diff mbox

[lucid] SRU: [PATCH] KVM: add schedule check to napi_enable call

Message ID 887199.99844.qm@web110310.mail.gq1.yahoo.com
State Accepted
Headers show

Commit Message

Ken Stailey Feb. 6, 2011, 11:20 p.m. UTC 
SRU Justification:

Impact: Under heavy network I/O load virtio-net driver crashes making VM guest unusable.

Testcase: I left a current Lucid VM running two concurrent "scp -r" of > 200 GB from NFS read-only source to a physical remote host overnight. VM quickly started emitting "page allocation errors" in the system log. Next morning when I checked the VM I could still ping it but could not establish an SSH connection.  

I put the patch in to ppa:nutznboltz/lucid-virtio-napi and applied it to the same machine and that VM did not crash as a result of copying the same data.

$ uname -a
Linux dubnium 2.6.32-28-server #55ubuntu1~ppa3~lucid1-Ubuntu SMP Sun Feb 6 01:03:25 UTC 2011 x86_64 GNU/Linux
diff mbox

Patch

diff -u linux-2.6.32/drivers/net/virtio_net.c linux-2.6.32/drivers/net/virtio_net.c
--- linux-2.6.32/drivers/net/virtio_net.c
+++ linux-2.6.32/drivers/net/virtio_net.c
@@ -391,6 +391,20 @@ 
     }
 }
 
+static void virtnet_napi_enable(struct virtnet_info *vi)
+{
+    napi_enable(&vi->napi);
+
+    /* If all buffers were filled by other side before we napi_enabled, we
+     * won't get another interrupt, so process any outstanding packets
+     * now.  virtnet_poll wants re-enable the queue, so we disable here.
+     * We synchronize against interrupts via NAPI_STATE_SCHED */
+    if (napi_schedule_prep(&vi->napi)) {
+        vi->rvq->vq_ops->disable_cb(vi->rvq);
+        __napi_schedule(&vi->napi);
+    }
+}
+
 static void refill_work(struct work_struct *work)
 {
     struct virtnet_info *vi;
@@ -399,7 +413,7 @@ 
     vi = container_of(work, struct virtnet_info, refill.work);
     napi_disable(&vi->napi);
     still_empty = !try_fill_recv(vi, GFP_KERNEL);
-    napi_enable(&vi->napi);
+    virtnet_napi_enable(vi);
 
     /* In theory, this can happen: if we don't get any buffers in
      * we will *never* try to fill again. */
@@ -591,16 +605,7 @@ 
 {
     struct virtnet_info *vi = netdev_priv(dev);
 
-    napi_enable(&vi->napi);
-
-    /* If all buffers were filled by other side before we napi_enabled, we
-     * won't get another interrupt, so process any outstanding packets
-     * now.  virtnet_poll wants re-enable the queue, so we disable here.
-     * We synchronize against interrupts via NAPI_STATE_SCHED */
-    if (napi_schedule_prep(&vi->napi)) {
-        vi->rvq->vq_ops->disable_cb(vi->rvq);
-        __napi_schedule(&vi->napi);
-    }
+    virtnet_napi_enable(vi);
     return 0;
 }