[PATCH] Fix MMC hierarchy

[Pierre Ossman]

David Zeuthen wrote:
> On Sat, 2005-12-10 at 00:29 +0100, Pierre Ossman wrote:
>   
>> There were some issues with how parents were selected for mmc stuff. The
>> attribute linux.sysfs_path_device was set for mmc_host:s, even though
>> they lack a device to be mapped too. This caused problems when a
>> physical card hade multiple mmc hosts.
>>
>> The second issue is similar by nature, and made worse by fixing the bug
>> above. Cards were associated with the last registered mmc_host for the
>> device. With the fix above they were associated with the physical
>> device. To fix this we have to look up a new parent based on the mmc
>> host number. I'm not sure if it's safe doing it this late, but it seems
>> to work.
>>     
>
> I don't have hardware to test with so if this works for you, I'd say go
> ahead and commit it since it obviously fixes a bug.
>
>
>   

There are more bugs I'm afraid since fstab-sync doesn't get linked to
the nodes on these machines. But I didn't have time to debug it further
so this would have to suffice for now. At least the hal tree structure
is correct now, and it doesn't break any hosts that were working previously.

Rgds
Pierre