Re: [RFC PATCH V3] xfsrestore: fix rootdir due to xfsdump bulkstat misuse

["Darrick J. Wong" <djwong@kernel.org>]

On Fri, Apr 21, 2023 at 11:10:06AM +0200, Carlos Maiolino wrote:
> Hi Darrick.
> 
> > > >> This patch adds a '-x' option (another awkward thing is that
> > > >> the codebase doesn't support long options) to address
> > > >> problamatic images by searching for the real dir, the reason
> > > >> that I don't enable it by default is that I'm not very confident
> > > >> with the xfsrestore codebase and xfsdump bulkstat issue will
> > > >> also be fixed immediately as well, so this function might be
> > > >> optional and only useful for pre-exist corrupted dumps.
> > > >
> > > > As far as fixing xfsdump -- wasn't XFS_BULK_IREQ_SPECIAL_ROOT supposed
> > > > to solve that problem by enabling dump to discover it it's really been
> > > > passed the fs root directory?
> > >
> > > Yes, but as I understand it this patch is to allow the user to recover
> > > from an already corrupted dump, at restore time, right?
> > 
> > Right, though I still haven't seen any patches to dump to employ
> > XFS_BULK_IREQ_SPECIAL_ROOT to avoid spitting out bad dumps in the first
> > place.  I think the heuristic that we applied is probably good enough,
> > but we might as well query the kernel when possible.
> > 
> > > This still feels like deep magic in xfsdump that most people struggle
> > > to understand, but it seems clear to me that the changes here are truly
> > > isolated to the new "-x" option - IOWs if "-x" is not specified, there is
> > > no behavior change at all.
> > >
> > > Since this is intended to attempt recovery from an already-corrupted
> > > dump image as a last resort, and given that there are already some xfstests
> > > in place to validate the behavior, I feel reasonably comfortable with
> > > merging this.
> > 
> > Documentation nit: Can restore detect that it's been given a corrupt
> > dump, and if so, should it warn the user to rerun with -x?
> > 
> > --D
> 
> I am assuming that even though you have concerns about not having
> XFS_BULK_IREQ_SPECIAL_ROOT employed in dump yet (and the documentation nit :),
> you are not opposed to have this patch merged?

Correct.  Consider this my somewhat passive-aggressive prod not to
forget to improve dump's "I'm doing something stupid" detector.

--D

> 
> -- 
> Carlos Maiolino