[PATCH 2.6.7] ext3 s_dirt for r/w

[PATCH 2.6.7] ext3 s_dirt for r/w

[FabF]

[-- Attachment #1: Type: text/plain, Size: 163 bytes --]

Andrew,

	Here is a patch setting s_dirt for read-write filesystems in ext3_init
(doing it in create_journal seems troublesome IMHO).

PS: untested

Regards,
FabF

[-- Attachment #2: s_dirt1.diff --]
[-- Type: text/x-patch, Size: 651 bytes --]

diff -Naur orig/fs/super.c edited/fs/super.c
--- orig/fs/super.c	2004-06-16 07:20:03.000000000 +0200
+++ edited/fs/super.c	2004-06-19 19:58:19.895637880 +0200
@@ -1569,7 +1569,10 @@
 		ext3_count_free_inodes(sb));
 	percpu_counter_mod(&sbi->s_dirs_counter,
 		ext3_count_dirs(sb));
-
+	if (!(sb->s_flags & MS_RDONLY)) {
+		mark_buffer_dirty(sbi->s_sbh);
+		sb->s_dirt = 1;
+	}
 	return 0;
 
 failed_mount3:
@@ -1841,7 +1844,6 @@
 	EXT3_SET_COMPAT_FEATURE(sb, EXT3_FEATURE_COMPAT_HAS_JOURNAL);
 
 	es->s_journal_inum = cpu_to_le32(journal_inum);
-	sb->s_dirt = 1;
 
 	/* Make sure we flush the recovery flag to disk. */
 	ext3_commit_super(sb, es, 1);

Re: [PATCH 2.6.7] ext3 s_dirt for r/w

[Andrew Morton]

FabF <fabian.frederick@skynet.be> wrote:
>
> 	Here is a patch setting s_dirt for read-write filesystems in ext3_init
>  (doing it in create_journal seems troublesome IMHO).

Why?

>  PS: untested

Please don't send untested patches.

>  diff -Naur orig/fs/super.c edited/fs/super.c
>  --- orig/fs/super.c	2004-06-16 07:20:03.000000000 +0200
>  +++ edited/fs/super.c	2004-06-19 19:58:19.895637880 +0200

fs/super.c does not contain ext3 code.

Re: [PATCH 2.6.7] ext3 s_dirt for r/w

[FabF]

On Sat, 2004-06-19 at 20:54, Andrew Morton wrote:
> FabF <fabian.frederick@skynet.be> wrote:
> >
> > 	Here is a patch setting s_dirt for read-write filesystems in ext3_init
> >  (doing it in create_journal seems troublesome IMHO).
> 
> Why?
ext3_create_journal is called with journal_inum parser option set which
means unjournaled r/w ext3 fs are s_dirt 0.
> 
> >  PS: untested
> 
> Please don't send untested patches.
Ok, tested.No problem here.

Re: [PATCH 2.6.7] ext3 s_dirt for r/w

[FabF]

On Sat, 2004-06-19 at 21:32, FabF wrote:
> On Sat, 2004-06-19 at 20:54, Andrew Morton wrote:
> > FabF <fabian.frederick@skynet.be> wrote:
> > >
> > > 	Here is a patch setting s_dirt for read-write filesystems in ext3_init
> > >  (doing it in create_journal seems troublesome IMHO).
> > 
> > Why?
> ext3_create_journal is called with journal_inum parser option set which
> means unjournaled r/w ext3 fs are s_dirt 0.
Oops ... ^has_journal is only ext2 readable so there's no problem at
that point ...Anyway, valid ext3 partition can be initialized with
load_journal or create_journal so s_dirt problem remains IMHO...

Regards,
FabF