X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=arch%2Fx86_64%2Fmm%2Ffault.c;h=9a46b46399dbf767df9230d8ac05d87902c67a9e;hb=97bf2856c6014879bd04983a3e9dfcdac1e7fe85;hp=2cdf34b990508d3f1a56f1b72872fe308e6ddadb;hpb=9213980e6a70d8473e0ffd4b39ab5b6caaba9ff5;p=linux-2.6.git diff --git a/arch/x86_64/mm/fault.c b/arch/x86_64/mm/fault.c index 2cdf34b99..9a46b4639 100644 --- a/arch/x86_64/mm/fault.c +++ b/arch/x86_64/mm/fault.c @@ -5,14 +5,13 @@ * Copyright (C) 2001,2002 Andi Kleen, SuSE Labs. */ -#include #include #include #include #include #include #include -#include +#include #include #include #include @@ -23,17 +22,53 @@ #include /* For unblank_screen() */ #include #include +#include +#include #include -#include #include -#include #include #include #include #include #include +/* Page fault error code bits */ +#define PF_PROT (1<<0) /* or no page found */ +#define PF_WRITE (1<<1) +#define PF_USER (1<<2) +#define PF_RSVD (1<<3) +#define PF_INSTR (1<<4) + +static ATOMIC_NOTIFIER_HEAD(notify_page_fault_chain); + +/* Hook to register for page fault notifications */ +int register_page_fault_notifier(struct notifier_block *nb) +{ + vmalloc_sync_all(); + return atomic_notifier_chain_register(¬ify_page_fault_chain, nb); +} +EXPORT_SYMBOL_GPL(register_page_fault_notifier); + +int unregister_page_fault_notifier(struct notifier_block *nb) +{ + return atomic_notifier_chain_unregister(¬ify_page_fault_chain, nb); +} +EXPORT_SYMBOL_GPL(unregister_page_fault_notifier); + +static inline int notify_page_fault(enum die_val val, const char *str, + struct pt_regs *regs, long err, int trap, int sig) +{ + struct die_args args = { + .regs = regs, + .str = str, + .err = err, + .trapnr = trap, + .signr = sig + }; + return atomic_notifier_call_chain(¬ify_page_fault_chain, val, &args); +} + void bust_spinlocks(int yes) { int loglevel_save = console_loglevel; @@ -58,23 +93,22 @@ void bust_spinlocks(int yes) /* Sometimes the CPU reports invalid exceptions on prefetch. Check that here and ignore. Opcode checker based on code by Richard Brunner */ -static int is_prefetch(struct pt_regs *regs, unsigned long addr) +static noinline int is_prefetch(struct pt_regs *regs, unsigned long addr, + unsigned long error_code) { - unsigned char *instr = (unsigned char *)(regs->rip); + unsigned char *instr; int scan_more = 1; int prefetch = 0; - unsigned char *max_instr = instr + 15; + unsigned char *max_instr; - /* Avoid recursive faults for this common case */ - if (regs->rip == addr) - return 0; - - /* Code segments in LDT could have a non zero base. Don't check - when that's possible */ - if (regs->cs & (1<<2)) + /* If it was a exec fault ignore */ + if (error_code & PF_INSTR) return 0; + + instr = (unsigned char __user *)convert_rip_to_linear(current, regs); + max_instr = instr + 15; - if ((regs->cs & 3) != 0 && regs->rip >= TASK_SIZE) + if (user_mode(regs) && instr >= (unsigned char *)TASK_SIZE64) return 0; while (scan_more && instr < max_instr) { @@ -82,7 +116,7 @@ static int is_prefetch(struct pt_regs *regs, unsigned long addr) unsigned char instr_hi; unsigned char instr_lo; - if (__get_user(opcode, instr)) + if (probe_kernel_address(instr, opcode)) break; instr_hi = opcode & 0xf0; @@ -106,7 +140,7 @@ static int is_prefetch(struct pt_regs *regs, unsigned long addr) /* Could check the LDT for lm, but for now it's good enough to assume that long mode only uses well known segments or kernel. */ - scan_more = ((regs->cs & 3) == 0) || (regs->cs == __USER_CS); + scan_more = (!user_mode(regs)) || (regs->cs == __USER_CS); break; case 0x60: @@ -120,7 +154,7 @@ static int is_prefetch(struct pt_regs *regs, unsigned long addr) case 0x00: /* Prefetch instruction is 0x0F0D or 0x0F18 */ scan_more = 0; - if (__get_user(opcode, instr)) + if (probe_kernel_address(instr, opcode)) break; prefetch = (instr_lo == 0xF) && (opcode == 0x0D || opcode == 0x18); @@ -136,30 +170,30 @@ static int is_prefetch(struct pt_regs *regs, unsigned long addr) static int bad_address(void *p) { unsigned long dummy; - return __get_user(dummy, (unsigned long *)p); + return probe_kernel_address((unsigned long *)p, dummy); } void dump_pagetable(unsigned long address) { - pml4_t *pml4; pgd_t *pgd; + pud_t *pud; pmd_t *pmd; pte_t *pte; - asm("movq %%cr3,%0" : "=r" (pml4)); - - pml4 = __va((unsigned long)pml4 & PHYSICAL_PAGE_MASK); - pml4 += pml4_index(address); - printk("PML4 %lx ", pml4_val(*pml4)); - if (bad_address(pml4)) goto bad; - if (!pml4_present(*pml4)) goto ret; + asm("movq %%cr3,%0" : "=r" (pgd)); - pgd = __pgd_offset_k((pgd_t *)pml4_page(*pml4), address); + pgd = __va((unsigned long)pgd & PHYSICAL_PAGE_MASK); + pgd += pgd_index(address); if (bad_address(pgd)) goto bad; - printk("PGD %lx ", pgd_val(*pgd)); - if (!pgd_present(*pgd)) goto ret; + printk("PGD %lx ", pgd_val(*pgd)); + if (!pgd_present(*pgd)) goto ret; - pmd = pmd_offset(pgd, address); + pud = pud_offset(pgd, address); + if (bad_address(pud)) goto bad; + printk("PUD %lx ", pud_val(*pud)); + if (!pud_present(*pud)) goto ret; + + pmd = pmd_offset(pud, address); if (bad_address(pmd)) goto bad; printk("PMD %lx ", pmd_val(*pmd)); if (!pmd_present(*pmd)) goto ret; @@ -210,29 +244,94 @@ static int is_errata93(struct pt_regs *regs, unsigned long address) int unhandled_signal(struct task_struct *tsk, int sig) { - /* Warn for strace, but not for gdb */ - if (!test_ti_thread_flag(tsk->thread_info, TIF_SYSCALL_TRACE) && - (tsk->ptrace & PT_PTRACED)) + if (is_init(tsk)) + return 1; + if (tracehook_consider_fatal_signal(tsk, sig)) return 0; return (tsk->sighand->action[sig-1].sa.sa_handler == SIG_IGN) || (tsk->sighand->action[sig-1].sa.sa_handler == SIG_DFL); } -int page_fault_trace; +static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs, + unsigned long error_code) +{ + unsigned long flags = oops_begin(); + struct task_struct *tsk; + + printk(KERN_ALERT "%s: Corrupted page table at address %lx\n", + current->comm, address); + dump_pagetable(address); + tsk = current; + tsk->thread.cr2 = address; + tsk->thread.trap_no = 14; + tsk->thread.error_code = error_code; + __die("Bad pagetable", regs, error_code); + oops_end(flags); + do_exit(SIGKILL); +} + +/* + * Handle a fault on the vmalloc area + * + * This assumes no large pages in there. + */ +static int vmalloc_fault(unsigned long address) +{ + pgd_t *pgd, *pgd_ref; + pud_t *pud, *pud_ref; + pmd_t *pmd, *pmd_ref; + pte_t *pte, *pte_ref; + + /* Copy kernel mappings over when needed. This can also + happen within a race in page table update. In the later + case just flush. */ + + pgd = pgd_offset(current->mm ?: &init_mm, address); + pgd_ref = pgd_offset_k(address); + if (pgd_none(*pgd_ref)) + return -1; + if (pgd_none(*pgd)) + set_pgd(pgd, *pgd_ref); + else + BUG_ON(pgd_page_vaddr(*pgd) != pgd_page_vaddr(*pgd_ref)); + + /* Below here mismatches are bugs because these lower tables + are shared */ + + pud = pud_offset(pgd, address); + pud_ref = pud_offset(pgd_ref, address); + if (pud_none(*pud_ref)) + return -1; + if (pud_none(*pud) || pud_page_vaddr(*pud) != pud_page_vaddr(*pud_ref)) + BUG(); + pmd = pmd_offset(pud, address); + pmd_ref = pmd_offset(pud_ref, address); + if (pmd_none(*pmd_ref)) + return -1; + if (pmd_none(*pmd) || pmd_page(*pmd) != pmd_page(*pmd_ref)) + BUG(); + pte_ref = pte_offset_kernel(pmd_ref, address); + if (!pte_present(*pte_ref)) + return -1; + pte = pte_offset_kernel(pmd, address); + /* Don't use pte_page here, because the mappings can point + outside mem_map, and the NUMA hash lookup cannot handle + that. */ + if (!pte_present(*pte) || pte_pfn(*pte) != pte_pfn(*pte_ref)) + BUG(); + return 0; +} + +int page_fault_trace = 0; int exception_trace = 1; /* * This routine handles page faults. It determines the address, * and the problem, and then passes it off to one of the appropriate * routines. - * - * error_code: - * bit 0 == 0 means no page found, 1 means protection fault - * bit 1 == 0 means read, 1 means write - * bit 2 == 0 means kernel, 1 means user-mode - * bit 3 == 1 means fault was an instruction fetch */ -asmlinkage void do_page_fault(struct pt_regs *regs, unsigned long error_code) +asmlinkage void __kprobes do_page_fault(struct pt_regs *regs, + unsigned long error_code) { struct task_struct *tsk; struct mm_struct *mm; @@ -240,23 +339,57 @@ asmlinkage void do_page_fault(struct pt_regs *regs, unsigned long error_code) unsigned long address; const struct exception_table_entry *fixup; int write; + unsigned long flags; siginfo_t info; -#ifdef CONFIG_CHECKING - { - unsigned long gs; - struct x8664_pda *pda = cpu_pda + stack_smp_processor_id(); - rdmsrl(MSR_GS_BASE, gs); - if (gs != (unsigned long)pda) { - wrmsrl(MSR_GS_BASE, pda); - printk("page_fault: wrong gs %lx expected %p\n", gs, pda); - } - } -#endif + tsk = current; + mm = tsk->mm; + prefetchw(&mm->mmap_sem); /* get the address */ __asm__("movq %%cr2,%0":"=r" (address)); + info.si_code = SEGV_MAPERR; + + + /* + * We fault-in kernel-space virtual memory on-demand. The + * 'reference' page table is init_mm.pgd. + * + * NOTE! We MUST NOT take any locks for this case. We may + * be in an interrupt or a critical region, and should + * only copy the information from the master page table, + * nothing more. + * + * This verifies that the fault happens in kernel space + * (error_code & 4) == 0, and that the fault was not a + * protection error (error_code & 9) == 0. + */ + if (unlikely(address >= TASK_SIZE64)) { + /* + * Don't check for the module range here: its PML4 + * is always initialized because it's shared with the main + * kernel text. Only vmalloc may need PML4 syncups. + */ + if (!(error_code & (PF_RSVD|PF_USER|PF_PROT)) && + ((address >= VMALLOC_START && address < VMALLOC_END))) { + if (vmalloc_fault(address) >= 0) + return; + } + if (notify_page_fault(DIE_PAGE_FAULT, "page fault", regs, error_code, 14, + SIGSEGV) == NOTIFY_STOP) + return; + /* + * Don't take the mm semaphore here. If we fixup a prefetch + * fault we could otherwise deadlock. + */ + goto bad_area_nosemaphore; + } + + if (notify_page_fault(DIE_PAGE_FAULT, "page fault", regs, error_code, 14, + SIGSEGV) == NOTIFY_STOP) + return; + if (likely(regs->eflags & X86_EFLAGS_IF)) local_irq_enable(); @@ -264,15 +397,8 @@ asmlinkage void do_page_fault(struct pt_regs *regs, unsigned long error_code) printk("pagefault rip:%lx rsp:%lx cs:%lu ss:%lu address %lx error %lx\n", regs->rip,regs->rsp,regs->cs,regs->ss,address,error_code); - tsk = current; - mm = tsk->mm; - info.si_code = SEGV_MAPERR; - - /* 5 => page not present and from supervisor mode */ - if (unlikely(!