From a8733f5dcec4928b3efdf2529f9fa8a4422fae60 Mon Sep 17 00:00:00 2001
From: Gerd Hoffmann <kraxel@redhat.com>
Date: Fri, 16 Jul 2010 11:24:58 -0300
Subject: [PATCH 2/4] spice: don't force fullscreen redraw on display resize.
RH-Author: Gerd Hoffmann <kraxel@redhat.com>
Message-id: <1279279498-12062-1-git-send-email-kraxel@redhat.com>
Patchwork-id: 10762
O-Subject: [RHEL-6 kvm PATCH] spice: don't force fullscreen redraw on display
resize.
Bugzilla: 612074
RH-Acked-by: Amit Shah <amit.shah@redhat.com>
RH-Acked-by: Alexander Larsson <alexl@redhat.com>
RH-Acked-by: Alex Williamson <alex.williamson@redhat.com>
It is pretty pointless. Screen content is undefined after mode
switching, so there is nothing to draw anyway. Once the guest filled
something in we get notified about it.
The only useful effect the code has is to make sure the dirty rectangle
isn't larger than the new screen size. Just zeroing the dirty rectangle
will do that too.
This especially avoids display updates for the initial 640x480 display
which is never ever used because the first thing the vga code does is
resizing the display to 720x400 (80x25 text mode). This in turn avoids
the segmentation fault reported in bug 612074.
No, this isn't the real fix. The real bug is a race condition, and this
patch only makes it much harder to actually hit it. Fixing it for real
isn't trivial though, so I'll put that at my upstream/rhel-6.1 todo list
and go with this non-invasive patch for rhel-6.0.
bugzilla: #612074
upstream: n/a
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
---
spice-display.c | 5 +----
1 files changed, 1 insertions(+), 4 deletions(-)
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
---
spice-display.c | 5 +----
1 files changed, 1 insertions(+), 4 deletions(-)
diff --git a/spice-display.c b/spice-display.c
index bfbde4a..17913ee 100644
--- a/spice-display.c
+++ b/spice-display.c
@@ -160,10 +160,7 @@ static void spice_display_resize(struct DisplayState *ds)
}
pthread_mutex_lock(&sdpy.lock);
- sdpy.dirty.left = 0;
- sdpy.dirty.right = ds_get_width(ds);
- sdpy.dirty.top = 0;
- sdpy.dirty.bottom = ds_get_height(ds);
+ memset(&sdpy.dirty, 0, sizeof(sdpy.dirty));
pthread_mutex_unlock(&sdpy.lock);
if (!sdpy.is_attached && sdpy.worker) {
--
1.7.0.3