Date: 21 Apr 1998 07:17:27 GMT From: peter@netplex.com.au (Peter Wemm) To: freebsd-scsi@FreeBSD.ORG Subject: Re: ahh, I think I see part of the problem.. (CAM bouncing) Message-ID: <6hhh67$7m7$2@haywire.dialix.com.au> References: <199804201933.DAA08958@spinner.netplex.com.au>
next in thread | previous in thread | raw e-mail | index | archive | help
In article <199804210624.AAA03372@pluto.plutotech.com>, gibbs@plutotech.com (Justin T. Gibbs) writes: >>I tried the previous patch that you posted but had a pretty spectacular >>explosion on a BT545S.. It "found" devices all the way up to da11 (at >>target 2 lun 5) and failed to mount root. The probes were reporting >>garbage.. Hmm.. Are all the allocations being zeroed? > > I don't think that is the problem. It is more likely caused by using > the wrong dmat on line 261 of bt_isa.c: > > if (bus_dmamem_alloc(bt->sense_dmat, <= Used to be mailbox_dmat > > This would allocate the incorrect amount of sense bounce space causing > untold trouble. I'll say.. :-) >>I added a 'if (addr == 0) return (1)' type test at the end of the vlfilter >>function, but the 445S doesn't work properly either. Things get SEGV >>during bootup to multi-user (eg: sendmail, nmbd etc), and the system >>usually panics with a 'vm_fault on nofault entry' type panic fairly >>shortly after that. I suspect the bios shadow avoidance has kicked in and >>a bug has been triggered (it doesn't work on the 545S at all, so that's >>why I suspect the bouncing). > > It is probably the sense buffering bug again, but it is only triggered > after more commands are run as the VL card has more CCBs than the ISA > card and the size of the mailbox_dmat is based on number of CCBs. > > I hate writing code blind. I'd rather catch my coding bugs myself than > have others suffer through it. 8-) That's what testers are for.. Although I'm quietly thankful that it hasn't scrambled block read commands to make them look like scsi 'format disk' commands... :-) Cheers, -Peter 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?6hhh67$7m7$2>