[PATCH, RFC] ext4: Remove i_ext_generation from ext4_inode_info structure

[PATCH, RFC] ext4: Remove i_ext_generation from ext4_inode_info structure

[Theodore Ts'o]

ext4_inode_info.i_ext_generation isn't used by anything.  Was there an
intent to use it for something in the future?

       	      	     	       	      - Ted

commit 51972792db1397cddd97e1ab9c8d2604453f5b4c
Author: Theodore Ts'o <tytso@mit.edu>
Date:   Tue Nov 4 18:46:03 2008 -0500

    ext4: Remove i_ext_generation from ext4_inode_info structure
    
    The i_ext_generation was incremented, but never used.  Remove it to
    slim down the ext4_inode_info structure.
    
    Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

diff --git a/fs/ext4/ext4_extents.h b/fs/ext4/ext4_extents.h
index bec7ce5..18cb67b 100644
--- a/fs/ext4/ext4_extents.h
+++ b/fs/ext4/ext4_extents.h
@@ -194,11 +194,6 @@ static inline unsigned short ext_depth(struct inode *inode)
 	return le16_to_cpu(ext_inode_hdr(inode)->eh_depth);
 }
 
-static inline void ext4_ext_tree_changed(struct inode *inode)
-{
-	EXT4_I(inode)->i_ext_generation++;
-}
-
 static inline void
 ext4_ext_invalidate_cache(struct inode *inode)
 {
diff --git a/fs/ext4/ext4_i.h b/fs/ext4/ext4_i.h
index 5c124c0..acc0b72 100644
--- a/fs/ext4/ext4_i.h
+++ b/fs/ext4/ext4_i.h
@@ -117,7 +117,6 @@ struct ext4_inode_info {
 	struct inode vfs_inode;
 	struct jbd2_inode jinode;
 
-	unsigned long i_ext_generation;
 	struct ext4_ext_cache i_cached_extent;
 	/*
 	 * File creation time. Its function is same as that of
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index ea2ce3c..501952f 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -1622,7 +1622,6 @@ cleanup:
 		ext4_ext_drop_refs(npath);
 		kfree(npath);
 	}
-	ext4_ext_tree_changed(inode);
 	ext4_ext_invalidate_cache(inode);
 	return err;
 }
@@ -2233,7 +2232,6 @@ static int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start)
 		}
 	}
 out:
-	ext4_ext_tree_changed(inode);
 	ext4_ext_drop_refs(path);
 	kfree(path);
 	ext4_journal_stop(handle);

Re: [PATCH, RFC] ext4: Remove i_ext_generation from ext4_inode_info structure

[Andreas Dilger]

On Nov 04, 2008  18:47 -0500, Theodore Ts'o wrote:
> ext4_inode_info.i_ext_generation isn't used by anything.  Was there an
> intent to use it for something in the future?

Alex was going to use this at one point in the future, but it never
happened.

Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.

Re: [PATCH, RFC] ext4: Remove i_ext_generation from ext4_inode_info structure

[Alex Zhuravlev]

Andreas Dilger wrote:
> On Nov 04, 2008  18:47 -0500, Theodore Ts'o wrote:
>> ext4_inode_info.i_ext_generation isn't used by anything.  Was there an
>> intent to use it for something in the future?
> 
> Alex was going to use this at one point in the future, but it never
> happened.

actually it was used by delayed allocation patch as walk_space() takes
truncate_sem, then, if block isn't allocated yet, start transaction which
we can't do holding truncate_sem. so the idea is to release truncate_sem,
start transaction, then acquire truncate_sem. but in this window the tree
can change making path structure obsolete. either we check tree hasn't
changed or use generation to avoid unneeded path revalidation?

thanks, Alex