[xfs-masters] Re: 2.6.22-rc1-mm1

[Nathan Scott <nscott@xxxxxxxxxx>]

On Tue, 2007-05-22 at 20:44 +1000, David Chinner wrote:
> 
> > xfs_buf_associate_memory is a mess.  My original plan was to get rid
> of
> > it, but I kept that out to keep that patchset small and easily
> reviable,
> > but it seems like that was a mistake.  My plan is the following:
> > 
> >  - xlog_bread and thus the whole buffer I/O path grows an iooffset
> >    paramater that specifies at which offset into the buffer we start
> >    the actual I/O.  That gets rid of all the
> xfs_buf_associate_memory
> >    memory uses in the log recovery code
> 
> Perhaps a new field in the xfs_buf structure - that way call paths
> don't need to grow extra parameters and potentially increase

Thatd be unfortunate - there are very few iclog buffers relative to
every other metadata buffer, so growing the struct for all of those
too would not be ideal (I remember Steve going on pagebuf shrinking
exercises in the distant past, to fit more of em in memory at once,
I can't remember what benchmark in particular he was using though).

cheers.

-- 
Nathan