X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=arch%2Fia64%2Fmm%2Finit.c;h=f6a035550aebbfe59cae49e55f448bc5bb56be19;hb=refs%2Fheads%2Fvserver;hp=c03b3980d9f01bf9ee215e40160ac88794818979;hpb=9213980e6a70d8473e0ffd4b39ab5b6caaba9ff5;p=linux-2.6.git diff --git a/arch/ia64/mm/init.c b/arch/ia64/mm/init.c index c03b3980d..f6a035550 100644 --- a/arch/ia64/mm/init.c +++ b/arch/ia64/mm/init.c @@ -4,7 +4,6 @@ * Copyright (C) 1998-2003 Hewlett-Packard Co * David Mosberger-Tang */ -#include #include #include @@ -19,9 +18,9 @@ #include #include #include +#include #include -#include #include #include #include @@ -39,6 +38,9 @@ DEFINE_PER_CPU(struct mmu_gather, mmu_gathers); +DEFINE_PER_CPU(unsigned long *, __pgtable_quicklist); +DEFINE_PER_CPU(long, __pgtable_quicklist_size); + extern void ia64_tlb_init (void); unsigned long MAX_DMA_ADDRESS = PAGE_OFFSET + 0x100000000UL; @@ -50,53 +52,86 @@ struct page *vmem_map; EXPORT_SYMBOL(vmem_map); #endif -static int pgt_cache_water[2] = { 25, 50 }; - -struct page *zero_page_memmap_ptr; /* map entry for zero page */ +struct page *zero_page_memmap_ptr; /* map entry for zero page */ EXPORT_SYMBOL(zero_page_memmap_ptr); +#define MIN_PGT_PAGES 25UL +#define MAX_PGT_FREES_PER_PASS 16L +#define PGT_FRACTION_OF_NODE_MEM 16 + +static inline long +max_pgt_pages(void) +{ + u64 node_free_pages, max_pgt_pages; + +#ifndef CONFIG_NUMA + node_free_pages = nr_free_pages(); +#else + node_free_pages = nr_free_pages_pgdat(NODE_DATA(numa_node_id())); +#endif + max_pgt_pages = node_free_pages / PGT_FRACTION_OF_NODE_MEM; + max_pgt_pages = max(max_pgt_pages, MIN_PGT_PAGES); + return max_pgt_pages; +} + +static inline long +min_pages_to_free(void) +{ + long pages_to_free; + + pages_to_free = pgtable_quicklist_size - max_pgt_pages(); + pages_to_free = min(pages_to_free, MAX_PGT_FREES_PER_PASS); + return pages_to_free; +} + void -check_pgt_cache (void) +check_pgt_cache(void) { - int low, high; - - low = pgt_cache_water[0]; - high = pgt_cache_water[1]; - - if (pgtable_cache_size > (u64) high) { - do { - if (pgd_quicklist) - free_page((unsigned long)pgd_alloc_one_fast(0)); - if (pmd_quicklist) - free_page((unsigned long)pmd_alloc_one_fast(0, 0)); - } while (pgtable_cache_size > (u64) low); + long pages_to_free; + + if (unlikely(pgtable_quicklist_size <= MIN_PGT_PAGES)) + return; + + preempt_disable(); + while (unlikely((pages_to_free = min_pages_to_free()) > 0)) { + while (pages_to_free--) { + free_page((unsigned long)pgtable_quicklist_alloc()); + } + preempt_enable(); + preempt_disable(); } + preempt_enable(); } void -update_mmu_cache (struct vm_area_struct *vma, unsigned long vaddr, pte_t pte) +lazy_mmu_prot_update (pte_t pte) { unsigned long addr; struct page *page; + unsigned long order; if (!pte_exec(pte)) return; /* not an executable page... */ page = pte_page(pte); - /* don't use VADDR: it may not be mapped on this CPU (or may have just been flushed): */ addr = (unsigned long) page_address(page); if (test_bit(PG_arch_1, &page->flags)) return; /* i-cache is already coherent with d-cache */ - flush_icache_range(addr, addr + PAGE_SIZE); + if (PageCompound(page)) { + order = (unsigned long) (page[1].lru.prev); + flush_icache_range(addr, addr + (1UL << order << PAGE_SHIFT)); + } + else + flush_icache_range(addr, addr + PAGE_SIZE); set_bit(PG_arch_1, &page->flags); /* mark page as clean */ } inline void ia64_set_rbs_bot (void) { - unsigned long stack_size = current->rlim[RLIMIT_STACK].rlim_max & -16; + unsigned long stack_size = current->signal->rlim[RLIMIT_STACK].rlim_max & -16; if (stack_size > MAX_USER_STACK_SIZE) stack_size = MAX_USER_STACK_SIZE; @@ -121,27 +156,39 @@ ia64_init_addr_space (void) * the problem. When the process attempts to write to the register backing store * for the first time, it will get a SEGFAULT in this case. */ - vma = kmem_cache_alloc(vm_area_cachep, SLAB_KERNEL); + vma = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL); if (vma) { memset(vma, 0, sizeof(*vma)); vma->vm_mm = current->mm; vma->vm_start = current->thread.rbs_bot & PAGE_MASK; vma->vm_end = vma->vm_start + PAGE_SIZE; vma->vm_page_prot = protection_map[VM_DATA_DEFAULT_FLAGS & 0x7]; - vma->vm_flags = VM_READ|VM_WRITE|VM_MAYREAD|VM_MAYWRITE|VM_GROWSUP; - insert_vm_struct(current->mm, vma); + vma->vm_flags = VM_DATA_DEFAULT_FLAGS|VM_GROWSUP|VM_ACCOUNT; + down_write(¤t->mm->mmap_sem); + if (insert_vm_struct(current->mm, vma)) { + up_write(¤t->mm->mmap_sem); + kmem_cache_free(vm_area_cachep, vma); + return; + } + up_write(¤t->mm->mmap_sem); } /* map NaT-page at address zero to speed up speculative dereferencing of NULL: */ if (!(current->personality & MMAP_PAGE_ZERO)) { - vma = kmem_cache_alloc(vm_area_cachep, SLAB_KERNEL); + vma = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL); if (vma) { memset(vma, 0, sizeof(*vma)); vma->vm_mm = current->mm; vma->vm_end = PAGE_SIZE; vma->vm_page_prot = __pgprot(pgprot_val(PAGE_READONLY) | _PAGE_MA_NAT); vma->vm_flags = VM_READ | VM_MAYREAD | VM_IO | VM_RESERVED; - insert_vm_struct(current->mm, vma); + down_write(¤t->mm->mmap_sem); + if (insert_vm_struct(current->mm, vma)) { + up_write(¤t->mm->mmap_sem); + kmem_cache_free(vm_area_cachep, vma); + return; + } + up_write(¤t->mm->mmap_sem); } } } @@ -155,7 +202,7 @@ free_initmem (void) eaddr = (unsigned long) ia64_imva(__init_end); while (addr < eaddr) { ClearPageReserved(virt_to_page(addr)); - set_page_count(virt_to_page(addr), 1); + init_page_count(virt_to_page(addr)); free_page(addr); ++totalram_pages; addr += PAGE_SIZE; @@ -164,12 +211,12 @@ free_initmem (void) (__init_end - __init_begin) >> 10); } -void +void __init free_initrd_mem (unsigned long start, unsigned long end) { struct page *page; /* - * EFI uses 4KB pages while the kernel can use 4KB or bigger. + * EFI uses 4KB pages while the kernel can use 4KB or bigger. * Thus EFI and the kernel may have different page sizes. It is * therefore possible to have the initrd share the same page as * the end of the kernel (given current setup). @@ -210,19 +257,27 @@ free_initrd_mem (unsigned long start, unsigned long end) continue; page = virt_to_page(start); ClearPageReserved(page); - set_page_count(page, 1); + init_page_count(page); free_page(start); ++totalram_pages; } } +int page_is_ram(unsigned long pagenr) +{ + //FIXME: implement w/efi walk + printk("page is ram is called!!!!!\n"); + return 1; +} + /* * This installs a clean page in the kernel's page table. */ -struct page * +static struct page * __init put_kernel_page (struct page *page, unsigned long address, pgprot_t pgprot) { pgd_t *pgd; + pud_t *pud; pmd_t *pmd; pte_t *pte; @@ -232,34 +287,34 @@ put_kernel_page (struct page *page, unsigned long address, pgprot_t pgprot) pgd = pgd_offset_k(address); /* note: this is NOT pgd_offset()! */ - spin_lock(&init_mm.page_table_lock); { - pmd = pmd_alloc(&init_mm, pgd, address); + pud = pud_alloc(&init_mm, pgd, address); + if (!pud) + goto out; + pmd = pmd_alloc(&init_mm, pud, address); if (!pmd) goto out; - pte = pte_alloc_map(&init_mm, pmd, address); + pte = pte_alloc_kernel(pmd, address); if (!pte) goto out; - if (!pte_none(*pte)) { - pte_unmap(pte); + if (!pte_none(*pte)) goto