Q: 'MacModel' replacement for radeon driver?

[Alex Deucher]

On Sun, Jan 27, 2013 at 6:38 AM, Albrecht Dreß <albrecht.dress at arcor.de> wrote:
> Hi Alex:
>
> I played a little more with the regs, but unfortunately could not find a
> solution.  I.e. I still don't have a usable signal on the connector.
>
> Maybe it would be helpful to check the radeon register settings from Mac OS
> 10.4.  What do you think?  Unfortunately, the radeonreg tool doesn't compile
> on OS X out of the box, and using some hacks I found on the internet, it
> doesn't produce any usable output.  Probably I have to dig into Apple's
> iokit for that.  Or do you know about a tool which would enlighten the
> situation?

We've checked all the relevant registers, dumping regs from OS X won't
help.  Perhaps the KMS driver is trying to point the display at an
invalid MC address or something like that.  Others have reported
r1xx/r2xx radeons working ok with KMS so it may be something mac or
ppc specific.

Alex


>
> Thanks,
> Albrecht.
>
>
>
> Am 05.01.13 12:31 schrieb(en) Albrecht Dreß:
>
>> Hi Alex:
>>
>> Am 03.01.13 20:36 schrieb(en) Alex Deucher:
>>>
>>> hmmm... everything else looks pretty good.  Try the attached patch.
>>
>>
>> This doesn't change the situation.
>>
>>> Maybe the pll isn't quite right.  You might try diffing the 2 reg dumps
>>> you sent me originally and using them as a guide to identify which registers
>>> are problematic.  Boot up to the broken state and then use radeonreg to set
>>> the registers to the values from the working dump.  e.g.,
>>> radeonreg regset 0x00000250 0x009f0013
>>>
>>> Some of the registers are related the memory configuration of each driver
>>> so try the following registers:
>>> 00000050 03000600
>>> 00000054 00008048
>>> 00000058 ff002102
>>> 00000204 008c051b
>>> 00000250 009f0013
>>> 00000254 03ff0026
>>> 00000278 009f03ff
>>> 0000027c 00000030
>>> 00000284 000300cd
>>> 000002c4 008c001e
>>> 000002c8 00860004
>>> 000002f0 20007c7c
>>> 000003f8 02000200
>>> 0000088c 07880100
>>> 00000d04 00000806
>>
>>
>> I set all these regs, one by one, to the values from the working fbdev
>> configuration of kernel 3.6.5, and double-checked with 'radeonreg regs
>> radeon' that they actually have these values.  However, the situation
>> doesn't change, i.e. these is still no output on the connector.
>>
>> Thanks,
>> Albrecht.