Re: [PATCH 6/6] xfs: don't zap bmbt forks if they are MAXLEVELS tall

On Tue, Jul 14, 2026 at 08:16:52AM +0200, Christoph Hellwig wrote:
> On Mon, Jul 13, 2026 at 11:07:30PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <[email protected]>
> > 
> > LOLLM noticed a discrepancy between the bmbt level checks in the libxfs
> > bmbt code vs. the inode repair code.  We do actually allow a bmbt root
> > that proclaims to have a height of XFS_BM_MAXLEVELS.
> 
> Looks good:
> 
> Reviewed-by: Christoph Hellwig <[email protected]>
> 
> I guess we need a test that actually creates such a deep tree.  But
> we'll probably run out space / extents before..

The bmapinflate command in xfs_db does this by writing a new bmap btree
with as many mappings pointing to the same "reflinked" block as you ask.
The only problem is that to hit maxlevels, you need to create a 1k
fsblock filesystem and then bmapinflate 2^54 records, which will take a
long time and require much memory.  Back in the day when Chandan was
working on nrext64 I tried this and it took 2 days and 400G of RAM to
write the huge bmap btree out to disk and generate the appropriate
rmap/reflink btrees.

--D