lkml.org 
[lkml]   [2011]   [Jun]   [13]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
SubjectRe: [PATCH] oprofile, x86: Fix nmi-unsafe callgraph support in, 2.6.27 and later
Date

On 06/09/11 08:57, Robert Richter wrote:
> On 08.06.11 13:43:56, John Lumby wrote:
>
> "EIP is at print_context_stack=0x45/0xb0
>
> and from a machine-code listing, I found that that offset corresponds
> to the line
>
> addr = *stack;
>
> in arch/x86/kernel/dumpstack.c"
>
> Actually this should not happen, because of checking the stack pointer
> in valid_stack_ptr(). So could you apply the change below and test if
> this throws a bug message?

I applied that change and, yes, it throws the BUG :
kernel BUG at arch/x86/kernel/dumpstack.c : 76!
It appears that the existing test in valid_stack_ptr makes too many assumptions
about the other variables involved to catch all cases.

Also, changing that test by adding a simplistic check for an absolute
range is not sufficient, as another crash then occurs in dump_trace at
the line
stack = (unsigned long *)context->previous_esp;

This patch to dumpstack_32.c seems to work for my system:
 (although it is probably more of a cover-up than a fix)

--- linux-2.6.39-FLG/arch/x86/kernel/dumpstack_32.c.orig 2011-05-19
00:06:34.000000000 -0400
+++ linux-2.6.39-FLG/arch/x86/kernel/dumpstack_32.c 2011-06-12
23:09:16.000000000 -0400
@@ -42,6 +42,19 @@ void dump_trace(struct task_struct *task

context = (struct thread_info *)
((unsigned long)stack & (~(THREAD_SIZE - 1)));
+
+ /* crude bugcheck */
+ if ( ((unsigned long)stack <= 1024)
+ || ((unsigned long)stack >= 0xffffe000)
+ || ((unsigned long)context <= 1024)
+ || ((unsigned long)context >= 0xffffe000)
+ ) {
+#ifdef NOISY_DUMP_TRACE
+ printk("dump_trace: pid %d stack= %p context= %p\n"
+ ,(task ? task->pid : -1),stack,context);
+#endif /* NOISY_DUMP_TRACE */
+ break;
+ }
bp = ops->walk_stack(context, stack, bp, ops, data, NULL, &graph);

stack = (unsigned long *)context->previous_esp;

> Thanks,
>
> -Robert
>
>
> diff --git a/arch/x86/kernel/dumpstack.c b/arch/x86/kernel/dumpstack.c
> index e2a3f06..37693f5 100644
> --- a/arch/x86/kernel/dumpstack.c
> +++ b/arch/x86/kernel/dumpstack.c
> @@ -73,6 +73,8 @@ static inline int valid_stack_ptr(struct thread_info *tinfo,
> void *p, unsigned int size, void *end)
> {
> void *t = tinfo;
> +
> + BUG_ON(p< (void *)THREAD_SIZE);
> if (end) {
> if (p< end&& p>= (end-THREAD_SIZE))
> return 1;
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2011-06-13 15:43    [W:3.101 / U:0.020 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site