x86/pv: fix MMUEXT_FLUSH_CACHE to flush all pCPU caches

The implementation of MMUEXT_FLUSH_CACHE is bogus, as it doesn't account to
flush the cache of any previous pCPU where the current vCPU might have run,
and hence is likely to not work as expected.

Fix this by resorting to use the same logic as MMUEXT_FLUSH_CACHE_GLOBAL,
which will be correct in all cases.

Fixes: 534ffcb416bb ("Fix WBINVD by adding a new hypercall.")
Signed-off-by: Roger Pau Monné <roger.pau@citrix.com>
---
Alternatively, the hypercall could be made correct by keeping track of the
pCPUs the vCPU has run on since the last cache flush.  That's however more
work.  See later in the series.

diff --git a/xen/arch/x86/mm.c b/xen/arch/x86/mm.c
index 66c15a3c864fbe27ede3db833eeb2123a9e606ed..38e2143522013334d046716a10a0c927ba892620 100644 (file)
--- a/xen/arch/x86/mm.c
+++ b/xen/arch/x86/mm.c
@@ -3805,14 +3805,11 @@ long do_mmuext_op(
             break;
 
         case MMUEXT_FLUSH_CACHE:
-            if ( unlikely(currd != pg_owner) )
-                rc = -EPERM;
-            else if ( unlikely(!cache_flush_permitted(currd)) )
-                rc = -EACCES;
-            else
-                wbinvd();
-            break;
-
+            /*
+             * Dirty pCPU caches where the current vCPU has been scheduled are
+             * not tracked, and hence we need to resort to a global cache
+             * flush for correctness.
+             */
         case MMUEXT_FLUSH_CACHE_GLOBAL:
             if ( unlikely(currd != pg_owner) )
                 rc = -EPERM;