This also marks the rpir_view for destroy in rpi_renderer_finish_frame()
---
src/rpi-renderer.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/src/rpi-renderer.c b/src/rpi-renderer.c
index 812e6e7..7f79410 100644
--- a/src/rpi-renderer.c
+++ b/src/rpi-renderer.c
@@ -1501,6 +1501,8 @@
The compositor will check if the client destroyed the wl_buffer
while it was in use in a display update, and delete the resource
itself once the update has finished.
---
src/rpi-bcm-stubs.h | 5
src/rpi-renderer.c | 74 +
2 files changed,
Othrewise a crash may happen because of an unitialized value of the
create_view field.
---
src/pixman-renderer.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/pixman-renderer.c b/src/pixman-renderer.c
index 85fcd4c..d6f638d 100644
--- a/src/pixman-renderer.c
+++
Also make sure backends destroy the renderers before shutting down the
compositor to avoid a double call to weston_binding_destroy().
This is a step towards making renderers switchable during runtime.
---
src/compositor-drm.c |4 ++--
src/compositor-fbdev.c |3 ++-
src/compositor-rpi.c
Remove create_surface() and destroy_surface() from the renderer
interface and change the renderers to create surface state on demand
and destroy it using the weston_surface's destroy signal.
Also make sure the surfaces' renderer state is reset to NULL on
destruction.
This is a step towards
When using the renderer switch feature, the normal start up path and
the loading of gl-renderer and dependency libraries will compete for
IO bandwidth. In systems where that is limited, this may impact the
time it takes to get an image to the screen negatively.
This patch adds an option to delay
Previously the renderers destroy function assumed they are only called
when the compositor is shutting down and that the compositor had
already destroyed all the surfaces. However, if a runtime renderer
switch would be done, the surface state would be leaked.
This patch adds a destroy_signal to
Option --fast-start will cause the backend to start using the pixman
renderer and at the same time spawn a thread to do the loading of the
gl renderer and creation of the gbm device (which loads a dri driver).
Once that thread signal it is done, a runtime switch to the gl renderer
happens.
---
When a renderer switch happen, it is possible that when the surface
state is created, a buffer for the given surface is already available.
In that case, run the attach routine so that the pixel contents are
properly set before a new attach request is made for that surface.
Also, make sure the DRM
Ok, here is version 4 of the patch taking into account the discussion
with Jason Ekstrand. The assumption is that if we have enough buffer
slots then we should always get a release event immediately after one
of the attaches. That means we can just rely on sending a sync request
after the commit
Ander,
When I was working on views, I really wanted to get rid of the
create_surface functions too... Thanks. That said, you should probably
also get rid of create_view (It's only used by rpi).
--Jason Ekstrand
On Fri, Oct 25, 2013 at 8:26 AM, Ander Conselvan de Oliveira
On Wed, Oct 23, 2013 at 02:56:04PM +1000, Peter Hutterer wrote:
The scanner is not very forgiving if the protocol doesn't match it's
expectations and crashes without much of a notice. Thus, validate the protocol
against a DTD.
Move the protocol subdir forward so we validate first before
On Wed, Oct 23, 2013 at 02:10:26PM +0200, Quentin Glidic wrote:
From: Quentin Glidic sardemff7+...@sardemff7.net
Signed-off-by: Quentin Glidic sardemff7+...@sardemff7.net
---
Here is a new version of my GLib main loop integration for Wayland.
Now everything is in libwayland-client (I
On Thu, Oct 24, 2013 at 03:38:28PM +0200, Tomeu Vizoso wrote:
Except for the first one, all are intended to fix the RPi backend after
the landing of the views work.
Most EGL clients won't work yet because of wl_buffer.release events being
always queued, though.
Thanks for fixing it up, all
On Fri, Oct 25, 2013 at 10:34:37AM +0200, Tomeu Vizoso wrote:
This also marks the rpir_view for destroy in rpi_renderer_finish_frame()
---
src/rpi-renderer.c | 2 ++
1 file changed, 2 insertions(+)
These two applied, thanks.
Kristian
diff --git a/src/rpi-renderer.c b/src/rpi-renderer.c
When we would build weston with libxkb 0.2.0 it would succesfully
compile, but the keybaord and desktop-shell wouldn't load, and
Weston wouldn't run.
Signed-off-by: Brian Lovin brian.j.lo...@intel.com
---
configure.ac | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git
On Fri, Oct 25, 2013 at 04:26:33PM +0300, Ander Conselvan de Oliveira wrote:
Remove create_surface() and destroy_surface() from the renderer
interface and change the renderers to create surface state on demand
and destroy it using the weston_surface's destroy signal.
