[WIP] Headless export-dmabuf #2268
Conversation
There was a problem hiding this comment.
I haven't looked into this in detail, but here are some general comments.
Something quite similar has been worked on for the DRM backend. I think ti would make sense to use the renderer v6's wlr_swapchain here too. See #2240
GBM buffer allocation. Would it be better to get the render node from DRM if the renderer belongs to DRM?
It may make sense to add wlr_renderer_get_fd. See #1376 (comment)
| @@ -164,6 +164,14 @@ bool wlr_renderer_blit_dmabuf(struct wlr_renderer *r, | |||
| return r->impl->blit_dmabuf(r, dst, src); | |||
| } | |||
|
|
|||
| GLuint wlr_renderer_renderbuffer_from_image(struct wlr_renderer *r, | |||
There was a problem hiding this comment.
This leaks EGL/GL implementation details into the generic renderer interface.
render/egl.c
Outdated
| @@ -419,6 +419,9 @@ void wlr_egl_save_context(struct wlr_egl_context *context) { | |||
| } | |||
|
|
|||
| bool wlr_egl_restore_context(struct wlr_egl_context *context) { | |||
| if (!context->display) | |||
There was a problem hiding this comment.
This doesn't correctly restore EGL_NO_DISPLAY.
#1376 now has |
any1
force-pushed
the
headless-export-dmabuf
branch
from
5454eaf
to
0db4b75
Compare
any1
force-pushed
the
headless-export-dmabuf
branch
from
0db4b75
to
55abc24
Compare
This is very much work in progress, but it does work quite well.
Things I'd like some feedback on:
glEGLImageTargetRenderbufferStorageOESinto the backend is rather ugly. Suggestions?