Systemd 230 Opens Up A New Graphics Vulnerability & FBDEV Still Should Die

[nanonyme]

systemd-consoled

EDIT: The lead developer dropped it: https://github.com/systemd/systemd/pull/747

I wonder how big a part the kdbus maintenance played in systemd-consoled being orphaned

[Krejzi]

as i understand it: without this change you can only run wayland on fb as root, so if you value security you wouldn't run wayland on fb anyway

Not really. Weston uses setuid wrapper to open devices as root, then drops privileges and runs as the user that started it. Similar to how Xorg does things since recently.

[pal666]

as i understand it: without this change you can only run wayland on fb as root, so if you value security you wouldn't run wayland on fb anyway

[ermo]

systemd-consoled

EDIT: The lead developer dropped it: https://github.com/systemd/systemd/pull/747

Just a thought, but FBDEV likely exists because some people find its functionality--if not the specific implementation details--useful. Seems to me that in an ideal world, KMSCON would need to be revived in some shape or form before it would be possible to kill off FBDEV?

In addition, it would likely end up being an additional driver for the ARM-people to just create a DRM/KMS driver for their hardware already and everyone would be happy?

[Ericg]

Speaking of the framebuffer and KMS:

Does anyone here know what happened to David Herrmann's plans to create a KMS console (KMSCON)?

EDIT: Oh, KMSCON's successor is in systemd?

systemd-consoled

EDIT: The lead developer dropped it: https://github.com/systemd/systemd/pull/747

[ermo]

Speaking of the framebuffer and KMS:

Does anyone here know what happened to David Herrmann's plans to create a KMS console (KMSCON)?

EDIT: Oh, KMSCON's successor is in systemd?

• KMSCON (Maintainer, 2011 – 2014): KMSCON is a terminal/console emulator for Linux. It was created as replacement for the in-kernel terminal emulator and VT subsystem. It is based on the Linux DRM subsystem and uses Kernel Mode Setting (KMS) to display the console to the screen. Similar to Wayland it is an attempt to replace the old Linux infrastructure with a modern approach and producing better maintainable code while dropping the 20-year old mess that is VT+X11. In 2014, I stopped working on it. We implemented a replacement in systemd, based on the lessons learned on KMSCON.

[unixfan2001]

Unless there's an alternative to fbdev I'm not aware of

That would be KMS.
Thankfully, that's the model embedded developers are now focusing on.

[schmidtbag]

Unless there's an alternative to fbdev I'm not aware of, I don't think it should ever be killed off. I personally find it very useful, and it can do a lot of interesting things. It allows you to get a working X session without proper GPU drivers (ideal for some ARM platforms). It allows you to graphically view websites, images, and PDFs through a tty (implying no graphical server). You can even do a VNC client through a tty. Are all of those features really all that practical? Maybe not, but it's this kind of stuff why I think Linux is so fascinating.

[darkbasic]

Just kill fbdev please.

[nanonyme]

gma500 is a DRM/KMS driver. However, like other in-kernel KMS drivers, it probably provides fbdev compatibility. Does your console switch to a higher resolution when gma500 loads? If yes, that's the fbdev compatibility at play.

Certain distros actually force it to use fbdev due to issues with KMS. This smells like the drivers were or are buggy