/arch/x86/mm/pf_in.h new file mode 100644 index 000000000000..e05341a51a27 --- /dev/null +++ b/arch/x86/mm/pf_in.h @@ -0,0 +1,39 @@ +/* + * Fault Injection Test harness (FI) + * Copyright (C) Intel Crop. + * + * This program is free software; you can redistribute it and/or + * modify it under the terms of the GNU General Public License + * as published by the Free Software Foundation; either version 2 + * of the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, + * USA. + * + */ + +#ifndef __PF_H_ +#define __PF_H_ + +enum reason_type { + NOT_ME, /* page fault is not in regions */ + NOTHING, /* access others point in regions */ + REG_READ, /* read from addr to reg */ + REG_WRITE, /* write from reg to addr */ + IMM_WRITE, /* write from imm to addr */ + OTHERS /* Other instructions can not intercept */ +}; + +enum reason_type get_ins_type(unsigned long ins_addr); +unsigned int get_ins_mem_width(unsigned long ins_addr); +unsigned long get_ins_reg_val(unsigned long ins_addr, struct pt_regs *regs); +unsigned long get_ins_imm_val(unsigned long ins_addr); + +#endif /* __PF_H_ */ diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c index 50159764f694..557b2abceef8 100644 --- a/arch/x86/mm/pgtable.c +++ b/arch/x86/mm/pgtable.c @@ -2,6 +2,7 @@ #include <asm/pgalloc.h> #include <asm/pgtable.h> #include <asm/tlb.h> +#include <asm/fixmap.h> pte_t *pte_alloc_one_kernel(struct mm_struct *mm, unsigned long address) { @@ -65,12 +66,6 @@ static inline void pgd_list_del(pgd_t *pgd) static void pgd_ctor(void *p) { pgd_t *pgd = p; - unsigned long flags; - - /* Clear usermode parts of PGD */ - memset(pgd, 0, KERNEL_PGD_BOUNDARY*sizeof(pgd_t)); - - spin_lock_irqsave(&pgd_lock, flags); /* If the pgd points to a shared pagetable level (either the ptes in non-PAE, or shared PMD in PAE), then just copy the @@ -90,8 +85,6 @@ static void pgd_ctor(void *p) /* list required to sync kernel mapping updates */ if (!SHARED_KERNEL_PMD) pgd_list_add(pgd); - - spin_unlock_irqrestore(&pgd_lock, flags); } static void pgd_dtor(void *pgd) @@ -119,6 +112,72 @@ static void pgd_dtor(void *pgd) #ifdef CONFIG_X86_PAE /* + * In PAE mode, we need to do a cr3 reload (=tlb flush) when + * updating the top-level pagetable entries to guarantee the + * processor notices the update. Since this is expensive, and + * all 4 top-level entries are used almost immediately in a + * new process's life, we just pre-populate them here. + * + * Also, if we're in a paravirt environment where the kernel pmd is + * not shared between pagetables (!SHARED_KERNEL_PMDS), we allocate + * and initialize the kernel pmds here. + */ +#define PREALLOCATED_PMDS UNSHARED_PTRS_PER_PGD + +void pud_populate(struct mm_struct *mm, pud_t *pudp, pmd_t *pmd) +{ + paravirt_alloc_pmd(mm, __pa(pmd) >> PAGE_SHIFT); + + /* Note: almost everything apart from _PAGE_PRESENT is + reserved at the pmd (PDPT) level. */ + set_pud(pudp, __pud(__pa(pmd) | _PAGE_PRESENT)); + + /* + * According to Intel App note "TLBs, Paging-Structure Caches, + * and Their Invalidation", April 2007, document 317080-001, + * section 8.1: in PAE mode we explicitly have to flush the + * TLB via cr3 if the top-level pgd is changed... + */ + if (mm == current->active_mm) + write_cr3(read_cr3()); +} +#else /* !CONFIG_X86_PAE */ + +/* No need to prepopulate any pagetable entries in non-PAE modes. */ +#define PREALLOCATED_PMDS 0 + +#endif /* CONFIG_X86_PAE */ + +static void free_pmds(pmd_t *pmds[]) +{ + int i; + + for(i = 0; i < PREALLOCATED_PMDS; i++) + if (pmds[i]) + free_page((unsigned long)pmds[i]); +} + +static int preallocate_pmds(pmd_t *pmds[]) +{ + int i; + bool failed = false; + + for(i = 0; i < PREALLOCATED_PMDS; i++) { + pmd_t *pmd = (pmd_t *)get_zeroed_page(GFP_KERNEL|__GFP_REPEAT); + if (pmd == NULL) + failed = true; + pmds[i] = pmd; + } + + if (failed) { + free_pmds(pmds); + return -ENOMEM; + } + + return 0; +} + +/* * Mop up any pmd pages which may still be attached to the pgd. * Normally they will be freed by munmap/exit_mmap, but any pmd we * preallocate which never got a corresponding vma will need to be @@ -128,7 +187,7 @@ static void pgd_mop_up_pmds(struct mm_struct *mm, pgd_t *pgdp) { int i; - for(i = 0; i < UNSHARED_PTRS_PER_PGD; i++) { + for(i = 0; i < PREALLOCATED_PMDS; i++) { pgd_t pgd = pgdp[i]; if (pgd_val(pgd) != 0) { @@ -142,32 +201,17 @@ static void pgd_mop_up_pmds(struct mm_struct *mm, pgd_t *pgdp) } } -/* - * In PAE mode, we need to do a cr3 reload (=tlb flush) when - * updating the top-level pagetable entries to guarantee the - * processor notices the update. Since this is expensive, and - * all 4 top-level entries are used almost immediately in a - * new process's life, we just pre-populate them here. - * - * Also, if we're in a paravirt environment where the kernel pmd is - * not shared between pagetables (!SHARED_KERNEL_PMDS), we allocate - * and initialize the kernel pmds here. - */ -static int pgd_prepopulate_pmd(struct mm_struct *mm, pgd_t *pgd) +static void pgd_prepopulate_pmd(struct mm_struct *mm, pgd_t *pgd, pmd_t *pmds[]) { pud_t *pud; unsigned long addr; int i; pud = pud_offset(pgd, 0); - for (addr = i = 0; i < UNSHARED_PTRS_PER_PGD; - i++, pud++, addr += PUD_SIZE) { - pmd_t *pmd = pmd_alloc_one(mm, addr); - if (!pmd) { - pgd_mop_up_pmds(mm, pgd); - return 0; - } + for (addr = i = 0; i < PREALLOCATED_PMDS; + i++, pud++, addr += PUD_SIZE) { + pmd_t *pmd = pmds[i]; if (i >= KERNEL_PGD_BOUNDARY) memcpy(pmd, (pmd_t *)pgd_page_vaddr(swapper_pg_dir[i]), @@ -175,61 +219,54 @@ static int pgd_prepopulate_pmd(struct mm_struct *mm, pgd_t *pgd) pud_populate(mm, pud, pmd); } - - return 1; } -void pud_populate(struct mm_struct *mm, pud_t *pudp, pmd_t *pmd) +pgd_t *pgd_alloc(struct mm_struct *mm) { - paravirt_alloc_pmd(mm, __pa(pmd) >> PAGE_SHIFT); + pgd_t *pgd; + pmd_t *pmds[PREALLOCATED_PMDS]; + unsigned long flags; - /* Note: almost everything apart from _PAGE_PRESENT is - reserved at the pmd (PDPT) level. */ - set_pud(pudp, __pud(__pa(pmd) | _PAGE_PRESENT)); + pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO); - /* - * According to Intel App note "TLBs, Paging-Structure Caches, - * and Their Invalidation", April 2007, document 317080-001, - * section 8.1: in PAE mode we explicitly have to flush the - * TLB via cr3 if the top-level pgd is changed... - */ - if (mm == current->active_mm) - write_cr3(read_cr3()); -} -#else /* !CONFIG_X86_PAE */ -/* No need to prepopulate any pagetable entries in non-PAE modes. */ -static int