Commit Graph

127 Commits

Author SHA1 Message Date
Simon Ser bedfec94bb backend/drm: use drmCloseBufferHandle
This has been added in [1] and allows us to close buffer handles
without manually calling drmIoctl.

[1]: https://gitlab.freedesktop.org/mesa/drm/-/merge_requests/192
2021-12-14 14:33:00 +01:00
Simon Ser fbaefd90fc backend/drm: poison buffers which cannot be scanned out
Rather than repeatedly trying to import DMA-BUFs which cannot be
scanned out, mark the failed ones with a special "poison" marker.
Inspired from [1].

[1]: https://gitlab.freedesktop.org/wayland/weston/-/merge_requests/731
2021-11-29 14:32:29 +00:00
Simon Ser e163a7cf48 backend/drm: fail on explicit modifier in drmModeAddFB2
drmModeAddFB2 doesn't support explicit modifiers. Only accept INVALID
which indicates an implicit modifier and LINEAR which may indicate
that GBM_BO_USE_LINEAR has been used.
2021-11-26 16:40:53 +00:00
Simon Ser affe9eda57 Require INVALID for implicit format modifiers
See [1] for the motivation.

[1]: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/75
2021-11-26 16:40:53 +00:00
Simon Ser a04cfca4da Remove support for DMA-BUF flags
They are never used in practice, which makes all of our flag
handling effectively dead code. Also, APIs such as KMS don't
provide a good way to deal with the flags. Let's just fail the
DMA-BUF import when clients provide flags.
2021-11-17 16:12:59 +00:00
Simon Ser 0817c52a21 backend/drm: get rid of BO handle table
The BO handle table exists to avoid double-closing a BO handle,
which aren't reference-counted by the kernel. But if we can
guarantee that there is only ever a single ref for each BO handle,
then we don't need the BO handle table anymore.

This is possible if we create the handle right before the ADDFB2
IOCTL, and close the handle right after. The handles are very
short-lived and we don't need to track their lifetime.

Because of multi-planar FBs, we need to be a bit careful: some
FB planes might share the same handle. But with a small check, it's
easy to avoid double-closing the same handle (which wouldn't be a
big deal anyways).

There's one gotcha though: drmModeSetCursor2 takes a BO handle as
input. Saving the handles until drmModeSetCursor2 time would require
us to track BO handle lifetimes, so we wouldn't be able to get rid
of the BO handle table. As a workaround, use drmModeGetFB to turn the
FB ID back to a BO handle, call drmModeSetCursor2 and then immediately
close the BO handle. The overhead should be minimal since these IOCTLs
are pretty cheap.

Closes: https://github.com/swaywm/wlroots/issues/3164
2021-10-29 11:38:37 +02:00
Simon Ser 61b83441a1 Revert "backend/drm: fail on explicit modifier in drmModeAddFB2"
This reverts commit d6be1d68b7.
2021-10-01 09:26:05 -06:00
Simon Ser d6be1d68b7 backend/drm: fail on explicit modifier in drmModeAddFB2
drmModeAddFB2 doesn't support explicit modifiers. Only accept INVALID
which indicates an implicit modifier and LINEAR which may indicate
that GBM_BO_USE_LINEAR has been used.
2021-10-01 09:21:50 -06:00
Simon Ser 5dfaf5ea9c backend/drm: introduce wlr_drm_bo_handle_table
Using GBM to import DRM dumb buffers tends to not work well. By
using GBM we're calling some driver-specific functions in Mesa.
These functions check whether Mesa can work with the buffer.
Sometimes Mesa has requirements which differ from DRM dumb buffers
and the GBM import will fail (e.g. on amdgpu).

Instead, drop GBM and use drmPrimeFDToHandle directly. But there's
a twist: BO handles are not ref'counted by the kernel and need to
be ref'counted in user-space [1]. libdrm usually performs this
bookkeeping and is used under-the-hood by Mesa.

We can't re-use libdrm for this task without using driver-specific
APIs. So let's just re-implement the ref'counting logic in wlroots.
The wlroots implementation is inspired from amdgpu's in libdrm [2].

