/*
- * linux/arch/i386/mm/fault.c
- *
* Copyright (C) 1995 Linus Torvalds
*/
#include <asm/desc.h>
#include <asm/segment.h>
-extern void die(const char *,struct pt_regs *,long);
+/*
+ * 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_INSTR (1<<4)
+
+extern void die(const char *, struct pt_regs *, long);
-#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() */
}
return ret;
-}
#else
-static inline int notify_page_fault(struct pt_regs *regs)
-{
return 0;
-}
#endif
+}
/*
* Return EIP plus the CS segment base. The segment limit is also
* If CS is no longer a valid code segment, or if EIP is beyond the
* limit, or if it is a kernel address when CS is not a kernel segment,
* then the returned value will be greater than *eip_limit.
- *
+ *
* This is slow, but is very rarely executed.
*/
static inline unsigned long get_segment_eip(struct pt_regs *regs,
/* The standard kernel/user address space limit. */
*eip_limit = user_mode(regs) ? USER_DS.seg : KERNEL_DS.seg;
-
+
/* By far the most common cases. */
if (likely(SEGMENT_IS_FLAT_CODE(seg)))
return ip;
return 1; /* So that returned ip > *eip_limit. */
}
- /* Get the GDT/LDT descriptor base.
+ /* Get the GDT/LDT descriptor base.
When you look for races in this code remember that
LDT and other horrors are only used in user space. */
if (seg & (1<<2)) {
desc = (void *)desc + (seg & ~7);
} else {
/* Must disable preemption while reading the GDT. */
- desc = (u32 *)get_cpu_gdt_table(get_cpu());
+ desc = (u32 *)get_cpu_gdt_table(get_cpu());
desc = (void *)desc + (seg & ~7);
}
/* Decode the code segment base from the descriptor */
base = get_desc_base((struct desc_struct *)desc);
- if (seg & (1<<2)) {
+ if (seg & (1<<2))
mutex_unlock(¤t->mm->context.lock);
- } else
+ else
put_cpu();
/* Adjust EIP and segment limit, and clamp at the kernel limit.
return ip + base;
}
-/*
+/*
* Sometimes AMD Athlon/Opteron CPUs report invalid exceptions on prefetch.
* Check that here and ignore it.
*/
static int __is_prefetch(struct pt_regs *regs, unsigned long addr)
-{
+{
unsigned long limit;
- unsigned char *instr = (unsigned char *)get_segment_eip (regs, &limit);
+ unsigned char *instr = (unsigned char *)get_segment_eip(regs, &limit);
int scan_more = 1;
- int prefetch = 0;
+ int prefetch = 0;
int i;
- for (i = 0; scan_more && i < 15; i++) {
+ for (i = 0; scan_more && i < 15; i++) {
unsigned char opcode;
unsigned char instr_hi;
unsigned char instr_lo;
if (instr > (unsigned char *)limit)
break;
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. */
+ /*
+ * 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..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 */
+ /* 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;
break;
prefetch = (instr_lo == 0xF) &&
(opcode == 0x0D || opcode == 0x18);
- break;
+ break;
default:
scan_more = 0;
break;
- }
+ }
}
return prefetch;
}
return __is_prefetch(regs, addr);
}
return 0;
-}
+}
static noinline void force_sig_info_fault(int si_signo, int si_code,
unsigned long address, struct task_struct *tsk)
* 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 use of reserved bit detected
- * bit 4 == 1 means fault was an instruction fetch
*/
void __kprobes do_page_fault(struct pt_regs *regs, unsigned long error_code)
{
struct task_struct *tsk;
struct mm_struct *mm;
- struct vm_area_struct * vma;
+ struct vm_area_struct *vma;
unsigned long address;
int write, si_code;
int fault;
trace_hardirqs_fixup();
/* get the address */
- address = read_cr2();
+ address = read_cr2();
tsk = current;
/*
* If we're in an interrupt, have no user context or are running in an
- * atomic region then we must not take the fault..
+ * atomic region then we must not take the fault.
*/
if (in_atomic() || !mm)
goto bad_area_nosemaphore;
* thus avoiding the deadlock.
*/
if (!down_read_trylock(&mm->mmap_sem)) {
- if ((error_code & 4) == 0 &&
+ if ((error_code & PF_USER) == 0 &&
!search_exception_tables(regs->ip))
goto bad_area_nosemaphore;
down_read(&mm->mmap_sem);
goto good_area;
if (!(vma->vm_flags & VM_GROWSDOWN))
goto bad_area;
- if (error_code & 4) {
+ if (error_code & PF_USER) {
/*
* Accessing the stack below %sp is always a bug.
