From: ignatz@homebru.chi.il.us (Dave Ihnat)
Subject: Re: Colorado 5Gb Patch Questionable
To: redhat-devel-list@redhat.com
Date: Mon, 30 Mar 1998 07:55:13 -0600 (CST)

Gentlefolk,

A quick followup.  I've got to verify this, but I believe it's related
to the known problems with pipeline mode, e.g., the block driver can't
report errors in a timely manner.  I tested this by simply putting a
write-protected tape in the drive, and then writing a small (<3K) block.

It acted as if it had successfully written the data.

The ramifications are pretty bad--you can't detect errors at all in some
cases, and you won't know just how *much* of your last few writes are bad
when you're near EOT.

Turning off pipelining is suggested if you "want accurate error reporting."
I think that's critical--the question is how badly does performance suffer?
New tests in order.
-- 
	Dave Ihnat
	ignatz@homebru.dminet.com	|| ignatz@advhealth.com
	312/315.1075 [home office]	|| 312/443.5860 [office]

-- 
To unsubscribe:
mail -s unsubscribe redhat-devel-list-request@redhat.com < /dev/null