Try to specify render target bindings flags first. If that fails, try
again with just sampler view binding. Note that we try to create the
texture resource with render target binding flags later when we allocate
the texture. Then, in FBO validation, we check if we can actually render
to the textures. If that fails, we generate GL_FRAMEBUFFER_UNSUPPORTED_EXT.
Changes suggested by Jose.
__glXInitializeVisualConfigFromTags doesn't skip the payload of
unrecognized tags. Instead, it treats the value as if it were the
next tag, which can happen if the server's GLX extension is not
Mesa's. For example, this falls down when NVIDIA sends a
GLX_FLOAT_COMPONENTS_NV = 0 pair, causing
__glXInitializeVisualConfigFromTags to bail out early.
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
After continuously running regression tests on Ubuntu for 2 days, shmget
mysteriously starts to fail. Even when the X server is reset.
This allow rendering to proceed, albeit using a slower presentation path.
If there is no depth buffer bound to current context don't
enable depth test. GL states that if depth test is enabled
without depth buffer it's as if depth buffer always pass.
Signed-off-by: Jerome Glisse <jglisse@redhat.com>
We would run into trouble due to the hardware using inclusive numbers
and the subtraction to handle that producing negative (meaning large
positive) coordinates.
Bug #27643.
a bit more involved than indirect addressing over consts, but still
fairly reasonable. we allocate an array instead of individual alloca's,
and we do it only if the shader does indirect addressing.
implement indirect addressing (ARL and ARR instructions) when used
with CONST's. indirect addressing over other vars (temps, inputs, outputs)
is not supported yet.
Put PIPE_BIND_ and PIPE_TEXTURE_GEOM_ prefixes on token names so
that they can be found with grep. This needs to be done in more places.
Corrected/improved a lot of information and grammer.
I don't know how to properly format everything - someone else can take
care of that.
Otherwise slightly difference order causes assertion failures.
Also remove mentions of PIPE_BIND_SCANOUT/PIPE_BIND_SHARED. They are not
propoer bind flags and will likely be deprecated. If surfaces should
be passed to the winsys then they should have the DISPLAY_TARGET flag
set, which is a proper bind flag.
