X-Git-Url: https://err.no/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=arch%2Fx86%2Fmm%2Ffault_64.c;h=dd26e680a43166ca9abe0ced6390989edddf4349;hb=3c1df68b848b39270752ff8d4b956cc4a4dce0f6;hp=0e26230669ca21aabb92a60f134100d8f453f5fc;hpb=aac9e28d2f3d3c1eacc4114d685864a2a6423b80;p=linux-2.6 diff --git a/arch/x86/mm/fault_64.c b/arch/x86/mm/fault_64.c index 0e26230669..dd26e680a4 100644 --- a/arch/x86/mm/fault_64.c +++ b/arch/x86/mm/fault_64.c @@ -1,6 +1,4 @@ /* - * linux/arch/x86-64/mm/fault.c - * * Copyright (C) 1995 Linus Torvalds * Copyright (C) 2001,2002 Andi Kleen, SuSE Labs. */ @@ -25,7 +23,6 @@ #include #include #include -#include #include #include @@ -34,16 +31,23 @@ #include #include -/* Page fault error code bits */ -#define PF_PROT (1<<0) /* or no page found */ +/* + * Page fault error code bits + * 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 use of reserved bit detected + * bit 4 == 1 means fault was an instruction fetch + */ +#define PF_PROT (1<<0) #define PF_WRITE (1<<1) -#define PF_USER (1<<2) -#define PF_RSVD (1<<3) +#define PF_USER (1<<2) +#define PF_RSVD (1<<3) #define PF_INSTR (1<<4) -#ifdef CONFIG_KPROBES static inline int notify_page_fault(struct pt_regs *regs) { +#ifdef CONFIG_KPROBES int ret = 0; /* kprobe_running() needs smp_processor_id() */ @@ -55,96 +59,128 @@ static inline int notify_page_fault(struct pt_regs *regs) } return ret; -} #else -static inline int notify_page_fault(struct pt_regs *regs) -{ return 0; -} #endif +} -/* Sometimes the CPU reports invalid exceptions on prefetch. - Check that here and ignore. - Opcode checker based on code by Richard Brunner */ -static noinline int is_prefetch(struct pt_regs *regs, unsigned long addr, - unsigned long error_code) -{ +/* + * X86_32 + * Sometimes AMD Athlon/Opteron CPUs report invalid exceptions on prefetch. + * Check that here and ignore it. + * + * X86_64 + * Sometimes the CPU reports invalid exceptions on prefetch. + * Check that here and ignore it. + * + * Opcode checker based on code by Richard Brunner + */ +static int is_prefetch(struct pt_regs *regs, unsigned long addr, + unsigned long error_code) +{ unsigned char *instr; int scan_more = 1; - int prefetch = 0; + int prefetch = 0; unsigned char *max_instr; +#ifdef CONFIG_X86_32 + if (unlikely(boot_cpu_data.x86_vendor == X86_VENDOR_AMD && + boot_cpu_data.x86 >= 6)) { + /* Catch an obscure case of prefetch inside an NX page. */ + if (nx_enabled && (error_code & PF_INSTR)) + return 0; + } else { + return 0; + } +#else /* If it was a exec fault ignore */ if (error_code & PF_INSTR) return 0; - - instr = (unsigned char __user *)convert_rip_to_linear(current, regs); +#endif + + instr = (unsigned char *)convert_ip_to_linear(current, regs); max_instr = instr + 15; if (user_mode(regs) && instr >= (unsigned char *)TASK_SIZE) return 0; - while (scan_more && instr < max_instr) { + while (scan_more && instr < max_instr) { unsigned char opcode; unsigned char instr_hi; unsigned char instr_lo; if (probe_kernel_address(instr, opcode)) - break; + break; - instr_hi = opcode & 0xf0; - instr_lo = opcode & 0x0f; + instr_hi = opcode & 0xf0; + instr_lo = opcode & 0x0f; instr++; - switch (instr_hi) { + switch (instr_hi) { case 0x20: case 0x30: - /* Values 0x26,0x2E,0x36,0x3E are valid x86 - prefixes. In long mode, the CPU will signal - invalid opcode if some of these prefixes are - present so we will never get here anyway */ + /* + * Values 0x26,0x2E,0x36,0x3E are valid x86 prefixes. + * In X86_64 long mode, the CPU will signal invalid + * opcode if some of these prefixes are present so + * X86_64 will never get here anyway + */ scan_more = ((instr_lo & 7) == 0x6); break; - +#ifdef CONFIG_X86_64 case 0x40: - /* In AMD64 long mode, 0x40 to 0x4F are valid REX prefixes - Need to figure out under what instruction mode the - instruction was issued ... */ - /* 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. */ + /* + * In AMD64 long mode 0x40..0x4F are valid REX prefixes + * Need to figure out under what instruction mode the + * instruction was issued. 