pgd_prepopulate_pmd(struct mm_struct *mm, pgd_t *pgd) -{ - return 1; -} + if (pgd == NULL) + goto out; -static void pgd_mop_up_pmds(struct mm_struct *mm, pgd_t *pgd) -{ -} -#endif /* CONFIG_X86_PAE */ + mm->pgd = pgd; -pgd_t *pgd_alloc(struct mm_struct *mm) -{ - pgd_t *pgd = (pgd_t *)__get_free_page(GFP_KERNEL | __GFP_ZERO); + if (preallocate_pmds(pmds) != 0) + goto out_free_pgd; - /* so that alloc_pmd can use it */ - mm->pgd = pgd; - if (pgd) - pgd_ctor(pgd); + if (paravirt_pgd_alloc(mm) != 0) + goto out_free_pmds; - if (pgd && !pgd_prepopulate_pmd(mm, pgd)) { - pgd_dtor(pgd); - free_page((unsigned long)pgd); - pgd = NULL; - } + /* + * Make sure that pre-populating the pmds is atomic with + * respect to anything walking the pgd_list, so that they + * never see a partially populated pgd. + */ + spin_lock_irqsave(&pgd_lock, flags); + + pgd_ctor(pgd); + pgd_prepopulate_pmd(mm, pgd, pmds); + + spin_unlock_irqrestore(&pgd_lock, flags); return pgd; + +out_free_pmds: + free_pmds(pmds); +out_free_pgd: + free_page((unsigned long)pgd); +out: + return NULL; } void pgd_free(struct mm_struct *mm, pgd_t *pgd) { pgd_mop_up_pmds(mm, pgd); pgd_dtor(pgd); + paravirt_pgd_free(mm, pgd); free_page((unsigned long)pgd); } @@ -255,7 +292,7 @@ int ptep_test_and_clear_young(struct vm_area_struct *vma, if (pte_young(*ptep)) ret = test_and_clear_bit(_PAGE_BIT_ACCESSED, - &ptep->pte); + (unsigned long *) &ptep->pte); if (ret) pte_update(vma->vm_mm, addr, ptep); @@ -274,3 +311,22 @@ int ptep_clear_flush_young(struct vm_area_struct *vma, return young; } + +int fixmaps_set; + +void __native_set_fixmap(enum fixed_addresses idx, pte_t pte) +{ + unsigned long address = __fix_to_virt(idx); + + if (idx >= __end_of_fixed_addresses) { + BUG(); + return; + } + set_pte_vaddr(address, pte); + fixmaps_set++; +} + +void native_set_fixmap(enum fixed_addresses idx, unsigned long phys, pgprot_t flags) +{ + __native_set_fixmap(idx, pfn_pte(phys >> PAGE_SHIFT, flags)); +} diff --git a/arch/x86/mm/pgtable_32.c b/arch/x86/mm/pgtable_32.c index 369cf065b6a4..b4becbf8c570 100644 --- a/arch/x86/mm/pgtable_32.c +++ b/arch/x86/mm/pgtable_32.c @@ -71,7 +71,7 @@ void show_mem(void) * Associate a virtual page frame with a given physical page frame * and protection flags for that frame. */ -static void set_pte_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags) +void set_pte_vaddr(unsigned long vaddr, pte_t pteval) { pgd_t *pgd; pud_t *pud; @@ -94,8 +94,8 @@ static void set_pte_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags) return; } pte = pte_offset_kernel(pmd, vaddr); - if (pgprot_val(flags)) - set_pte_present(&init_mm, vaddr, pte, pfn_pte(pfn, flags)); + if (pte_val(pteval)) + set_pte_present(&init_mm, vaddr, pte, pteval); else pte_clear(&init_mm, vaddr, pte); @@ -141,22 +141,9 @@ void set_pmd_pfn(unsigned long vaddr, unsigned long pfn, pgprot_t flags) __flush_tlb_one(vaddr); } -static int fixmaps; unsigned long __FIXADDR_TOP = 0xfffff000; EXPORT_SYMBOL(__FIXADDR_TOP); -void __set_fixmap (enum fixed_addresses idx, unsigned long phys, pgprot_t flags) -{ - unsigned long address = __fix_to_virt(idx); - - if (idx >= __end_of_fixed_addresses) { - BUG(); - return; - } - set_pte_pfn(address, phys >> PAGE_SHIFT, flags); - fixmaps++; -} - /** * reserve_top_address - reserves a hole in the top of kernel address space * @reserve - size of hole to reserve @@ -164,11 +151,44 @@ void __set_fixmap (enum fixed_addresses idx, unsigned long phys, pgprot_t flags) * Can be used to relocate the fixmap area and poke a hole in the top * of kernel address space to make room for a hypervisor. */ -void reserve_top_address(unsigned long reserve) +void __init reserve_top_address(unsigned long reserve) { - BUG_ON(fixmaps > 0); + BUG_ON(fixmaps_set > 0); printk(KERN_INFO "Reserving virtual address space above 0x%08x\n", (int)-reserve); __FIXADDR_TOP = -reserve - PAGE_SIZE; __VMALLOC_RESERVE += reserve; } + +/* + * vmalloc=size forces the vmalloc area to be exactly 'size' + * bytes. This can be used to increase (or decrease) the + * vmalloc area - the default is 128m. + */ +static int __init parse_vmalloc(char *arg) +{ + if (!arg) + return -EINVAL; + + __VMALLOC_RESERVE = memparse(arg, &arg); + return 0; +} +early_param("vmalloc", parse_vmalloc); + +/* + * reservetop=size reserves a hole at the top of the kernel address space which + * a hypervisor can load into later. Needed for dynamically loaded hypervisors, + * so relocating the fixmap can be done before paging initialization. + */ +static int __init parse_reservetop(char *arg) +{ + unsigned long address; + + if (!arg) + return -EINVAL; + + address = memparse(arg, &arg); + reserve_top_address(address); + return 0; +} +early_param("reservetop", parse_reservetop); diff --git a/arch/x86/mm/srat_32.c b/arch/x86/mm/srat_32.c new file mode 100644 index 000000000000..1eb2973a301c --- /dev/null +++ b/arch/x86/mm/srat_32.c @@ -0,0 +1,279 @@ +/* + * Some of the code in this file has been gleaned from the 64 bit + * discontigmem support code base. + * + * Copyright (C) 2002, IBM Corp. + * + * All rights reserved. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, but + * WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, GOOD TITLE or + * NON INFRINGEMENT. See the GNU General Public License for more + * details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * + * Send feedback to Pat Gaughen <gone@us.ibm.com> + */ +#include <linux/mm.h> +#include <linux/bootmem.h> +#include <linux/mmzone.h> +#include <linux/acpi.h> +#include <linux/nodemask.h> +#include <asm/srat.h> +#include <asm/topology.h> +#include <asm/smp.h> +#include <asm/e820.h> + +/* + * proximity macros and definitions + */ +#define NODE_ARRAY_INDEX(x) ((x) / 8) /* 8 bits/char */ +#define NODE_ARRAY_OFFSET(x) ((x) % 8) /* 8 bits/char */ +#define BMAP_SET(bmap, bit) ((bmap)[NODE_ARRAY_INDEX(bit)] |= 1 << NODE_ARRAY_OFFSET(bit)) +#define BMAP_TEST(bmap, bit) ((bmap)[NODE_ARRAY_INDEX(bit)] & (1 << NODE_ARRAY_OFFSET(bit))) +/* bitmap length; _PXM is at most 255 */ +#define PXM_BITMAP_LEN (MAX_PXM_DOMAINS / 8) +static u8 __initdata pxm_bitmap[PXM_BITMAP_LEN]; /* bitmap of proximity domains */ + +#define MAX_CHUNKS_PER_NODE 3 +#define MAXCHUNKS (MAX_CHUNKS_PER_NODE * MAX_NUMNODES) +struct node_memory_chunk_s { + unsigned long start_pfn; + unsigned long end_pfn; + u8 pxm; // proximity domain of node + u8 nid; // which cnode contains this chunk? + u8 bank; // which mem bank on this node +}; +static struct node_memory_chunk_s __initdata node_memory_chunk[MAXCHUNKS]; + +static int __initdata num_memory_chunks; /* total number of memory chunks */ +static u8 __initdata apicid_to_pxm[MAX_APICID]; + +int numa_off __initdata; +int acpi_numa __initdata; + +static __init void bad_srat(void) +{ + printk(KERN_ERR "SRAT: SRAT not used.