[Spice-devel] Fwd: [PATCH 0/3] spice server vdi_port interface fixes

[Alon Levy]

It seems I never actually sent this to the list. Three single line
patches with incredibly detailed explanation.

----- Forwarded Message -----
From: "Alon Levy" <alevy at redhat.com>
To: alevy at redhat.com
Sent: Sunday, August 22, 2010 10:28:35 PM (GMT+0200) Auto-Detected
Subject: [PATCH 0/3] spice server vdi_port interface fixes

These patches fix issues preventing large clipboard to guest and to client
from succeeding. Specifically they fix issues related to the change from the
old vdi_port device to the new virtio-serial based solution.

Issues left that are not addressed:
 * we still call spice-server from two threads. The second patch just
 prevents one of the paths, but it isn't clear this can't happen otherwise.
 no mutex has been added.
 * still no clipboard channel. It is clear from testing that mouse doesn't
 respond while sending a large (1MB) buffer. However, this doesn't
 seem to be because of spice (we send chunks of 2056 bytes, which
 can be interposed with mouse movement messages, and in the other direction
 it is a totally separate queue), but perhaps driver or qemu issues. Not
 clear if they will go away when using a separate channel.

Alon Levy (3):
  server: bugfix - make vdi_port_write_retry reiterate if write_queue
    still not empty
  server: fix race when data arrives from guest through vdi interface
  server: retry mouse send on queue full - required after previous race
    fix

 server/reds.c |    5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)