From: SMTP%"ben21@advsyscon.com" 2-APR-1997 22:58:27.11 To: everhart@star.zko.dec.com CC: Subj: Re: corruption problem... Hi Glenn, Your mail message is very illuminating. I thought that was happening which is why I kept reasking the question everytime someone said "..no.. all I/Os go through START I/O." This type of behaviour was first done in a DKDRIVER in V5.4-x where the "attempt_reorder" routine did the same crap. As I remember DEC provided a patch that prevented it (the routine) from doing that. In V6.0, attempt_reorder was changed to "behave". How do we get DKDRIVER to behave? Any reason why DKDRIVER can't call through the vector (or is that the kernel stack problem you alluded to?)? I wonder how the disk caching products get around this since they also tend to use START_IO. I'm dialed in from Canada (Ottawa actually) so I've capturedyour mail message and will read it again (Sorry the disruption... TCP from Canada isn't so great). Any ideas? This problem isn't going to go away I'm afraid. Regards, Ben ================== RFC 822 Headers ================== Return-Path: ben21@advsyscon.com Received: by dimond.zko.dec.com (UCX V4.1-12A, OpenVMS V6.2 VAX); Wed, 2 Apr 1997 22:58:19 -0500 Received: from spock.advsyscon.com by mail11.digital.com (8.7.5/UNX 1.5/1.0/WV) id WAA14105; Wed, 2 Apr 1997 22:50:33 -0500 (EST) Received: from SULU (SULU::SYSTEM) by SPOCK (MX V4.1 VAX) with SMTP (DECnet); Wed, 02 Apr 1997 20:11:20 EST Received: by AdvSysCon.COM (MX V4.1 VAX) id 3; Wed, 02 Apr 1997 20:10:56 EST Date: Wed, 02 Apr 1997 20:10:56 EST From: "Ben Rosenberg, ASCI" To: everhart@star.zko.dec.com CC: ben21@AdvSysCon.COM Message-ID: <009B232A.6FD0B8A0.3@AdvSysCon.COM> Subject: Re: corruption problem...