Date: Fri, 17 Apr 1998 10:14:52 +0800 From: Peter Wemm <peter@netplex.com.au> To: Kerry Morse <kerry.morse@metro.tas.com.au> Cc: "Justin T. Gibbs" <gibbs@narnia.plutotech.com>, Tom <tom@uniserve.com>, current@FreeBSD.ORG Subject: Re: Help - please... Adaptec 1542 and Exabyte tape problem... Message-ID: <199804170214.KAA24737@spinner.netplex.com.au> In-Reply-To: Your message of "Thu, 16 Apr 1998 21:15:37 %2B1000." <02B01380C828D1119ED70020AF641C53068558@MTTMail.metro.tas.gov.au>
next in thread | previous in thread | raw e-mail | index | archive | help
Kerry Morse wrote: > > Thanks for the help... > > > So basically it's not going to work for at least a while or a few fixes > so swapping the controller out over with a PCI controller while no ones > looking is my best solution at present... I've just found a whole mess of bounce buffer bitrot in the existing scsi code and applied some bandaids. I can't test some of the more obscure drivers, but an aha1542B + sd0/sd1/st0/cd0 now seems to work fine. Before, talking to the tape would fail 50% of the time at startup. Ironically, I think I to blame for a good chunk of the original breakage. The scsi base code (scsi_scsi_cmd() specifically) had some patchwork to detect data transfers to/from the per-process kernel stack and how to manually bounce them. When I moved the kernel stack out of user space last year some time, the glue code that checked for requests on the kstack no longer worked as it used assumptions about the VM space layout that were no longer true. However, there was bitrot in other drivers too, notably worm, ch and st. Cheers, -Peter To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199804170214.KAA24737>