[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: segfaults and other murders

Ron Farrer <rbf@farrer.net> writes:

> I still get segfaults in nmap and several other programs. I also still
> have problems with licq and gnomeicu. I would have thought these would
> have been fixed by now... Does anyone have any suggestions? 

You should follow proper bug reporting procedure - figure out how to
reproduce the segfaults, then file bugs against the packages.  If you
can track down the actual bug and produce a patch, then that's great,
but it's not necessary.

It's more important to have accurate information on how to reproduce a
bug than possibly incorrect speculation on the cause of the bug or a
possibly incorrect patch (the GCC bug reporting guidelines are a must
read for everyone).

In my experience these sorts of things are either:

1) Non-Alpha-related bugs in the programs that don't manifest
   themselves on i386 purely out of chance (uninitialized auto
   variables seem to cause this a lot, as do some buffer overruns)

2) 64-bit-uncleanliness bugs, which are usually easily fixed by making
   the program actually use a 32-bit type where it needs a 32-bit
   type.  These can sometimes be fairly subtle in the case where
   people are making assumptions about the sizes of system types.

3) Other bugs related to some features of Linux on Alpha that are just
   different from i386 (usually OSF/1 compatibility stuff)

4) Compiler bugs :-(  (these are actually less frequent than you'd think)

These are mostly easy to solve with the exception of (4)...

David Huggins-Daines, GNU/Linux Consultant, Linuxcare, Inc.
613.223.0225 mobile
dhd@linuxcare.com, http://www.linuxcare.com/
Linuxcare. Support for the revolution.

Reply to: