*BSD News Article 17285


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!elroy.jpl.nasa.gov!usc!howland.reston.ans.net!noc.near.net!ceylon!genesis!steve2
From: steve2@genesis.nred.ma.us (Steve Gerakines)
Newsgroups: comp.os.386bsd.development
Subject: Re: QIC 80 driver
Message-ID: <C8snr3.BK4@genesis.nred.ma.us>
Date: 18 Jun 93 01:49:49 GMT
References: <740187774snz@maxwell.demon.co.uk>
Organization: Genesis Public Access Unix +1 508 664 0149
Lines: 72

In article <740187774snz@maxwell.demon.co.uk> David@maxwell.demon.co.uk writes:
>I downloaded the QIC 80 driver from hrd769.brooks.af.mil:/pub/alpha and
>successfully installed it in 386bsd (with patchkit 0.2.3 applied previously).
>I rebuilt the kernal (no errors) and rebooted.

Great!  So long as the patchkit has an accurate timer for DELAY() and
possibly fixes to reduce interrupt latency, this configuration should
be fine.

>After seeing my only floppy I got the following messages (I commented out
>the tape command debug line in ft.c)

>tape_start start
>tape_recal start
>tape_recal end
>tape_recal end
>qic_status has dead drive ... st3 = $28
>qic_status has dead drive ... st3 = $28
>qic_status has dead drive ... st3 = $28

So far this is normal.  In the attach routine, the drive enable commands
for Colorado drives are tried first.  Seeing that this failed, the
driver goes on and tries the mountain enable commands.

>tape_recal start
>tape_recal start
>tape_recal end
>tape_recal end
>qic_status returned $01
>tape_status got $0001
>tape_end start
>tape_recal start
>tape_recal end
>tape_end end

The driver received a status code from the tape drive.  Since it was
enabled using the mountain commands, it is assumed to be a Mountain
type drive.

>fd0: drives 0: 1.44M, 2: Summit tape at 0x3f0 irq 6 drq 2 on isa

(This is how the attach line should normally appear with debug messages
turned off.)

>My config file has nfs commented out along with the scsi drivers and devices
>if that's important.

Shouldn't matter really.  The driver depends only on the floppy
controller.

>I next tarred the contents of a directory and attempted to list the tape using
>"tar tv". This seemed to list the tape and then hung. I can give further
>details if anyone is interested.

I am definitely interested.  Yours is the first (and only!) status
report that I've received.  Unfortunately I really need much more
feedback to make the driver stable for everyone.

How far along did you get when you read back the tape?  Does it hang
every time?

If you could send me the debug listing up to where the hang takes place
it would be helpful.  I suspect I have a race condition in the read
routine, where the driver top sleeps waiting for the read bottom, but the
bottom is in the middle of ending.  You might want to scan the driver
and search for debug messages that I have commented out.  (There are
places where debug messages appear so frequently that it causes the
driver to "miss" a block and have to rewind.)

Thanks,
- Steve
steve2@genesis.nred.ma.us