Bringing up xf86-video-intel 2.2.1 without a monitor connected

[Barry Scott]

Adam Jackson wrote:
> On Tue, 2008-03-18 at 08:05 -0700, Jesse Barnes wrote:
>   
>> On Tuesday, March 18, 2008 5:58 am Barry Scott wrote:
>>     
>>> Simon Farnsworth wrote:
>>>       
>>>> Hello,
>>>>
>>>> We've recently upgraded to the 2.2.1 version of the intel driver (from
>>>> git b31bef1a8effa9acb6de7edd206b9d8c48d88144), and now find that we
>>>> must plug a monitor into our systems in order to get them to start X.
>>>>
>>>> I can't see any changes in the code that would cause this, although
>>>> there has been quite a lot of churn in that area.
>>>>
>>>> Adding a modeline hasn't helped; I've attached my config file and
>>>> Xorg.log in the hope that someone knows what I need to do to fix things.
>>>>         
>>> This is a regression from the earlier Intel driver behaviour that is
>>> very important to us.
>>> Our user often power up before connecting VGA cables.
>>>
>>> We are happy to develop a patch to fix this regression. But would
>>> appreciate being
>>> pointed in the right general area.
>>>       
>> Yeah, I can't think of any commits offhand that might have caused this either.  
>> Can you bisect the driver down to the offending commit?  Also, is there a bug 
>> filed for the issue?
>>     
>
> It's not really the intel driver's fault, afaict.  If no outputs show up
> as connected (or unknown), then there'll be no mode list to work with,
> and xf86InitialConfiguration will give you the finger.
>
> - ajax
>   
I'll let the folks that understand the semantics of the code to figure 
out what needs fixing.

 From the outside we see a regression in the intel driver behaviour. The 
ideal would be that
we can configure any driver to assume that an output is connected (or 
unknown) rather then
deal with this issue driver by driver .

Barry