From: Greg Banks Subject: Re: [PATCH] knfsd: remove unnecessary BUG_ON's Date: Sat, 15 Sep 2007 02:58:01 +1000 Message-ID: <20070914165801.GJ25610@sgi.com> References: <20070912234317.GF13792@fieldses.org> <20070914164058.GK21965@sgi.com> <20070914165013.GE26622@fieldses.org> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Cc: Neil Brown , nfs@lists.sourceforge.net To: "J. Bruce Fields" Return-path: Received: from sc8-sf-mx2-b.sourceforge.net ([10.3.1.92] helo=mail.sourceforge.net) by sc8-sf-list2-new.sourceforge.net with esmtp (Exim 4.43) id 1IWEUA-0001AB-SD for nfs@lists.sourceforge.net; Fri, 14 Sep 2007 09:57:42 -0700 Received: from netops-testserver-4-out.sgi.com ([192.48.171.29] helo=relay.sgi.com) by mail.sourceforge.net with esmtp (Exim 4.44) id 1IWEUF-0008TT-JO for nfs@lists.sourceforge.net; Fri, 14 Sep 2007 09:57:47 -0700 In-Reply-To: <20070914165013.GE26622@fieldses.org> List-Id: "Discussion of NFS under Linux development, interoperability, and testing." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: nfs-bounces@lists.sourceforge.net Errors-To: nfs-bounces@lists.sourceforge.net On Fri, Sep 14, 2007 at 12:50:13PM -0400, J. Bruce Fields wrote: > On Sat, Sep 15, 2007 at 02:40:58AM +1000, Greg Banks wrote: > > On Wed, Sep 12, 2007 at 07:43:17PM -0400, J. Bruce Fields wrote: > > > Trivial patch, but--any objection to removing these assertions? > > > > > > We set svsk->sk_pool = pool just 30-some lines above, and there's no > > > obvious reason this should have changed, or any other particular reason > > > to assert this here. > > > > Yeah, you'd think that. However we saw these BUG_ON()s tripped > > when attempting to port the 2.6.18-based NFS/RDMA patches to a > > 2.6.19-like kernel. All it takes is for some transport code to get > > confused about the proper way to clear SK_BUSY, and *blam*. So I > > think they're useful. > > OK, I'll take your word for it--dropped. (But, I'm just curious--do you > remember what the sequence of events was?) It's been a while, but IIRC the new transport code had copied out the internals of svc_request_received() pre .19, and were manually clearing SK_BUSY and calling svc_sock_enqueue(). In .19 when sk_pool was added, svc_request_received() began setting sk_pool to NULL, and the transport code wasn't doing that. Or maybe that was a different failure mode ;-) Greg. -- Greg Banks, R&D Software Engineer, SGI Australian Software Group. Apparently, I'm Bedevere. Which MPHG character are you? I don't speak for SGI. ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs