4.9-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Vineeth Remanan Pillai <[email protected]>

commit 538d92912d3190a1dd809233a0d57277459f37b2 upstream.

The commit 90c311b0eeea ("xen-netfront: Fix Rx stall during network
stress and OOM") caused the refill timer to be triggerred almost on
all invocations of xennet_alloc_rx_buffers for certain workloads.
This reworks the fix by reverting to the old behaviour and taking into
consideration the skb allocation failure. Refill timer is now triggered
on insufficient requests or skb allocation failure.

Signed-off-by: Vineeth Remanan Pillai <[email protected]>
Fixes: 90c311b0eeea (xen-netfront: Fix Rx stall during network stress and OOM)
Reported-by: Boris Ostrovsky <[email protected]>
Reviewed-by: Boris Ostrovsky <[email protected]>
Signed-off-by: David S. Miller <[email protected]>
Cc: Eduardo Valentin <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>

---
 drivers/net/xen-netfront.c |   14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

--- a/drivers/net/xen-netfront.c
+++ b/drivers/net/xen-netfront.c
@@ -281,6 +281,7 @@ static void xennet_alloc_rx_buffers(stru
 {
        RING_IDX req_prod = queue->rx.req_prod_pvt;
        int notify;
+       int err = 0;
 
        if (unlikely(!netif_carrier_ok(queue->info->netdev)))
                return;
@@ -295,8 +296,10 @@ static void xennet_alloc_rx_buffers(stru
                struct xen_netif_rx_request *req;
 
                skb = xennet_alloc_one_rx_buffer(queue);
-               if (!skb)
+               if (!skb) {
+                       err = -ENOMEM;
                        break;
+               }
 
                id = xennet_rxidx(req_prod);
 
@@ -320,8 +323,13 @@ static void xennet_alloc_rx_buffers(stru
 
        queue->rx.req_prod_pvt = req_prod;
 
-       /* Not enough requests? Try again later. */
-       if (req_prod - queue->rx.sring->req_prod < NET_RX_SLOTS_MIN) {
+       /* Try again later if there are not enough requests or skb allocation
+        * failed.
+        * Enough requests is quantified as the sum of newly created slots and
+        * the unconsumed slots at the backend.
+        */
+       if (req_prod - queue->rx.rsp_cons < NET_RX_SLOTS_MIN ||
+           unlikely(err)) {
                mod_timer(&queue->rx_refill_timer, jiffies + (HZ/10));
                return;
        }


Reply via email to