Getting started with OpenCL using mesa/rusticl

Mesa, the open source low level graphics stack, has featured support for Open Compute Language (OpenCL) for some time via a front end called Clover. The problem was that the GPUs that it supported were limited, it didn’t have Image support, and wasn’t really under active development. Around a year ago Karol Herbst filed a merge request adding rusticl to Mesa 22.3 release, and soon after that I enabled it optionally in Fedora.. What is rusticl? It’s a OpenCL API implementation written in rust for Mesa, it will eventually replace clover. One advantage it has is image support from the outset and also works on much wider range of OpenCL capable GPUs, including some ARM GPUs, and the support is actively improving all the time.

In the year since it landed in a stable mesa release rusticl keeps evolving, faster, more HW support, more features, less crashes. I’ve tinkered with it as I’ve had spare time on the weekends and evenings as well as trying to work out details of how you’d use it to run higher level stacks.

As of writing it works with following gallium drivers: iris (Intel), nouveau (Nvidia), radeonsi/r600 (AMD ATI), and panfrost (Arm MALI). There’s other drivers that are various stages of development but are not yet upstream.

So let’s get the basics up and running on Fedora if you have supported hardware. First install the core software stack:

$ sudo dnf install -y mesa-libOpenCL mesa-dri-drivers spirv-llvm-translator spirv-tools-libs clinfo clpeak

Next we run clinfo and clpeak with required parameters, from the driver names above (in this case an Intel laptop), to enable rusticl. The two commands output a lot of information so I’m not going to post them here but the output shows OpenCL running and some details about what’s supported:

$ RUSTICL_ENABLE=iris clinfo --list
WARNING: OpenCL support via iris+clover is incomplete.
For a complete and conformant OpenCL implementation, use
https://github.com/intel/compute-runtime instead
Platform #0: rusticl
 `-- Device #0: Mesa Intel(R) Xe Graphics (TGL GT2)
Platform #1: Portable Computing Language
 `-- Device #0: cpu-11th Gen Intel(R) Core(TM) i7-1185G7 @ 3.00GHz
Platform #2: Clover
$ RUSTICL_ENABLE=iris clinfo
output a lot of details
$ RUSTICL_ENABLE=iris clpeak
output a lot of details

This gets up the basic pieces up and running for OpenCL, there’s of course more to do and not all use cases are complete. Eventually we won’t need to have the environment variables to enable rusticl. The details of drivers and other features enabled by environment variables are documented here. I plan to do some more posts as follow ups to build on this basis.

HW video offload on Fedora Arm

There’s been the starting pieces of hardware video offload with the stateless engine support for some time and it now supports at least H264/HEVC/VP8/VP9/mepg2 decode offload depending on the hardware capabilities. The problem has been support for software/userspace has taken longer then the initial drivers but now that’s catching up now with gstreamer support landing in 2020 and with apps like clapper now using it. I’ve been meaning to play with this and work out how to make it work in Fedora as it’s useful for devices based on the AllWinner/Rockchip/NXP i.MX8 devices like the Pine64 laptops/phones plus a bunch of other devices, even NVIDIA Jetson devices should work before long.

You’ll need to ensure you have gstreamer1-plugins-bad-free installed and the video application I was testing with is clapper:

$ sudo dnf install -y gstreamer1-plugins-bad-free clapper

Seeing what hardware offload is supported:

$ gst-inspect-1.0 v4l2codecs
Plugin Details:
  Name                     v4l2codecs
  Description              V4L2 CODEC Accelerators plugin
  Filename                 /usr/lib64/gstreamer-1.0/libgstv4l2codecs.so
  Version                  1.20.0
  License                  LGPL
  Source module            gst-plugins-bad
  Source release date      2022-02-03
  Binary package           Fedora GStreamer-plugins-bad package
  Origin URL               http://download.fedoraproject.org

  v4l2slh264dec: V4L2 Stateless H.264 Video Decoder
  v4l2slmpeg2dec: V4L2 Stateless Mpeg2 Video Decoder
  v4l2slvp8alphadecodebin: VP8 Alpha Decoder
  v4l2slvp8dec: V4L2 Stateless VP8 Video Decoder
  v4l2slvp9alphadecodebin: VP9 Alpha Decoder
  v4l2slvp9dec: V4L2 Stateless VP9 Video Decoder

  6 features:
  +-- 6 elements

Finally in Clapper you need to enable playbin3 option, I also enabled Pipewire audio support:
Clapper Preferences

We will also be enabling decode support in Chromium and Chromium freeworld before long, there’s a little more work to do here, but as usual once it lands it’ll all just start to work in Chromium too!

Fedora on the Pinebook Pro

First thing to note here is that this is not limited to the Pinebook Pro, I’m just using it as the example for 64 bit Rockchip devices with SPI flash on Fedora. This post is focused on devices with SPI but I’ll do a separate follow-up post for other devices including details for writing to eMMC over USB.

