latest git xserver fails to startx on x86_64

Justin P. Mattock justinmattock at gmail.com
Sun Aug 9 12:34:37 PDT 2009


Aaron Plattner wrote:
> On Fri, Aug 07, 2009 at 11:28:57AM -0700, Justin P. Mattock wrote:
>    
>> Dave Airlie wrote:
>>      
>>> On Fri, Aug 7, 2009 at 3:52 PM, Justin P.
>>> Mattock<justinmattock at gmail.com>   wrote:
>>>
>>>        
>>>> Dave Airlie wrote:
>>>>
>>>>          
>>>>> On Fri, Aug 7, 2009 at 2:37 PM, Justin P.
>>>>> Mattock<justinmattock at gmail.com>     wrote:
>>>>>
>>>>>            
>>>>>> with a quick glance, is this type of error due to something in the
>>>>>> xserver, or is this something that I did with the build of either the
>>>>>> xserver or nvidia
>>>>>>
>>>>>> dlopen: /usr/lib64/xorg/modules/drivers/nvidia_drv.so: undefined symbol:
>>>>>> resVgaShared
>>>>>>
>>>>>>              
>>>>> you seem to have accidentally binary driver.
>>>>>
>>>>> Dave.
>>>>>
>>>>>
>>>>>            
>>>> yeah, I was trying to determine if this was caused by a change in the
>>>> xserver, or a problem with the build of nvidia.  (creating a pure64 is a
>>>> bit of a pain)
>>>>
>>>>   From what it seems xf86-video-vesa activates the xserver.
>>>>
>>>> I think what I'll do is downgrade everything with the xserver(since nvidia
>>>> seems to have an abi v5 and the xserver is at abi 7) just to be safe.
>>>>          
>
> It's 6, not 7.  You might be looking at the XINPUT ABI version.
>
>    
>>> nvidia don't support master X servers usually until they are released and
>>> shipping in Ubuntu.
>>>        
>
> I should jump in here and mention that while we don't *officially* support
> master X servers until the ABI is declared stable, I do try to keep support
> relatively up to date in the shipping drivers.  I bake the git commit ID of the
> snapshot the driver was built against into the driver, so if you start X with
> -ignoreABI and -logverbose 5, it will (if it loads at all) tell you which
> version of the prerelease SDK it was built against.
>
> We wait for whoever the release manager happens to be to declare the ABI stable
> before flipping the switch in the driver to make it not require -ignoreABI.
> It's not when any particular distro ships that determines it.
>
> Support for Dave's RAC removal change should be in the next beta release.  I
> know it would be nice if you could just use the driver with any custom-built X
> server, but hopefully this is the next best thing.
>
> -- Aaron
>
>    
Id have to say I am really impressed with
how nvidia keeps up.

At the moment one machine is running
the latest git xserver(before the RAC removal)
with the latest nvidia.
there is the abi error message but the xserver does start,
and 3D is enabled.

and on the other machine I compiled a stable version
of the xserver so things can be right with the nvidia module,
only due to the behalf  is(with the machine that has the latest git xserver)
I'm noticing mplayer has for some reason
awkward opaque window moving(trails behind the cursor when
moving the window during playback of t.v.) probably nothing harmful.

But In this case being patient for things to get looked over
is I think a better choice.(then again if you have to have the latest 
xserver
then that option is there). And knowing that somebody is there
keeping track of the kernel, xserver, etc.. is a comforting feeling.
(from past experiences having a module(no names mentioned)
  that is not maintained, or is but only released 6/9 months
without the fixes you need  is a bit difficult especially
if your not good at code writing).

Justin P. mattock











More information about the xorg mailing list