\n"); + acpi_numa = -1; + num_memory_chunks = 0; +} + +static __init inline int srat_disabled(void) +{ + return numa_off || acpi_numa < 0; +} + +/* Identify CPU proximity domains */ +void __init +acpi_numa_processor_affinity_init(struct acpi_srat_cpu_affinity *cpu_affinity) +{ + if (srat_disabled()) + return; + if (cpu_affinity->header.length != + sizeof(struct acpi_srat_cpu_affinity)) { + bad_srat(); + return; + } + + if ((cpu_affinity->flags & ACPI_SRAT_CPU_ENABLED) == 0) + return; /* empty entry */ + + /* mark this node as "seen" in node bitmap */ + BMAP_SET(pxm_bitmap, cpu_affinity->proximity_domain_lo); + + apicid_to_pxm[cpu_affinity->apic_id] = cpu_affinity->proximity_domain_lo; + + printk(KERN_DEBUG "CPU %02x in proximity domain %02x\n", + cpu_affinity->apic_id, cpu_affinity->proximity_domain_lo); +} + +/* + * Identify memory proximity domains and hot-remove capabilities. + * Fill node memory chunk list structure. + */ +void __init +acpi_numa_memory_affinity_init(struct acpi_srat_mem_affinity *memory_affinity) +{ + unsigned long long paddr, size; + unsigned long start_pfn, end_pfn; + u8 pxm; + struct node_memory_chunk_s *p, *q, *pend; + + if (srat_disabled()) + return; + if (memory_affinity->header.length != + sizeof(struct acpi_srat_mem_affinity)) { + bad_srat(); + return; + } + + if ((memory_affinity->flags & ACPI_SRAT_MEM_ENABLED) == 0) + return; /* empty entry */ + + pxm = memory_affinity->proximity_domain & 0xff; + + /* mark this node as "seen" in node bitmap */ + BMAP_SET(pxm_bitmap, pxm); + + /* calculate info for memory chunk structure */ + paddr = memory_affinity->base_address; + size = memory_affinity->length; + + start_pfn = paddr >> PAGE_SHIFT; + end_pfn = (paddr + size) >> PAGE_SHIFT; + + + if (num_memory_chunks >= MAXCHUNKS) { + printk(KERN_WARNING "Too many mem chunks in SRAT." + " Ignoring %lld MBytes at %llx\n", + size/(1024*1024), paddr); + return; + } + + /* Insertion sort based on base address */ + pend = &node_memory_chunk[num_memory_chunks]; + for (p = &node_memory_chunk[0]; p < pend; p++) { + if (start_pfn < p->start_pfn) + break; + } + if (p < pend) { + for (q = pend; q >= p; q--) + *(q + 1) = *q; + } + p->start_pfn = start_pfn; + p->end_pfn = end_pfn; + p->pxm = pxm; + + num_memory_chunks++; + + printk(KERN_DEBUG "Memory range %08lx to %08lx" + " in proximity domain %02x %s\n", + start_pfn, end_pfn, + pxm, + ((memory_affinity->flags & ACPI_SRAT_MEM_HOT_PLUGGABLE) ? + "enabled and removable" : "enabled" ) ); +} + +/* Callback for SLIT parsing */ +void __init acpi_numa_slit_init(struct acpi_table_slit *slit) +{ +} + +void acpi_numa_arch_fixup(void) +{ +} +/* + * The SRAT table always lists ascending addresses, so can always + * assume that the first "start" address that you see is the real + * start of the node, and that the current "end" address is after + * the previous one. + */ +static __init void node_read_chunk(int nid, struct node_memory_chunk_s *memory_chunk) +{ + /* + * Only add present memory as told by the e820. + * There is no guarantee from the SRAT that the memory it + * enumerates is present at boot time because it represents + * *possible* memory hotplug areas the same as normal RAM. + */ + if (memory_chunk->start_pfn >= max_pfn) { + printk(KERN_INFO "Ignoring SRAT pfns: %08lx - %08lx\n", + memory_chunk->start_pfn, memory_chunk->end_pfn); + return; + } + if (memory_chunk->nid != nid) + return; + + if (!node_has_online_mem(nid)) + node_start_pfn[nid] = memory_chunk->start_pfn; + + if (node_start_pfn[nid] > memory_chunk->start_pfn) + node_start_pfn[nid] = memory_chunk->start_pfn; + + if (node_end_pfn[nid] < memory_chunk->end_pfn) + node_end_pfn[nid] = memory_chunk->end_pfn; +} + +int __init get_memcfg_from_srat(void) +{ + int i, j, nid; + + + if (srat_disabled()) + goto out_fail; + + if (num_memory_chunks == 0) { + printk(KERN_WARNING + "could not finy any ACPI SRAT memory areas.\n"); + goto out_fail; + } + + /* Calculate total number of nodes in system from PXM bitmap and create + * a set of sequential node IDs starting at zero. (ACPI doesn't seem + * to specify the range of _PXM values.) + */ + /* + * MCD - we no longer HAVE to number nodes sequentially. PXM domain + * numbers could go as high as 256, and MAX_NUMNODES for i386 is typically + * 32, so we will continue numbering them in this manner until MAX_NUMNODES + * approaches MAX_PXM_DOMAINS for i386. + */ + nodes_clear(node_online_map); + for (i = 0; i < MAX_PXM_DOMAINS; i++) { + if (BMAP_TEST(pxm_bitmap, i)) { + int nid = acpi_map_pxm_to_node(i); + node_set_online(nid); + } + } + BUG_ON(num_online_nodes() == 0); + + /* set cnode id in memory chunk structure */ + for (i = 0; i < num_memory_chunks; i++) + node_memory_chunk[i].nid = pxm_to_node(node_memory_chunk[i].pxm); + + printk(KERN_DEBUG "pxm bitmap: "); + for (i = 0; i < sizeof(pxm_bitmap); i++) { + printk(KERN_CONT "%02x ", pxm_bitmap[i]); + } + printk(KERN_CONT "\n"); + printk(KERN_DEBUG "Number of logical nodes in system = %d\n", + num_online_nodes()); + printk(KERN_DEBUG "Number of memory chunks in system = %d\n", + num_memory_chunks); + + for (i = 0; i < MAX_APICID; i++) + apicid_2_node[i] = pxm_to_node(apicid_to_pxm[i]); + + for (j = 0; j < num_memory_chunks; j++){ + struct node_memory_chunk_s * chunk = &node_memory_chunk[j]; + printk(KERN_DEBUG + "chunk %d nid %d start_pfn %08lx end_pfn %08lx\n", + j, chunk->nid, chunk->start_pfn, chunk->end_pfn); + node_read_chunk(chunk->nid, chunk); + e820_register_active_regions(chunk->nid, chunk->start_pfn, + min(chunk->end_pfn, max_pfn)); + } + + for_each_online_node(nid) { + unsigned long start = node_start_pfn[nid]; + unsigned long end = min(node_end_pfn[nid], max_pfn); + + memory_present(nid, start, end); + node_remap_size[nid] = node_memmap_size_bytes(nid, start, end); + } + return 1; +out_fail: + printk(KERN_ERR "failed to get NUMA memory information from SRAT" + " table\n"); + return 0; +} diff --git a/arch/x86/mm/srat_64.c b/arch/x86/mm/srat_64.c index 99649dccad28..1b4763e26ea9 100644 --- a/arch/x86/mm/srat_64.c +++ b/arch/x86/mm/srat_64.c @@ -100,7 +100,19 @@ static __init inline int srat_disabled(void) /* Callback for SLIT parsing */ void __init acpi_numa_slit_init(struct acpi_table_slit *slit) { - acpi_slit = slit; + unsigned length; + unsigned long phys; + + length = slit->header.length; + phys = find_e820_area(0, max_pfn_mapped<<PAGE_SHIFT, length, + PAGE_SIZE); + + if (phys == -1L) + panic(" Can not save slit!