[fedora-arm] omap_hsmmc

Jon Masters jonathan at jonmasters.org
Sat Jul 14 13:53:20 UTC 2012


On 07/14/2012 04:32 AM, Peter Robinson wrote:
> On Sat, Jul 14, 2012 at 7:53 AM, Jon Masters <jonathan at jonmasters.org> wrote:
>> Folks,
>>
>> The latest F18 kernel is configured with CONFIG_MMC_OMAP_HS=m (which was
>> previously configured as a builtin driver, as it probably should be).
>> While in theory this would be working, it doesn't actually seem to
>> result in dracut picking up the module as a necessary platform device.
>> It is best fixed by building in the module again, which isn't happening
>> because some other config option is forcing this to be modular too. I
>> haven't tracked down what the other option was yet, feel free to help.
> 
> 
> Interesting because the config bits don't appear to have changed:
> 
> CONFIG_MMC_OMAP=m
> CONFIG_MMC_OMAP_HS=y
> 
> These two bits are the same on both kernels, maybe some more generic
> config that's been moved around that's causing the regression.

Yea.

>> In the interim, the following dracut config fileis a workaround:
>>
>> # /etc/dracut.conf.d/omap_hsmmc.conf
>> # Dracut is not picking up the platform device driver for the HSMMC
>> # interface
>>
>> add_drivers+="omap_hsmmc"
> 
> So surely that's a bug in dracut that needs to detect that the module
> is needed for booting.

Well yes and no. Yes, if the driver is providing alias data that is also
provided by the platform at runtime that ought to get matched. My
concern is that - given the specific driver implementation - this is not
the case, but I didn't look into that yet. Still, let's build it back in
for now. We just need to track down why Kbuild change the config due to
whatever Kconfig rules were in place at build time.

Jon.


More information about the arm mailing list