* [Patch] remove some rsm psr.dt in ivt.S
@ 2006-10-17 22:10 Zou Nan hai
2006-10-17 23:04 ` Zou Nan hai
2006-10-18 0:30 ` Chen, Kenneth W
0 siblings, 2 replies; 3+ messages in thread
From: Zou Nan hai @ 2006-10-17 22:10 UTC (permalink / raw)
To: linux-ia64
The rsm psr.dt in page_not_present, key_permission, iaccess_rights and
daccess_right are not necessary.
Since there is no data access between those points to page_fault, and
page_fault will enable psr.dt at the beginning.
Sign-off-by: Zou Nan hai <nanhai.zou@intel.com>
--- linux-2.6.18/arch/ia64/kernel/ivt.S 2006-10-18 00:15:11.000000000 +0800
+++ linux-2.6.18-fix/arch/ia64/kernel/ivt.S 2006-10-18 00:39:48.000000000 +0800
@@ -1201,7 +1201,6 @@ END(dispatch_to_fault_handler)
ENTRY(page_not_present)
DBG_FAULT(20)
mov r16=cr.ifa
- rsm psr.dt
/*
* The Linux page fault handler doesn't expect non-present pages to be in
* the TLB. Flush the existing entry now, so we meet that expectation.
@@ -1211,7 +1210,6 @@ ENTRY(page_not_present)
ptc.l r16,r17
;;
mov r31=pr
- srlz.d
br.sptk.many page_fault
END(page_not_present)
@@ -1221,10 +1219,8 @@ END(page_not_present)
ENTRY(key_permission)
DBG_FAULT(21)
mov r16=cr.ifa
- rsm psr.dt
mov r31=pr
;;
- srlz.d
br.sptk.many page_fault
END(key_permission)
@@ -1234,10 +1230,8 @@ END(key_permission)
ENTRY(iaccess_rights)
DBG_FAULT(22)
mov r16=cr.ifa
- rsm psr.dt
mov r31=pr
;;
- srlz.d
br.sptk.many page_fault
END(iaccess_rights)
@@ -1247,10 +1241,8 @@ END(iaccess_rights)
ENTRY(daccess_rights)
DBG_FAULT(23)
mov r16=cr.ifa
- rsm psr.dt
mov r31=pr
;;
- srlz.d
br.sptk.many page_fault
END(daccess_rights)
^ permalink raw reply [flat|nested] 3+ messages in thread
* RE: [Patch] remove some rsm psr.dt in ivt.S
2006-10-17 22:10 [Patch] remove some rsm psr.dt in ivt.S Zou Nan hai
@ 2006-10-17 23:04 ` Zou Nan hai
2006-10-18 0:30 ` Chen, Kenneth W
1 sibling, 0 replies; 3+ messages in thread
From: Zou Nan hai @ 2006-10-17 23:04 UTC (permalink / raw)
To: linux-ia64
On Wed, 2006-10-18 at 08:30, Chen, Kenneth W wrote:
> Zou Nan hai wrote on Tuesday, October 17, 2006 3:10 PM
> > The rsm psr.dt in page_not_present, key_permission, iaccess_rights
> and
> > daccess_right are not necessary.
> > Since there is no data access between those points to page_fault,
> and
> > page_fault will enable psr.dt at the beginning.
> >
> > @@ -1221,10 +1219,8 @@ END(page_not_present)
> > ENTRY(key_permission)
> > DBG_FAULT(21)
> > mov r16=cr.ifa
> > - rsm psr.dt
> > mov r31=pr
> > ;;
> > - srlz.d
> > br.sptk.many page_fault
> > END(key_permission)
>
>
> You can get rid of that stop bit too. Same for two other instances.
> And your changelog mentioned daccess_right, but patch missed that
> hunk.
>
> - Ken
>
Ok, daccess_right hunk was in the patch...
--- linux-2.6.18/arch/ia64/kernel/ivt.S 2006-10-18 00:15:11.000000000 +0800
+++ linux-2.6.18-fix/arch/ia64/kernel/ivt.S 2006-10-18 01:30:14.000000000 +0800
@@ -1201,7 +1201,6 @@ END(dispatch_to_fault_handler)
ENTRY(page_not_present)
DBG_FAULT(20)
mov r16=cr.ifa
- rsm psr.dt
/*
* The Linux page fault handler doesn't expect non-present pages to be in
* the TLB. Flush the existing entry now, so we meet that expectation.
@@ -1209,9 +1208,7 @@ ENTRY(page_not_present)
mov r17=PAGE_SHIFT<<2
;;
ptc.l r16,r17
- ;;
mov r31=pr
- srlz.d
br.sptk.many page_fault
END(page_not_present)
@@ -1221,10 +1218,7 @@ END(page_not_present)
ENTRY(key_permission)
DBG_FAULT(21)
mov r16=cr.ifa
- rsm psr.dt
mov r31=pr
- ;;
- srlz.d
br.sptk.many page_fault
END(key_permission)
@@ -1234,10 +1228,7 @@ END(key_permission)
ENTRY(iaccess_rights)
DBG_FAULT(22)
mov r16=cr.ifa
- rsm psr.dt
mov r31=pr
- ;;
- srlz.d
br.sptk.many page_fault
END(iaccess_rights)
@@ -1247,10 +1238,7 @@ END(iaccess_rights)
ENTRY(daccess_rights)
DBG_FAULT(23)
mov r16=cr.ifa
- rsm psr.dt
mov r31=pr
- ;;
- srlz.d
br.sptk.many page_fault
END(daccess_rights)
^ permalink raw reply [flat|nested] 3+ messages in thread
* RE: [Patch] remove some rsm psr.dt in ivt.S
2006-10-17 22:10 [Patch] remove some rsm psr.dt in ivt.S Zou Nan hai
2006-10-17 23:04 ` Zou Nan hai
@ 2006-10-18 0:30 ` Chen, Kenneth W
1 sibling, 0 replies; 3+ messages in thread
From: Chen, Kenneth W @ 2006-10-18 0:30 UTC (permalink / raw)
To: linux-ia64
Zou Nan hai wrote on Tuesday, October 17, 2006 3:10 PM
> The rsm psr.dt in page_not_present, key_permission, iaccess_rights and
> daccess_right are not necessary.
> Since there is no data access between those points to page_fault, and
> page_fault will enable psr.dt at the beginning.
>
> @@ -1221,10 +1219,8 @@ END(page_not_present)
> ENTRY(key_permission)
> DBG_FAULT(21)
> mov r16=cr.ifa
> - rsm psr.dt
> mov r31=pr
> ;;
> - srlz.d
> br.sptk.many page_fault
> END(key_permission)
You can get rid of that stop bit too. Same for two other instances.
And your changelog mentioned daccess_right, but patch missed that hunk.
- Ken
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2006-10-18 0:30 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-10-17 22:10 [Patch] remove some rsm psr.dt in ivt.S Zou Nan hai
2006-10-17 23:04 ` Zou Nan hai
2006-10-18 0:30 ` Chen, Kenneth W
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox