diff options
author | Jeff Garzik <jgarzik@pobox.com> | 2005-10-30 20:37:44 -0500 |
---|---|---|
committer | Jeff Garzik <jgarzik@pobox.com> | 2005-10-30 20:37:44 -0500 |
commit | 9e0cb06b17be7e562cbdaba2768649f025826dc6 (patch) | |
tree | aaf5ef8c6cd11764d222df9c446ad9af17e0020e /arch | |
parent | 23da0c20ef1c1f0432f373e0e2233a6b6ab2678f (diff) | |
parent | 6e9d6b8ee4e0c37d3952256e6472c57490d6780d (diff) |
Merge branch 'master'
Diffstat (limited to 'arch')
60 files changed, 390 insertions, 1544 deletions
diff --git a/arch/alpha/mm/numa.c b/arch/alpha/mm/numa.c index c7481d59b6d..6d5251254f6 100644 --- a/arch/alpha/mm/numa.c +++ b/arch/alpha/mm/numa.c @@ -371,6 +371,8 @@ show_mem(void) show_free_areas(); printk("Free swap: %6ldkB\n", nr_swap_pages<<(PAGE_SHIFT-10)); for_each_online_node(nid) { + unsigned long flags; + pgdat_resize_lock(NODE_DATA(nid), &flags); i = node_spanned_pages(nid); while (i-- > 0) { struct page *page = nid_page_nr(nid, i); @@ -384,6 +386,7 @@ show_mem(void) else shared += page_count(page) - 1; } + pgdat_resize_unlock(NODE_DATA(nid), &flags); } printk("%ld pages of RAM\n",total); printk("%ld free pages\n",free); diff --git a/arch/alpha/mm/remap.c b/arch/alpha/mm/remap.c index 19817ad3d89..a78356c3ead 100644 --- a/arch/alpha/mm/remap.c +++ b/arch/alpha/mm/remap.c @@ -2,7 +2,6 @@ #include <asm/pgalloc.h> #include <asm/cacheflush.h> -/* called with the page_table_lock held */ static inline void remap_area_pte(pte_t * pte, unsigned long address, unsigned long size, unsigned long phys_addr, unsigned long flags) @@ -31,7 +30,6 @@ remap_area_pte(pte_t * pte, unsigned long address, unsigned long size, } while (address && (address < end)); } -/* called with the page_table_lock held */ static inline int remap_area_pmd(pmd_t * pmd, unsigned long address, unsigned long size, unsigned long phys_addr, unsigned long flags) @@ -46,7 +44,7 @@ remap_area_pmd(pmd_t * pmd, unsigned long address, unsigned long size, if (address >= end) BUG(); do { - pte_t * pte = pte_alloc_kernel(&init_mm, pmd, address); + pte_t * pte = pte_alloc_kernel(pmd, address); if (!pte) return -ENOMEM; remap_area_pte(pte, address, end - address, @@ -70,7 +68,6 @@ __alpha_remap_area_pages(unsigned long address, unsigned long phys_addr, flush_cache_all(); if (address >= end) BUG(); - spin_lock(&init_mm.page_table_lock); do { pmd_t *pmd; pmd = pmd_alloc(&init_mm, dir, address); @@ -84,7 +81,6 @@ __alpha_remap_area_pages(unsigned long address, unsigned long phys_addr, address = (address + PGDIR_SIZE) & PGDIR_MASK; dir++; } while (address && (address < end)); - spin_unlock(&init_mm.page_table_lock); return error; } diff --git a/arch/arm/kernel/signal.c b/arch/arm/kernel/signal.c index a94d75fef59..a917e3dd366 100644 --- a/arch/arm/kernel/signal.c +++ b/arch/arm/kernel/signal.c @@ -139,93 +139,33 @@ struct iwmmxt_sigframe { unsigned long storage[0x98/4]; }; -static int page_present(struct mm_struct *mm, void __user *uptr, int wr) -{ - unsigned long addr = (unsigned long)uptr; - pgd_t *pgd = pgd_offset(mm, addr); - if (pgd_present(*pgd)) { - pmd_t *pmd = pmd_offset(pgd, addr); - if (pmd_present(*pmd)) { - pte_t *pte = pte_offset_map(pmd, addr); - return (pte_present(*pte) && (!wr || pte_write(*pte))); - } - } - return 0; -} - -static int copy_locked(void __user *uptr, void *kptr, size_t size, int write, - void (*copyfn)(void *, void __user *)) -{ - unsigned char v, __user *userptr = uptr; - int err = 0; - - do { - struct mm_struct *mm; - - if (write) { - __put_user_error(0, userptr, err); - __put_user_error(0, userptr + size - 1, err); - } else { - __get_user_error(v, userptr, err); - __get_user_error(v, userptr + size - 1, err); - } - - if (err) - break; - - mm = current->mm; - spin_lock(&mm->page_table_lock); - if (page_present(mm, userptr, write) && - page_present(mm, userptr + size - 1, write)) { - copyfn(kptr, uptr); - } else - err = 1; - spin_unlock(&mm->page_table_lock); - } while (err); - - return err; -} - static int preserve_iwmmxt_context(struct iwmmxt_sigframe *frame) { - int err = 0; + char kbuf[sizeof(*frame) + 8]; + struct iwmmxt_sigframe *kframe; /* the iWMMXt context must be 64 bit aligned */ - WARN_ON((unsigned long)frame & 7); - - __put_user_error(IWMMXT_MAGIC0, &frame->magic0, err); - __put_user_error(IWMMXT_MAGIC1, &frame->magic1, err); - - /* - * iwmmxt_task_copy() doesn't check user permissions. - * Let's do a dummy write on the upper boundary to ensure - * access to user mem is OK all way up. - */ - err |= copy_locked(&frame->storage, current_thread_info(), - sizeof(frame->storage), 1, iwmmxt_task_copy); - return err; + kframe = (struct iwmmxt_sigframe *)((unsigned long)(kbuf + 8) & ~7); + kframe->magic0 = IWMMXT_MAGIC0; + kframe->magic1 = IWMMXT_MAGIC1; + iwmmxt_task_copy(current_thread_info(), &kframe->storage); + return __copy_to_user(frame, kframe, sizeof(*frame)); } static int restore_iwmmxt_context(struct iwmmxt_sigframe *frame) { - unsigned long magic0, magic1; - int err = 0; + char kbuf[sizeof(*frame) + 8]; + struct iwmmxt_sigframe *kframe; - /* the iWMMXt context is 64 bit aligned */ - WARN_ON((unsigned long)frame & 7); - - /* - * Validate iWMMXt context signature. - * Also, iwmmxt_task_restore() doesn't check user permissions. - * Let's do a dummy write on the upper boundary to ensure - * access to user mem is OK all way up. - */ - __get_user_error(magic0, &frame->magic0, err); - __get_user_error(magic1, &frame->magic1, err); - if (!err && magic0 == IWMMXT_MAGIC0 && magic1 == IWMMXT_MAGIC1) - err = copy_locked(&frame->storage, current_thread_info(), - sizeof(frame->storage), 0, iwmmxt_task_restore); - return err; + /* the iWMMXt context must be 64 bit aligned */ + kframe = (struct iwmmxt_sigframe *)((unsigned long)(kbuf + 8) & ~7); + if (__copy_from_user(kframe, frame, sizeof(*frame))) + return -1; + if (kframe->magic0 != IWMMXT_MAGIC0 || + kframe->magic1 != IWMMXT_MAGIC1) + return -1; + iwmmxt_task_restore(current_thread_info(), &kframe->storage); + return 0; } #endif diff --git a/arch/arm/kernel/traps.c