The `rna_Scene_ray_cast()` function tried to find the current depsgraph. To
this end, it required the scene, the view layer, and bmain. Scene has a cache
of per-view-layer depsgraphs, to speed up switching between view layers. This
cache does not contain render depsgraphs, and evaluated view layers also don't
have a depsgraph here.
When a suitable depsgraph cannot be found, a new depsgraph is created. However,
this depsgraph is not evaluated, and has an unexpanded scene pointer with a
`NULL` `view_layer`. Using this then crashes Blender. Also, there was no way
for the code to get the render depsgraph.
The solution is to pass the depsgraph to the `ray_cast()` function, instead of
the view layer. This avoids the depsgraph lookup, and also works correctly when
rendering.
Some add-ons also need updating, which I'll do in the `addons`
repository soon.
Reviewed By: Sergey
Differential Revision: https://developer.blender.org/D8475
This uses the vertices per grid instead of quads to set the limit of
grids per PBVH Node. This should create more leaf nodes in lower
subdivisions levels where the duplicates count is high, producing more
uniform performance across different levels.
Reviewed By: sergey
Differential Revision: https://developer.blender.org/D8454
There were two issues. First, I made a mistake when I switched from unsigned
to signed integers. Second, two classes with the same name were defined in
separate files. Those classes are in an anonymus namespace now, so that they
don't leak into other files.
ad4928a171 disabled alignment for too many cases. Still try to avoid
aligning many items, to avoid thousands of redundant alignment
calculations. But now we're much more picky adding an sub-row with
alignment.
On windows, spacebar would be passed as UTF-8 text input, despite the
control key being pressed. On macOS, there already was an explicit
exception for this (command key in this case), on Linux XInput already
handled this case for us.
Note that Alt should still allow text input, for special character
sequences.
Issue also happened in the Text Editor if a text data-block was set.
Optimization was disabled in this function to work around a bug in MSVC, use
a different solution that does not come with such a big performance regression.
In Blender 2.90 EEVEE materials were refactored that introduced crashes on Intel
GPUs on Windows. The crash happened in the `local_context_workaround` that temporary
stored compiled materials in a binary form to reload it in the main GL context.
It has been tested that the workaround isn't needed anymore for HD6xx GPUs, but it
is still needed for HD4000.
After several unsuccesfull fixes we came to the conclusion that we could not support
the local context workaround and needed to come with a different workaround. The idea
of this patch is that in these cases there is only a single context that is used for
rendering. Threads that uses these contextes are guarded by a mutex and will block.
Impact on User Level:
* Due to main mutex lock the UI freezes when rendering or baking or feel less snappy
Reviewed By: Clément Foucault, Brecht van Lommel
Differential Revision: https://developer.blender.org/D8410
As far as I can see, this problem was introduced with the gpu_free_unused_buffers() changes in rB97b597c.
The code checks the returned pointer to stop looping, but the last iteration will return the last GPUTexture and set the queue to NULL, meaning that the next pop() will crash.
Differential Revision: https://developer.blender.org/D8468
Some implementation have different maximum texture size.
This patch avoid crash when texture allocation fails when:
- trying to bake a lightcache too big for the OpenGL imeplementaion.
- loading a cache from file that is too big for the OpenGL imeplementation.
