diff options
| author | Zachary Amsden <zach@vmware.com> | 2007-02-13 13:26:21 +0100 |
|---|---|---|
| committer | Andi Kleen <andi@basil.nowhere.org> | 2007-02-13 13:26:21 +0100 |
| commit | 7b3552024380f306a6c50d5105d18d9d4258fa4e (patch) | |
| tree | 904a408768b216be7f759b4a190809f851a914d5 /arch/i386 | |
| parent | bbab4f3bb7f528d2b8ccb5de9ae5f6ff3fb29684 (diff) | |
[PATCH] i386: Profile pc badness
Profile_pc was broken when using paravirtualization because the
assumption the kernel was running at CPL 0 was violated, causing
bad logic to read a random value off the stack.
The only way to be in kernel lock functions is to be in kernel
code, so validate that assumption explicitly by checking the CS
value. We don't want to be fooled by BIOS / APM segments and
try to read those stacks, so only match KERNEL_CS.
I moved some stuff in segment.h to make it prettier.
Signed-off-by: Zachary Amsden <zach@vmware.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Diffstat (limited to 'arch/i386')
| -rw-r--r-- | arch/i386/kernel/time.c | 10 |
1 files changed, 4 insertions, 6 deletions
diff --git a/arch/i386/kernel/time.c b/arch/i386/kernel/time.c index 9603ccaba99..a4f67a6e682 100644 --- a/arch/i386/kernel/time.c +++ b/arch/i386/kernel/time.c @@ -131,15 +131,13 @@ unsigned long profile_pc(struct pt_regs *regs) unsigned long pc = instruction_pointer(regs); #ifdef CONFIG_SMP - if (!user_mode_vm(regs) && in_lock_functions(pc)) { + if (!v8086_mode(regs) && SEGMENT_IS_KERNEL_CODE(regs->xcs) && + in_lock_functions(pc)) { #ifdef CONFIG_FRAME_POINTER return *(unsigned long *)(regs->ebp + 4); #else - unsigned long *sp; - if ((regs->xcs & 3) == 0) - sp = (unsigned long *)®s->esp; - else - sp = (unsigned long *)regs->esp; + unsigned long *sp = (unsigned long *)®s->esp; + /* Return address is either directly at stack pointer or above a saved eflags. Eflags has bits 22-31 zero, kernel addresses don't. */ |