The patch titled
inodes forever dirty fix
has been removed from the -mm tree. Its filename is
inodes-forever-dirty-fix.patch
This patch was dropped because it is obsolete
------------------------------------------------------
Subject: inodes forever dirty fix
From: Andreas Gruenbacher <agruen@xxxxxxx>
inodes like /dev/ram0 have a double life both as device inodes, and as inodes
describing a ramdisk. This has lead to a bug where inodes would clog the
s_io list of a superblock, and no further inodes would make it to disk. Fixed
in [1]. Later, [2] re-introduced an optimization that [1] removed. The result
is that some inodes remain dirty forever, bouncing between s_io and s_dirty.
This happens as follows:
- An inode like /dev/ram0 is opened, which sets BDI_CAP_NO_WRITEBACK in
the inode's backing_dev_info->capabilities. bdi_cap_writeback_dirty
becomes false.
- The inode changes, e.g., chmod 0660 /dev/ram0, so the inode goes on
s_dirty.
- sync (which speeds up things, but makes no functional difference).
- sync_sb_inodes() moves it from s_dirty to s_io, then finds it cannot
writepage the inode, and moves it back to s_dirty. It will see the
same inode again at some later point.
- The inode never gets to __writeback_single_inode(), and never becomes
clean again.
Signed-off-by: Andrew Morton <akpm@xxxxxxxx>
---
fs/fs-writeback.c | 8 +-------
1 files changed, 1 insertion(+), 7 deletions(-)
diff -puN fs/fs-writeback.c~inodes-forever-dirty-fix fs/fs-writeback.c
--- 25/fs/fs-writeback.c~inodes-forever-dirty-fix Mon Oct 24 16:35:44 2005
+++ 25-akpm/fs/fs-writeback.c Mon Oct 24 16:38:35 2005
@@ -320,20 +320,14 @@ sync_sb_inodes(struct super_block *sb, s
long pages_skipped;
if (!bdi_cap_writeback_dirty(bdi)) {
- list_move(&inode->i_list, &sb->s_dirty);
if (sb == blockdev_superblock) {
/*
* Dirty memory-backed blockdev: the ramdisk
* driver does this. Skip just this inode
*/
+ list_move(&inode->i_list, &sb->s_dirty);
continue;
}
- /*
- * Dirty memory-backed inode against a filesystem other
- * than the kernel-internal bdev filesystem. Skip the
- * entire superblock.
- */
- break;
}
if (wbc->nonblocking && bdi_write_congested(bdi)) {
_
Patches currently in -mm which might be from agruen@xxxxxxx are
origin.patch
git-kbuild.patch
-
To unsubscribe from this list: send the line "unsubscribe mm-commits" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
