Hi Greg,

Thanks for the response! It's always good to get notes on a patch.
Some responses to your points:

> Why __le32?  Does this variable go across the user/kernel boundry
> somehow?  If not, just use le32.

Good point, this should probably have been le32.

> At first glance, you aren't doing ths same logic in this function as the
> original did, please look at this very closely again and verify that you
> are doing this correctly.
>
> Don't just blindly quiet tools like sparse, it is warning for a reason,
> but be careful about your fix.

On a second, closer look at the code I am not doing this correctly: the
buffer I am converting to le32 needs to persist (which a local variable
would not). On my first glance at this code I saw the same buffer being
used for both little- and big-endian storage of the same data -- it's
correct, but a little ugly.

I am going to leave this code as is, since it was functioning properly
before my patch.

Thanks,
Jake