Closes: https://github.com/swaywm/wlroots/issues/2916

[1]: https://gitlab.freedesktop.org/mesa/drm/-/merge_requests/110
[2]: 1a4c0ec9ae/amdgpu/handle_table.c
2021-08-25 10:05:37 -04:00
Simon Ser 3ce2ea9e16 Move allocator stuff into new directory
Add render/allocator/ and include/render/allocator/ to hold
everything allocator-related.
2021-08-25 09:57:20 -04:00
Simon Ser 7544b7abf9 backend/drm: use addon for wlr_drm_fb 2021-08-17 11:08:45 +02:00
Simon Ser c55f70c8b7 backend/drm: stop initializing renderer for parent backend
Unless we're dealing with a multi-GPU setup and the backend being
initialized is secondary, we don't need a renderer nor an allocator.
Stop initializing these.
2021-07-28 22:52:35 +02:00
Simon Ser 9b99570869 backend/drm: drop drm_surface_{make,unset}_current 2021-07-28 22:52:35 +02:00
Simon Ser f5900c1f00 backend/drm: remove primary swapchain
We can't nuke it completely, we still need it for multi-GPU.
2021-07-28 22:52:35 +02:00
Simon Zeni 04d4fb536d render/wlr_texture: put wlr_texture_from_buffer into the public API 2021-07-22 22:28:24 +02:00
Simon Ser c1b27cc499 backend/drm: stop using drm_surface_make_current in drm_surface_blit
drm_surface_make_current and drm_surface_unset_current set implicit
state and are an unnecessary mid-layer. Prefer to use directly
wlr_renderer_begin_with_buffer, which automatically unsets the back
buffer on wlr_renderer_end.

I'd like to get rid of drm_surface_make_current once we stop using
it for the primary swapchain.
2021-07-08 14:44:42 -04:00
Simon Ser b2f6db3533 render: drop wlr_ prefix from wlr_renderer_bind_buffer
Make it clear this function is a private wlroots API and will stay
that way.
2021-06-09 10:26:09 +02:00
Simon Ser 2b0a1aeed5 output: take a wlr_buffer in set_cursor
Instead of passing a wlr_texture to the backend, directly pass a
wlr_buffer. Use get_cursor_size and get_cursor_formats to create
a wlr_buffer that can be used as a cursor.

We don't want to pass a wlr_texture because we want to remove as
many rendering bits from the backend as possible.
2021-06-02 11:08:52 +02:00
Simon Ser 01e0f51fad backend/drm: introduce drm_plane_pick_render_format
This is a new helper function to pick a render format suitable for
a plane.

The next commit will use it to initialize the cursor multi-GPU
surface.
2021-06-02 11:08:52 +02:00
Simon Ser e06ea4e84a backend/drm: remove format arg from drm_plane_init_surface
This was always set to ARGB8888.
2021-06-02 11:08:52 +02:00
Simon Ser 9ca743f9fd backend/drm: use wlr_texture_from_buffer 2021-05-17 16:22:43 +02:00
Simon Ser e7f68ba081 backend/drm: fix allocator DRM FD on multi-GPU setups
On multi-GPU setups, there is a primary DRM backend and secondary
DRM backends. wlr_backend_get_drm_fd will always return the parent
DRM FD even on secondary backends, so that users always use the
primary device for rendering.

However, for our internal rendering we want to use the secondary
device. Use allocator_autocreate_with_drm_fd to make sure the
allocator will create buffers on the secondary device.

We do something similar to ensure our internal rendering will
happen on the secondary device with renderer_autocreate_with_drm_fd.

Fixes: cc1b66364c ("backend: use wlr_allocator_autocreate")
2021-04-29 15:58:56 +02:00
Simon Ser 619a975025 render: remove wlr_ prefix from wlr_renderer_autocreate_with_drm_fd
This function is only required because the DRM backend still needs
to perform multi-GPU magic under-the-hood. Remove the wlr_ prefix
to make it clear it's not a candidate for being made public.
2021-04-29 09:46:34 +02:00
Simon Ser 6bf2406dbf backend/drm: reword wlr_renderer failure message
Remove the assumption about EGL.
2021-04-29 09:38:10 +02:00
Simon Zeni cc1b66364c backend: use wlr_allocator_autocreate 2021-04-28 20:55:57 +02:00
Simon Zeni c75aa71816 render/gbm_allocator: make wlr_gbm_allocator_create return a wlr_allocator 2021-04-28 20:55:57 +02:00
Simon Zeni 84dea55b20 render: rename get_dmabuf_render_formats into get_render_formats 2021-04-15 17:10:06 +02:00
Simon Ser 6bfbf35618 backend/drm: reject DMA-BUFs with flags
We cannot scan-out DMA-BUFs with any flag right now.
2021-04-06 15:02:57 +02:00
Simon Ser 1ec97bdf4f backend/drm: improve logs in drm_fb_create
Downgrade errors to DEBUG level, because drm_fb_create is used in
test_buffer, so errors aren't always fatal. Add ERROR logs at call
sites where a failure is fatal, to make it clear something wrong
happened.
2021-04-06 14:57:59 +02:00
Simon Zeni c8b3536b01 backend/drm: use pixel format table in renderer 2021-03-25 10:55:54 +01:00
Simon Ser c740fccc9d Fix buffer blit matrices
There was a missing wlr_matrix_scale call, so we ended up with black
frames.

Closes: https://github.com/swaywm/wlroots/issues/2780
2021-03-16 17:57:51 +01:00
Simon Zeni 9601a2abf0 output: improve transform matrix calculation
Compute only the transform matrix in the output. The projection matrix
will be calculated inside the gles2 renderer when we start rendering.