(error_code & 5) && - ((address >= VMALLOC_START && address <= VMALLOC_END) || - (address >= MODULES_VADDR && address <= MODULES_END)))) - goto vmalloc_fault; + if (unlikely(error_code & PF_RSVD)) + pgtable_bad(address, regs, error_code); /* * If we're in an interrupt or have no user @@ -282,7 +408,27 @@ asmlinkage void do_page_fault(struct pt_regs *regs, unsigned long error_code) goto bad_area_nosemaphore; again: - down_read(&mm->mmap_sem); + /* When running in the kernel we expect faults to occur only to + * addresses in user space. All other faults represent errors in the + * kernel and should generate an OOPS. Unfortunatly, in the case of an + * erroneous fault occurring in a code path which already holds mmap_sem + * we will deadlock attempting to validate the fault against the + * address space. Luckily the kernel only validly references user + * space from well defined areas of code, which are listed in the + * exceptions table. + * + * As the vast majority of faults will be valid we will only perform + * the source reference check when there is a possibilty of a deadlock. + * Attempt to lock the address space, if we cannot we then validate the + * source. If this is invalid we can skip the address space check, + * thus avoiding the deadlock. + */ + if (!down_read_trylock(&mm->mmap_sem)) { + if ((error_code & PF_USER) == 0 && + !search_exception_tables(regs->rip)) + goto bad_area_nosemaphore; + down_read(&mm->mmap_sem); + } vma = find_vma(mm, address); if (!vma) @@ -292,8 +438,10 @@ asmlinkage void do_page_fault(struct pt_regs *regs, unsigned long error_code) if (!(vma->vm_flags & VM_GROWSDOWN)) goto bad_area; if (error_code & 4) { - // XXX: align red zone size with ABI - if (address + 128 < regs->rsp) + /* Allow userspace just enough access below the stack pointer + * to let the 'enter' instruction work. + */ + if (address + 65536 + 32 * sizeof(unsigned long) < regs->rsp) goto bad_area; } if (expand_stack(vma, address)) @@ -305,18 +453,18 @@ asmlinkage void do_page_fault(struct pt_regs *regs, unsigned long error_code) good_area: info.si_code = SEGV_ACCERR; write = 0; - switch (error_code & 3) { + switch (error_code & (PF_PROT|PF_WRITE)) { default: /* 3: write, present */ /* fall through */ - case 2: /* write, not present */ + case PF_WRITE: /* write, not present */ if (!(vma->vm_flags & VM_WRITE)) goto bad_area; write++; break; - case 1: /* read, present */ + case PF_PROT: /* read, present */ goto bad_area; - case 0: /* read, not present */ - if (!(vma->vm_flags & (VM_READ | VM_EXEC))) + case 0: /* read, not present */ + if (!(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE))) goto bad_area; } @@ -326,13 +474,13 @@ good_area: * the fault. */ switch (handle_mm_fault(mm, vma, address, write)) { - case 1: + case VM_FAULT_MINOR: tsk->min_flt++; break; - case 2: + case VM_FAULT_MAJOR: tsk->maj_flt++; break; - case 0: + case VM_FAULT_SIGBUS: goto do_sigbus; default: goto out_of_memory; @@ -349,20 +497,9 @@ bad_area: up_read(&mm->mmap_sem); bad_area_nosemaphore: - -#ifdef CONFIG_IA32_EMULATION - /* 32bit vsyscall. map on demand. */ - if (test_thread_flag(TIF_IA32) && - address >= 0xffffe000 && address < 0xffffe000 + PAGE_SIZE) { - if (map_syscall32(mm, address) < 0) - goto out_of_memory2; - return; - } -#endif - /* User mode accesses just cause a SIGSEGV */ - if (error_code & 4) { - if (is_prefetch(regs, address)) + if (error_code & PF_USER) { + if (is_prefetch(regs, address, error_code)) return; /* Work around K8 erratum #100 K8 in compat mode @@ -376,10 +513,11 @@ bad_area_nosemaphore: return; if (exception_trace && unhandled_signal(tsk, SIGSEGV)) { - printk(KERN_INFO - "%s[%d]: segfault at %016lx rip %016lx rsp %016lx error %lx\n", - tsk->comm, tsk->pid, address, regs->rip, - regs->rsp, error_code); + printk( + "%s%s[%d:#%u]: segfault at %016lx rip %016lx rsp %016lx error %lx\n", + tsk->pid > 1 ? KERN_INFO : KERN_EMERG, + tsk->comm, tsk->pid, tsk->xid, address, + regs->rip, regs->rsp, error_code); } tsk->thread.cr2 = address; @@ -389,7 +527,7 @@ bad_area_nosemaphore: info.si_signo = SIGSEGV; info.si_errno = 0; /* info.si_code has been set above */ - info.si_addr = (void *)address; + info.si_addr = (void __user *)address; force_sig_info(SIGSEGV, &info, tsk); return; } @@ -407,7 +545,7 @@ no_context: * Hall of shame of CPU/BIOS bugs. */ - if (is_prefetch(regs, address)) + if (is_prefetch(regs, address, error_code)) return; if (is_errata93(regs, address)) @@ -418,7 +556,7 @@ no_context: * terminate things with extreme prejudice. */ - oops_begin(); + flags = oops_begin(); if (address < PAGE_SIZE) printk(KERN_ALERT "Unable to handle kernel NULL pointer dereference"); @@ -426,12 +564,14 @@ no_context: printk(KERN_ALERT "Unable to handle kernel paging request"); printk(" at %016lx RIP: \n" KERN_ALERT,address); printk_address(regs->rip); - printk("\n"); dump_pagetable(address); + tsk->thread.cr2 = address; + tsk->thread.trap_no = 14; + tsk->thread.error_code = error_code; __die("Oops", regs, error_code); /* Executive summary in case the body of the oops scrolled away */ printk(KERN_EMERG "CR2: %016lx\n", address); - oops_end(); + oops_end(flags); do_exit(SIGKILL); /* @@ -440,12 +580,12 @@ no_context: */ out_of_memory: up_read(&mm->mmap_sem); -out_of_memory2: - if (current->pid == 1) { + if (is_init(current)) { yield(); goto again; } - printk("VM: killing process %s\n", tsk->comm); + printk("VM: killing process %s(%d:#%u)\n", + tsk->comm, tsk->pid, tsk->xid); if (error_code & 4) do_exit(SIGKILL); goto no_context; @@ -454,7 +594,7 @@ do_sigbus: up_read(&mm->mmap_sem); /* Kernel mode? Handle exceptions or die */ - if (!(error_code & 4)) + if (!(error_code & PF_USER)) goto no_context; tsk->thread.cr2 = address; @@ -463,36 +603,56 @@ do_sigbus: info.si_signo = SIGBUS; info.si_errno = 0; info.si_code = BUS_ADRERR; - info.si_addr = (void *)address; + info.si_addr = (void __user *)address; force_sig_info(SIGBUS, &info, tsk); return; +} -vmalloc_fault: - { - pgd_t *pgd; - pmd_t *pmd; - pte_t *pte; +DEFINE_SPINLOCK(pgd_lock); +struct page *pgd_list; - /* - * x86-64 has the same kernel 3rd level pages for all CPUs. - * But for vmalloc/modules the TLB synchronization works lazily, - * so it can happen that we get a page fault for something - * that is really already in the page table. Just check if it - * is really there and when yes flush the local TLB. - */ - pgd = pgd_offset_k(address); - if (pgd != current_pgd_offset_k(address)) - BUG(); - if (!pgd_present(*pgd)) - goto bad_area_nosemaphore; - pmd = pmd_offset(pgd, address); - if (!pmd_present(*pmd)) - goto bad_area_nosemaphore; - pte = pte_offset_kernel(pmd, address); - if (!pte_present(*pte)) - goto bad_area_nosemaphore; +void vmalloc_sync_all(void) +{ + /* 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 address; - __flush_tlb_all(); - return; + 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); + struct page *page; + + if (pgd_none(*pgd_ref)) + continue; + spin_lock(&pgd_lock); + for (page = pgd_list; page; + page = (struct page *)page->index) { + 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(&pgd_lock); + set_bit(pgd_index(address), insync); + } + if (address == start) + start = address + PGDIR_SIZE; } + /* Check that there is no need to do the same for the modules area. */ + BUILD_BUG_ON(!(MODULES_VADDR > __START_KERNEL)); + BUILD_BUG_ON(!(((MODULES_END - 1) & PGDIR_MASK) == + (__START_KERNEL & PGDIR_MASK))); +} + +static int __init enable_pagefaulttrace(char *str) +{ + page_fault_trace = 1; + return 1; } +__setup("pagefaulttrace", enable_pagefaulttrace);