[PATCH] Print error code in page faults

[PATCH] Print error code in page faults

[Glauber de Oliveira Costa]

[-- Attachment #1: Type: text/plain, Size: 312 bytes --]

When a page faults comes from a kernel space, the printed summary
leaves us clueless about what kind of access was being tried (which
is encoded in the error_code variable).

Having it promply available may ease debugging in a bunch of
situations.

Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>



[-- Attachment #2: patch_err --]
[-- Type: text/plain, Size: 542 bytes --]

diff --git a/arch/x86_64/mm/fault.c b/arch/x86_64/mm/fault.c
index 6ada723..e65522e 100644
--- a/arch/x86_64/mm/fault.c
+++ b/arch/x86_64/mm/fault.c
@@ -539,7 +539,7 @@ 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(" at %016lx (error=0x%02lx) RIP: \n" KERN_ALERT, error_code, address);
 	printk_address(regs->rip);
 	dump_pagetable(address);
 	tsk->thread.cr2 = address;

Re: [PATCH] Print error code in page faults

[Chuck Ebbert]

Glauber de Oliveira Costa wrote:
> When a page faults comes from a kernel space, the printed summary
> leaves us clueless about what kind of access was being tried (which
> is encoded in the error_code variable).
> 
> Having it promply available may ease debugging in a bunch of
> situations.
> 
> Signed-off-by: Glauber de Oliveira Costa <gcosta@redhat.com>
> 
> 
> 
> ------------------------------------------------------------------------
> 
> diff --git a/arch/x86_64/mm/fault.c b/arch/x86_64/mm/fault.c
> index 6ada723..e65522e 100644
> --- a/arch/x86_64/mm/fault.c
> +++ b/arch/x86_64/mm/fault.c
> @@ -539,7 +539,7 @@ 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(" at %016lx (error=0x%02lx) RIP: \n" KERN_ALERT, error_code, address);
>  	printk_address(regs->rip);
>  	dump_pagetable(address);
>  	tsk->thread.cr2 = address;

Umm, it's already there, right after the word "Oops".


Oops: 0002 [1] SMP
      ^^^^

Re: [PATCH] Print error code in page faults

[Glauber de Oliveira Costa]

On Sat, Feb 24, 2007 at 02:12:29PM -0500, Chuck Ebbert wrote:
> 
> Umm, it's already there, right after the word "Oops".
> 
> 
> Oops: 0002 [1] SMP
>       ^^^^
Oops! ;-) 
 

-- 
Glauber de Oliveira Costa
Red Hat Inc.
"Free as in Freedom"

Re: [PATCH] Print error code in page faults

[Andi Kleen]

On Saturday 24 February 2007 06:31, Glauber de Oliveira Costa wrote:
> When a page faults comes from a kernel space, the printed summary
> leaves us clueless about what kind of access was being tried (which
> is encoded in the error_code variable).
> 
> Having it promply available may ease debugging in a bunch of
> situations.

It's in the __die output. I don't think outputting it twice is needed

-Andi