Re: Big Endian fix patch

[Lennart Sorensen]

On Wed, Jul 28, 2010 at 04:10:58PM -0400, Doug Nazar wrote:
>  On 2010-07-28 2:49 PM, Lennart Sorensen wrote:
>>
>> mdadm: device 1 in /dev/md/0 has wrong state in superblock, but /dev/sdb2 
>> seems
> Ok, I think this is a bug in mdadm. I couldn't replicate it on a x86 box  
> but then tried it in the ppc image and it got the same message. Not sure  
> if it's because of an endian issue or version issue my x86 boxes have  
> version 3.1.2 while the Ubuntu ppc is running 2.6.7.1.
>
> The error seems kinda strange as devices are 0 numbered so sdb2 is  
> probably device 1 (I assume sda2 is device 0). Unfortunately the mdadm  
> source has half a dozen structs with the same members so it's almost  
> impossible to grep for where things are. It looks like that error  
> happens when it's detecting if the drives are in the right roles.  
> There's a __cpu_to_le32 that seems suspicious to me (super1.c:676).
>
> What version of mdadm do you have?

3.0.3.

>> That's the first raid1.
>>
> The superblocks are correct.

Well that's good.  They do appear to work.

-- 
Len Sorensen