SpacemiT K1's BSP has a version of FFmpeg that contains both
v4l2m2m and their own *_stcodec decoders. Not only is their
V4L2 driver for their mvx decoder seemingly totally broken
(output buffers are all zeros), but it doesn't have any of
the DRM_PRIME support work that is maintained out of tree.
The result is we use a broken copyback decoder instead of
a (mostly) working zero-copy decoder because we didn't
include any logic to rank non-hwaccel decoders by pixfmt
like we have with device type for hwaccel decoders.
There are at least 2 out-of-tree OMX decoder implementations
that both lack AV_CODEC_CAP_HARDWARE, so they trick us into
thinking that neither of them is hardware accelerated.
Fixes false decoder warnings when linked against the patched
FFmpeg builds shipping on VisionFive 2 and LicheePi 4A SBCs.
This binds us against the exact compile-time version of Qt with no
forwards or backwards compatibility, so it must not be used when we
could load Qt libraries from a location outside of our app bundle.
This technique seems to actually make performance worse on some AMD GPUs (RX 480)
and causes rendering errors on others (HD 5570). These might be AMD-specific bugs
but let's not risk a behavior change for AMD/NVIDIA where nobody was having perf
issues with the old copy method anyway.
Performing this each time DrmRenderer is initialized leads to long
delays when choosing a decoder on embedded platforms, particularly
those like TH1520/JH7110 which lack accelerated GL drivers.
Replace the generic starfive hack with proper logic to examine the
supported enum values to select a colorspace. This fixes incorrect
colors with vs-drm on the TH1520.
Planes seem to be listed in descending order of capabilities
with later planes sometimes lacking scaling capabilities, which
is the case with vs-drm on the TH1520.
This kills performance on some Intel iGPUs (particularly Atom chips like N100),
so let's remove the copy and solve this issue a different way instead.
This reverts commit a6fccf93d149a8b67eeac0b0fe109a142f0937d8.
These are only really useful for Wayland scenarios, but:
- Wayland is explicitly disabled for AppImage due to EGL issues
- VDPAU now works under XWayland with 545 and later drivers
- Moonlight now has a Vulkan Video backend which works with 535 and later drivers
Fixes#1314
libva 2.20 is good enough at detecting driver names (even under XWayland)
that we don't need to use the fallback name list anymore. This saves time
probing drivers, avoids excessive log output from failed probes, and avoids
tickling bugs in VAAPI drivers that are installed but unused.
