RE: [PATCH v2 07/25] pnfsblock: add blocklayout Kconfig option, Makefile, and stubs

From: "Myklebust, Trond" <Trond.Myklebust@netapp.com>
To: "Benny Halevy" <bhalevy@tonian.com>
Cc: "Jim Rees" <rees@umich.edu>, <linux-nfs@vger.kernel.org>,
	"peter honeyman" <honey@citi.umich.edu>
Subject: RE: [PATCH v2 07/25] pnfsblock: add blocklayout Kconfig option, Makefile, and stubs
Date: Mon, 25 Jul 2011 10:25:38 -0700	[thread overview]
Message-ID: <2E1EB2CF9ED1CB4AA966F0EB76EAB4430A517C9D@SACMVEXC2-PRD.hq.netapp.com> (raw)
In-Reply-To: <4E2D82AB.6060502@tonian.com>

> -----Original Message-----
> From: Benny Halevy [mailto:bhalevy@tonian.com]
> Sent: Monday, July 25, 2011 10:50 AM
> To: Myklebust, Trond
> Cc: Jim Rees; linux-nfs@vger.kernel.org; peter honeyman
> Subject: Re: [PATCH v2 07/25] pnfsblock: add blocklayout Kconfig
> option, Makefile, and stubs
> 
> On 2011-07-25 10:38, Myklebust, Trond wrote:
> >> -----Original Message-----
> >> From: Benny Halevy [mailto:bhalevy@tonian.com]
> >> Sent: Monday, July 25, 2011 10:31 AM
> >> To: Jim Rees
> >> Cc: Myklebust, Trond; linux-nfs@vger.kernel.org; peter honeyman
> >> Subject: Re: [PATCH v2 07/25] pnfsblock: add blocklayout Kconfig
> >> option, Makefile, and stubs
> >>
> >> On 2011-07-21 15:34, Jim Rees wrote:
> >>> From: Fred Isaman <iisaman@citi.umich.edu>
> >>>
> >>> Define a configuration variable to enable/disable compilation of
> the
> >>> block driver code.
> >>>
> >>> Add the minimal structure for a pnfs block layout driver, and
empty
> >>> list-heads that will hold the extent data
> >>>
> >>> [pnfsblock: make NFS_V4_1 select PNFS_BLOCK]
> >>> Signed-off-by: Peng Tao <peng_tao@emc.com>
> >>> Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
> >>> Signed-off-by: Benny Halevy <bhalevy@panasas.com>
> >>> [pnfs-block: fix CONFIG_PNFS_BLOCK dependencies]
> >>> Signed-off-by: Benny Halevy <bhalevy@panasas.com>
> >>> Signed-off-by: Benny Halevy <benny@tonian.com>
> >>> [pnfsblock: SQUASHME: port block layout code]
> >>> Signed-off-by: Peng Tao <peng_tao@emc.com>
> >>> [pnfsblock: SQUASHME: adjust to API change]
> >>> Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
> >>> [pnfs: move pnfs_layout_type inline in nfs_inode]
> >>> Signed-off-by: Benny Halevy <bhalevy@panasas.com>
> >>> [blocklayout: encode_layoutcommit implementation]
> >>> Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
> >>> Signed-off-by: Benny Halevy <bhalevy@panasas.com>
> >>> Signed-off-by: Benny Halevy <benny@tonian.com>
> >>> [pnfsblock: layout alloc and free]
> >>> Signed-off-by: Fred Isaman <iisaman@citi.umich.edu>
> >>> [pnfs: move pnfs_layout_type inline in nfs_inode]
> >>> Signed-off-by: Benny Halevy <bhalevy@panasas.com>
> >>> Signed-off-by: Benny Halevy <benny@tonian.com>
> >>> [pnfsblock: define module alias]
> >>> Signed-off-by: Peng Tao <peng_tao@emc.com>
> >>> ---
> >>>  fs/nfs/Kconfig                   |    8 ++-
> >>>  fs/nfs/Makefile                  |    1 +
> >>>  fs/nfs/blocklayout/Makefile      |    5 +
> >>>  fs/nfs/blocklayout/blocklayout.c |  175
> >> ++++++++++++++++++++++++++++++++++++++
> >>>  fs/nfs/blocklayout/blocklayout.h |   91 ++++++++++++++++++++
> >>>  5 files changed, 279 insertions(+), 1 deletions(-)
> >>>  create mode 100644 fs/nfs/blocklayout/Makefile
> >>>  create mode 100644 fs/nfs/blocklayout/blocklayout.c
> >>>  create mode 100644 fs/nfs/blocklayout/blocklayout.h
> >>>
> >>> diff --git a/fs/nfs/Kconfig b/fs/nfs/Kconfig
> >>> index 2cde5d9..be02077 100644
> >>> --- a/fs/nfs/Kconfig
> >>> +++ b/fs/nfs/Kconfig
> >>> @@ -79,15 +79,21 @@ config NFS_V4_1
> >>>  	depends on NFS_FS && NFS_V4 && EXPERIMENTAL
> >>>  	select SUNRPC_BACKCHANNEL
> >>>  	select PNFS_FILE_LAYOUT
> >>> +	select PNFS_BLOCK
> >>> +	select MD
> >>> +	select BLK_DEV_DM
> >>
> >> Why is PNFS_BLOCK enabled automatically in all cases?
> >> That renders the use of modules for layout drivers totally useless.
> >> I sort of understand that for PNFS_FILE_LAYOUT (when my
> >> arm is twisted really hard behind my back :) since it
> >> is an integral part of RFC5661 but what's the justification
> >> for PNFS_BLOCK? and why blocks and not objects?
> >
> > The question is rather why did objects add a selectable compile
> option?
> 
> Just good citizenship :)
> 
> > What is the point of not compiling a given layout driver if all the
> > dependencies are met?
> 
> Reducing build times...
> Building a smaller kernel when modules are disabled...

You can add a line with
	depends on m

to ensure that it is always compiled as a module. I think that might be
a good thing until we have nailed down all the issues with pNFS.

> We're fine in terms of memory consumption when CONFIG_MODULES=y since
> the
> layout driver is loaded on demand but shouldn't be worried about
> the other case?
> 
> >
> > IOW: The only thing I'd change above is the select MD and select
> > BLK_DEV_DM: I'd prefer something like
> >
> > config PNFS_BLOCK
> > 	depends on NFS_V4_1 && MD && BLK_DEV_DM
> > 	default y
> 
> This is closer to the original version.
> However, selecting MD and BLK_DEV_DM was proven useful to
automatically
> take
> care of the module dependencies without having to dive into details.

Yes, but since the MD is a completely different layer that is not under
our control (well, OK, Neil is still an NFS maintainer and an MD
maintainer) then I'd prefer to leave it as a dependency.

We can always add something like

comment
	depends on NFS_V4_1 && !BLK_DEV_DM
	Please enable BLK_DEV_MD if you wish to enable the pNFS block
driver.

Cheers
  Trond