i_version mount option
PATCH 1/1 i_version mount option
hi,
This patch apply on the top of the ext4 git tree. It adds a mount option to enable i_version feature for ext4.
Signed-off-by Jean noel Cordenner <[email protected]>
fs/ext4/super.c +5 -1
@@ -736,7 +736,7 @@ Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota, Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_quota, Opt_noquota, Opt_ignore, Opt_barrier, Opt_err, Opt_resize, Opt_usrquota, - Opt_grpquota, Opt_extents, Opt_noextents, + Opt_grpquota, Opt_extents, Opt_noextents, Opt_i_version, Opt_delayed_alloc, Opt_nodelayed_alloc, };
@@ -793,6 +793,7 @@ {Opt_noextents, "noextents"}, {Opt_delayed_alloc, "delalloc"}, {Opt_nodelayed_alloc, "nodelalloc"}, + {Opt_i_version, "i_version"}, {Opt_err, NULL}, {Opt_resize, "resize"}, };
@@ -1144,6 +1145,9 @@case Opt_noextents:clear_opt (sbi->s_mount_opt,EXTENTS );break ; +case Opt_i_version: +clear_opt (sbi->s_mount_opt,I_VERSION ); +break ;default :printk (KERN_ERR "EXT4-fs: Unrecognized mount option \"%s\" "
include/linux/ext4_fs.h +1 -0
@@ -489,6 +489,7 @@#define EXT4_MOUNT_JOURNAL_CHECKSUM 0x800000/* Journal checksums */ #define EXT4_MOUNT_JOURNAL_ASYNC_COMMIT 0x1000000/* Journal Async Commit */ #define EXT4_MOUNT_DELAYED_ALLOC 0x2000000/* Delayed allocation support*/ +#define EXT4_MOUNT_I_VERSION 0x4000000/* i_version support */ /* Compatibility, for having both ext2_fs.h and ext4_fs.h included at once */ #ifndef _LINUX_EXT2_FS_H
fs/ext4/inode.c +3 -1
@@ -3107,7 +3107,9 @@ {int err =0 ; - inode->i_version++; +if (test_opt (inode->i_sb,I_VERSION )) + inode->i_version++; +/* the do_update_inode consumes one bh->b_count */ get_bh (iloc->bh);
Quoted 5 lines · Jean noel Cordenner
hi,
This patch apply on the top of the ext4 git tree. It adds a mount option to enable
i_versionfeature for ext4.
I think we should fix overwrite case first. Per lkml discussion, if we fix the overwrite case by updating the counter in file_update_time() all the time, then the concern is this will impact other filesystems that don't want inode versioning support.
I believe that's how leads to put this patch back (adding a mount option).
It's better to have a VFS level super block flag to indicating if the filesystem as inode versioning support or not.
Thanks
Mingming
Quoted 2 lines
Signed-off-by Jean noel Cordenner <[email protected]>