clients die on xrandr 1.2 switch

[Ken Mandelberg]

Bryce Harrington wrote:
> On Thu, Aug 23, 2007 at 10:18:33AM -0400, Ken Mandelberg wrote:
>> I'm now running on Ubuntu Feisty with
>>
>> X Window System Version 1.3.0
>> X.Org Video Driver: 1.2
>>
>> on a 915GM laptop.
>>
>> xrandr 1.2 switches "work", the Gnome desktop switches between the 
>> 1024x768 LVDS and a 1920x1200 VGA and resizes.
>>
>> In the VGA->LVDS switch all the clients survive.
>>
>> In the LDVS->VGA switch the gnome-terminals always segfault. It's hit 
>> and miss on other clients. xterms survive.
>>
>> Am I missing some library updates that need to be xrandr savvy?
> 
> Hi Ken,
> 
> You'll want to make sure you are using the -intel 2.0 or later driver.
> Feisty by default has an older version with incomplete xrandr support.
> 
> You might also want to try out the Gutsy LiveCD, which has much better
> xrandr 1.2 support.  http://cdimage.ubuntu.com/daily-live/current/
> 
> Hope this helps,
> Bryce
> 

Bryce

I have been tracking with the latest intel driver, whats new is moving 
up to the 1.3 server with randr 1.2. But thanks for the suggestion about 
trying the live Gutsy cd. The problem I'm seeing going LDVS->VGA doesn't 
show on the live CD.

Still, I'm not ready to move to Gutsy beta and really would like to be 
able to use randr 1.2 on Feisty.

One thing I noticed was a difference in the desktop behavior on Gutsy. 
On Feisty when I go VGA->LVDS all the windows get repositioned to stay 
on the smaller desktop. On Gutsy they stay where they are, with some 
being pushed off into other virtual desktops. I guess this must be 
metacity code.

In a different posting I noted that the client segfaults trace back to

0  init_xinerama_support (screen=<value optimized out>) at 
gdkscreen-x11.c:617
#1  0xb7c06090 in _gdk_x11_screen_size_changed (screen=0x80680d0,
     event=0xbffdb618) at gdkscreen-x11.c:753
#2  0xb7bf85d9 in gdk_event_translate (display=0x8065098, event=0x80695f8,
     xevent=0xbffdb618, return_exposes=0) at gdkevents-x11.c:1825
#3  0xb7bf9cfb in _gdk_events_queue (display=0x8065098) at 
gdkevents-x11.c:2252
#4  0xb7bfa0ff in gdk_event_dispatch (source=0x806cec0, callback=0,
     user_data=0x0) at gdkevents-x11.c:2312
#5  0xb78a7df2 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#6  0xb78aadcf in ?? () from /usr/lib/libglib-2.0.so.0

I'm wondering if the different placement strategy shortcuts the trip 
through gdk_x11_screen_size_changed which is apparently causing the 
client segfaults.