Hello,
I got to spend some hours on VSTa this weekend, and fixed a number of
mouldering bugs. First, KA9Q was locking up if you tried to turn on a
physical interface (looping packets to itself works fine, for all the good
that'd do you...). Fixed; it was a regression caused by fixing some edge
conditions with when it chooses to run its scheduling loop.
I turned on its packet tracing, and saw that it was ARP'ing for very
nearly every packet out of the box. Hunted this back to a bug with how it
ran its timer for ARP entries. Fixed. With these, it's running
reasonably well. FreeBSD's complaining on the PORT command it generates
(FTP), so I have a BPF tracing kernel ready to boot on the box so I can
see what's going wrong on the BSD side.
A long standing (as in, from the original port) bug in the Bourne shell
port was finally exterminated. There was an edge condition in the local
shell handling of variables which it had received out of the /env server.
Fixed. This would show up as command in a shell script which would fail
with something like "<command>: not found" even though they were valid
commands in your PATH. If you run with the standard path, you probably
wouldn't see this.
And the emulation of environ[] (the UNIX thing which kind of assumes that
the environment is static, sigh) had a bug where a junk character could
end up at the end of some array entries. Missing null termination, oops.
Fixed.
Raw booting is working OK. I had occasion to make myself a boot floppy
which ran a menu.lst and secondary loader off of the hard disk. Works
fine, and the GRUB instructions make it easy to do.
gdb is working very well! I just kick myself for not doing the port a LOT
sooner. Hope some of you break it out and give it a spin for your own
code.
Regards,
Andy
Received on Fri Nov 28 12:48:28 1997
This archive was generated by hypermail 2.1.8 : Thu Sep 22 2005 - 15:12:43 PDT