Create a DRM driver for VGA video cards
a project by tdz
Updated over 2 years ago. 6 hacker ♥️. 4 followers.

Yes, those VGA video cards. The goal of this project is to implement a DRM graphics driver for such devices. While actual hardware is hard to obtain or even run today, qemu emulates VGA output.

VGA has a number of limitations, which make this project interesting.

  • There are only 640x480 pixels (or less) on the screen. That resolution is also a soft lower limit imposed by DRM. It's mostly a problem for desktop environments though.
  • Desktop environments assume 16 million colors, but there are only 16 colors with VGA. VGA's 256 color palette is not available at 640x480. We can choose those 16 colors freely. The interesting part is how to choose them. We have to build a palette for the displayed frame and map each color to one of the palette's 16 entries. This is called dithering, and VGA's limitations are a good opportunity to learn about dithering algorithms.
  • VGA has an interesting memory layout. Most graphics devices use linear framebuffers, which store the pixels byte by byte. VGA uses 4 bitplanes instead. Plane 0 holds all bits 0 of all pixels. Plane 1 holds all bits 1 of all pixels, and so on.

The driver will probably not be useful to many people. But, if finished, it can serve as test environment for low-level hardware. There's some interest in supporting old Amiga and Atari framebuffers in DRM. Those systems have similar limitations as VGA, but are harder to obtain and test with. With qemu, the VGA driver could fill this gap.

Apart from the Wikipedia entry, good resources on VGA are at osdev.net and FreeVGA

Join this project

Looking for hackers with the skills:

graphics drivers kernel drm

This project is part of:

Hack Week 22 Hack Week 24

Activity

  • 7 months ago: Pharaoh_Atem liked this project.
  • 7 months ago: llansky3 liked this project.
  • 8 months ago: michals liked this project.
  • over 2 years ago: vliaskovitis liked this project.
  • over 2 years ago: wfrisch liked this project.
  • over 2 years ago: achaudhary2 joined this project.
  • over 2 years ago: tdz added keyword "graphics" to this project.
  • over 2 years ago: tdz added keyword "drivers" to this project.
  • over 2 years ago: tdz added keyword "kernel" to this project.
  • over 2 years ago: tdz added keyword "drm" to this project.
  • over 2 years ago: tdz started this project.
  • over 2 years ago: tdz liked this project.
  • over 2 years ago: tdz originated this project.


    • Comments

    • tdz
      over 2 years ago by tdz | Reply

      I have pushed a skeleton driver to Gitlab. The does not display anything yet, but binds to a provided VGA device. The rsp kernel config for i586 is here.

      I use 32-bit Tumbleweed on qemu as testing environment. Here are some short setup instructions.

      • Download TW for 32-bit and install in a qemu guest
      • There are various emulated graphics devices available with qemu's 32-bit emulation. I use virtio, but anything with VGA should work
      • On the installed system, open /etc/default/grub and set 'GRUB_TERMINAL=console' to enable textmode. Also append 'console=ttyS0,115200n8' to 'GRUB_CMDLINE_LINUX_DEFAULT' to enable serial-console output.
      • Recreate grub's config with ' sudo grub2-mkconfig -o /boot/grub2/grub.cfg'. These steps will set grub to VGA text mode.
      • Build and install the provided kernel. I use 'make O=build-i586/' and then again with module_install and install to get it installed. This requires kernel development tools.
      • After a reboot, grub should start in text mode. The development kernel should load the vgadrm driver. It does not display anything. If you switch to the serial console and login, 'dmesg | grep drm' should mention vgadrm.

    • tdz
      over 2 years ago by tdz | Reply

      Day 1: I started with the skeleton driver that binds to the VGA device without doing proper output. I added blitting code to copy the framebuffer data into the video memory. The driver also installs a default palette on each pageflip. The screen now turns dark, but at least it's obvious that something changed. Updating the video memory is somewhat complicated due to VGA's awkward memory layout and a number of register settings that modify the written framebuffer data.

      See https://gitlab.suse.de/tzimmermann/linux/-/commits/vgadrm-day1.

    • tdz
      over 2 years ago by tdz | Reply

      Amit joined the project and is about to install and try the currently available driver code.

    • tdz
      over 2 years ago by tdz | Reply

      Day 2: I've been fixing register settings for the whole day. The driver is still not functional, but there's at least noise displayed on the screen. The existing material at FreeVGA and osdev.net has been extremely helpful.

      See https://gitlab.suse.de/tzimmermann/linux/-/tree/vgadrm-day2.