* The large cushion allows instructions like enter
good_area:
si_code = SEGV_ACCERR;
write = 0;
- switch (error_code & 3) {
- default: /* 3: write, present */
- /* fall through */
- case 2: /* write, not present */
- if (!(vma->vm_flags & VM_WRITE))
- goto bad_area;
- write++;
- break;
- case 1: /* read, present */
+ 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 */
+ 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;
}
survive:
bad_area_nosemaphore:
/* User mode accesses just cause a SIGSEGV */
- if (error_code & 4) {
+ if (error_code & PF_USER) {
/*
* It's possible to have interrupts off here.
*/
local_irq_enable();
- /*
- * Valid to do another page fault here because this one came
+ /*
+ * Valid to do another page fault here because this one came
* from user space.
*/
if (is_prefetch(regs, address, error_code))
*/
if (boot_cpu_data.f00f_bug) {
unsigned long nr;
-
+
nr = (address - idt_descr.address) >> 3;
if (nr == 6) {
if (fixup_exception(regs))
return;
- /*
+ /*
* Valid to do another page fault here, because if this fault
- * had been triggered by is_prefetch fixup_exception would have
+ * had been triggered by is_prefetch fixup_exception would have
* handled it.
*/
- if (is_prefetch(regs, address, error_code))
- return;
+ if (is_prefetch(regs, address, error_code))
+ return;
/*
* Oops. The kernel tried to access some bad page. We'll have to
else
printk(KERN_ALERT "BUG: unable to handle kernel paging"
" request");
- printk(" at virtual address %08lx\n",address);
+ printk(" at virtual address %08lx\n", address);
printk(KERN_ALERT "printing ip: %08lx ", regs->ip);
page = read_cr3();
up_read(&mm->mmap_sem);
/* Kernel mode? Handle exceptions or die */
- if (!(error_code & 4))
+ if (!(error_code & PF_USER))
goto no_context;
/* User space => ok to do another page fault */
/*
- * linux/arch/x86-64/mm/fault.c
- *
* Copyright (C) 1995 Linus Torvalds
* Copyright (C) 2001,2002 Andi Kleen, SuSE Labs.
*/
#include <asm/proto.h>
#include <asm-generic/sections.h>
-/* 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_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() */
}
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)
-{
+{
unsigned char *instr;
int scan_more = 1;
- int prefetch = 0;
+ int prefetch = 0;
unsigned char *max_instr;
/* 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 (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. */
scan_more = !instr_lo || (instr_lo>>1) == 1;
- break;
+ break;
case 0x00:
/* Prefetch instruction is 0x0F0D or 0x0F18 */
scan_more = 0;
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 int bad_address(void *p)
+{
unsigned long dummy;
return probe_kernel_address((unsigned long *)p, dummy);
-}
+}
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;
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;
printk("BAD\n");
}
-static const char errata93_warning[] =
+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"
/* 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)
+static int is_errata93(struct pt_regs *regs, unsigned long address)
{
static int warned;
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->ip = address;
return 1;
}
return 0;
-}
+}
static noinline void pgtable_bad(unsigned long address, struct pt_regs *regs,
unsigned long error_code)
{
struct task_struct *tsk;
struct mm_struct *mm;
- struct vm_area_struct * vma;
+ struct vm_area_struct *vma;
unsigned long address;
int write, fault;
unsigned long flags;
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;
goto good_area;
if (!(vma->vm_flags & VM_GROWSDOWN))
goto bad_area;
- if (error_code & 4) {
+ if (error_code & PF_USER) {
/* Allow userspace just enough access below the stack pointer
* to let the 'enter' instruction work.
*/
info.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;
}
/*
tsk->comm, tsk->pid, address, regs->ip,
regs->sp, error_code);
}
-
+
tsk->thread.cr2 = address;
/* Kernel addresses are always protection faults */
tsk->thread.error_code = error_code | (address >= TASK_SIZE);
}
no_context:
-
/* Are we prepared to handle this kernel fault? */
- if (fixup_exception(regs)) {
+ 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
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(" at %016lx RIP: \n" KERN_ALERT, address);
printk_address(regs->ip);
dump_pagetable(address);
tsk->thread.cr2 = address;
void vmalloc_sync_all(void)
{
- /* Note that races in the updates of insync and start aren't
+ /* 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). */
}
/* 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)));
}
projects / linux-2.6 / commitdiff