[Desktop12.04-Topic] Working VNC (or no VNC)

[Bryce Harrington]

Hi Alberto,

Can you escalate bug #772873 with AMD in your next call?

The bug has to do with the use of XDamage in a remote X context using
VNC with compositing enabled.  Nvidia fixed this in their driver in
version 256.25, and it works fine with all the open drivers, but fglrx
seems to still be affected.

I know there's not time to resolve this for the oneiric release, but
perhaps once it's fixed we could include it fixed driver in x-updates.

I've added the issue to the Oneiric release notes as well.
https://wiki.ubuntu.com/OneiricOcelot/TechnicalOverview

Thanks,
Bryce

On Thu, Oct 06, 2011 at 05:17:56PM -0700, Bryce Harrington wrote:
> On Thu, Oct 06, 2011 at 06:46:10PM +0200, Sebastien Bacher wrote:
> > Le mercredi 05 octobre 2011 à 12:41 -0700, Bryce Harrington a écrit :
> > 
> > > But I think as a next action, I think I have all the hardware I need to
> > > test it myself, so I'll try and do that today or tomorrow.  I would hate
> > > to spin up a bunch of people to do testing if there really are obvious
> > > breakages...
> > 
> > Thanks Bryce!
> 
> Ok, here's what I've learned.
> 
> I can confirm the bug that on fglrx you can connect and get an initial
> display of the desktop, but it never updates.  I tested this on oneiric.
> 
> I was unable to test the workaround of enabling xdamage in vino, because
> that option appears to have been removed from gconf in oneiric.
> 
> On -nvidia, using the old nvidia-173 driver on natty, I also reproduced
> the bug.
> 
> On the same system I switched to the nvidia-current driver (still on
> natty), and it worked with no bug.
> 
> So, this all seems to confirm my hypothesis.  The issue existed for both
> nvidia and fglrx, but NVIDIA fixed it (presumably as of version 256.25 -
> http://www.nvnews.net/vbulletin/showthread.php?p=2255561).  Older
> alternative nvidia drivers may still experience the bug, but not current
> nvidia.
> 
> I tested -ati, -intel, and -nouveau on both natty and oneiric, and it
> worked in all cases I tried.
> 
> 
> So, my conclusion is that this seems to confirm the issue is a bug in
> fglrx and should be escalated to AMD for resolution.
> 
> Even if AMD is able to fix this swiftly, there isn't going to be time to
> roll out a new fglrx driver, and since binary drivers can't be SRU'd, we
> won't be able to fix this bug.  We should list it in the Release Notes,
> escalate to AMD, get it fixed in Precise and in x-updates.
> 
> As to putting out a call for testing, I'll leave that to your discretion
> but don't think it's necessary.  Can't hurt, but I strongly suspect it's
> merely going to just confirm that it's broken only on fglrx and older
> nvidia.
> 
> Bryce
> 
> -- 
> ubuntu-desktop mailing list
> ubuntu-desktop at lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/ubuntu-desktop