[RESEND,2/8] gnttab: avoid spurious maptrack handle allocation failures

Commit Message

Jan Beulich Aug. 15, 2017, 2:39 p.m. UTC

When no memory is available in the hypervisor, rather than immediately
failing the request, try to steal a handle from another vCPU.

Reported-by: George Dunlap <[e-mail]>
Signed-off-by: Jan Beulich <[e-mail]>

Comments

Andrew Cooper Aug. 15, 2017, 4:52 p.m. UTC | #1

On 15/08/17 15:39, Jan Beulich wrote:
> @@ -422,8 +422,13 @@ get_maptrack_handle(
>      /*
>       * If we've run out of frames, try stealing an entry from another
>       * VCPU (in case the guest isn't mapping across its VCPUs evenly).
> +     * Also use this path in case we're out of memory, to avoid spurious
> +     * failures.

This comment isn't strictly correct any more.  It is now "If we've run
out of handles and still have frame headroom, try allocating a new
maptrack frame.  If there is no headroom, or Xen is out of memory, try
stealing an entry from another vcpu".

~Andrew

>       */
> -    if ( nr_maptrack_frames(lgt) >= max_maptrack_frames )
> +    if ( nr_maptrack_frames(lgt) < max_maptrack_frames )
> +        new_mt = alloc_xenheap_page();
> +
> +    if ( !new_mt )
>      {
>          spin_unlock(&lgt->maptrack_lock);
>

Patch

--- a/xen/common/grant_table.c
+++ b/xen/common/grant_table.c
@@ -411,7 +411,7 @@  get_maptrack_handle(
     struct vcpu          *curr = current;
     unsigned int          i, head;
     grant_handle_t        handle;
-    struct grant_mapping *new_mt;
+    struct grant_mapping *new_mt = NULL;
 
     handle = __get_maptrack_handle(lgt, curr);
     if ( likely(handle != -1) )
@@ -422,8 +422,13 @@  get_maptrack_handle(
     /*
      * If we've run out of frames, try stealing an entry from another
      * VCPU (in case the guest isn't mapping across its VCPUs evenly).
+     * Also use this path in case we're out of memory, to avoid spurious
+     * failures.
      */
-    if ( nr_maptrack_frames(lgt) >= max_maptrack_frames )
+    if ( nr_maptrack_frames(lgt) < max_maptrack_frames )
+        new_mt = alloc_xenheap_page();
+
+    if ( !new_mt )
     {
         spin_unlock(&lgt->maptrack_lock);
 
@@ -446,12 +451,6 @@  get_maptrack_handle(
         return steal_maptrack_handle(lgt, curr);
     }
 
-    new_mt = alloc_xenheap_page();
-    if ( !new_mt )
-    {
-        spin_unlock(&lgt->maptrack_lock);
-        return -1;
-    }
     clear_page(new_mt);
 
     /*