Failure growing xfs with linux 3.10.5

[Eric Sandeen]

On 8/14/13 1:20 AM, Dave Chinner wrote:

> Hence looks like we'll need some new xfs_repair functionality to fix
> this. It might take me a little while to get you a fix - perhaps
> someone else with a little bit of spare time could get it done
> sooner than I can. Anyone?

Pff, not like I have the time, but it also looks pretty trivial.

secondary_sb_wack() only conditionally checks & zeroes unused space;
making it unconditional is probably all it takes.

I think it also needs to _not_ zero sb_bad_features2 as it does
today, because that is supposed to be kept in sync w/ sb_features2.

I have a patch I'll test; it'd be great to get a metadump from Michael
to test it on his original failure case.

-Eric