From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Sun, 1 Apr 2012 09:55:39 -0400 From: Jeff Layton To: Steve French Cc: Greg KH , linux-kernel@vger.kernel.org, stable@vger.kernel.org, torvalds@linux-foundation.org, akpm@linux-foundation.org, alan@lxorguk.ukuu.org.uk, Pavel Shilovsky , linux-cifs@vger.kernel.org Subject: Re: [ 105/175] CIFS: Respect negotiated MaxMpxCount Message-ID: <20120401095539.31323c95@corrin.poochiereds.net> In-Reply-To: References: <20120330195801.GA31806@kroah.com> <20120330194846.517484801@linuxfoundation.org> <20120331201825.036fe1d8@corrin.poochiereds.net> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8BIT Sender: linux-kernel-owner@vger.kernel.org List-ID: On Sun, 1 Apr 2012 02:26:58 -0500 Steve French wrote: > On Sat, Mar 31, 2012 at 7:18 PM, Jeff Layton wrote: > > On Fri, 30 Mar 2012 12:50:10 -0700 > > Greg KH wrote: > > > >> 3.3-stable review patch.  If anyone has any objections, please let me know. > >> > > > > I'm not sure it's a good idea to put this patch into any stable release > > just yet. I think it's a correct and necessary fix, but it has some > > potential to cause regressions too (or uncover other preexisting > > problems with this code). Might it be best to wait until we have some > > more experience with this before we push it into stable? > > Delaying it a little is probably ok, but weigh that against the > frequency/severity > of the problem that Pavel's fix addresses: ie sequential writes to some > versions of Windows 7 and Windows Vista can fail partway > through large file copy (which also causes those servers > to become unresponsive to subsequent connection requests > as well) so the fix does address a hot problem. > > > I'm not disputing whether this patch is correct. It's clearly a bug that cifs.ko has never respected this value, and it's quite problematic with certain servers. The problem is that we don't have any confirmation that: a) this fixes any of the problems that we think it will b) this doesn't introduce any regressions The request slot allocation code is quite fiddly and fragile, and I think the potential for "b" is somewhat high. I'd feel more comfortable if we waited until we have more experience with this patch before merging it into stable. -- Jeff Layton