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 = (!user_mode(regs)) || (regs->cs == __USER_CS); break; - +#endif case 0x60: /* 0x64 thru 0x67 are valid prefixes in all modes. */ scan_more = (instr_lo & 0xC) == 0x4; - break; + break; case 0xF0: - /* 0xF0, 0xF2, and 0xF3 are valid prefixes in all modes. */ + /* 0xF0, 0xF2, 0xF3 are valid prefixes in all modes. */ scan_more = !instr_lo || (instr_lo>>1) == 1; - break; + break; case 0x00: /* Prefetch instruction is 0x0F0D or 0x0F18 */ scan_more = 0; + if (probe_kernel_address(instr, opcode)) break; prefetch = (instr_lo == 0xF) && (opcode == 0x0D || opcode == 0x18); - break; + break; default: scan_more = 0; break; - } + } } return prefetch; } -static int bad_address(void *p) -{ +static void force_sig_info_fault(int si_signo, int si_code, + unsigned long address, struct task_struct *tsk) +{ + siginfo_t info; + + info.si_signo = si_signo; + info.si_errno = 0; + info.si_code = si_code; + info.si_addr = (void __user *)address; + force_sig_info(si_signo, &info, tsk); +} + +static int bad_address(void *p) +{ unsigned long dummy; return probe_kernel_address((unsigned long *)p, dummy); -} +} void dump_pagetable(unsigned long address) { @@ -155,11 +191,11 @@ void dump_pagetable(unsigned long address) pgd = (pgd_t *)read_cr3(); - pgd = __va((unsigned long)pgd & PHYSICAL_PAGE_MASK); + 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; + if (!pgd_present(*pgd)) goto ret; pud = pud_offset(pgd, address); if (bad_address(pud)) goto bad; @@ -173,7 +209,7 @@ void dump_pagetable(unsigned long address) pte = pte_offset_kernel(pmd, address); if (bad_address(pte)) goto bad; - printk("PTE %lx", pte_val(*pte)); + printk("PTE %lx", pte_val(*pte)); ret: printk("\n"); return; @@ -181,39 +217,44 @@ bad: printk("BAD\n"); } -static const char errata93_warning[] = +#ifdef CONFIG_X86_64 +static const char errata93_warning[] = KERN_ERR "******* Your BIOS seems to not contain a fix for K8 errata #93\n" KERN_ERR "******* Working around it, but it may cause SEGVs or burn power.\n" KERN_ERR "******* Please consider a BIOS update.\n" KERN_ERR "******* Disabling USB legacy in the BIOS may also help.\n"; +#endif /* Workaround for K8 erratum #93 & buggy BIOS. BIOS SMM functions are required to use a specific workaround - to avoid corruption of the 64bit RIP register on C stepping K8. - A lot of BIOS that didn't get tested properly miss this. + to avoid corruption of the 64bit RIP register on C stepping K8. + A lot of BIOS that didn't get tested properly miss this. The OS sees this as a page fault with the upper 32bits of RIP cleared. Try to work around it here. - Note we only handle faults in kernel here. */ - -static int is_errata93(struct pt_regs *regs, unsigned long address) + Note we only handle faults in kernel here. + Does nothing for X86_32 + */ +static int is_errata93(struct pt_regs *regs, unsigned long address) { +#ifdef CONFIG_X86_64 static int warned; - if (address != regs->rip) + if (address != regs->ip) return 0; - if ((address >> 32) != 0) + if ((address >> 32) != 0) return 0; address |= 0xffffffffUL << 32; - if ((address >= (u64)_stext && address <= (u64)_etext) || - (address >= MODULES_VADDR && address <= MODULES_END)) { + if ((address >= (u64)_stext && address <= (u64)_etext) || + (address >= MODULES_VADDR && address <= MODULES_END)) { if (!warned) { - printk(errata93_warning); + printk(errata93_warning); warned = 1; } - regs->rip = address; + regs->ip = address; return 1; } +#endif return 0; -} +} static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs, unsigned long error_code) @@ -228,9 +269,9 @@ static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs, 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); + if (__die("Bad pagetable", regs, error_code)) + regs = NULL; + oops_end(flags, regs, SIGKILL); } /* @@ -240,6 +281,26 @@ static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs, */ static int vmalloc_fault(unsigned long address) { +#ifdef CONFIG_X86_32 + unsigned long pgd_paddr; + pmd_t *pmd_k; + pte_t *pte_k; + /* + * Synchronize this task's top level page-table + * with the 'reference' page table. + * + * Do _not_ use "current" here. We might be inside + * an interrupt in the middle of a task switch.. + */ + pgd_paddr = read_cr3(); + pmd_k = vmalloc_sync_one(__va(pgd_paddr), address); + if (!pmd_k) + return -1; + pte_k = pte_offset_kernel(pmd_k, address); + if (!pte_present(*pte_k)) + return -1; + return 0; +#else pgd_t *pgd, *pgd_ref; pud_t *pud, *pud_ref; pmd_t *pmd, *pmd_ref; @@ -283,6 +344,7 @@ static int vmalloc_fault(unsigned long address) if (!pte_present(*pte) || pte_pfn(*pte) != pte_pfn(*pte_ref)) BUG(); return 0; +#endif } int show_unhandled_signals = 1; @@ -297,12 +359,11 @@ asmlinkage void __kprobes do_page_fault(struct pt_regs *regs, { struct task_struct *tsk; struct mm_struct *mm; - struct vm_area_struct * vma; + struct vm_area_struct *vma; unsigned long address; - const struct exception_table_entry *fixup; int write, fault; unsigned long flags; - siginfo_t info; + int si_code; /* * We can fault from pretty much anywhere, with unknown IRQ state. @@ -316,8 +377,10 @@ asmlinkage void __kprobes do_page_fault(struct pt_regs *regs, /* get the address */ address = read_cr2(); - info.si_code = SEGV_MAPERR; + si_code = SEGV_MAPERR; + if (notify_page_fault(regs)) + return; /* * We fault-in kernel-space virtual memory on-demand. The @@ -343,8 +406,6 @@ asmlinkage void __kprobes do_page_fault(struct pt_regs *regs, if (vmalloc_fault(address) >= 0) return; } - if (notify_page_fault(regs)) - return; /* * Don't take the mm semaphore here. If we fixup a prefetch * fault we could otherwise deadlock. @@ -352,18 +413,15 @@ asmlinkage void __kprobes do_page_fault(struct pt_regs *regs, goto bad_area_nosemaphore; } - if (notify_page_fault(regs)) - return; - - if (likely(regs->eflags & X86_EFLAGS_IF)) + if (likely(regs->flags & X86_EFLAGS_IF)) local_irq_enable(); if (unlikely(error_code & PF_RSVD)) pgtable_bad(address, regs, error_code); /* - * If we're in an interrupt or have no user - * context, we must not take the fault.. + * If we're in an interrupt, have no user context or are running in an + * atomic region then we must not take the fault. */ if (unlikely(in_atomic() || !mm)) goto bad_area_nosemaphore; @@ -393,7 +451,7 @@ asmlinkage void __kprobes do_page_fault(struct pt_regs *regs, */ if (!down_read_trylock(&mm->mmap_sem)) { if ((error_code & PF_USER) == 0 && - !search_exception_tables(regs->rip)) + !search_exception_tables(regs->ip)) goto bad_area_nosemaphore; down_read(&mm->mmap_sem); } @@ -405,11 +463,14 @@ asmlinkage void __kprobes do_page_fault(struct pt_regs *regs, goto good_area; if (!(vma->vm_flags & VM_GROWSDOWN)) goto bad_area; - if (error_code & 4) { - /* Allow userspace just enough access below the stack pointer - * to let the 'enter' instruction work. + if (error_code & PF_USER) { + /* + * Accessing the stack below %sp is always a bug. + * The large cushion allows instructions like enter + * and pusha to work. ("enter $65535,$31" pushes + * 32 pointers and then decrements %sp by 65535.) */ - if (address + 65536 + 32 * sizeof(unsigned long) < regs->rsp) + if (address + 65536 + 32 * sizeof(unsigned long) < regs->sp) goto bad_area; } if (expand_stack(vma, address)) @@ -419,21 +480,21 @@ asmlinkage void __kprobes do_page_fault(struct pt_regs *regs, * we can handle it.. */ good_area: - info.si_code = SEGV_ACCERR; + si_code = SEGV_ACCERR; write = 0; switch (error_code & (PF_PROT|PF_WRITE)) { - default: /* 3: write, present */ - /* fall through */ - case PF_WRITE: /* write, not present */ - if (!(vma->vm_flags & VM_WRITE)) - goto bad_area; - write++; - break; - case PF_PROT: /* read, present */ + default: /* 3: write, present */ + /* fall through */ + case PF_WRITE: /* write, not present */ + if (!