Subject: Odds and ends with 1.3K => 1.4
To: None <netbsd-users@netbsd.org>
From: Hal Murray <murray@pa.dec.com>
List: netbsd-users
Date: 04/12/1999 03:52:16
I'm trying to learn more about NetBSD and/or help with 1.4 testing 
on alphas (without adding too much noise).  I started with the latest/only 
snapshot which is 1.3K.  I pulled over some source tar files and 
built a 1.4 kernel.  It seems to work. 

1) "netstat -i" hangs.  "netstat -in" prints out garbage for the 
network part when it finds an IP address.

    Name  Mtu   Network       Address              Ipkts Ierrs    Opkts Oerrs Colls
    de0   1500  <Link>        00:00:f8:76:2c:54     3249     0     1792     1     4
    de0   1500  1.1.8         16.4.80.82            3249     0     1792     1     4
    ....        ^^^^^

[That 1.1.8 should be 16.4.80/24]

I tried rebuilding netstat but the new one does the same thing.  
Has anybody else noticed this?  Did something change in some network 
control block?  (I thought I diffed the old/new /usr/include but 
I could have missed something or botched it.)  Will this all get 
better if I just go through more of the dance for installing 1.4?  
(I haven't done anything other than make a new kernel so far.) 


2) Pinging a nearby machine typically prints out 3 or 6 microseconds. 
(These are fast machines, but not that fast. :)  Is this normal?  
If so, what's going on?  I'd expect either 0 or ~1ms.  It looks reasonable 
if I ping a machine farther away - at least the decimal point is 
right.  

    foraker# ping logan
    PING logan.pa.dec.com (16.4.80.81): 48 data bytes
    64 bytes from 16.4.80.81: icmp_seq=0 ttl=64 time=0.003 ms
    64 bytes from 16.4.80.81: icmp_seq=1 ttl=64 time=0.006 ms
    64 bytes from 16.4.80.81: icmp_seq=2 ttl=64 time=0.006 ms
    64 bytes from 16.4.80.81: icmp_seq=3 ttl=64 time=0.006 ms
    ...


3) I noticed a few minor quirks in the install procedures.  This 
is from the 1.3K snapshot. 

    The printout from FTP isn't quite right on a VT100.  The status 
    message is supposed to overwrite the current line but the return 
    is getting lost so the test that should repaint overflows to 
    the next line, or something like that.  Things get pretty confusing 
    if ftp isn't making any progress.  I have some output tee-ed 
    from a tip connection if that will help.  It's not a problem 
    if things are working right on a fast net.

    The password isn't printed when it gets read in, but it does 
    show on command printout at the to of the screen while you are 
    FTPing something over.  (Also, it ends up in a log file in case 
    your console is tip-ed from someplace else.) 

    The install process ftped over several X related tar files but 
    I can't fine any X stuff on this machine.  Did they get unpacked, 
    or just dropped on the floor?  (There aren't any displays on 
    these machines.  But remote X might be interesting.) 
    
    The in-memory install system has ping but no netstat.  How do 
    people find out what's wrong with a network setup without netstat? 

    My name server is on the other side of a router.  The net-check 
    stuff pings the name server first, then the default router.  
    It's not clear what's the right thing to do there, but I got 
    a bit confused when my cables were scrambled. 

----

I've got one machine that gets a machine check when I boot NetBSD 
1.3K from floppies.  That machine may be flaky.  I'll investigate 
more when I get some time.  (It did it several times, as soon as 
it starts probing the PCI bus.)

  Ahh.  I just figured out something different about that machine.  
  It's got a tga on it.  (The ones that work OK don't have any display 
  installed.) 


Is there some system call to find out how fast the CPU clock is?