Re: [PATCH] xfs_db: allow recalculating CRCs on invalid metadata

[Eric Sandeen <sandeen@xxxxxxxxxxx>]

On 5/12/16 6:28 PM, Dave Chinner wrote:
>>> +
>>> > > +       if (corrupt) {
>>> > >                 nowrite_ops.verify_write = xfs_dummy_verify;
>>> > > -               iocur_top->bp->b_ops = &nowrite_ops;
>>> > > -               dbprintf(_("Allowing write of corrupted data\n"));
>>> > > +               dbprintf(_("Allowing write of corrupted data and bad 
>>> > > CRC\n"));
>>> > > +       } else {
>> > 
>> > Maybe a helpful/redundant comment about /* invalid_data */ alongside } 
>> > else { ?
> I though the dbprintf() documented it well enough? maybe move that
> to the top of each branch?

Oh, I suppose it does.  By the time we returned early on !corrupt & ! 
invalid_data,
then tested explicitly for (corrupt), it requires a decent brain-stack to 
remember what the } else { clause is for, IMHO.

But like I said, just a thought.

-Eric