Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 31 Mar 1997 22:06:20 +0200
From:      j@uriah.heep.sax.de (J Wunsch)
To:        dbx@atmos.washington.edu ("Doug Burks")
Cc:        bugs@freebsd.org, toor@dyson.iquest.net (John S. Dyson)
Subject:   Re: kern/3156: Floppy disk copy freezes
Message-ID:  <19970331220620.KM38633@uriah.heep.sax.de>
In-Reply-To: <Pine.NEB.3.95q.970331101621.266B-100000@cargpc5.atmos.washington.edu>; from "Doug Burks" on Mar 31, 1997 10:27:28 -0800
References:  <19970331195000.TX25580@uriah.heep.sax.de> <Pine.NEB.3.95q.970331101621.266B-100000@cargpc5.atmos.washington.edu>

next in thread | previous in thread | raw e-mail | index | archive | help
As "Doug Burks" wrote:

>     Also, you should never, _never_ use the block device for raw file
>     data.
> 
> When I use the character device for the floppy copy, everything works
> smoothly.  Maybe my setup is the prime example of why one "should never,
> _never_ use the block device for raw file data."  :(

There are multiple reasons for it:

	. you blow the buffer cache for no good reason,

	. you miss the possibility to get write errors reported (since
	  the writes are cached, and the application has already got
	  a ``good'' indication of the write),

	. you can never be sure when the data are being actually
	  written; the worst i've seen so far was on Data General
	  DG/UX where it required a shutdown to get a few KB actually
	  written (since the OS was apparently waiting for a more
	  data to arrive).

Of course, a machine freeze shouldn't happen either.  I have no idea
why it happens.

> As for the original problem, it still exists, but this workaround is
> certainly quite easy.  As for more information, I can't add much.  ANY
> time under ANY conditions whenever I try to copy a large file onto a raw
> block floppy device, the computer completely freezes after a few
> seconds.

Can i convince you to compile a kernel with DDB, and see whether you
can break into DDB once it freezes?  This might give us a clue for
what is happening.

I have never been able to reproduce this kind of problem myself.
Maybe it's some strange interaction between VM buffers and the floppy
driver.

-- 
cheers, J"org

joerg_wunsch@uriah.heep.sax.de -- http://www.sax.de/~joerg/ -- NIC: JW11-RIPE
Never trust an operating system you don't have sources for. ;-)



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?19970331220620.KM38633>