[kmscon-devel] radeonsi scrambled graphics

David Herrmann dh.herrmann at gmail.com
Tue Jan 21 10:22:08 PST 2014


Hi

On Tue, Jan 21, 2014 at 7:16 PM, glphvgacs <darwinskernel at gmail.com> wrote:
> On Tue, Jan 21, 2014 at 07:12:21PM +0100, David Herrmann wrote:
>> Hi
>>
>> On Tue, Jan 21, 2014 at 7:08 PM, glphvgacs <darwinskernel at gmail.com> wrote:
>> > hello,
>> > scrambled is the best i could come up with for what i see, or not see
>> > really, when i pass --hwaccel --drm to kmscon. i'm including portions of
>> > the logs that are relavent but if you need more let me know.
>>
>> What happens if you don't pass --hwaccel?
>
> i works, but scrolling is *jittery*.

Ugh, sure? That shouldn't happen. Can you run kmscon with --debug and
append the log? And what exactly does "jittery" mean here? Does it not
update the screen atomically? Or does it lag? Or do you see artifacts?

>> If it works, then it's a bug
>> in the radeonsi mesa driver and you should report it to their bugzilla
>> (bugs.freedesktop.org). However, I recommend just not using hw-accel
>> at all. The mesa drivers still regularly fail outside of Xorg and I
>> cannot do anything about it.
>>
>> Thanks
>> David
>
> i don't see any useful msg to send to them. can you tell me what to look
> for?

There's not much information to send to them, sadly. The easiest way
to debug this is to test:
  https://github.com/robclark/kmscube

Simply clone the repository, compile it and run it from a VT(!). If it
shows the same artifacts, then the driver is broken. It's usually
enough to open an upstream bug-report with a link to kmscube and tell
them that it's broken (a photo of the screen helps them a lot!).
kmscube is a know simple example, so they should be able to figure it
out.

Thanks
David


More information about the kmscon-devel mailing list