Since 1.6 seems to now have its own -current (though 1.6 itself isn't
released), at least as I understand the situation, it no longer seems
appropriate to post this to type of thing to current-users.  I don't know
if this is at all port-dependant, but I'm using port i386, so I'm posting
here.  (^&

(Maybe we need a list for discussing pending releases?  Or should I take
posts like this to current-users in the future?  netbsd-help?)


First, some background.

Under 1.5.2, I would get lockups trying to use my DVD player on the
motherboard IDE/ATAPI controller.  Using it on a PCI IDE controller card
worked, but apparently the card didn't support ATAPI and performance
degraded to PIO mode 4.  (Despite dire statements by others, this
generally doesn't seem to have been too bad, though it did hurt my
framerate somewhat.  Still...UDMA seemed desirable.)

So, I grabbed the most recent GENERIC 1.6 BETA kernel I could find.  (The
August 10 BETA only had an Alpha directory; August 9 had an i386...)  I
haven't upgraded my OS in general, I've just stuck the new kernel in
/netbsd-1.6 and booted from that.

Also, my router is another NetBSD box.  Because of an ifconfig problem,
I've knocked my main systems ethernet MTU down to 1400.


Things I've noticed: My DVD player works on the motherboard controller,
now, and seems to be giving UDMA support (which helps relieve the CPU
somewhat, improving framerate it seems).  This is good news.

However, I have some problems:

I seem to be getting more kernel messages of the form:

auvia0: codec invalid

...while playing sounds.  (At least while playing DVD's.  I haven't tried
playing from my ogg-vorbis files, etc.)  These are not obviously
problematic, though I notice more pops/skips in the sound from DVD's,
which may correspond to these errors.  (I didn't see any of these errors
happening, so I'm speculating.)


Second problem, when I decided to get online to mail about the above two:
I found that I couldn't get out of my system onto even my LAN.  Mozilla is
telling me that connections are refused (even now after finding a partial
workaround---see below).  Ping seems to be stopping up until it gets 6 or
7 packets, then it gets all packets (first one with 7-second delay, next
with 6-second delay, ...).  ssh seems to simply be unable to get a
connection.

Except for the mozilla problem, these are all resolved by setting my MTU
back up to 1500.  Whether this will resume causing problems for me when I
ftp, etc., through my router, I don't know.

(And, yes, lynx works.  So does Netscape.  And I tried more than one site
with Mozilla (www.NetBSD.org, the Daemon News site, and CNN at least).
Mozilla is something like release-candidate-3(?) for 1.0 built out of
pkgsrc a couple of months back.)

Any ideas on Mozilla?  A "shot in the dark" would be to upgrade my whole
system and rebuild lynx against the new system.  (^&  But, maybe someone's
encountered this already and can suggest a less drastic fix if I want to
wait for 1.6 to be fully released.  (It's not so much that I don't trust
the BETA release, as I just don't want to upgrade twice in a short period
of time, and I don't really have the time to fiddle with two upgrades
right now...though I might upgrade my router box if there's reason to
expect that doing so would fix things.)


For the most part, things look okay.  Though I'll need to get the kernel
sources and build a custom kernel to use ogle again.  (Ogle wants more
SYSV-style shared memory than GENERIC provides---I assume that there's no
runtime way to tweak that.)  I would do that anyway, after I get an actual
1.6 release installed.  (Also, mplayer generally works better for
me---though I can't get Disney's Fantasia to play sound with mplayer.)

I haven't been using 1.6's kernel long enough to see if there are any
other points of note, but I think that I'll keep a 1.6 kernel running, if
I can.  (I haven't seen the benefits/penalties of UBC, yet.  (^&)



In sum: My DVD player is happier with UDMA, now.  Non-standard MTU's seem
to cause problems (even with small-payloads such as ping).  Mozilla's got
some kind of deeper problems.


  ``I probably don't know what I'm talking about.'' --rauch@math.rice.edu