ESIX Networking Question(s) - fairly long

Avi Freedman freedman at euclid.math.temple.edu
Mon Jul 2 12:57:44 AEST 1990


I've got a probelm with ESIX that I was wondering if anyone could help
me with.  First, let me say that once civilization (i.e. bash, gdb,
emacs, gcc, g++, kermit, xmodem, etc...) is added to ESIX, it's a 
wonderful bargain for the price.

Now onto the question... I've got my ESIX system set up as 129.32.32.98,
a node called 'bigboy.cis.temple.edu', on the 'cis.temple.edu' domain.
Bigboy is a Northgate 386/25, with a DTC caching controller, 330 MB Hard
Disk, and WD8003E ethernet card.  From within temple.edu, there no problems
getting into or out of bigboy.  (More than I can say for the Xenix TCP/IP
which I was running).

However, as soon as I try to contact a service (telnet, rlogin, ftp) from
_outside_ Temple, or try to get out to the net from bigboy, that particular
service dies and cannot be reset by running the appropriate daemon short
of rebooting the computer.  HELP!!!  This is most distressing.  I don't
recall what the daemon says, but it says something about 't_connect' or
't_accept'.  A possibly related issue is that whatever I tell ifconfig,
I _cannot_ set the netmask or broadcast address on lan0.  At least, it
takes it as legal syntax, but doesn't tell me that it's changed the status
of either of those variables.  (If I'm correct, I want netmask to be
255.255.0.0 and broadcast to be 129.32.255.255).

Also, as another issue, but less important, is the socket library included
with ESIX supposed to work?  (I think /usr/lib/libnsl_s.a or something).
I got the X mandelbrot display program that communicates with multiple
clients from comp.sources.x, and, after the usual ministrations to the
header file paths and finding where fd_set was declared, got both the
client and server parts of the thing compiled.  Now, I have no trouble
using either on Suns or RS/6000s or SGIs, and the server part (displays
the result of computations happening on other machines) displays just fine
under Xwindows, and even communicates over sockets to the compute clients,
the compute client will not run on ESIX itself.

The compute client (called mserver), dies in the listen(sock, 1) command,
according to both sdb and gdb.  Sock (I checked) is 3, and appropriate
checks for -1 return values from system calls were made.  If I comment
out the listen, then it dies in the accepts.  Either way, I get a bus
error or memory access violation somewhere inside of a shared library.

		Thanks for any help,
			Avi Freedman
			freedman at euclid.math.temple.edu



More information about the Comp.unix.i386 mailing list