Date: Tue, 28 Jul 1998 11:12:51 -0500 From: Karl Denninger <karl@mcs.net> To: Warner Losh <imp@village.org> Cc: The Hermit Hacker <scrappy@hub.org>, scsi@FreeBSD.ORG Subject: Re: New problem - Adaptec cards with REV E on the sequencer chipset Message-ID: <19980728111251.29545@mcs.net> In-Reply-To: <199807281612.KAA14864@harmony.village.org>; from Warner Losh on Tue, Jul 28, 1998 at 10:12:06AM -0600 References: <19980728090821.09162@mcs.net> <19980728071337.07188@mcs.net> <Pine.BSF.3.96.980728093345.15349H-200000@hub.org> <19980728090821.09162@mcs.net> <199807281612.KAA14864@harmony.village.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Jul 28, 1998 at 10:12:06AM -0600, Warner Losh wrote: > In message <19980728090821.09162@mcs.net> Karl Denninger writes: > : That's it. It looks like there was a change in the count of the number > : of bytes moved to or from the SCB buffer from 32 to 28. > > That is correct. > > : I don't know a thing about the sequencer hardware or why this is significant, > : but this is what I see as a difference in the two versions. > > Justin explained it to me, so I'll give it a shot. On the PCI bus > there are two parameters that control how long a card can remain on > the bus. One of them is how long a DMA can take to complete before it > is forced off the bus. This value is set to be 28 bytes by most > BIOSes. There appears to be some problem with the newer revs of the > chips asserting control of the bus. Since the SCB is DMA'd to the > card, any larger than 28 bytes will cause the hangs. > > : FYI, it appears that the NON-CAM source is working with 28 bytes as well, > : but that code is *radically* different than the CAM version and I haven't > : studied the differences in depth. > > I would imagine, given the nature of the problem, changing 32 to 28 in > the NON-CAM code would work, iff those last 4 bytes aren't used for > anything. I don't know the sequencer code. > > BTW, if Justin contradicts anything that I've said here, he's right > and I'm misremembering... > > Warner Well, what I saw in the non-CAM code was that it appears that a fix was made recently for this, in that "one instruction" was changed. The CAM code has a completely different set of execution paths than the non-CAM code; the sequencer code for each is significnatly different. -- -- Karl Denninger (karl@MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin http://www.mcs.net/ | T1's from $600 monthly / All Lines K56Flex/DOV | NEW! Corporate ISDN Prices dropped by up to 50%! Voice: [+1 312 803-MCS1 x219]| EXCLUSIVE NEW FEATURE ON ALL PERSONAL ACCOUNTS Fax: [+1 312 803-4929] | *SPAMBLOCK* Technology now included at no cost To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-scsi" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19980728111251.29545>