[patch] i386: show_registers(): try harder to print failing code

[patch] i386: show_registers(): try harder to print failing code

[Chuck Ebbert]

show_registers() tries to dump failing code starting 43 bytes
before the offending instruction, but this address can be bad,
for example in a device driver where the failing instruction is
less than 43 bytes from the start of the driver's code.  When that
happens, try to dump code starting at the failing instruction
instead of printing no code at all.

Signed-off-by: Chuck Ebbert <76306.1226@compuserve.com>

--- 2.6.18-rc1-32.orig/arch/i386/kernel/traps.c
+++ 2.6.18-rc1-32/arch/i386/kernel/traps.c
@@ -307,6 +307,8 @@ void show_registers(struct pt_regs *regs
 	 */
 	if (in_kernel) {
 		u8 __user *eip;
+		int code_bytes = 64;
+		unsigned char c;
 
 		printk("\n" KERN_EMERG "Stack: ");
 		show_stack_log_lvl(NULL, regs, (unsigned long *)esp, KERN_EMERG);
@@ -314,9 +316,12 @@ void show_registers(struct pt_regs *regs
 		printk(KERN_EMERG "Code: ");
 
 		eip = (u8 __user *)regs->eip - 43;
-		for (i = 0; i < 64; i++, eip++) {
-			unsigned char c;
-
+		if (eip < (u8 __user *)PAGE_OFFSET || __get_user(c, eip)) {
+			/* try starting at EIP */
+			eip = (u8 __user *)regs->eip;
+			code_bytes = 32;
+		}
+		for (i = 0; i < code_bytes; i++, eip++) {
 			if (eip < (u8 __user *)PAGE_OFFSET || __get_user(c, eip)) {
 				printk(" Bad EIP value.");
 				break;
-- 
Chuck

Re: [patch] i386: show_registers(): try harder to print failing code

[Jesper Juhl]

On 18/07/06, Chuck Ebbert <76306.1226@compuserve.com> wrote:
> show_registers() tries to dump failing code starting 43 bytes
> before the offending instruction, but this address can be bad,
> for example in a device driver where the failing instruction is
> less than 43 bytes from the start of the driver's code.  When that
> happens, try to dump code starting at the failing instruction
> instead of printing no code at all.
>
Shouldn't the kernel be printing some info noting that this fallback
is in use then? Or will that be completely obvious and I'm just not
able to see that?

-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html

Re: [patch] i386: show_registers(): try harder to print failing code

[Keith Owens]

"Jesper Juhl" (on Wed, 19 Jul 2006 00:12:32 +0200) wrote:
>On 18/07/06, Chuck Ebbert <76306.1226@compuserve.com> wrote:
>> show_registers() tries to dump failing code starting 43 bytes
>> before the offending instruction, but this address can be bad,
>> for example in a device driver where the failing instruction is
>> less than 43 bytes from the start of the driver's code.  When that
>> happens, try to dump code starting at the failing instruction
>> instead of printing no code at all.
>>
>Shouldn't the kernel be printing some info noting that this fallback
>is in use then? Or will that be completely obvious and I'm just not
>able to see that?

The instruction at the EIP is bracketed, which makes it obvious if you
got any preceding instructions or not.

Re: [patch] i386: show_registers(): try harder to print failing code

[Chuck Ebbert]

In-Reply-To: <9a8748490607181512t11e9970eu1a7aa1ad1644ec54@mail.gmail.com>

On Wed, 19 Jul 2006 00:12:32 +0200, Jesper Juhl wrote:
> 
> > show_registers() tries to dump failing code starting 43 bytes
> > before the offending instruction, but this address can be bad,
> > for example in a device driver where the failing instruction is
> > less than 43 bytes from the start of the driver's code.  When that
> > happens, try to dump code starting at the failing instruction
> > instead of printing no code at all.
> >
> Shouldn't the kernel be printing some info noting that this fallback
> is in use then? Or will that be completely obvious and I'm just not
> able to see that?

The code byte at EIP is marked with '<>', so it's obvious:

Code: <a1> 00 00 00 00 c7 04 24 05 30 b5 de 89 44 24 04 e8 f5 6f 5c e1 c9 31 c0 c3 00 00 00 00 00 00 00

-- 
Chuck

Re: [patch] i386: show_registers(): try harder to print failing code

[Jesper Juhl]

On 19/07/06, Chuck Ebbert <76306.1226@compuserve.com> wrote:
> In-Reply-To: <9a8748490607181512t11e9970eu1a7aa1ad1644ec54@mail.gmail.com>
>
> On Wed, 19 Jul 2006 00:12:32 +0200, Jesper Juhl wrote:
> >
> > > show_registers() tries to dump failing code starting 43 bytes
> > > before the offending instruction, but this address can be bad,
> > > for example in a device driver where the failing instruction is
> > > less than 43 bytes from the start of the driver's code.  When that
> > > happens, try to dump code starting at the failing instruction
> > > instead of printing no code at all.
> > >
> > Shouldn't the kernel be printing some info noting that this fallback
> > is in use then? Or will that be completely obvious and I'm just not
> > able to see that?
>
> The code byte at EIP is marked with '<>', so it's obvious:
>
> Code: <a1> 00 00 00 00 c7 04 24 05 30 b5 de 89 44 24 04 e8 f5 6f 5c e1 c9 31 c0 c3 00 00 00 00 00 00 00
>

Ahh, ok. I was not aware of that. Thank you for the info.

-- 
Jesper Juhl <jesper.juhl@gmail.com>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html