Re: [RFC][E2FSPROGS][PATCH 1/8]
Andreas Dilger <[email protected]>
| Newsgroups | gmane.comp.file-systems.ext2.devel |
|---|---|
| Message-ID | <[email protected]> |
On May 29, 2006 17:13 +0200, Alexandre Ratchov wrote: > On Wed, May 24, 2006 at 03:52:13PM -0600, Andreas Dilger wrote: > > On May 24, 2006 16:36 +0200, Alexandre Ratchov wrote: > > > move the new 'bg_flags' field of the ext2_group_desc structure in > > > order to make room for the MSB of the 64bit fields. > > > > Since the BIG_BG support is a mke2fs-time option (i.e. it is impossible > > to convert an existing filesystem to this layout) Laurent and I had > > previously discussed changing the ext3_group_desc struct to be 256 bytes > > in size in order to more easily accomodate the larger fields. > > > > The reason I'm concerned about using up these fields in the existing > > group descriptor is that some other features such as inode and block > > bitmap checksums, the group descriptor checksum, and the itable "high > > watermark" value (which could be enabled on existing filesystems) all > > need space in the existing group descriptor also. > > > > That is at least 4 + 4 + 1 + 2 = 11 bytes, and there are only 7 left > > after the BIG_BG changes... > > I like the idea of changing descriptor size; perhaps we could allow group > descriptors to have variable size (in an analogous way to the inode variable > size)? Sure - have a superblock field which indicates the descriptor size, given that this is already an incompatible format-time change. That's a great idea. Cheers, Andreas -- Andreas Dilger Principal Software Engineer Cluster File Systems, Inc. ------------------------------------------------------- All the advantages of Linux Managed Hosting--Without the Cost and Risk! Fully trained technicians. The highest number of Red Hat certifications in the hosting industry. Fanatical Support. Click to learn more http://sel.as-us.falkag.net/sel?cmd=lnk&kid=107521&bid=248729&dat=121642