- Jul 11, 2015
-
-
Lina Versace authored
All APIs have been updated. This eliminates the diff between the work-in-progress header and the 0.130 header.
-
Lina Versace authored
-
Lina Versace authored
-
Lina Versace authored
Replace the anv_device parameter with anv_physical_device, because this needs querying before vkCreateDevice.
-
Lina Versace authored
During anv_physical_device_init(), we opend the DRM device to do some queries, then promptly closed it. Now we keep it open for the lifetime of the anv_physical_device so that we can query it some more during vkGetPhysicalDevice*Properties() [which will happen in follow-up commits].
-
Lina Versace authored
Because in a follow-up patch I need to do some non-trival teardown on anv_physical_device. Currently, however, anv_physical_device_finish() is currently a no-op that's just called in the right place. Also, rename function fill_physical_device -> anv_physical_device_init for symmetry.
-
- Jul 10, 2015
-
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Lina Versace authored
Don't assume that $(top_srcdir)/.git is a directory. It may be a gitlink file [1] if $(top_srcdir) is a submodule checkout or a linked worktree [2]. [1] A "gitlink" is a text file that specifies the real location of the gitdir. [2] Linked worktrees are a new feature in Git 2.5. Cc: "10.6, 10.5" <mesa-stable@lists.freedesktop.org> Reviewed-by: Iago Toral Quiroga <itoral@igalia.com> (cherry picked from commit 75784243)
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
We will need these as soon as we turn on type saftey. We might as well define and start using them now rather than later.
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
- Jul 09, 2015
-
-
Faith Ekstrand authored
-
Faith Ekstrand authored
We do nothing with it at the moment, so this is a no-op.
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Faith Ekstrand authored
-
Lina Versace authored
The Vulkan spec says that pPhysicalDeviceCount is an out parameter if pPhysicalDevices is NULL; otherwise it's an inout parameter. Mesa incorrectly treated it unconditionally as an inout parameter, which could have lead to reading unitialized data.
-
Lina Versace authored
Don't enumerate devices in vkCreateInstance(). That's where global, device-independent initialization should happen. Move device enumeration to the more logical location, vkEnumeratePhysicalDevices().
-
Lina Versace authored
Function fill_physical_device() has a 'path' parameter, and struct anv_physical_device has a 'path' member. Sometimes these are used; sometimes hardcoded "/dev/dri/renderD128" is used instead. Be consistent. Hardcode "/dev/dri/renderD128" in exactly one location, during initialization of the physical device.
-
Connor Abbott authored
Prevents problems when initializing the sanity_param_count.
-
Connor Abbott authored
-
Connor Abbott authored
Fixes valgrind errors with func.depthstencil.basic.
-
Connor Abbott authored
Before values are pushed or annotated with a name, decoration, etc., they need to have an invalid type, NULL name, NULL decoration, etc. ralloc zero's everything by accident, so this wasn't an issue in practice, but we should be explicitly zero'ing it.
-
Connor Abbott authored
This fixes Valgrind errors and gets all the tests to pass with --use-spir-v.
-
Connor Abbott authored
This doesn't handle inputs and outputs larger than a vec4, but we plan to add a varyiing splitting/packing pass to handle those anyways.
-
Faith Ekstrand authored
We now have the exact same structs which require pType.
-