diff mbox

[03/26] rtl8139: flush queued packets when RxBufPtr is written

Message ID 1371073298-14519-4-git-send-email-mdroth@linux.vnet.ibm.com
State New
Headers show

Commit Message

Michael Roth June 12, 2013, 9:41 p.m. UTC 
From: Stefan Hajnoczi <stefanha@redhat.com>

Net queues support efficient "receive disable".  For example, tap's file
descriptor will not be polled while its peer has receive disabled.  This
saves CPU cycles for needlessly copying and then dropping packets which
the peer cannot receive.

rtl8139 is missing the qemu_flush_queued_packets() call that wakes the
queue up when receive becomes possible again.

As a result, the Windows 7 guest driver reaches a state where the
rtl8139 cannot receive packets.  The driver has actually refilled the
receive buffer but we never resume reception.

The bug can be reproduced by running a large FTP 'get' inside a Windows
7 guest:

  $ qemu -netdev tap,id=tap0,...
         -device rtl8139,netdev=tap0

The Linux guest driver does not trigger the bug, probably due to a
different buffer management strategy.

Reported-by: Oliver Francke <oliver.francke@filoo.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
(cherry picked from commit 00b7ade807b5ce6779ddd86ce29c5521ec5c529a)

Signed-off-by: Michael Roth <mdroth@linux.vnet.ibm.com>
---
 hw/net/rtl8139.c |    3 +++
 1 file changed, 3 insertions(+)
diff mbox

Patch

diff --git a/hw/net/rtl8139.c b/hw/net/rtl8139.c
index 9369507..7993f9f 100644
--- a/hw/net/rtl8139.c
+++ b/hw/net/rtl8139.c
@@ -2575,6 +2575,9 @@  static void rtl8139_RxBufPtr_write(RTL8139State *s, uint32_t val)
     /* this value is off by 16 */
     s->RxBufPtr = MOD2(val + 0x10, s->RxBufferSize);
 
+    /* more buffer space may be available so try to receive */
+    qemu_flush_queued_packets(qemu_get_queue(s->nic));
+
     DPRINTF(" CAPR write: rx buffer length %d head 0x%04x read 0x%04x\n",
         s->RxBufferSize, s->RxBufAddr, s->RxBufPtr);
 }