Re: Bug#6398: joe: segfaults on large files

[kai@khms.westfalen.de (Kai Henningsen)]

maor@ece.utexas.edu (Guy Maor)  wrote on 05.01.97 in <[🔎] 87eng0oa3k.fsf@slip-60-8.ots.utexas.edu>:

> Dale Scheetz <dwarf@polaris.net> writes:
>
> > If so, it should go away when I rebuild it with the new libc5.
>
> No, it won't.  I don't know why everyone has this misconception.

That's because everyone still thinks it's a problem in the malloc, when in  
fact it's a problem with the program.

"It worked before, so the program must be right".

No, the program isn't right. The old malloc didn't catch the buggy  
behaviour.

A frequent cause for this is code that does

  a = malloc(...);
  free(a);
  free(a);

or

  a = malloc(...);
  free(a);
  do_something(*a);

Of course, it's usually obfuscated by a lot of additional code around.

Try using a malloc debugger (electric fence or similar) on programs that  
break using the new malloc. It will probably find one bug or another.


MfG Kai


--
This message was delayed because the list mail delivery agent was down.