Subject: My experience with the _RCs so far
To: None <port-i386@NetBSD.org>
From: Miguel Mendez <flynn@energyhq.es.eu.org>
List: port-i386
Date: 10/06/2004 16:08:27
Hi all,

I posted my 2.0_BETA experience some months ago and, when the RC was
announced, decided to install NetBSD again. Unfortunately some of the
bugs are still there.

First I can't manage to finish the install process. The kernel will
always panic in the middle of the tarballs extraction in ffs_balloc so,
again, I did it the manual way, mounted / in sync mode, which I'm still
keeping until I can allocate some spare time to obtain a crash dump and
try to find out what's going on. I'm confident I can reproduce the
problem. High disk activity on / leads to a panic unless I mount it in
sync mode. Surprisingly, it does never happen on my /usr partition, a
striped (with RAIDframe) 16GB partition spread over two disks.

Then there's the umass thing. This is known issue, and I wonder if
somebody is already working on it. Basically that problem is that if you
attempt to mount an umass device and your system has been up for some
time ( > 12 hours) you're very likely to get the dreaded buffer memory
error.

Oct  3 17:04:37 scienide /netbsd: umass0 at uhub0 port 2 configuration 1
interface 0
Oct  3 17:04:37 scienide /netbsd: umass0: Packard Bell Portable MP3
Player, rev 1.10/10.00, addr 2
Oct  3 17:04:37 scienide /netbsd: umass0: using SCSI over Bulk-Only
Oct  3 17:04:37 scienide /netbsd: umass0: no buffer memory

This problem doesn't show up on a recently booted system. I'd be willing
to help debug this as well, but my knowledge of both the USB and scsi
subsystems is almost null. I'm a fast learner though :)

Other than that 2.0 works very well and performance is actually very
good.

Cheers,
-- 
	Miguel Mendez <flynn@energyhq.es.eu.org>
	http://www.energyhq.es.eu.org
	PGP Key: 0xDC8514F1
	Note: All HTML and non-english mail goes to /dev/null