ReiserFS v3 changesets resend

Hello!

It seems our first attempt of sending these to Linus have failed, here is another one.

Bye,
Oleg

reiserfs updates:

These first two changesets contain fixes for reiserfs. They fix issue with handling of displacing_large_files allocator option and a problem with remounting from readwrite to readwrite mode if FS holds some deleted but not yet closed files.

Next three changesets implement reiserfs_file_write. Also third one exports generic_osync_inode,block_commit_write and remove_suid because these are now needed for reiserfs.
There was no reiserfs_file_write in the 2.4 port of reiserfs (and Hans was very unhappy about it).
With current 'one block at a time' algorithm, writes past the end of a file are slow because each new file block is separately added into the tree causing shifting of other items which is CPU expensive.
With this new implementation if you write into file with big enough chunks, it uses half as much CPU. Also this version is more SMP friendly than the current one.

Next four changesets replace recently added lock_kernels with reiserfs wrappers (that would eventually evolve into real separate locks), includes more C99 designated initialisers cleanups (this time from Art Haas) and updates some reiserfs help entries. Also fixing a buglet in reiserfs_file_write code discovered during some 2.4.20-pre testing.

You can pull these from bk://thebsh.namesys.com/bk/reiser3-linux-2.5

Diffstats:

 fs/reiserfs/inode.c         |    2 +-
 include/linux/reiserfs_fs.h |    3 ++-
 2 files changed, 3 insertions(+), 2 deletions(-)

 super.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletion(-)

 do_balan.c        |  105 +++++++++++++++++++++++++++---------------------------
 inode.c           |   48 ++++++++++++++++++++----
 tail_conversion.c |    5 +-
 3 files changed, 95 insertions(+), 63 deletions(-)

 fs/reiserfs/bitmap.c           |   21
 fs/reiserfs/file.c             | 1080 ++++++++++++++++++++++++++++++++++++++++-
 fs/reiserfs/inode.c            |    4
 fs/reiserfs/super.c            |    1
 include/linux/reiserfs_fs.h    |    1
 include/linux/reiserfs_fs_sb.h |    1
 6 files changed, 1104 insertions(+), 4 deletions(-)

 kernel/ksyms.c |    3 +++
 mm/filemap.c   |    2 +-
 2 files changed, 4 insertions(+), 1 deletion(-)

 journal.c |    5 ++---
 1 files changed, 2 insertions(+), 3 deletions(-)

 dir.c   |    6 +++---
 inode.c |   14 +++++++-------
 namei.c |   18 +++++++++---------
 super.c |   30 +++++++++++++++---------------
 4 files changed, 34 insertions(+), 34 deletions(-)

 Config.help |   33 ++++++++++++---------------------
 1 files changed, 12 insertions(+), 21 deletions(-)

 file.c |    2 +-
 1 files changed, 1 insertion(+), 1 deletion(-)

[email protected], 2002-10-02 22:16:12+04:00, [email protected]
Small buglet in reiserfs_file_write fixed, that was found during 2.4 stage testing, but somehow it was lost in 2.5 version

[email protected], 2002-10-01 17:29:34+04:00, [email protected]
Update reiserfs config help entries.

[email protected], 2002-10-01 17:13:13+04:00, [email protected]
reiserfs: C99 designated initializers, by Art Haas

[email protected], 2002-10-01 17:12:19+04:00, [email protected]
lock_kernel is replaced with per superblock lock (kind of)

[email protected], 2002-10-01 17:11:32+04:00, [email protected]
export generic_osync_inode,block_commit_write, remove_suid

[email protected], 2002-10-01 17:09:56+04:00, [email protected]
reiserfs_file_write() implemenation. Ported from 2.4

[email protected], 2002-10-01 17:08:55+04:00, [email protected]
reiserfs: Implement insertion of more than one unformatted pointer insertion at a time. Considerably speedup hole creation.

[email protected], 2002-10-01 17:07:40+04:00, [email protected]
reiserfs: Fix a problem with delayed unlinks and remounting RW filesystem RW.

[email protected], 2002-10-01 17:06:40+04:00, [email protected]
reiserfs: Take into account file information even when not doing preallocation. Fixes a bug with displacing_large_files option.

You guys need to use a better format for emailing Linus... see his posts on the subject, or just use Documentation/BK-usage/bk-make-sum script. Having N diffstat outputs followed by N changeset descriptions is just non-sensical. Linus should not need to count down diffstats to figure out which one corresponds to which changeset.

Further, an overall [single] diffstat gives a better picture of what he will pull when he does a 'bk pull'...

Jeff

Hans,

Please post your patches to lkml in one big file so that we can apply the patch. Don't know how to access bk://thebsh.namesys.com/bk/reiser3-linux-2.5

Thanks,
Jeff.

On Wed, 9 Oct 2002, Hans Reiser wrote:
Hello!

It seems our first attempt of sending these to Linus have failed, here is another one.

Bye,
Oleg

reiserfs updates:

Oleg sent only the changesets on the resend, since he figured that everyone but Linus would have applied it after the first send, and they could look through the archives if not. I sort of suspected that was too logical to be true.;-) We'll send the full patch in the future. I don't really like the idea of doing anything in a way that makes people feel disadvantaged unless they use bk. Oleg, send Jeff the patch.

Hans