Re: MMC broken on beagleboard with current git

[Tony Lindgren <tony@atomide.com>]

* Koen Kooi <k.kooi@student.utwente.nl> [080925 13:25]:
>
> Op 25 sep 2008, om 11:47 heeft Tony Lindgren het volgende geschreven:
>
>> * Koen Kooi <k.kooi@student.utwente.nl> [080925 12:23]:
>>>
>>> Op 25 sep 2008, om 10:51 heeft Tony Lindgren het volgende geschreven:
>>>
>>>> * Koen Kooi <k.kooi@student.utwente.nl> [080925 11:42]:
>>>>> Hi,
>>>>>
>>>>> I just built a kernel from current git[1] and the mmc controller
>>>>> doesn't
>>>>> get detected on boot on my beagleboards. The omap mmc system has
>>>>> seen a
>>>>> lot of changes lately, does anyone have a hint where to start  
>>>>> looking
>>>>> before attemping a bisect?
>>>>
>>>> Sounds like the recent mmc init changes we've done:
>>>>
>>>> 535e1808d1c77a580815f7d7daacc6c030e29cb1 -  12 september 2008
>>>
>>>> 8e6e06c71b9d21d85a3ae652868e2802c054a08a - 12 september 2008
>>>
>>>> c434c15d28c82d92e55897bd265c423e9ab69362 - 24 september 2008
>>>
>>>
>>>> I've verified that it works on omap3430sdp. But mach-omap2/hsmmc.c
>>>> needs
>>>> some more work done to support other mmc controllers besides the  
>>>> first
>>>> one.
>>>> I don't think other controllers ever worked before these patches
>>>> either though..
>>>>
>>>> Let me know if you can narrow it down to some commit and I may be  
>>>> able
>>>> to help further from there.
>>>
>>> The previous working version I built was based of
>>> de1121fdb899f762b9e717f44eaf3fae7c00cd3e which was commited at 13
>>> september 2008, so 535e18.. and 8e6e06c.. don't break it.
>>
>> OK, good to know. Can you try to git-bisect it?
>
> That proved to be a lot less work than expected :)
>
> Last good: 9d1dff8638c9e96a401e1885f9948662e9ff9636
> First bad: c434c15d28c82d92e55897bd265c423e9ab69362
>
> So 
> http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=commit;h=c434c15d28c82d92e55897bd265c423e9ab69362 
>  breaks MMC on beagle.

Hmm, I don't understand how it could work on 3430sdp...

Looks like board-omap3beagle.c calls hsmmc_init() just like
board-3430sdp.c. Maybe timings have changed and you need
a longer delay somewhere during the init?

Maybe sprinkle some printks to hsmmc.c and see if that
makes it to work?

Also make sure omap_hsmmc.c probe finishes OK.

Tony

>>>>> [1] http://source.mvista.com/git/gitweb.cgi?p=linux-omap-2.6.git;a=commit;h=c434c15d28c82d92e55897bd265c423e9ab69362