[Openchrome-devel] regression in kms_branch r955 and xorg-1.11.0

[James Simmons]

> > > I have not bisected it but last I tested (r955) has a regression.
> > > 
> > > The icons in notification area in xfce gets scrambled as showed on
> > > those screenshots:
> > > http://dev.alpinelinux.org/~ncopa/openchrome/
> > > 
> > > Those are both with xorg-server-1.11.0 and the problem happens with
> > > r955 build both with normal kernel and the drm-openchrome kernel.
> > > 
> > > Other than that and the normal "slowness" it seems ok.
> > 
> > Do you get EXA unalignment errors?
> 
> Not that I can see. (Xorg.0.log attached)
>
> > Is this regression still present with the latest code.
> 
> Yes. It's there with r977.

I looked at your logs and didn't see any problems as well. In my gnome and 
kde environments I haven't seen this so I'm going to have to pull out a 
render testers to see if I can replicate the problem. Strange this bug 
would show up for mode set handling code.
 
> > BTW you will need to use the soft cursor since
> > it appears to be broken for the VX900.
> 
> I just noticed :)
> 
> In case someone else is trying the driver, here is how:
> 
> alpine-netbook:~$ cat /etc/X11/xorg.conf.d/30-cursor.conf 
> Section "Device"
> 	Identifier "device0"
> 	Driver "openchrome"
> 	Option "SWCursor"  "true"
> EndSection

Thank you for the info.

> > Thanks for being brave enough to test it.
> 
> Thanks to you for working on this driver.
> 
> I'd be happy to run it on daily basis, as my primary display
> driver, but currently it is too slow for that. :-(

I need to get it working with the new TTM code :-)

> What is needed to fix the speed issues? Needs to add support for my
> chipset to the kernel drm module? (I know a little C but very little
> about xorg drivers. I might be able to figure things out if you point me
> in the right direction)

The standard via driver in the kernel doesn't support any acceleration 
which makes it slow. With new drm-openchrome DRM driver we can support 
acceleration. I need to implement the support on the xorg side. You just 
can't turn on the KMS because of the lack of LVDS support kernel side.
We need to create a buffer manager that can act to support TTM and non-TTM 
support.