[pnfs] [PATCH] Removed sessions hard-coded 32678 byte max sizes

[Marc Eshel]

I think that the priority should be show a complete implementation for the 
bakeathon so the draft can verify that the draft is complete and than go 
back and reorganize the code for future development on latest kernels.
Marc.




"William A. (Andy) Adamson" <andros at citi.umich.edu> 
Sent by: pnfs-bounces at linux-nfs.org
03/16/2007 10:03 AM

To
"J. Bruce Fields" <bfields at fieldses.org>
cc
"Iyer, Rahul" <Rahul.Iyer at netapp.com>, pnfs at linux-nfs.org
Subject
Re: [pnfs] [PATCH] Removed sessions hard-coded 32678 byte max sizes








On 3/16/07, J. Bruce Fields <bfields at fieldses.org> wrote:
On Fri, Mar 16, 2007 at 09:44:35AM -0700, Iyer, Rahul wrote:
> Btw, isn't largeio part of the mainline now?

Yeah, so there's a decision to make here:

We could update our current code to something more recent.  That would 
have the advantage that we could dump some temporary fixes like that
bigio patch.  But it would have the disadvantage that it will require
some tedious fixing up, and may introduce a new bug or two.  Also, it's 
more work sunk into something that (as long as it's not reviewed and
split up into small patches) looks like a bit of a dead end.

So I'm tempted by the idea of freezing this prototype at 2.6.18, and
saying that any new development start from the ground up with the latest
kernel.  I liked Trond's idea of starting with a patch that literally
just clones the nfs_rpc_ops struct.  Interested in starting that?


last phone conf rahul said he would send out his 'battle plan' for getting 
client sessions into the kernel after the client slot table and callback 
channel code etc. needed for 4.1 bakeathon testing was added to the 
current 2.6.18.3 tree.

we need to address 4.1 bakeathon testing first - agreed?

-->Andy
--b.

_______________________________________________
pNFS mailing list
pNFS at linux-nfs.org
http://linux-nfs.org/cgi-bin/mailman/listinfo/pnfs