out; - } set_pte(pte, mk_pte(page, pgprot)); - pte_unmap(pte); } - out: spin_unlock(&init_mm.page_table_lock); + out: /* no need for flush_tlb */ return page; } -static void +static void __init setup_gate (void) { struct page *page; /* - * Map the gate page twice: once read-only to export the ELF headers etc. and once - * execute-only page to enable privilege-promotion via "epc": + * Map the gate page twice: once read-only to export the ELF + * headers etc. and once execute-only page to enable + * privilege-promotion via "epc": */ page = virt_to_page(ia64_imva(__start_gate_section)); put_kernel_page(page, GATE_ADDR, PAGE_READONLY); @@ -268,6 +323,20 @@ setup_gate (void) put_kernel_page(page, GATE_ADDR + PAGE_SIZE, PAGE_GATE); #else put_kernel_page(page, GATE_ADDR + PERCPU_PAGE_SIZE, PAGE_GATE); + /* Fill in the holes (if any) with read-only zero pages: */ + { + unsigned long addr; + + for (addr = GATE_ADDR + PAGE_SIZE; + addr < GATE_ADDR + PERCPU_PAGE_SIZE; + addr += PAGE_SIZE) + { + put_kernel_page(ZERO_PAGE(0), addr, + PAGE_READONLY); + put_kernel_page(ZERO_PAGE(0), addr + PERCPU_PAGE_SIZE, + PAGE_READONLY); + } + } #endif ia64_patch_gate(); } @@ -277,7 +346,6 @@ ia64_mmu_init (void *my_cpu_data) { unsigned long psr, pta, impl_va_bits; extern void __devinit tlb_init (void); - int cpu; #ifdef CONFIG_DISABLE_VHPT # define VHPT_ENABLE_BIT 0 @@ -321,13 +389,22 @@ ia64_mmu_init (void *my_cpu_data) if (impl_va_bits < 51 || impl_va_bits > 61) panic("CPU has bogus IMPL_VA_MSB value of %lu!\n", impl_va_bits - 1); + /* + * mapped_space_bits - PAGE_SHIFT is the total number of ptes we need, + * which must fit into "vmlpt_bits - pte_bits" slots. Second half of + * the test makes sure that our mapped space doesn't overlap the + * unimplemented hole in the middle of the region. + */ + if ((mapped_space_bits - PAGE_SHIFT > vmlpt_bits - pte_bits) || + (mapped_space_bits > impl_va_bits - 1)) + panic("Cannot build a big enough virtual-linear page table" + " to cover mapped address space.\n" + " Try using a smaller page size.\n"); + /* place the VMLPT at the end of each page-table mapped region: */ pta = POW2(61) - POW2(vmlpt_bits); - if (POW2(mapped_space_bits) >= pta) - panic("mm/init: overlap between virtually mapped linear page table and " - "mapped kernel space!"); /* * Set the (virtually mapped linear) page table address. Bit * 8 selects between the short and long format, bits 2-7 the @@ -342,31 +419,73 @@ ia64_mmu_init (void *my_cpu_data) ia64_set_rr(HPAGE_REGION_BASE, HPAGE_SHIFT << 2); ia64_srlz_d(); #endif +} - cpu = smp_processor_id(); +#ifdef CONFIG_VIRTUAL_MEM_MAP +int vmemmap_find_next_valid_pfn(int node, int i) +{ + unsigned long end_address, hole_next_pfn; + unsigned long stop_address; + pg_data_t *pgdat = NODE_DATA(node); - /* mca handler uses cr.lid as key to pick the right entry */ - ia64_mca_tlb_list[cpu].cr_lid = ia64_getreg(_IA64_REG_CR_LID); + end_address = (unsigned long) &vmem_map[pgdat->node_start_pfn + i]; + end_address = PAGE_ALIGN(end_address); - /* insert this percpu data information into our list for MCA recovery purposes */ - ia64_mca_tlb_list[cpu].percpu_paddr = pte_val(mk_pte_phys(__pa(my_cpu_data), PAGE_KERNEL)); - /* Also save per-cpu tlb flush recipe for use in physical mode mca handler */ - ia64_mca_tlb_list[cpu].ptce_base = local_cpu_data->ptce_base; - ia64_mca_tlb_list[cpu].ptce_count[0] = local_cpu_data->ptce_count[0]; - ia64_mca_tlb_list[cpu].ptce_count[1] = local_cpu_data->ptce_count[1]; - ia64_mca_tlb_list[cpu].ptce_stride[0] = local_cpu_data->ptce_stride[0]; - ia64_mca_tlb_list[cpu].ptce_stride[1] = local_cpu_data->ptce_stride[1]; -} + stop_address = (unsigned long) &vmem_map[ + pgdat->node_start_pfn + pgdat->node_spanned_pages]; -#ifdef CONFIG_VIRTUAL_MEM_MAP + do { + pgd_t *pgd; + pud_t *pud; + pmd_t *pmd; + pte_t *pte; -int + pgd = pgd_offset_k(end_address); + if (pgd_none(*pgd)) { + end_address += PGDIR_SIZE; + continue; + } + + pud = pud_offset(pgd, end_address); + if (pud_none(*pud)) { + end_address += PUD_SIZE; + continue; + } + + pmd = pmd_offset(pud, end_address); + if (pmd_none(*pmd)) { + end_address += PMD_SIZE; + continue; + } + + pte = pte_offset_kernel(pmd, end_address); +retry_pte: + if (pte_none(*pte)) { + end_address += PAGE_SIZE; + pte++; + if ((end_address < stop_address) && + (end_address != ALIGN(end_address, 1UL << PMD_SHIFT))) + goto retry_pte; + continue; + } + /* Found next valid vmem_map page */ + break; + } while (end_address < stop_address); + + end_address = min(end_address, stop_address); + end_address = end_address - (unsigned long) vmem_map + sizeof(struct page) - 1; + hole_next_pfn = end_address / sizeof(struct page); + return hole_next_pfn - pgdat->node_start_pfn; +} + +int __init create_mem_map_page_table (u64 start, u64 end, void *arg) { unsigned long address, start_page, end_page; struct page *map_start, *map_end; int node; pgd_t *pgd; + pud_t *pud; pmd_t *pmd; pte_t *pte; @@ -381,7 +500,11 @@ create_mem_map_page_table (u64 start, u64 end, void *arg) pgd = pgd_offset_k(address); if (pgd_none(*pgd)) pgd_populate(&init_mm, pgd, alloc_bootmem_pages_node(NODE_DATA(node), PAGE_SIZE)); - pmd = pmd_offset(pgd, address); + pud = pud_offset(pgd, address); + + if (pud_none(*pud)) + pud_populate(&init_mm, pud, alloc_bootmem_pages_node(NODE_DATA(node), PAGE_SIZE)); + pmd = pmd_offset(pud, address); if (pmd_none(*pmd)) pmd_populate_kernel(&init_mm, pmd, alloc_bootmem_pages_node(NODE_DATA(node), PAGE_SIZE)); @@ -408,7 +531,6 @@ virtual_memmap_init (u64 start, u64 end, void *arg) struct page *map_start, *map_end; args = (struct memmap_init_callback_data *) arg; - map_start = vmem_map + (__pa(start) >> PAGE_SHIFT); map_end = vmem_map + (__pa(end) >> PAGE_SHIFT); @@ -427,20 +549,23 @@ virtual_memmap_init (u64 start, u64 end, void *arg) / sizeof(struct page)); if (map_start < map_end) - memmap_init_zone(map_start, (unsigned long) (map_end - map_start), - args->nid, args->zone, page_to_pfn(map_start)); + memmap_init_zone((unsigned long)(map_end - map_start), + args->nid, args->zone, page_to_pfn(map_start), + MEMMAP_EARLY); return 0; } void -memmap_init (struct page *start, unsigned long size, int nid, - unsigned long zone, unsigned long start_pfn) +memmap_init (unsigned long size, int nid, unsigned long zone, + unsigned long start_pfn) { if (!vmem_map) - memmap_init_zone(start, size, nid, zone, start_pfn); + memmap_init_zone(size, nid, zone, start_pfn, MEMMAP_EARLY); else { + struct page *start; struct memmap_init_callback_data args; + start = pfn_to_page(start_pfn); args.start = start; args.end = start + size; args.nid = nid; @@ -456,13 +581,13 @@ ia64_pfn_valid (unsigned long pfn) char byte; struct page *pg = pfn_to_page(pfn); - return (__get_user(byte, (char *) pg) == 0) + return (__get_user(byte, (char __user *) pg) == 0) && ((((u64)pg & PAGE_MASK) == (((u64)(pg + 1) - 1) & PAGE_MASK)) - || (__get_user(byte, (char *) (pg + 1) - 1) == 0)); + || (__get_user(byte, (char __user *) (pg + 1) - 1) == 0)); } EXPORT_SYMBOL(ia64_pfn_valid); -int +int __init find_largest_hole (u64 start, u64 end, void *arg) { u64 *max_gap = arg; @@ -476,9 +601,16 @@ find_largest_hole (u64 start, u64 end, void *arg) last_end = end; return 0; } + +int __init +register_active_ranges(u64 start, u64 end, void *arg) +{ + add_active_range(0, __pa(start) >> PAGE_SHIFT, __pa(end) >> PAGE_SHIFT); + return 0; +} #endif /* CONFIG_VIRTUAL_MEM_MAP */ -static int +static int __init count_reserved_pages (u64 start, u64 end, void *arg) { unsigned long num_reserved = 0; @@ -499,7 +631,7 @@ count_reserved_pages (u64 start, u64 end, void *arg) * purposes. */ -static int nolwsys; +static int nolwsys __initdata; static int __init nolwsys_setup (char *s) @@ -510,15 +642,18 @@ nolwsys_setup (char *s) __setup("nolwsys", nolwsys_setup); -void +void __init mem_init (void) { long reserved_pages, codesize, datasize, initsize; - unsigned long num_pgt_pages; pg_data_t *pgdat; int i; static struct kcore_list kcore_mem, kcore_vmem, kcore_kernel; + BUG_ON(PTRS_PER_PGD * sizeof(pgd_t) != PAGE_SIZE); + BUG_ON(PTRS_PER_PMD * sizeof(pmd_t) != PAGE_SIZE); + BUG_ON(PTRS_PER_PTE * sizeof(pte_t) != PAGE_SIZE); + #ifdef CONFIG_PCI /* * This needs to be called _after_ the command line has been parsed but _before_ @@ -528,7 +663,7 @@ mem_init (void) platform_dma_init(); #endif -#ifndef CONFIG_DISCONTIGMEM +#ifdef CONFIG_FLATMEM if (!mem_map) BUG(); max_mapnr = max_low_pfn; @@ -540,8 +675,9 @@ mem_init (void) kclist_add(&kcore_vmem, (void *)VMALLOC_START, VMALLOC_END-VMALLOC_START); kclist_add(&kcore_kernel, _stext, _end - _stext); - for_each_pgdat(pgdat) - totalram_pages += free_all_bootmem_node(pgdat); + for_each_online_pgdat(pgdat) + if (pgdat->bdata->node_bootmem_map) + totalram_pages += free_all_bootmem_node(pgdat); reserved_pages = 0; efi_memmap_walk(count_reserved_pages, &reserved_pages); @@ -555,18 +691,6 @@ mem_init (void) num_physpages << (PAGE_SHIFT - 10), codesize >> 10, reserved_pages << (PAGE_SHIFT - 10), datasize >> 10, initsize >> 10); - /* - * Allow for enough (cached) page table pages so that we can map the entire memory - * at least once. Each task also needs a couple of page tables pages, so add in a - * fudge factor for that (don't use "threads-max" here; that would be wrong!). - * Don't allow the cache to be more than 10% of total memory, though. - */ -# define NUM_TASKS 500 /* typical number of tasks */ - num_pgt_pages = nr_free_pages() / PTRS_PER_PGD + NUM_TASKS; - if (num_pgt_pages > nr_free_pages() / 10) - num_pgt_pages = nr_free_pages() / 10; - if (num_pgt_pages > (u64) pgt_cache_water[1]) - pgt_cache_water[1] = num_pgt_pages; /* * For fsyscall entrpoints with no light-weight handler, use the ordinary @@ -580,9 +704,46 @@ mem_init (void) if (!fsyscall_table[i] || nolwsys) fsyscall_table[i] = sys_call_table[i] | 1; } - setup_gate(); /* setup gate pages before we free up boot memory... */ + setup_gate(); #ifdef CONFIG_IA32_SUPPORT - ia32_boot_gdt_init(); + ia32_mem_init(); #endif } + +#ifdef CONFIG_MEMORY_HOTPLUG +void online_page(struct page *page) +{ + ClearPageReserved(page); + init_page_count(page); + __free_page(page); + totalram_pages++; + num_physpages++; +} + +int arch_add_memory(int nid, u64 start, u64 size) +{ + pg_data_t *pgdat; + struct zone *zone; + unsigned long start_pfn = start >> PAGE_SHIFT; + unsigned long nr_pages = size >> PAGE_SHIFT; + int ret; + + pgdat = NODE_DATA(nid); + + zone = pgdat->node_zones + ZONE_NORMAL; + ret = __add_pages(zone, start_pfn, nr_pages); + + if (ret) + printk("%s: Problem encountered in __add_pages() as ret=%d\n", + __FUNCTION__, ret); + + return ret; +} + +int remove_memory(u64 start, u64 size) +{ + return -EINVAL; +} +EXPORT_SYMBOL_GPL(remove_memory); +#endif