\n"); + + acpi_slit = __va(phys); + memcpy(acpi_slit, slit, length); + reserve_early(phys, phys + length, "ACPI SLIT"); } /* Callback for Proximity Domain -> LAPIC mapping */ @@ -299,7 +311,7 @@ static int __init nodes_cover_memory(const struct bootnode *nodes) pxmram = 0; } - e820ram = end_pfn - absent_pages_in_range(0, end_pfn); + e820ram = max_pfn - absent_pages_in_range(0, max_pfn); /* We seem to lose 3 pages somewhere. Allow a bit of slack. */ if ((long)(e820ram - pxmram) >= 1*1024*1024) { printk(KERN_ERR @@ -376,7 +388,7 @@ int __init acpi_scan_nodes(unsigned long start, unsigned long end) if (node == NUMA_NO_NODE) continue; if (!node_isset(node, node_possible_map)) - numa_set_node(i, NUMA_NO_NODE); + numa_clear_node(i); } numa_init_array(); return 0; @@ -495,6 +507,7 @@ int __node_distance(int a, int b) EXPORT_SYMBOL(__node_distance); +#if defined(CONFIG_MEMORY_HOTPLUG_SPARSE) || defined(CONFIG_ACPI_HOTPLUG_MEMORY) int memory_add_physaddr_to_nid(u64 start) { int i, ret = 0; @@ -506,4 +519,4 @@ int memory_add_physaddr_to_nid(u64 start) return ret; } EXPORT_SYMBOL_GPL(memory_add_physaddr_to_nid); - +#endif diff --git a/arch/x86/mm/testmmiotrace.c b/arch/x86/mm/testmmiotrace.c new file mode 100644 index 000000000000..d877c5b423ef --- /dev/null +++ b/arch/x86/mm/testmmiotrace.c @@ -0,0 +1,71 @@ +/* + * Written by Pekka Paalanen, 2008 <pq@iki.fi> + */ +#include <linux/module.h> +#include <linux/io.h> + +#define MODULE_NAME "testmmiotrace" + +static unsigned long mmio_address; +module_param(mmio_address, ulong, 0); +MODULE_PARM_DESC(mmio_address, "Start address of the mapping of 16 kB."); + +static void do_write_test(void __iomem *p) +{ + unsigned int i; + for (i = 0; i < 256; i++) + iowrite8(i, p + i); + for (i = 1024; i < (5 * 1024); i += 2) + iowrite16(i * 12 + 7, p + i); + for (i = (5 * 1024); i < (16 * 1024); i += 4) + iowrite32(i * 212371 + 13, p + i); +} + +static void do_read_test(void __iomem *p) +{ + unsigned int i; + for (i = 0; i < 256; i++) + ioread8(p + i); + for (i = 1024; i < (5 * 1024); i += 2) + ioread16(p + i); + for (i = (5 * 1024); i < (16 * 1024); i += 4) + ioread32(p + i); +} + +static void do_test(void) +{ + void __iomem *p = ioremap_nocache(mmio_address, 0x4000); + if (!p) { + pr_err(MODULE_NAME ": could not ioremap, aborting.\n"); + return; + } + do_write_test(p); + do_read_test(p); + iounmap(p); +} + +static int __init init(void) +{ + if (mmio_address == 0) { + pr_err(MODULE_NAME ": you have to use the module argument " + "mmio_address.\n"); + pr_err(MODULE_NAME ": DO NOT LOAD THIS MODULE UNLESS" + " YOU REALLY KNOW WHAT YOU ARE DOING!\n"); + return -ENXIO; + } + + pr_warning(MODULE_NAME ": WARNING: mapping 16 kB @ 0x%08lx " + "in PCI address space, and writing " + "rubbish in there.\n", mmio_address); + do_test(); + return 0; +} + +static void __exit cleanup(void) +{ + pr_debug(MODULE_NAME ": unloaded.\n"); +} + +module_init(init); +module_exit(cleanup); +MODULE_LICENSE("GPL"); |