b/arch/arm/kernel/traps.c index baa09601a64..66e5a0516f2 100644 --- a/arch/arm/kernel/traps.c +++ b/arch/arm/kernel/traps.c @@ -483,29 +483,33 @@ asmlinkage int arm_syscall(int no, struct pt_regs *regs) unsigned long addr = regs->ARM_r2; struct mm_struct *mm = current->mm; pgd_t *pgd; pmd_t *pmd; pte_t *pte; + spinlock_t *ptl; regs->ARM_cpsr &= ~PSR_C_BIT; - spin_lock(&mm->page_table_lock); + down_read(&mm->mmap_sem); pgd = pgd_offset(mm, addr); if (!pgd_present(*pgd)) goto bad_access; pmd = pmd_offset(pgd, addr); if (!pmd_present(*pmd)) goto bad_access; - pte = pte_offset_map(pmd, addr); - if (!pte_present(*pte) || !pte_write(*pte)) + pte = pte_offset_map_lock(mm, pmd, addr, &ptl); + if (!pte_present(*pte) || !pte_write(*pte)) { + pte_unmap_unlock(pte, ptl); goto bad_access; + } val = *(unsigned long *)addr; val -= regs->ARM_r0; if (val == 0) { *(unsigned long *)addr = regs->ARM_r1; regs->ARM_cpsr |= PSR_C_BIT; } - spin_unlock(&mm->page_table_lock); + pte_unmap_unlock(pte, ptl); + up_read(&mm->mmap_sem); return val; bad_access: - spin_unlock(&mm->page_table_lock); + up_read(&mm->mmap_sem); /* simulate a write access fault */ do_DataAbort(addr, 15 + (1 << 11), regs); return -1; diff --git a/arch/arm/mm/consistent.c b/arch/arm/mm/consistent.c index 82f4d5e27c5..47b0b767f08 100644 --- a/arch/arm/mm/consistent.c +++ b/arch/arm/mm/consistent.c @@ -397,8 +397,6 @@ static int __init consistent_init(void) pte_t *pte; int ret = 0; - spin_lock(&init_mm.page_table_lock); - do { pgd = pgd_offset(&init_mm, CONSISTENT_BASE); pmd = pmd_alloc(&init_mm, pgd, CONSISTENT_BASE); @@ -409,7 +407,7 @@ static int __init consistent_init(void) } WARN_ON(!pmd_none(*pmd)); - pte = pte_alloc_kernel(&init_mm, pmd, CONSISTENT_BASE); + pte = pte_alloc_kernel(pmd, CONSISTENT_BASE); if (!pte) { printk(KERN_ERR "%s: no pte tables\n", __func__); ret = -ENOMEM; @@ -419,8 +417,6 @@ static int __init consistent_init(void) consistent_pte = pte; } while (0); - spin_unlock(&init_mm.page_table_lock); - return ret; } diff --git a/arch/arm/mm/fault-armv.c b/arch/arm/mm/fault-armv.c index be4ab3d73c9..7fc1b35a674 100644 --- a/arch/arm/mm/fault-armv.c +++ b/arch/arm/mm/fault-armv.c @@ -26,6 +26,11 @@ static unsigned long shared_pte_mask = L_PTE_CACHEABLE; /* * We take the easy way out of this problem - we make the * PTE uncacheable. However, we leave the write buffer on. + * + * Note that the pte lock held when calling update_mmu_cache must also + * guard the pte (somewhere else in the same mm) that we modify here. + * Therefore those configurations which might call adjust_pte (those + * without CONFIG_CPU_CACHE_VIPT) cannot support split page_table_lock. */ static int adjust_pte(struct vm_area_struct *vma, unsigned long address) { @@ -127,7 +132,7 @@ void __flush_dcache_page(struct address_space *mapping, struct page *page); * 2. If we have multiple shared mappings of the same space in * an object, we need to deal with the cache aliasing issues. * - * Note that the page_table_lock will be held. + * Note that