Re: GFS2 and DLM

[Steven Whitehouse <swhiteho@redhat.com>]

Hi,

On Wed, 2006-06-21 at 00:04 +1000, Nick Piggin wrote:
> Thanks.
> 
> Steven Whitehouse wrote:
> >  kernel/printk.c                    |    1 
> >  mm/filemap.c                       |    1 
> >  mm/readahead.c                     |    1 
> 
> These EXPORTs are a bit unfortunate.
> 
> BTW. you have one function that calls file_ra_state_init but never appears
> to use the initialized ra_state.
> 
> Why is gfs2_internal_read() called the "external read function" in the
> changelog?
> 
> The internal_read function doesn't look like a great candidate for passing
> a ra_state to, which invokes all the mechanism expecting a regular file
> being accessed by a user program.
> 
> It seems as though you could explicitly control readahead more optimally,
> but I don't know what the best way to do that would be. I assume Andrew
> has had a quick look and doesn't know either.
> 
> The part where you needed file_read_actor looks like pretty much a stright
> cut and paste from __generic_file_aio_read, which indicates that you might
> be exporting at the wrong level.
> 
Having thought about this a bit more and taken some advice, I've changed
this export (file_read_actor) to _GPL so that all three of the exports
you mention are now _GPL exports.

The updated tree is at:

git://git.kernel.org/pub/scm/linux/kernel/git/steve/gfs2-2.6.git

Let me know if you still feel that there are further issues to look at,

Steve.