*BSD News Article 35200


Return to BSD News archive

Newsgroups: comp.os.386bsd.development
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!agate!tfs.com!julian
From: julian@tfs.com (Julian Elischer)
Subject: Re: freeBSD 1.1.5.1 bug
Message-ID: <CvDLK7.LzA@tfs.com>
Organization: TRW Financial Systems, Oakland, CA
References: <54559@toad.com>
Date: Wed, 31 Aug 1994 01:31:19 GMT
X-Original-Newsgroups: comp.os.386bsd.development
Lines: 45

In article <54559@toad.com>, curt mayer <curt@mofo.emergent.com> wrote:
>I upgraded to FreeBSD 1.1.5.1 from the WC 1.1 cdrom last weekend, and the
>user files seem ok. However, when running the new kernel, I get freeing
>free inode, freeing free block, etc panics. this is under moderate disk load.
>
>is there anyone running an adaptec 154[02]A with freebsd 1.1.5.1?
>
[..]
>aha0: aha_scsi_cmd, more than 17 DMA segs
>sd2: oops not queuedbad block -1, ino 3456
>uid 0 on /usr/oldsrc: bad block
>aha0: aha_scsi_cmd, more than 17 DMA segs
>sd2: oops not queuedbad block -1, ino 3456
>uid 0 on /usr/oldsrc: bad block
>aha0: aha_scsi_cmd, more than 17 DMA segs
>sd2: oops not queuedbad block -1, ino 3456
>uid 0 on /usr/oldsrc: bad block
>bad block -1, ino 3456
oh dear.
well, something is trying to write chunks that are too big
to the adapter. (and they are being rejected)

I can't think of the likely reason, but note that
the limit of 17 DMA segs was actually 16 on some early
1542 cards.
(this is a limit in the driver.. you may want to try
recompile aha1542.c with this limit lowered..
but that won't help THIS problem..
for This problem, the request never even gets to the card.
The driver is rejecting it as bad
(The region to be written extends over > 17 pages).

though, it sounds as thought the entire transfer is strange..
(block -1?)

looks as if the inode writing code has for some reason
decided to write strange objects.

the error mesages should be:
>aha0: aha_scsi_cmd, more than 17 DMA segs <---from aha driver
>sd2: oops not queued    <------from disk driver
>bad block -1, ino 3456  <------ who writes THIS message?

julian