Fw: 32k Blocksize Support

[Eric Sandeen]

D Landy wrote:
> Hi again, 
> 
> First of all, thanks to Eric Sandeen for his offline support. 
> 
> I'm coming back here at his suggestion as we haven't managed to resolve it. 
> 
> So far, we've established that it *is* an ext2 filesystem (using file -s), 
> and that resize2fs reports that it has an invalid superblock. 
> 
> Eric wrote: 
> 
>> I'd probably dig into why resize2fs says it's corrupt; large block
>> should not mean corrupt, AFAIK, even if the running kernel can't
>> actually mount it. 
>>
>> You might get this back on-list, too, so future generations can benefit
>> from your pain (and in case someone else knows these answers).
> 
> Does anyone know if a 32k blocksize would cause resize2fs to report an 
> invalid superblock? I've downloaded the source code and from what I can see 
> the maximum block size is 64k, so I wouldn't have thought so - but I'm not a 
> C programmer and have trouble following the source sometimes. 
> 
> I'd appreciate another set of eyes going over the code... 
> 
> Any help greatly appreciated. 

I don't know if they're using a standard ext3 fs or not; perhaps it is
adultrated in some way for their needs that makes it incompatible w/ the
upstream tools.

You could go through the code to find where that message is printed,
then work backwards to why (either via gdb, or printf insertions, or
whatever you're comfortable with...)

-Eric