[HEADS UP] Radeon r6/7xx drm code to be committed

O. Hartmann ohartman at mail.zedat.fu-berlin.de
Fri Mar 13 13:36:48 PDT 2009


Kip Macy wrote:
> Are you using the radeonhd from git? I needed to use that to get it to work.
>
> -Kip
>
> On Fri, Mar 13, 2009 at 2:09 AM, O. Hartmann
> <ohartman at zedat.fu-berlin.de> wrote:
>   
>> My private box is a UP FreeBSD 8.0-CURRENT/amd64 equipted with a MSI HD4830
>> (RV770). On this machine with the most recent OS and the most recent
>> xorg-7.4 compiled from the ports,
>>
>> Driver 'radeonhd'
>>
>> does not work, the tty does never show up login screen. It seems the driver
>> does not recognise the GPU. Also on this box, enabling
>>
>> Option 'DRI'
>>
>> doesn't work. When getty tries to start xdm on the specific tty, Xserver
>> can't be started - there is a error message on console coming from [drm],
>> saying something about 'idle'. I never caught this.
>>
>> On my lab's box, a SMP FreeBSD 8.0-CURRENT/amd64 with the most recent
>> xorg-7.4 as mentioned above, equippted with  a MSI HD4670 (RV730),
>>
>> Driver 'radeonhd'
>>
>> works, but shows up weird behaviour sometimes (crashing), so I switchd to
>>
>> Driver 'radeon'.
>>
>> But also on this box, enabling
>>
>> Option 'DRI'
>>
>> does not let X come up.
>>
>> Either way, with radeonhd or radeon, ending a X session (I use windowmaker),
>> seems crashing X and then locking up/crashing the whole box. Also very
>> boring is the fact, that I can't swtch to the console via
>>
>> Ctrl-Alt-F{1-8}
>>
>> althought this is explicitely enabled in the config. I see 'nothing' but a
>> black screen, the console is definitely changing, I can change back to the X
>> screen, but never to a console. This behaviour is both on FreeBSD 7.1-STABLE
>> (most recent before I switched these days to 8.0-CURRENT on that box) and
>> 8.0-CURRENT! I guess this is a serious issue in Xorg 7.4 and/or the Radeon
>> driver.
>>
>> It is also very surprising, that on the slower UP FreeBSD box with the
>> faster GPU (RV770), X is much(!) faster than on the faster SMP FreeBSD box
>> with the slower GPU (RV730). On both systems DRI is not enabled and both
>> boxes do not load modules 'drm' and 'radeon'.
>>
>> Kip Macy wrote:
>>     
>>> radeon locks up my system, use radeonhd
>>>
>>> On Thu, Mar 12, 2009 at 12:25 PM, O. Hartmann
>>> <ohartman at mail.zedat.fu-berlin.de> wrote:
>>>       
>>>> Kip Macy wrote:
>>>>         
>>>>> I have an HD4850 and it works well. Unfortunately it is the one from
>>>>> ASUS where only one port does dual-link.
>>>>>
>>>>> -Kip
>>>>>
>>>>> On Mon, Mar 9, 2009 at 10:27 AM, Nenhum_de_Nos
>>>>> <matheus at eternamente.info> wrote:
>>>>>
>>>>>           
>>>>>> On Mon, March 9, 2009 13:53, Robert Noland wrote:
>>>>>>
>>>>>>             
>>>>>>> On Mon, 2009-03-09 at 12:16 -0400, Robert Huff wrote:
>>>>>>>
>>>>>>>               
>>>>>>>> Gary Jennejohn writes:
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>>>>  > >  This should work on pretty much all Radeon r600+ cards (and
>>>>>>>>> the
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>> older
>>>>>>>>
>>>>>>>>                 
>>>>>>>>>  > >  stuff as well).  With the appropriate X drivers, you should
>>>>>>>>> get
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>> EXA
>>>>>>>>
>>>>>>>>                 
>>>>>>>>>  > >  accelleration and Xv.  3d is not available yet, afaik.
>>>>>>>>>  >
>>>>>>>>>
>>>>>>>>>  This is kernel code and has nothing directly to do with ports.
>>>>>>>>>
>>>>>>>>>                   
>>>>>>>>     Let me be more specific.
>>>>>>>>     Assuming I'm using a 3850, which I believe is the RV670
>>>>>>>> chipset, is there anything I will need to change in xorg.conf (or
>>>>>>>> elsewhere) to take advantage of/not break the new code?
>>>>>>>>
>>>>>>>>                 
>>>>>>> rv670 is what I'm testing on... I tried to put a few details in the