The story of Fedora on the Pinebook Pro, and other Rockchip devices, has been a sordid story of a lack of time, bugs, rabbit holes, more bugs and various other things. Not at all sordid at all really, mostly just a lack of time on my behalf, and nobody else stepping up to assist in a way to benefit all Fedora users, mostly they do one time hacks to sort themselves. Overall the support in Fedora for Rockchip devices has been quite solid for a number of releases. The problem has been with the early boot firmware, notable because without SPI flash it wants to splat itself across the first 8Mb of the disk, and if there was SPI flash it generally wasn’t overly stable/straight forward.

Anyway we’re now in a place where devices with SPI flash should mostly work just fine, those devices without it will work with a little manual intervention, and while the support isn’t complete, and will need more polish, they’re all details we can polish with little interruption to users by standard package updates. By default users will have accelerated graphics and from my testing on GNOME 40 it’s by all accounts a pretty decent experience!

Setting up the firmware

First step is to get the firmware written to SPI flash. This is a two step process, the first is to write out a micro SD card from another device, the second is to boot that mSD card on the Pinebook Pro, or another device like the Rockpro64, and write the firmware to the SPI flash.

There’s some nuances to this process, and the way the early boot firmware works, if another version of U-Boot takes precedence that is likely OK as it should still be able to work, the fall back is to use the internal switch to turn off the eMMC temporarily. I also have no idea if the Pine64 shipped U-Boot has any display output, the Fedora build does, if not you’ll need to use the option to disable eMMC or use a serial console cable. Anyway on to the steps:

Set up the mSD card
Use a mSD card that has no data you wish to keep, this process will wipe it out. You want at least U-Boot build 2021.04-3.fc34, you can adjust the umount to be more specific, and you need to substitute XXX for the media, otherwise it’s a relatively quick and straightforward process:

sudo dnf install --enablerepo=updates-testing -y arm-image-installer uboot-images-armv8
sudo umount /run/media//*
sudo spi-flashing-disk --target=pinebook-pro-rk3399 --media=/dev/XXX

Write the firmware to flash
Now remove the mSD card from your host and put it into the Pinebook Pro. Press the power button, from experience you likely need to press and momentarily hold and in a second or two the display will light up with text output. Interrupt the boot by pressing space. Next up we write out the flash:

Hit any key to stop autoboot:  0 
=> ls mmc 1:1
   167509   idbloader.img
   335872   idbloader.spi
   975872   u-boot.itb
  9331712   u-boot-rockchip.bin

4 file(s), 0 dir(s)

=> sf probe
SF: Detected gd25q128 with page size 256 Bytes, erase size 4 KiB, total 16 MiB

=> load mmc 1:1 ${fdt_addr_r} idbloader.spi
335872 bytes read in 39 ms (8.2 MiB/s)

=> sf update ${fdt_addr_r} 0 ${filesize}
device 0 offset 0x0, size 0x52000
61440 bytes written, 274432 bytes skipped in 0.803s, speed 427777 B/s

=> load mmc 1:1 ${fdt_addr_r} u-boot.itb
975872 bytes read in 107 ms (8.7 MiB/s)

=> sf update ${fdt_addr_r} 60000 ${filesize}
device 0 offset 0x60000, size 0xee400
914432 bytes written, 61440 bytes skipped in 9.415s, speed 106127 B/s

Once the last command above has completed eject the mSD card and type reset at the => prompt and the device should reboot and you should see output similar to before but running from the SPI flash!

If you had to turn off the eMMC you can now turn it back on.

Installing Fedora

The nice thing with the firmware on SPI flash it should now work mostly like any other laptop and you can use either the pre canned desktop images (Workstation, KDE, XFCE, Sugar), the Workstation LiveCD iso or the standard everything network installer.

To run the arm Workstation image off a micro SD card or USB stick you can do the following:

arm-image-installer --media=/dev/XXX --resizefs --target=none --image=Fedora-Workstation-34-1.2.aarch64.raw.xz

Note ATM you’ll need to use the USB port on the right hand side, I need to investigate the USB/USB-C port on the left as it appears not to currently work in firmware, but works fine once Fedora is running.

Next steps and improvements

The two biggest issues remaining for the Pinebook Pro is enabling PCIe support (supported from June 2021) and the lack of the brcmfmac firmware, both WiFi and bluetooth, being upstream. For the later issue if there’s anyone from Synaptics that can assist in resolving that problem please reach out to me! A interim WiFi firmware to use is here.

Some things at the Fedora level I’ve not really tested and will do so more, and likely polish with OS updates, in the coming weeks include sound, USB-C port (charging and display output). On the firmware level there’s still some more improvements to be done, tweaks to improve the USB support, turning on the power LED as early as possible to give an indicator, improvements to the EFI framebuffer to ensure consistent early boot output, support for UEFI BGRT to enable smooth boot etc.

For support please email the Fedora Arm mailing list or reach out on IRC via #fedora-arm on Libera.Chat.