the pte lock will be held. */ void update_mmu_cache(struct vm_area_struct *vma, unsigned long addr, pte_t pte) { diff --git a/arch/arm/mm/ioremap.c b/arch/arm/mm/ioremap.c index 6fb1258df1b..0f128c28fee 100644 --- a/arch/arm/mm/ioremap.c +++ b/arch/arm/mm/ioremap.c @@ -75,7 +75,7 @@ remap_area_pmd(pmd_t * pmd, unsigned long address, unsigned long size, pgprot = __pgprot(L_PTE_PRESENT | L_PTE_YOUNG | L_PTE_DIRTY | L_PTE_WRITE | flags); do { - pte_t * pte = pte_alloc_kernel(&init_mm, pmd, address); + pte_t * pte = pte_alloc_kernel(pmd, address); if (!pte) return -ENOMEM; remap_area_pte(pte, address, end - address, address + phys_addr, pgprot); @@ -97,7 +97,6 @@ remap_area_pages(unsigned long start, unsigned long phys_addr, phys_addr -= address; dir = pgd_offset(&init_mm, address); BUG_ON(address >= end); - spin_lock(&init_mm.page_table_lock); do { pmd_t *pmd = pmd_alloc(&init_mm, dir, address); if (!pmd) { @@ -114,7 +113,6 @@ remap_area_pages(unsigned long start, unsigned long phys_addr, dir++; } while (address && (address < end)); - spin_unlock(&init_mm.page_table_lock); flush_cache_vmap(start, end); return err; } diff --git a/arch/arm/mm/mm-armv.c b/arch/arm/mm/mm-armv.c index 61bc2fa0511..1221fdde176 100644 --- a/arch/arm/mm/mm-armv.c +++ b/arch/arm/mm/mm-armv.c @@ -180,11 +180,6 @@ pgd_t *get_pgd_slow(struct mm_struct *mm) if (!vectors_high()) { /* - * This lock is here just to satisfy pmd_alloc and pte_lock - */ - spin_lock(&mm->page_table_lock); - - /* * On ARM, first page must always be allocated since it * contains the machine vectors. */ @@ -201,23 +196,14 @@ pgd_t *get_pgd_slow(struct mm_struct *mm) set_pte(new_pte, *init_pte); pte_unmap_nested(init_pte); pte_unmap(new_pte); - - spin_unlock(&mm->page_table_lock); } return new_pgd; no_pte: - spin_unlock(&mm->page_table_lock); pmd_free(new_pmd); - free_pages((unsigned long)new_pgd, 2); - return NULL; - no_pmd: - spin_unlock(&mm->page_table_lock); free_pages((unsigned long)new_pgd, 2); - return NULL; - no_pgd: return NULL; } @@ -243,6 +229,7 @@ void free_pgd_slow(pgd_t *pgd) pte = pmd_page(*pmd); pmd_clear(pmd); dec_page_state(nr_page_table_pages); + pte_lock_deinit(pte); pte_free(pte); pmd_free(pmd); free: diff --git a/arch/arm/oprofile/backtrace.c b/arch/arm/oprofile/backtrace.c index df35c452a8b..7c22c12618c 100644 --- a/arch/arm/oprofile/backtrace.c +++ b/arch/arm/oprofile/backtrace.c @@ -49,42 +49,22 @@ static struct frame_tail* kernel_backtrace(struct frame_tail *tail) static struct frame_tail* user_backtrace(struct frame_tail *tail) { - struct frame_tail buftail; + struct frame_tail buftail[2]; - /* hardware pte might not be valid due to dirty/accessed bit emulation - * so we use copy_from_user and benefit from exception fixups */ - if (copy_from_user(&buftail, tail, sizeof(struct frame_tail))) + /* Also check accessibility of one struct frame_tail beyond */ + if (!access_ok(VERIFY_READ, tail, sizeof(buftail))) + return NULL; + if (__copy_from_user_inatomic(buftail, tail, sizeof(buftail))) return