Re: ad0: FAILURE - WRITE_DMA

On Tue, Dec 07, 2004 at 01:38:32AM -0500, Garance A Drosihn wrote:
>About all I can do is guess at some possible problems.  What is the
>capacity of the Seagate drive that you're having trouble with?  Are
>you using vinum or gvinum on it?  Is your machine set up with a serial
>console?  If the machine panics, can you get a core-dump of it, or do
>you just get a disk error when it tries to write the dump to disk?

My disks are 200MB, but I am/was only booting with one at a time.  No
*vinum or anything fancy.  FreeBSD with UFS2 is on one disk, OpenBSD
on the other.  Both occupy their whole disk.

At least one incident was a freeze, where it might have been trying to
panic, but didn't shutdown.  With the real panics, there were several
kinds, with different messages which I can no longer remember.

Gigabyte 7N400Pro2
AMD Barton 2600+
0.5GB memory
SiI 3512A disk controlled
2 x 200MB Seagate 7200.7 SATA disks (1 in use)
re0 GigE onboard
onboard AC97 Audio not used
Radeon 9200SE currently in text mode
no floppy - I forgot to order one

I'm not running FreeBSD on this machine any more, so there won't be
anything to make a FreeBSD coredump from.  I'm not even going to try
again unless someone convinces me that it's fixed and available on
installation media.  Maybe 5.4 if I still care enough to keep my CD
subscription alive.  I'll keep 5.3 on the disk it's on until I need
the space, though, so it could be updated if there's reason later.

I only got it to 5.3 by installing 5.2 from CD onto a different
machine, cvsup'ing to 5.3-R and compiling that before moving the disk
to this machine.

The OpenBSD message I get occasionally (less frequently than the
FreeBSD WRITE_DMA failure) is "aborted command, interface CRC error
writing fsbn [...]".

My understanding is that the SiI 3*1* stuff just sux, but it's a new
motherboard, and I did get what I ordered, so I'd feel funny trying to
replace it just now.

-- 
Christopher Vance