2007-09-18 17:09:53

by Avantika Mathur

[permalink] [raw]
Subject: Ext4 devel interlock meeting minutes (September 17, 2007)

Ext4 Developer Interlock Call: September 17, 2007 meeting minutes
Attendees: Mingming Cao, Andreas Dilger, Eric Sandeen, Dave Kleikamp,
Ted Ts'o,
Avantika Mathur

Minutes can be accessed at:

Reviewed patches in the patch queue, identifying those that are ready to
be merged with 2.6.24-rc1

- Journal_Checksum: Not yet ready for mainline, Avantika and Girish are
looking into kernel oops on fsstress.

- Uninitialized Block Groups: Ready to be pushed upstream, Avantika
will post this patch to lkml

- Large Block Support: Mingming tested these patches on various
architectures. There should be new testcases added to e2fsprogs to test
large block sizes. Andreas also described a possible error case with
empty directories. Otherwise the feature is ready to be pushed to lkml

- JBD Cleanups: Ready to be pushed upstream

- Flex_BG Patches: should be pushed upstream asap. The required flag in
e2fsprogs is already reserved for this feature.

- Delalloc: Patch should be sent to lkml for detailed review, since not
only ext4 code is being touched. Delalloc will be pushed with the other
patches that are being pushed upstream now.
- Shaggy will look into whether the delalloc framework can be used
for JFS
- We will also try porting the patch to ext3, with a search/replace,
and try testing on ext3.

- Mballoc: Not ready for 2.6.23. We will wait for the next release, as
it hasn't been in the patch queue for very long. The main issue is to
break the patch into smaller pieces making it easier to understand.
This is tricky to break such a large patch into smaller patches.

- Mingming will re-order the patch queue, putting the patches that are
ready to be pushed upstream at the top.

- Ted sent out his initial extents support patches to the mailing list;
he is continuing on ToDo items which includes a lot of error checking,
and waiting for feedback on the patches. In his patches, any knowledge
of the on-disk format of the filesystem is isolated to a small part of
the library, making it easier to change format in the future.

- 64-bit support in e2fsprogs is background priority, while adding
support for other ext4 features (e.g. unintialized block groups) is
higher priority right now. Some of Valerie's pre-work/cleanup patches
are ready to be merged now.

- Avantika and Aneesh are working on automated testing of e2fsprogs.