NULL; - oprofile_add_trace(buftail.lr); + oprofile_add_trace(buftail[0].lr); /* frame pointers should strictly progress back up the stack * (towards higher addresses) */ - if (tail >= buftail.fp) + if (tail >= buftail[0].fp) return NULL; - return buftail.fp-1; -} - -/* Compare two addresses and see if they're on the same page */ -#define CMP_ADDR_EQUAL(x,y,offset) ((((unsigned long) x) >> PAGE_SHIFT) \ - == ((((unsigned long) y) + offset) >> PAGE_SHIFT)) - -/* check that the page(s) containing the frame tail are present */ -static int pages_present(struct frame_tail *tail) -{ - struct mm_struct * mm = current->mm; - - if (!check_user_page_readable(mm, (unsigned long)tail)) - return 0; - - if (CMP_ADDR_EQUAL(tail, tail, 8)) - return 1; - - if (!check_user_page_readable(mm, ((unsigned long)tail) + 8)) - return 0; - - return 1; + return buftail[0].fp-1; } /* @@ -118,7 +98,6 @@ static int valid_kernel_stack(struct frame_tail *tail, struct pt_regs *regs) void arm_backtrace(struct pt_regs * const regs, unsigned int depth) { struct frame_tail *tail; - unsigned long last_address = 0; tail = ((struct frame_tail *) regs->ARM_fp) - 1; @@ -132,13 +111,6 @@ void arm_backtrace(struct pt_regs * const regs, unsigned int depth) return; } - while (depth-- && tail && !((unsigned long) tail & 3)) { - if ((!CMP_ADDR_EQUAL(last_address, tail, 0) - || !CMP_ADDR_EQUAL(last_address, tail, 8)) - && !pages_present(tail)) - return; - last_address = (unsigned long) tail; + while (depth-- && tail && !((unsigned long) tail & 3)) tail = user_backtrace(tail); - } } - diff --git a/arch/arm26/mm/memc.c b/arch/arm26/mm/memc.c index 8e8a2bb2487..34def6397c3 100644 --- a/arch/arm26/mm/memc.c +++ b/arch/arm26/mm/memc.c @@ -79,12 +79,6 @@ pgd_t *get_pgd_slow(struct mm_struct *mm) goto no_pgd; /* - * This lock is here just to satisfy pmd_alloc and pte_lock - * FIXME: I bet we could avoid taking it pretty much altogether - */ - spin_lock(&mm->page_table_lock); - - /* * On ARM, first page must always be allocated since it contains * the machine vectors. */ @@ -92,7 +86,7 @@ pgd_t *get_pgd_slow(struct mm_struct *mm) if (!new_pmd) goto no_pmd; - new_pte = pte_alloc_kernel(mm, new_pmd, 0); + new_pte = pte_alloc_map(mm, new_pmd, 0); if (!new_pte) goto no_pte; @@ -101,6 +95,7 @@ pgd_t *get_pgd_slow(struct mm_struct *mm) init_pte = pte_offset(init_pmd, 0); set_pte(new_pte, *init_pte); + pte_unmap(new_pte); /* * the page table entries are zeroed @@ -112,23 +107,14 @@ pgd_t *get_pgd_slow(struct mm_struct *mm) memcpy(new_pgd + FIRST_KERNEL_PGD_NR, init_pgd + FIRST_KERNEL_PGD_NR, (PTRS_PER_PGD - FIRST_KERNEL_PGD_NR) * sizeof(pgd_t)); - spin_unlock(&mm->page_table_lock); - /* update MEMC tables */ cpu_memc_update_all(new_pgd); return new_pgd; no_pte: - spin_unlock(&mm->page_table_lock); pmd_free(new_pmd); - free_pgd_slow(new_pgd); - return NULL; - no_pmd: - spin_unlock(&mm->page_table_lock); free_pgd_slow(new_pgd); - return NULL; - no_pgd: return NULL; } diff --git a/arch/cris/arch-v32/mm/tlb.c b/arch/cris/arch-v32/mm/tlb.c