Re: [RFC 1/2] ext3: enlarge blocksize and fix rec_len overflow

[Andreas Dilger <adilger@clusterfs.com>]

On Jun 29, 2006  11:10 -0700, Daniel Phillips wrote:
> Andreas Dilger wrote:
> >On Jun 28, 2006  17:50 +0200, Johann Lombardi wrote:
> >>ext2/ext3_dir_entry_2 has a 16-bit entry(rec_len) and it would overflow
> >>with 64KB blocksize.  This patch prevent from overflow by limiting
> >>rec_len to 65532.
> >
> >Having a max rec_len of 65532 is rather unfortunate, since the dir
> >blocks always need to filled with dir entries.  65536 - 65532 = 4, and
> >the minimum ext3_dir_entry size is 8 bytes.  I would instead make this
> >maybe 64 bytes less so that there is room for a filename in the "tail"
> >dir_entry.
> 
> Then why not introduce a little symmetry by making max rec_len 2**15 and
> treat big directory blocks as an array of smaller ones?  I dimly recall
> the page-cache oriented Ext2 dir code already does this.

I have no objection to this at all, but I think it will lead to a slightly
more complex implementation.  We even discussed in the distant past to
make large directories a series of 4kB "chunks", for fs blocksize >= 4kB.
This has negative implications for large filenames because the internal
free space fragmentation is high, but has the advantage that it might
eventually still be usable if we can get blocksize > PAGE_SIZE.

The difficulty is that when freeing dir entires you would have to be
concerned with a merging a dir_entry that is spanning the middle
of a 2^16 block.

Cheers, Andreas
--
Andreas Dilger
Principal Software Engineer
Cluster File Systems, Inc.