    • tdz
      over 2 years ago by tdz | Reply

      Day 3: I only did register debugging today, as I still don't see an image on the screen. VGA has plenty of settings for text mode; plus compatibility with the even older EGA and CGA cards. I guess those could interfere or I'm simply missing something trivial. The qemu emulator doesn't seems to tell me about possible problems. So I'm going to set up a regular BIOS-based PC to test the driver. Connecting a monitor should give a signal and a video mode. If the signal is there, the problem is in the screen update; if not it's in the mode setting.

      See https://gitlab.suse.de/tzimmermann/linux/-/tree/vgadrm-day3.

      • tdz
        over 2 years ago by tdz | Reply

        And just after I typed this comment, I tried the driver on a PC and could not get a signal on the monitor. So the mode-setting code appears to be the problem.

    • tdz
      over 2 years ago by tdz | Reply

      Day 4: I stepped through my commits and made them into smaller pieces until I found the reason why the screen remained dark. It turns out that it's necessary to activate the palette after filling it with colors. The bit the to that is non-intuitively hidden in the attribute controller's index register, which also serve as a data register. The VGA device's interface is often like this and small errors can quickly backfire.

      So there's still the noise on the screen, but with repeating patterns. It looks like the final problem is in the way the VGA chips reads from video memory.

      See https://gitlab.suse.de/tzimmermann/linux/-/tree/vgadrm-day4.

    • tdz
      over 2 years ago by tdz | Reply

      Day 5: Today I finally got a working output on the screen! Upon boot, the kernel console is being displayed and it is possible to log in and use the command line. The screen flickers, has a few artifacts and sometimes the colors are off, but it mostly works. I assume that the remaining problems come from many little issue within the kernel's DRM framework. The helpers for color-format conversion and blitting are not prepared to handle 16-color palette modes. So a few fixes might be required. Nevertheless, the display output is there and I count this hackweek project as a success. If I find the time to fix the remaining bugs, I'll send the driver to upstream for inclusion in the official Linux kernel.

      See https://gitlab.suse.de/tzimmermann/linux/-/tree/vgadrm-day5.

    Similar Projects

    early stage kdump support by mbrugger

    Project Description

    When we experience a early boot crash, we are not able to analyze the kernel dump, as user-space wasn't able to load the crash system. The idea is to make the crash system compiled into the host kernel (think of initramfs) so that we can create a kernel dump really early in the boot process.

    Goal for the Hackweeks

    1. Investigate if this is possible and the implications it would have (done in HW21)
    2. Hack up a PoC (done in HW22 and HW23)
    3. Prepare RFC series (giving it's only one week, we are entering wishful thinking territory here).

    update HW23

    • I was able to include the crash kernel into the kernel Image.
    • I'll need to find a way to load that from init/main.c:start_kernel() probably after kcsan_init()
    • I workaround for a smoke test was to hack kexec_file_load() systemcall which has two problems:
      1. My initramfs in the porduction kernel does not have a new enough kexec version, that's not a blocker but where the week ended
      2. As the crash kernel is part of init.data it will be already stale once I can call kexec_file_load() from user-space.

    The solution is probably to rewrite the POC so that the invocation can be done from init.text (that's my theory) but I'm not sure if I can reuse the kexec infrastructure in the kernel from there, which I rely on heavily.

    update HW24

    • Day1
      • rebased on v6.12 with no problems others then me breaking the config
      • setting up a new compilation and qemu/virtme env
      • getting desperate as nothing works that used to work
    • Day 2
      • getting to call the invocation of loading the early kernel from __init after kcsan_init()

    • Day 3

      • fix problem of memdup not being able to alloc so much memory... use 64K page sizes for now
      • code refactoring
      • I'm now able to load the crash kernel
      • When using virtme I can boot into the crash kernel, also it doesn't boot completely (major milestone!), crash in elfcorehdr_read_notes()

    • Day 4

      • crash systems crashes (no pun intended) in copy_old_mempage() link; will need to understand elfcorehdr...
      • call path vmcore_init() -> parse_crash_elf_headers() -> elfcorehdr_read() -> read_from_oldmem() -> copy_oldmem_page() -> copy_to_iter()

    • Day 5

      • hacking arch/arm64/kernel/crash_dump.c:copy_old_mempage() to see if crash system really starts. It does.
      • fun fact: retested with more reserved memory and with UEFI FW, host kernel crashes in init but directly starts the crash kernel, so it works (somehow) \o/

    • TODOs

      • fix elfcorehdr so that we actually can make use of all this...
      • test where in the boot __init() chain we can/should call kexec_early_dump()