index 8233406798d..b08a28bb58a 100644 --- a/arch/cris/arch-v32/mm/tlb.c +++ b/arch/cris/arch-v32/mm/tlb.c @@ -175,6 +175,8 @@ init_new_context(struct task_struct *tsk, struct mm_struct *mm) return 0; } +static DEFINE_SPINLOCK(mmu_context_lock); + /* Called in schedule() just before actually doing the switch_to. */ void switch_mm(struct mm_struct *prev, struct mm_struct *next, @@ -183,10 +185,10 @@ switch_mm(struct mm_struct *prev, struct mm_struct *next, int cpu = smp_processor_id(); /* Make sure there is a MMU context. */ - spin_lock(&next->page_table_lock); + spin_lock(&mmu_context_lock); get_mmu_context(next); cpu_set(cpu, next->cpu_vm_mask); - spin_unlock(&next->page_table_lock); + spin_unlock(&mmu_context_lock); /* * Remember the pgd for the fault handlers. Keep a seperate copy of it diff --git a/arch/cris/mm/ioremap.c b/arch/cris/mm/ioremap.c index ebba11e270f..a92ac987758 100644 --- a/arch/cris/mm/ioremap.c +++ b/arch/cris/mm/ioremap.c @@ -52,7 +52,7 @@ static inline int remap_area_pmd(pmd_t * pmd, unsigned long address, unsigned lo if (address >= end) BUG(); do { - pte_t * pte = pte_alloc_kernel(&init_mm, pmd, address); + pte_t * pte = pte_alloc_kernel(pmd, address); if (!pte) return -ENOMEM; remap_area_pte(pte, address, end - address, address + phys_addr, prot); @@ -74,7 +74,6 @@ static int remap_area_pages(unsigned long address, unsigned long phys_addr, flush_cache_all(); if (address >= end) BUG(); - spin_lock(&init_mm.page_table_lock); do { pud_t *pud; pmd_t *pmd; @@ -94,7 +93,6 @@ static int remap_area_pages(unsigned long address, unsigned long phys_addr, address = (address + PGDIR_SIZE) & PGDIR_MASK; dir++; } while (address && (address < end)); - spin_unlock(&init_mm.page_table_lock); flush_tlb_all(); return error; } diff --git a/arch/frv/mm/dma-alloc.c b/arch/frv/mm/dma-alloc.c index cfc4f97490c..342823aad75 100644 --- a/arch/frv/mm/dma-alloc.c +++ b/arch/frv/mm/dma-alloc.c @@ -55,21 +55,18 @@ static int map_page(unsigned long va, unsigned long pa, pgprot_t prot) pte_t *pte; int err = -ENOMEM; - spin_lock(&init_mm.page_table_lock); - /* Use upper 10 bits of VA to index the first level map */ pge = pgd_offset_k(va); pue = pud_offset(pge, va); pme = pmd_offset(pue, va); /* Use middle 10 bits of VA to index the second-level map */ - pte = pte_alloc_kernel(&init_mm, pme, va); + pte = pte_alloc_kernel(pme, va); if (pte != 0) { err = 0; set_pte(pte, mk_pte_phys(pa & PAGE_MASK, prot)); } - spin_unlock(&init_mm.page_table_lock); return err; } diff --git a/arch/frv/mm/pgalloc.c b/arch/frv/mm/pgalloc.c index 4eaec0f3525..2c67dfe5a6b 100644 --- a/arch/frv/mm/pgalloc.c +++ b/arch/frv/mm/pgalloc.c @@ -87,14 +87,14 @@ static inline void pgd_list_add(pgd_t *pgd) if (pgd_list) pgd_list->private = (unsigned long) &page->index; pgd_list = page; - page->private = (unsigned long) &pgd_list; + set_page_private(page, (unsigned long)&pgd_list); } static inline void pgd_list_del(pgd_t *pgd) { struct page *next, **pprev, *page = virt_to_page(pgd); next = |