(vma->vm_flags & VM_WRITE)) + goto bad_area; + write++; + break; + case PF_PROT: /* read, present */ + goto bad_area; + case 0: /* read, not present */ + if (!(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE))) goto bad_area; - case 0: /* read, not present */ - if (!(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE))) - goto bad_area; } /* @@ -453,6 +514,17 @@ good_area: tsk->maj_flt++; else tsk->min_flt++; + +#ifdef CONFIG_X86_32 + /* + * Did it hit the DOS screen memory VA from vm86 mode? + */ + if (v8086_mode(regs)) { + unsigned long bit = (address - 0xA0000) >> PAGE_SHIFT; + if (bit < 32) + tsk->thread.screen_bitmap |= 1 << bit; + } +#endif up_read(&mm->mmap_sem); return; @@ -488,42 +560,41 @@ bad_area_nosemaphore: if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) && printk_ratelimit()) { printk( - "%s%s[%d]: segfault at %lx rip %lx rsp %lx error %lx\n", - tsk->pid > 1 ? KERN_INFO : KERN_EMERG, - tsk->comm, tsk->pid, address, regs->rip, - regs->rsp, error_code); +#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 + task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG, + tsk->comm, task_pid_nr(tsk), address, regs->ip, + regs->sp, error_code); + print_vma_addr(" in ", regs->ip); + printk("\n"); } - + tsk->thread.cr2 = address; /* Kernel addresses are always protection faults */ tsk->thread.error_code = error_code | (address >= TASK_SIZE); tsk->thread.trap_no = 14; - info.si_signo = SIGSEGV; - info.si_errno = 0; - /* info.si_code has been set above */ - info.si_addr = (void __user *)address; - force_sig_info(SIGSEGV, &info, tsk); + + force_sig_info_fault(SIGSEGV, si_code, address, tsk); return; } no_context: - /* Are we prepared to handle this kernel fault? */ - fixup = search_exception_tables(regs->rip); - if (fixup) { - regs->rip = fixup->fixup; + if (fixup_exception(regs)) return; - } - /* + /* * Hall of shame of CPU/BIOS bugs. */ - if (is_prefetch(regs, address, error_code)) - return; + if (is_prefetch(regs, address, error_code)) + return; if (is_errata93(regs, address)) - return; + return; /* * Oops. The kernel tried to access some bad page. We'll have to @@ -536,17 +607,17 @@ no_context: printk(KERN_ALERT "Unable to handle kernel NULL pointer dereference"); else printk(KERN_ALERT "Unable to handle kernel paging request"); - printk(" at %016lx RIP: \n" KERN_ALERT,address); - printk_address(regs->rip); + printk(" at %016lx RIP: \n" KERN_ALERT, address); + printk_address(regs->ip, 1); dump_pagetable(address); tsk->thread.cr2 = address; tsk->thread.trap_no = 14; tsk->thread.error_code = error_code; - __die("Oops", regs, error_code); + if (__die("Oops", regs, error_code)) + regs = NULL; /* Executive summary in case the body of the oops scrolled away */ printk(KERN_EMERG "CR2: %016lx\n", address); - oops_end(flags); - do_exit(SIGKILL); + oops_end(flags, regs, SIGKILL); /* * We ran out of memory, or some other thing happened to us that made @@ -559,7 +630,7 @@ out_of_memory: goto again; } printk("VM: killing process %s\n", tsk->comm); - if (error_code & 4) + if (error_code & PF_USER) do_group_exit(SIGKILL); goto no_context; @@ -573,11 +644,7 @@ do_sigbus: tsk->thread.cr2 = address; tsk->thread.error_code = error_code; tsk->thread.trap_no = 14; - info.si_signo = SIGBUS; - info.si_errno = 0; - info.si_code = BUS_ADRERR; - info.si_addr = (void __user *)address; - force_sig_info(SIGBUS, &info, tsk); + force_sig_info_fault(SIGBUS, BUS_ADRERR, address, tsk); return; } @@ -586,10 +653,12 @@ LIST_HEAD(pgd_list); 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). */ + /* + * 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; @@ -618,6 +687,6 @@ void vmalloc_sync_all(void) } /* 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) == + BUILD_BUG_ON(!(((MODULES_END - 1) & PGDIR_MASK) == (__START_KERNEL & PGDIR_MASK))); }