Re: [PATCH] simplify b_inode usage

[Andrew Morton]

Christoph Hellwig wrote:
> 
> Current the b_inode of struct buffer_head is a pointer to an inode, but
> it only always used as bool value.  This patch changes it to an simple
> int (yes, I know some people have ideas for a flag that uses less space,
> but that can be easily done ontop of this cleanup).  The advantage is
> that we don't have to pass in the inode into buffer_insert_inode_queue/
> buffer_insert_inode_data_queue and can merge them into a more general
> buffer_insert_list, with inline wrappers around it.  reiserfs can now
> use buffer_insert_list directly and embedd a simple list_head instead
> of a full static inode into it's journal.
> 
> A similar cleanup has already been done in early 2.5, but the b_inode
> flag is completly gone there now.
> 

Current ext3 CVS (ie: 2.4.20 candidate code) is using b_inode
as an inode *.   Stephen has acked a proposal to stop doing that,
but let's double check with him first.

Also, Joe Thornber needs to add another pointer to struct buffer_head
for LVM2 reasons.  If we collapse b_inode into a b_flags bit then
Joe gets his pointer for free (bh stays at 48 bytes on ia32).

So I'd suggest you just go ahead and do it that way.  (I had a patch
for that but seem to have misplaced it).