handle_dev_update does area_update, i.e. rendering to a surface (usually
the primary, surface 0) on request of the driver. Since we only use a single
canvas for each surface on the device memory (the rest are in host memory),
this patch may not really be required.
Testing with and without this patch shows that: (tested with two clients, winxp
vm)
* without this patch update area is correct only for one client. The other
clearly shows wrong rendering artifacts (window drag shows traces of
intermediate points with wrong background).
* with this patch there are still wrong rendering (image on client correct,
image via printscreen wrong) but less so.
TODO: more testing. Clearly something is very wrong.
---
server/red_worker.c | 15 ++++++++-------
1 files changed, 8 insertions(+), 7 deletions(-)
diff --git a/server/red_worker.c b/server/red_worker.c
index 43373af..c9d6211 100644
--- a/server/red_worker.c
+++ b/server/red_worker.c
@@ -10271,6 +10271,8 @@ static inline void handle_dev_update(RedWorker *worker)
uint32_t num_dirty_rects;
uint32_t surface_id;
uint32_t clear_dirty_region;
+ RingItem *link;
+ RedRender *render;
receive_data(worker->channel, &surface_id, sizeof(uint32_t));
receive_data(worker->channel, &rect, sizeof(SpiceRect *));
@@ -10285,18 +10287,17 @@ static inline void handle_dev_update(RedWorker
*worker)
validate_surface(&worker->render, surface_id);
red_update_area(worker, rect, surface_id);
- surface = &worker->render.surfaces[surface_id];
- region_ret_rects(&surface->draw_dirty_region, dirty_rects,
num_dirty_rects);
-
- if (clear_dirty_region) {
- region_clear(&surface->draw_dirty_region);
+ RENDER_FOREACH(link, render, worker) {
+ surface = &render->surfaces[surface_id];
+ region_ret_rects(&surface->draw_dirty_region, dirty_rects,
num_dirty_rects);
+ if (clear_dirty_region) {
+ region_clear(&surface->draw_dirty_region);
+ }
}
-
message = RED_WORKER_MESSAGE_READY;
write_message(worker->channel, &message);
}
-
static inline void handle_dev_add_memslot(RedWorker *worker)
{
RedWorkerMessage message;
--
1.7.5.1
_______________________________________________
Spice-devel mailing list
Spice-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/spice-devel
