Hello together,
I am trying to get running the cl-alleg library: http://cl-alleg.sourceforge.net/
But in cmucl, sbcl and clisp I either get segfaults or cause the abort handler to be called.
My set-up is a recent debian lenny: Linux XXX 2.6.21.070708-1354 #1 SMP clisp 1:2.41-1 cmucl 19d-20061116-2 sbcl 1:1.0.7.0-1 slime 1:20070628-1 liballegro4.2-dev 2:4.2.0-5 cffi-070620.tar.gz
I've started to write lisp-unit test cases to track down the root of the problem, but without success.
I would really appreciate if you could tell me about your approach to debugging such problems.
I started to modify cl-alleg a bit, because it was originally written for windows and if you would like to see which problems I am talking about you can checkout the following CVS module:
cvs -d:pserver:anonymous@sablaster.cvs.sourceforge.net:/cvsroot/sablaster login cvs -z3 -d:pserver:anonymous@sablaster.cvs.sourceforge.net:/cvsroot/sablaster co -P project-workspace/
In project-workspace/lisp/README I've put a detailed description of how to arrive at the test situation.
I would be glad about any helping hand,
Hello together,
I am trying to get running the cl-alleg library: http://cl-alleg.sourceforge.net/
But in cmucl, sbcl and clisp I either get segfaults or cause the abort handler to be called.
To debug segfaults, I generally fire up gdb. Since SBCL (and maybe the others) internally uses many things which gdb likes to catch, its good to "detach" gdb whenever you're not using it.
My workflow looks like this: ("L:" is a lisp prompt, ">" is a bash shell prompt) L: (defun bad-cffi () ... ) L: (bad-cffi) ... segfault detected, abort
ps ax | grep sbcl
... shows pid 1553
gdb sbcl 1553
... lots of stuff loads gdb> c ... (short for 'continue' letting sbcl run) L: (bad-cffi) ... lisp hangs at the segfault (which is caught by gdb) gdb> bt full ... look at what's going on gdb> detach L: (defun better-cffi () ... ) gdb> attach 1553 L: (better-cffi) ...
Hope that helps, Daniel
-
Christian Schuhegger -
dherring＠tentpost.com