Right now cogl does all rendering and modesetting using
/dev/dri/card0. If rendering were moved to /dev/dri/renderD128
then it would be in a better position to support offloading
rendering operations to a second gpu.

Also, some versions of the vmware driver don't support using
card0 when running on an inactive VT.  The result is crashes
low in the rendering stack.

This commit leaves card0 for modesetting operations, but defers
rendering operations to a render node.  The output is synchronized
to card0 using dma-buf. render nodes can be used even when VT
switched away, so this will fix crashes on vmware.

https://bugzilla.gnome.org/show_bug.cgi?id=753531

At the moment the drm fd is stored in the renderer structure twice:
once for reading and once for closing. This is a little messy, and
will only get worse when we start throwing render nodes into the mix,
too.

This commit abstracts the device handling out to another structure.
Rather than having two members for each fd, this commit employees a
boolean to decide whether or not the fd needs to get explicitly closed.

https://bugzilla.gnome.org/show_bug.cgi?id=753531

The variable device_name is currently used to hold
the name of the drm device to open ("/dev/dri/card0").
We're going to be opening other drm devices in the future
(render nodes), so device_name will become ambiguous.

This commit renames it to card_device_name

https://bugzilla.gnome.org/show_bug.cgi?id=753531

If cogl fails to open the drm device, initialize gbm, or open the
egl display, then it closes the drm fd, uninitializes gbm, closes the
display and then calls _cogl_winsys_renderer_disconnect which does
most of those things again, on the, now deinitialized, members.

This commit removes the explicit failure handling in renderer_connect and
defers cleanup to disconnect.

https://bugzilla.gnome.org/show_bug.cgi?id=753531

gbm confusingly has two different format types, and cogl
is using the wrong one in some of its calls to gbm_surface_create

This commit fixes the calls that are wrong.

https://bugzilla.gnome.org/show_bug.cgi?id=753531

Right now cogl tries to accomodate older gbm versions that
have the function gbm_bo_get_pitch() instead of the more
recently named gbm_bo_get_stride().

This adds an ugly #ifdef in the code. Furthermore, we are
going to rely on newer gbm for dma-buf support anyway.

This commit drops the #ifdef.

https://bugzilla.gnome.org/show_bug.cgi?id=753531

Mesa packagers disabled it explicitly on Fedora since 2013, and it makes
sense not to have people target an obsolete, barely supported API.

glGetIntegerv (GL_DEPTH_BITS, ...) and friends are deprecated in GL3; we
have to use glGetFramebufferAttachmentParameteriv() instead, like we do
for offscreen framebuffers.

Based on a patch by: Adel Gadllah <adel.gadllah@gmail.com>
Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>

https://bugzilla.gnome.org/show_bug.cgi?id=753295

Marek Chalupa authored Jul 15, 2015 and Emmanuele Bassi committed Jul 28, 2015

Otherwise we just append NULL output which is wrong and can
break things

Signed-off-by: Emmanuele Bassi <ebassi@gnome.org>

https://bugzilla.gnome.org/show_bug.cgi?id=752406

Both create_gl3_context and glXCreateNewContext can fail with an X error.

https://bugzilla.gnome.org/show_bug.cgi?id=742678

Bisection would be useful.

We don't need the hack any more; CoglRenderer has the appropriate API
for the Xlib-based winsys backends.

Instead of using the winsys vtable.

Instead of peeking into the winsys vtable.

Instead of freeing it after creating the relative GL contexts, store it
in the CoglXlibRenderer data.

We want to be able to retrieve the XVisualInfo used when creating the
GL context under GLX and EGL-X11, so that we can use the visual before
we have an onscreen frame buffer.

Fixes Cogl compilation on build.gnome.org.

POSIX does not define 'source' shell command, so we should use a dot instead.

https://bugzilla.gnome.org/show_bug.cgi?id=750723

It seems we really suck at that, or clang is really strict.

https://bugzilla.gnome.org/show_bug.cgi?id=750725

We care especially about the violations of C89, for portability on some
compilers.

Stereo mode changed two fields in the winsys vtable, so we need to
update the signature of the implementation in the EGL-X11 winsys.

The conformance test suite exercises deprecated API, so we don't need to
get compiler warnings for it.

Examples should be as idiomatic as possible.

When compiling against a newer version of Pango we can avoid the
deprecation warning.

Initialize variables; GCC does not always catch all cases where the
variables are used after being initialized, especially when it comes to
out parameters.

Avoids a compiler warning for a missing declaration.

The explicit type system initialization has been deprecated since
GLib 2.36. It's done automatically, now.

Simple enumeration checks.

Looks like a thinko introduced by commit
79719347 .

Found by a coverity scan.

https://bugzilla.gnome.org/show_bug.cgi?id=749111

Some drivers ( like mgag200 ) don't yet support drmModePageFlip.

This commit forgoes waiting for vblank and flips right away
in those cases. That prevents the hardware from freezing up the screen,
but does mean there will be some visible tearing.

https://bugzilla.gnome.org/show_bug.cgi?id=746042