Date: Sat, 16 Sep 2000 14:22:49 -0600 From: "Justin T. Gibbs" <gibbs@plutotech.com> To: "Brandon D. Valentine" <bandix@looksharp.net> Cc: stable@freebsd.org, gibbs@freebsd.org, steve@freebsd.org, jkh@freebsd.org Subject: Re: MFC of ahc driver updates Message-ID: <200009162021.OAA02336@pluto.plutotech.com> In-Reply-To: Your message of "Sat, 16 Sep 2000 13:32:05 EDT." <Pine.BSF.4.21.0009161306260.58081-100000@turtle.looksharp.net>
next in thread | previous in thread | raw e-mail | index | archive | help
>I've recently been struggling with problems in r1.41.2.1 of >src/sys/dev/aic7xxx/aic7xxx.c which is currently in RELENG_4 and has >been since the branch. This revision of the file is equivalent to r1.42 >in HEAD. The revision in RELENG_4 incorrectly handles SCBs on some >Adaptec 7892-based (29160) controllers. In actuality, I'm not sure exactly what in the RELENG-4 driver is causing problems with the 7892. The driver in -current (before today's commit) had several enhancements for U2 and U160 chips, but did not contain any corrections for "SCB handling" as you put it. >I have experienced these problems firsthand and was able to solve them >by installing a -current running r1.47 of aic7xxx.c. This system is >incredibly stable and I've been pounding it with bonnie and rawio over >the past week and seen not a single SCSI error or panic. Unfortunately, the changes that were made to increase performance on the U2/U3 and even some older chips in current exposed some chip->chipset interactions to which I had previously been unaware. So, the driver that works so well for you, will fail on some systems with onboard aic7880s and just about any system with a 3940AUW, just to name the two most prominent issues. This was the main reason the driver in -current has not been merged to -stable. After more than a month of investigation, experimentation, and development, I believe that the driver I committed to -current today will address the issues with non U2/U3 chips. Unfortunately, I don't have hardware to test against of the problems I discovered through investigation of Adaptec's system testing data. So the question still remains as to whether the driver in -current is stable enough to be merged back into -stable. Don't get me wrong. I want these issues to be resolved for -stable and release users just as much as anyone else, but the only way to do this safely and not wind up in a situation worse than we have now is through expansive testing. I'd be more than happy to make patches available relative to -stable (sys/dev/aic7xxx/... can simply be copied to a 4.X system and it should work with an added register define in sys/pci/pcireg.h and some minor changes to sys/conf/files), but I can only sanction the merge to stable if adequate testing occurrs. -- Justin To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200009162021.OAA02336>