[PATCH] kvm: qemu: net: unbreak tap networking

[PATCH] kvm: qemu: net: unbreak tap networking

[Mark McLoughlin]

A recent merge broke tap networking because qemu_send_packet()
now always returns -EAGAIN causing tap to try and resend the
same packet forever.

Fix by having qemu_send_packet() return the status from the
qemu_deliver_packet() of the packet in question, rather than
any packets which happened to have been queued up during the
delivery of the original packet.

Signed-off-by: Mark McLoughlin <markmc@redhat.com>
---
 net.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/net.c b/net.c
index 16f6e48..b5ebcea 100644
--- a/net.c
+++ b/net.c
@@ -445,9 +445,9 @@ int qemu_send_packet(VLANClientState *vc, const uint8_t *buf, int size)
         vlan->send_queue = packet;
     } else {
         vlan->delivering = 1;
-        qemu_deliver_packet(vc, buf, size);
+        ret = qemu_deliver_packet(vc, buf, size);
         while ((packet = vlan->send_queue) != NULL) {
-            ret = qemu_deliver_packet(packet->sender, packet->data, packet->size);
+            qemu_deliver_packet(packet->sender, packet->data, packet->size);
             vlan->send_queue = packet->next;
             qemu_free(packet);
         }
-- 
1.6.0.6

Re: [PATCH] kvm: qemu: net: unbreak tap networking

[Avi Kivity]

Mark McLoughlin wrote:
> A recent merge broke tap networking because qemu_send_packet()
> now always returns -EAGAIN causing tap to try and resend the
> same packet forever.
>
> Fix by having qemu_send_packet() return the status from the
> qemu_deliver_packet() of the packet in question, rather than
> any packets which happened to have been queued up during the
> delivery of the original packet.
>   

Applied, thanks.

What's the outlook on converging with upstream here?  Otherwise I'll 
keep breaking net on every merge.

-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.

Re: [PATCH] kvm: qemu: net: unbreak tap networking

[Mark McLoughlin]

On Tue, 2009-04-28 at 11:58 +0300, Avi Kivity wrote:
> Mark McLoughlin wrote:
> > A recent merge broke tap networking because qemu_send_packet()
> > now always returns -EAGAIN causing tap to try and resend the
> > same packet forever.
> >
> > Fix by having qemu_send_packet() return the status from the
> > qemu_deliver_packet() of the packet in question, rather than
> > any packets which happened to have been queued up during the
> > delivery of the original packet.
> >   
> 
> Applied, thanks.
> 
> What's the outlook on converging with upstream here?  Otherwise I'll 
> keep breaking net on every merge.

Yeah, I was getting close to something sane and then got distracted by
other things. Soon, hopefully.

Cheers,
Mark.