>>>>>>> commit message.
>>>>>>>
>>>>>>> Basically, for R600+ you have to force DRI on for the moment.  You
>>>>>>> should also be using exa.
>>>>>>>
>>>>>>> So, right now, I'm using radeonhd from git master, with :
>>>>>>>
>>>>>>>        Driver      "radeonhd"
>>>>>>>        VendorName  "ATI Technologies Inc"
>>>>>>>        BoardName   "RV670PRO [Radeon HD 3850]"
>>>>>>>        BusID       "PCI:1:0:0"
>>>>>>>        Option      "DRI"
>>>>>>>        Option      "AccelMethod" "EXA"
>>>>>>>        Option      "Audio"
>>>>>>> EndSection
>>>>>>>
>>>>>>> robert.
>>>>>>>
>>>>>>>
>>>>>>>               
>>>>>>>>                             Robert Huff
>>>>>>>>
>>>>>>>>
>>>>>>>>                 
>>>>>>> --
>>>>>>> Robert Noland <rnoland at FreeBSD.org>
>>>>>>> FreeBSD
>>>>>>>
>>>>>>>               
>>>>>> I'm planning to move from nVidia to ATi, and all this work is what
>>>>>> makes
>>>>>> me do that. This code you're talking about, would benefit a HD4850 ?
>>>>>>
>>>>>> thanks,
>>>>>>
>>>>>> matheus
>>>>>>
>>>>>> --
>>>>>> We will call you cygnus,
>>>>>> The God of balance you shall be
>>>>>>
>>>>>> _______________________________________________
>>>>>> freebsd-current at freebsd.org mailing list
>>>>>> http://lists.freebsd.org/mailman/listinfo/freebsd-current
>>>>>> To unsubscribe, send any mail to
>>>>>> "freebsd-current-unsubscribe at freebsd.org"
>>>>>>
>>>>>>
>>>>>>             
>>>>> _______________________________________________
>>>>> freebsd-current at freebsd.org mailing list
>>>>> http://lists.freebsd.org/mailman/listinfo/freebsd-current
>>>>> To unsubscribe, send any mail to
>>>>> "freebsd-current-unsubscribe at freebsd.org"
>>>>>
>>>>>           
>>>> I use a MSI HD4830 on a UP FreeBSD 8.0-CURRENT/amd64 box with
>>>>
>>>> Driver   "radeon"
>>>>
>>>> from ports/xorg-7.4. Without EXA and DRI the graphics card works well,
>>>> but when enabling EXA and/or DRI, system goes mustang. Ctrl-Alt-FX
>>>> doesn't bring back a console, the tty is 'blinking' as it looks like the
>>>> X server tries to start and fails.
>>>>
>>>> On another box, running FreeBSD 8.0-CURRENT/amd64, but SMP and MSI
>>>> HD4670, with EXA and DRI enabled the whole box crashes immediately when
>>>> loading Xserver. Without EXA and DRI, leaving a windowmaker session and
>>>> therefore ending the Xserver kills the box, the OS dies and coredumps or
>>>> hangs forever with a tiled screen showing 3 rows of 6 columns of the
>>>> console screen on a 20,1 inch TFT.
>>>>
>>>>         
>>> _______________________________________________
>>> freebsd-current at freebsd.org mailing list
>>> http://lists.freebsd.org/mailman/listinfo/freebsd-current
>>> To unsubscribe, send any mail to "freebsd-current-unsubscribe at freebsd.org"
>>>       
>>
>>     
> _______________________________________________
> freebsd-current at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-current
> To unsubscribe, send any mail to "freebsd-current-unsubscribe at freebsd.org"
>   


No, it is the driver taken from ports. I do not use drivers from the
development tree aside the FreeBSD development tree, including ports.

By the way, whenever I use 'radeon' on my lab's box (FreeBSD
8.0-CURRENT/amd64 SMP, Xorg 7.4), every reboot (shutdown -r or
reboot(1)) crashes the system, leaving the box unusable (filesystem /var
gets dirty and so the OpenLDAP server doesn't start anymore). This issue
is due to the xorg-7.4-driver (radeon, radeonhd doesn't work as I said).
The graphics card is a HD4670. Without xdm/Xorg started the box reboots
clean. This also happened when this box ran FreeBSD 7.1-STABLE before
the upgrade last week! With the new drm-code, DRI isn't usable as option
anymore on all FreeBSD 8.0 boxes I have around here equipted with ATI
HD4670/HD4830 cards. The situation is really nasty this moment.




More information about the freebsd-x11 mailing list