Ex Bibliotheca

The life and times of Zack Weinberg.

Thursday, 11 April 2002

# 5:30 PM

alive! it's alive!

-> ld < test.o
Loading test.o |
value = 134765376 = 0x8085b40
-> test
hello world
2+2 = 7
value = 0 = 0x0

(Two and two are seven? See Trurl's Machine in The Cyberiad, by Stanislaw Lem.)

The last bug, incidentally, was that the system has no concept of a machine with two network interfaces, only one of which can talk to the embedded board. This happens to be how I arranged the local network.

# 3:30 AM

Recompiling and LD_PRELOAD-ing one tiny chunk of the C library causes the bug to mysteriously vanish. I love this sort of thing. Hopefully a coherent C-library bug report can be extracted from it.

Now I have a new bug to fix, or two bugs:

     T  O  R  N  A  D  O 
     Development  System
     Host  Based   Shell
     Version  2.2

Cannot start the asynchronous event notification
-> 
-> Segmentation fault

One hopes that the first bug will be easy to find and that fixing it will make the second bug go away, but I'm not exactly sanguine.

Time to kick it in the head for tonight, though, I think.

# 3:10 AM

Still can't get my "target server" to attach to the board. The bug is firmly in This Can't Possibly Be Happening territory; something is going wrong deep in the guts of the C library, for apparently correct input data and operation sequences. I can't run the program under the debugger to find out what's wrong, because it uses threads, and GDB does not like threads. (When "break main" doesn't work, you know you're in trouble.)