[PATCH v3 07/12] x86/shadow: OOS doesn't track VAs anymore

Jan Beulich posted 12 patches 2 years, 8 months ago
[PATCH v3 07/12] x86/shadow: OOS doesn't track VAs anymore
Posted by Jan Beulich 2 years, 8 months ago
The tracking lasted only for about two weeks, but the related comment
parts were never purged.

Fixes: 50b74f55e0c0 ("OOS cleanup: Fixup arrays instead of fixup tables")
Signed-off-by: Jan Beulich <jbeulich@suse.com>
---
I'm heavily inclined to fold this into "x86/shadow: move OOS functions
to their own file".

This largely removes the justification for the per-vCPU hash. Is there
any other reason to name there in exchange?
---
v3: New.

--- a/xen/arch/x86/mm/shadow/oos.c
+++ b/xen/arch/x86/mm/shadow/oos.c
@@ -51,13 +51,10 @@
  *
  * Currently out-of-sync pages are listed in a simple open-addressed
  * hash table with a second chance (must resist temptation to radically
- * over-engineer hash tables...)  The virtual address of the access
- * which caused us to unsync the page is also kept in the hash table, as
- * a hint for finding the writable mappings later.
+ * over-engineer hash tables...).
  *
  * We keep a hash per vcpu, because we want as much as possible to do
- * the re-sync on the save vcpu we did the unsync on, so the VA hint
- * will be valid.
+ * the re-sync on the same vcpu we did the unsync on.
  */
 
 #if SHADOW_AUDIT & SHADOW_AUDIT_ENTRIES_FULL
Re: [PATCH v3 07/12] x86/shadow: OOS doesn't track VAs anymore
Posted by Roger Pau Monné 2 years, 8 months ago
On Tue, May 16, 2023 at 09:40:50AM +0200, Jan Beulich wrote:
> The tracking lasted only for about two weeks, but the related comment
> parts were never purged.
> 
> Fixes: 50b74f55e0c0 ("OOS cleanup: Fixup arrays instead of fixup tables")
> Signed-off-by: Jan Beulich <jbeulich@suse.com>

Acked-by: Roger Pau Monné <roger.pau@citrix.com>

Thanks, Roger.