Bug#981722: DRM: Segfault on Etnaviv

[Guido Günther]

Hi,
On Wed, Feb 03, 2021 at 09:40:20AM +0000, Simon McVittie wrote:
> On Wed, 03 Feb 2021 at 09:56:21 +0100, Guido Günther wrote:
> > SDL2 currently segfaults on imx8 hardware with the DRM backend when trying to
> > use osk-sdl (currenty in NEW) when trying to use a cursor plane (so
> > other applications should hit this as well)
> 
> The KMSDRM backend is quite new and I suspect none of the maintainers
> either upstream or in Debian use it (X11 or Wayland are a lot more
> interesting for most SDL users), so if it works, that's lucky, and if
> it doesn't work, we are not really well-placed to resolve that.
> 
> > The buster version works
> 
> With which backend? I didn't think we had KMSDRM enabled until August 2020?
> (Unfortunately I can't find the bug report or merge request that asked for
> it to be enabled.)
> 
> > and above changes were reverted in
> > current hg so 2.0.15 is working as well but a backport is non trivial
> > since most of the code there got rolled back so backporting that would
> > be rather invasive.
> 
> If you can prepare a tested patch, I'd consider it: hopefully the changes
> would be isolated to KMSDRM files and so could not cause regressions for
> X11 or Wayland?

It's isolated but pretty invasive:

https://source.puri.sm/guido.gunther/libsdl2/-/tree/kmsdrm/debian/patches/backport

I'll test that for a couple of days but I'm not sure this qualifies for
bullseye?

> Or, if the KMSDRM backend doesn't actually work in its current state, we
> could disable it to stop it from making people think it ought to work?
> 
> If this functionality is important to you, an autopkgtest would also be
> welcome, although I realise that probably isn't feasible because it would
> need to be run on suitable (real or virtual) hardware.

I'll look at adding a patch for this down the read. Likely by adding a
dry-run mode to osk-sdl that would exercise most bits of the graphics
pipeline so uses can still unlock their phones.

Cheers,
 -- Guido