The goal is to help the pixman rendering process.
2021-03-10 15:33:36 +01:00
Simon Ser f3758d1d0a backend: add error messages in attach_render impls 2021-02-18 22:14:19 +01:00
Simon Ser 73ffab70b0 backend/drm: don't blit in drm_fb_import
Instead blit in drm_plane_lock_surface. This makes drm_fb_import simpler
and better fits its name.
2021-01-24 18:33:56 +01:00
Simon Ser 4af85f4c19 backend/drm: simplify drm_fb_lock_surface
Make it take a plane instead, and rename to drm_plane_lock_surface.
2021-01-24 18:22:03 +01:00
Simon Ser 9e98f497af backend/drm: use local DRM FD for wlr_rend
The new wlr_renderer_autocreate API is great for compositors, however
it causes some issues with DRM multi-GPU support.

A DRM child backend wants the compositor to use the parent GPU, so it
exposes the parent's DRM FD in get_drm_fd. However, in order to be able
to perform multi-GPU buffer copies, the child DRM backend still needs to
create a local renderer.

Use the new private wlr_renderer_autocreate_with_drm_fd function to
avoid creating a renderer for the parent GPU.

Fixes: e128e6c08d ("render: drop egl parameters from wlr_renderer_autocreate")
2021-01-16 22:52:26 +01:00
Simon Ser afdf4dc890 render/gles2: make EGL context current in bind_buffer
Instead of requiring callers to manually make the EGL context current
before binding a buffer and unsetting it after unbinding a buffer, do
it inside wlr_renderer_bind_buffer.

This hides renderer-specific implementation details inside the
wlr_renderer interface. Non-GLES2 renderers may not use EGL.
This removes all EGL dependencies from the backends.

References: https://github.com/swaywm/wlroots/issues/2618
References: https://github.com/swaywm/wlroots/pull/2615#issuecomment-756687006
2021-01-16 09:06:17 +01:00
Simon Zeni 08a4c62aac render: remove egl include from wlr_renderer 2021-01-16 08:57:42 +01:00
Simon Zeni e128e6c08d render: drop egl parameters from wlr_renderer_autocreate 2021-01-16 08:57:42 +01:00
Simon Ser 3f7e0cf5f0 render/egl: remove surface and buffer age args from make_current
These aren't used anymore.
2021-01-12 11:31:04 +01:00
Simon Ser d9bbc416a6 backend/drm: re-use FBs
Instead of importing buffers to GBM and KMS at each frame, cache them
and re-use them while the wlr_buffer is alive.

This is the same as [1] and [2] but for the DRM backend.

[1]: https://github.com/swaywm/wlroots/pull/2538
[2]: https://github.com/swaywm/wlroots/pull/2539
2021-01-10 11:24:35 +01:00
Simon Ser 91cb0fc443 backend/drm: only keep track of local buffer
Stop keeping track of buffers on the parent GPU when multi-GPU is used.

This removes support for export_dmabuf on secondary GPUs, but renderer
v6 will bring this back by managing the swapchains in wlr_output instead
of the backends.
2021-01-10 11:24:35 +01:00
Simon Ser 5bd86b94f9 backend/drm: add wlr_drm_buf.local_wlr_buf 2021-01-10 11:24:35 +01:00
Simon Ser cd64610c66 backend/drm: introduce drm_fb_create 2021-01-10 11:24:35 +01:00
Simon Ser 5b1b43c68c backend/drm: make wlr_drm_plane.{pending,queued,current}_fb pointers
This will be useful once we start re-using wlr_drm_fb.
2021-01-10 11:24:35 +01:00
Simon Zeni 9192c0480a remove unnecessary egl includes 2021-01-07 17:11:22 +01:00
Simon Zeni b899a412e3 backend: remove wlr_egl from all backends 2021-01-07 17:11:22 +01:00
Simon Ser 07d75c99db render: remove EGL config and visual from wlr_renderer_autocreate
This isn't used anymore by any backend.

Some examples still provide an EGL config to wlr_egl_init, so we can't
drop it yet there.
2021-01-06 12:05:51 +01:00
Simon Ser 4b03bdc3ab Remove wlr_create_renderer_func_t
This callback allowed compositors to customize the EGL config used by
the renderer. However with renderer v6 EGL configs aren't used anymore.
Instead, buffers are allocated via GBM and GL FBOs are rendered to. So
customizing the EGL config is a no-op.
2020-12-30 17:09:40 +01:00
Simon Ser bec1e6b149 backend/drm: remove special linear case for cursor plane
We now properly mark the cursor plane's formats as linear-only, and we
now have a version of wlr_drm_format_intersect that handles the case of
linear-only formats and implicit modifiers.

We can remove the special drm_plane_init_surface flag we had for cursor
planes. This also allows us to use a non-linear layout for cursor planes
on drivers that support it.

Tested on amdgpu GFX9.
2020-12-30 11:17:43 +01:00