Also make sure the
On 25/10/2013 20:01, Kristian Høgsberg wrote:
On Wed, Oct 23, 2013 at 02:10:26PM +0200, Quentin Glidic wrote:
From: Quentin Glidic sardemff7+...@sardemff7.net
Signed-off-by: Quentin Glidic sardemff7+...@sardemff7.net
---
Here is a new version of my GLib main loop integration for Wayland.
Now
On Fri, Oct 25, 2013 at 10:27:08AM -0500, Jason Ekstrand wrote:
Ander,
When I was working on views, I really wanted to get rid of the
create_surface functions too... Thanks. That said, you should probably
also get rid of create_view (It's only used by rpi).
--Jason Ekstrand
Yeah, I agree,
Hi,
this is an update to the patch set posted earlier.
It fixes unconditional build of DirectFB clients.
From 85397933453e3d080b740ac74c2180ad87fdc012 Mon Sep 17 00:00:00 2001
From: Denis Oliver Kropp d...@directfb.org
Date: Fri, 25 Oct 2013 23:07:16 +0200
Subject: [PATCH] clients: fix
Signed-off-by: Jonas Ådahl jad...@gmail.com
---
tests/.gitignore | 1 +
1 file changed, 1 insertion(+)
diff --git a/tests/.gitignore b/tests/.gitignore
index 28d6576..aba378c 100644
--- a/tests/.gitignore
+++ b/tests/.gitignore
@@ -1,4 +1,5 @@
*.test
+*.trs
*.weston
logs
matrix-test
--
Hi,
This series contains, except for some minor fixes, a patch that fixes
some issues related with ending grabs, for example when unplugging some
input device.
I have only tested with with a pointer device (since I don't have an
external keyboard nor touch device to unplug).
Jonas
Jonas
Signed-off-by: Jonas Ådahl jad...@gmail.com
---
configure.ac | 1 +
1 file changed, 1 insertion(+)
diff --git a/configure.ac b/configure.ac
index ac3878b..51c6a8e 100644
--- a/configure.ac
+++ b/configure.ac
@@ -506,6 +506,7 @@ AC_MSG_RESULT([
Cairo Renderer
Signed-off-by: Jonas Ådahl jad...@gmail.com
---
src/input.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/src/input.c b/src/input.c
index 49788b2..ffb7b35 100644
--- a/src/input.c
+++ b/src/input.c
@@ -1062,8 +1062,6 @@ WL_EXPORT void
notify_pointer_focus(struct weston_seat *seat, struct
Signed-off-by: Jonas Ådahl jad...@gmail.com
---
src/shell.c | 12 +++-
1 file changed, 7 insertions(+), 5 deletions(-)
diff --git a/src/shell.c b/src/shell.c
index a5e653e..badfd0c 100644
--- a/src/shell.c
+++ b/src/shell.c
@@ -1130,12 +1130,14 @@ touch_move_grab_down(struct
A grab can potentially allocate memory and would normally end the grab
itself, freeing the allocated memory in the process. However at in some
situations the compositor may want to abort a grab. The grab owner still
needs to free some memory and abort the grab properly. To do this a new
function
Quentin Glidic writes:
On 25/10/2013 20:01, Kristian Høgsberg wrote:
On Wed, Oct 23, 2013 at 02:10:26PM +0200, Quentin Glidic wrote:
From: Quentin Glidic sardemff7+...@sardemff7.net
Signed-off-by: Quentin Glidic sardemff7+...@sardemff7.net
---
Here is a new version of my GLib main loop
On 25/10/2013 23:26, Tarnyko wrote:
Hi Quentin,
Just to let you know that there are people interested in your work
(including me). I'm currently backporting your patches to 1.2 to test
some command-line apps using GLib with libwayland-client.
Regards,
Tarnyko
In this case you should simply use
On Thu, Oct 17, 2013 at 02:20:31PM +0200, David Herrmann wrote:
If we call wl_event_source_check() on a timerfd, our dispatcher will block
on the timerfd read(). Avoid calling read() if no epoll-event was
reported.
Note: The internal tick-count of a timerfd is never decreased. So once we
My knowledge of mesa and of the wl_display_prepare_read stuff is a bit
lacking. From what I can see, it looks good.
On Fri, Oct 25, 2013 at 9:50 AM, Neil Roberts n...@linux.intel.com wrote:
Ok, here is version 4 of the patch taking into account the discussion
with Jason Ekstrand. The
30 matches
Mail list logo