* [PATCH] Flush stale TLBs
@ 2007-10-30 14:18 Ben Guthro
2007-10-30 14:45 ` Ben Guthro
0 siblings, 1 reply; 2+ messages in thread
From: Ben Guthro @ 2007-10-30 14:18 UTC (permalink / raw)
To: xen-devel; +Cc: Robert Phillips
[-- Attachment #1.1: Type: text/plain, Size: 743 bytes --]
This patch causes a flush of the local host TLBs after any
modification of the linearpagetable mapping.
This was not needed when vmenter/vmexit always had the side effect of
flushing host TLBs.
But, with SVM ASIDs, it is possible to:
(1) Update CR3 update,
(2) vmenter the guest, and
(3) and vmexit due to a page fault
all without an intervening host TLB flush.
Then the page fault code could use the linear pagetable
to read a top-level shadow page table entry.
But, without this change, it would fetch the wrong value
due to a stale TLB.
This code applies to unstable.
We also have a 3.1.2 version available upon request.
Signed-off-by: Robert Phillips <rphillips@virtualiron.com>
Signed-off-by: Ben Guthro <bguthro@virtualiron.com>
[-- Attachment #1.2: Type: text/html, Size: 1088 bytes --]
[-- Attachment #2: xen-stale-tlb.patch --]
[-- Type: text/x-patch, Size: 989 bytes --]
diff -r 5612da277300 xen/arch/x86/mm/shadow/multi.c
--- a/xen/arch/x86/mm/shadow/multi.c Mon Oct 29 07:54:17 2007 -0400
+++ b/xen/arch/x86/mm/shadow/multi.c Mon Oct 29 07:56:15 2007 -0400
@@ -3392,6 +3392,20 @@ sh_update_linear_entries(struct vcpu *v)
#else
#error this should not happen
#endif
+ if ( shadow_mode_external(d) )
+ {
+ /* Having modified the linear pagetable mapping, flush local host TLBs.
+ * This was not needed when vmenter/vmexit always had the side effect of
+ * flushing host TLBs but, with SVM ASIDs, it is possible to
+ * finish this CR3 update, vmenter the guest, vmexit due to a
+ * page fault, without an intervening host TLB flush.
+ * Then the page fault code could use the linear pagetable
+ * to read a top-level shadow page table entry.
+ * But, without this change, it would fetch the wrong value
+ * due to a stale TLB.
+ */
+ flush_local_tlb();
+ }
}
[-- Attachment #3: Type: text/plain, Size: 138 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [PATCH] Flush stale TLBs
2007-10-30 14:18 [PATCH] Flush stale TLBs Ben Guthro
@ 2007-10-30 14:45 ` Ben Guthro
0 siblings, 0 replies; 2+ messages in thread
From: Ben Guthro @ 2007-10-30 14:45 UTC (permalink / raw)
To: xen-devel; +Cc: Robert Phillips
[-- Attachment #1.1: Type: text/plain, Size: 910 bytes --]
Apologies. The prior version caused build errors in unstable.
Please use the attached version instead
Ben Guthro wrote:
> This patch causes a flush of the local host TLBs after any
> modification of the linearpagetable mapping.
>
> This was not needed when vmenter/vmexit always had the side effect of
> flushing host TLBs.
>
> But, with SVM ASIDs, it is possible to:
> (1) Update CR3 update,
> (2) vmenter the guest, and
> (3) and vmexit due to a page fault
> all without an intervening host TLB flush.
>
> Then the page fault code could use the linear pagetable
> to read a top-level shadow page table entry.
>
> But, without this change, it would fetch the wrong value
> due to a stale TLB.
>
> This code applies to unstable.
> We also have a 3.1.2 version available upon request.
>
> Signed-off-by: Robert Phillips <rphillips@virtualiron.com>
> Signed-off-by: Ben Guthro <bguthro@virtualiron.com>
>
[-- Attachment #1.2: Type: text/html, Size: 1443 bytes --]
[-- Attachment #2: xen-stale-tlb.patch --]
[-- Type: text/x-patch, Size: 989 bytes --]
diff -r 5612da277300 xen/arch/x86/mm/shadow/multi.c
--- a/xen/arch/x86/mm/shadow/multi.c Mon Oct 29 07:54:17 2007 -0400
+++ b/xen/arch/x86/mm/shadow/multi.c Mon Oct 29 07:56:15 2007 -0400
@@ -3392,6 +3392,20 @@ sh_update_linear_entries(struct vcpu *v)
#else
#error this should not happen
#endif
+ if ( shadow_mode_external(d) )
+ {
+ /* Having modified the linear pagetable mapping, flush local host TLBs.
+ * This was not needed when vmenter/vmexit always had the side effect of
+ * flushing host TLBs but, with SVM ASIDs, it is possible to
+ * finish this CR3 update, vmenter the guest, vmexit due to a
+ * page fault, without an intervening host TLB flush.
+ * Then the page fault code could use the linear pagetable
+ * to read a top-level shadow page table entry.
+ * But, without this change, it would fetch the wrong value
+ * due to a stale TLB.
+ */
+ flush_tlb_local();
+ }
}
[-- Attachment #3: Type: text/plain, Size: 138 bytes --]
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2007-10-30 14:45 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-10-30 14:18 [PATCH] Flush stale TLBs Ben Guthro
2007-10-30 14:45 ` Ben Guthro
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.