Release criteria for X11 releases

[Brice Goglin]

Dave Airlie wrote:
>> My main concern as a distributor (OpenBSD), is the lack of coherence
>> (and information) on the level of XRandR 1.2 support in the various
>> drivers. This is probably going to hurt several users who had working
>> pre xrandr setups.
>>     
>
> As far as I know in Xorg 7.3 only one driver supports randr 1.2 that
> being the Intel driver, the released ATI driver is still 6.6.3 which
> is the old style driver. We've done major testing on the randr-1.2 ATI
> driver but a final gold release wasn't shipped with Xorg 7.3.
>
> For intel users old-style dualhead config barely worked before..
>   

Apart from single-board dual-head issues, there is apparently also a
pretty big problem for people having 2 boards, one with a RandR 1.2
driver, the other one with non RandR driver. It looks like the latter
wants Xinerama-like syntax, but it crashes the former... These setups
seem totally broken these days. I guess anybody with the Intel driver
2.x and a random other board will hit the same problem.

There's an example with ATI 6.7.192 and MGA 1.4.7 at
    http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=443274
(this bug report is pretty messy since it involves some ATI-specific
breakage first).

The backtrace of the crash is an apparently-common one:
    2: X(xf86RandR12SetRotations+0x6b) [0x80f752b]
    3: X(xf86CrtcScreenInit+0x9e) [0x80f334e]
    4:
/usr/lib/xorg/modules/drivers//radeon_drv.so(RADEONScreenInit+0x1865)
[0xb7b51c65]
    5: X(AddScreen+0x1fc) [0x8073c9c]
    6: X(InitOutput+0x21e) [0x80a7eae]
    7: X(main+0x296) [0x8074426]
(from the end of the log at  
http://bugs.debian.org/cgi-bin/bugreport.cgi?msg=84;filename=Xorg.0.log.segfault2;att=1;bug=443274)

>From my point of view (i.e. looking at X-related bugs from Debian users
since Xorg 7.3 arrived in unstable), it is the worst bug with current
server and drivers...

Brice