[patch] 4 level page table bug fix in vhpt_miss

[patch] 4 level page table bug fix in vhpt_miss

[Chen, Kenneth W]

From source code inspection, I think there is a bug with 4 level
page table with vhpt_miss handler.  In the code path of rechecking
page table entry against previously read value after tlb insertion,
*pte value in register r18 was overwritten with value newly read
from pud pointer, render the check of new *pte against previous
*pte completely wrong.  Though the bug is none fatal and the penalty
is to purge the entry and retry.  For functional correctness, it
should be fixed.  The fix is to use a different register so new
*pud don't trash *pte.  (btw, the comments in the cmp statement is
wrong as well, which I will address in the next patch).

Signed-off-by: Ken Chen <kenneth.w.chen@intel.com> 

--- ./arch/ia64/kernel/ivt.S.orig	2005-11-16 23:35:27.086799703 -0800
+++ ./arch/ia64/kernel/ivt.S	2005-11-17 00:14:51.247903555 -0800
@@ -209,13 +209,13 @@ ENTRY(vhpt_miss)
 	ld8 r25=[r21]				// read L4 entry again
 	ld8 r26=[r17]				// read L3 PTE again
 #ifdef CONFIG_PGTABLE_4
-	ld8 r18=[r28]				// read L2 entry again
+	ld8 r19=[r28]				// read L2 entry again
 #endif
 	cmp.ne p6,p7=r0,r0
 	;;
 	cmp.ne.or.andcm p6,p7=r26,r20		// did L3 entry change
 #ifdef CONFIG_PGTABLE_4
-	cmp.ne.or.andcm p6,p7=r29,r18		// did L4 PTE change
+	cmp.ne.or.andcm p6,p7=r19,r29		// did L4 PTE change
 #endif
 	mov r27=PAGE_SHIFT<<2
 	;;

Re: [patch] 4 level page table bug fix in vhpt_miss

[Robin Holt]

This brings performance back in line with the three-level case (approx
0.2% slowdown).  Thanks for finding this.

Robin

On Thu, Nov 17, 2005 at 01:38:42AM -0800, Chen, Kenneth W wrote:
> >From source code inspection, I think there is a bug with 4 level
> page table with vhpt_miss handler.  In the code path of rechecking
> page table entry against previously read value after tlb insertion,
> *pte value in register r18 was overwritten with value newly read
> from pud pointer, render the check of new *pte against previous
> *pte completely wrong.  Though the bug is none fatal and the penalty
> is to purge the entry and retry.  For functional correctness, it
> should be fixed.  The fix is to use a different register so new
> *pud don't trash *pte.  (btw, the comments in the cmp statement is
> wrong as well, which I will address in the next patch).
> 
> Signed-off-by: Ken Chen <kenneth.w.chen@intel.com> 
> 
> --- ./arch/ia64/kernel/ivt.S.orig	2005-11-16 23:35:27.086799703 -0800
> +++ ./arch/ia64/kernel/ivt.S	2005-11-17 00:14:51.247903555 -0800
> @@ -209,13 +209,13 @@ ENTRY(vhpt_miss)
>  	ld8 r25=[r21]				// read L4 entry again
>  	ld8 r26=[r17]				// read L3 PTE again
>  #ifdef CONFIG_PGTABLE_4
> -	ld8 r18=[r28]				// read L2 entry again
> +	ld8 r19=[r28]				// read L2 entry again
>  #endif
>  	cmp.ne p6,p7=r0,r0
>  	;;
>  	cmp.ne.or.andcm p6,p7=r26,r20		// did L3 entry change
>  #ifdef CONFIG_PGTABLE_4
> -	cmp.ne.or.andcm p6,p7=r29,r18		// did L4 PTE change
> +	cmp.ne.or.andcm p6,p7=r19,r29		// did L4 PTE change
>  #endif
>  	mov r27=PAGE_SHIFT<<2
>  	;;
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-ia64" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html