home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.wizards
- Path: sparky!uunet!mcsun!sun4nl!dutrun!donau!donau.et.tudelft.nl!bas
- From: bas@phys.uva.nl (Bas de Bakker)
- Subject: Re: Questions About Unix
- In-Reply-To: jhelberg@nl.oracle.com's message of 26 Aug 92 11: 40:22 GMT
- Message-ID: <BAS.92Aug27083352@carlo.phys.uva.nl>
- Sender: news@donau.et.tudelft.nl (UseNet News System)
- Nntp-Posting-Host: carlo.phys.uva.nl
- Organization: Institute for Theoretical Physics, Amsterdam, the Netherlands
- References: <37090@sdcc12.ucsd.edu> <2749@nlsun1.oracle.nl>
- Date: Thu, 27 Aug 1992 13:33:52 GMT
- Lines: 25
-
- >>>>> On 26 Aug 92 11:40:22 GMT, jhelberg@nl.oracle.com (Joost Helberg) said:
-
- >> : 1. can a program crash the processor on which it is running ? if so how?.
-
- jh> No, it shouldn't be capable to do so. You might try generating a lot of
- jh> signals (preferably recursive) to let the kernel-stack of the process grow,
- jh> this crashes some machines.
-
- By a bit of playing around with multiprocess ptrace(), one could crash
- an IBM RS6000 under AIX 3.1. I haven't tried this with 3.2 yet.
-
- To be more precise, I used dbx with multiprocess turned on, on the
- following program:
-
- #include <sys/ptrace.h>
- main()
- {
- if (fork()==0) ptrace(PT_TRACE_ME,0,0,0,0);
- }
-
- In dbx, continuing both processes after the fork() and then sending
- one of them a signal (anyone I tried) the computer crashed. Must have
- been some kernel bug.
-
- Bas de Bakker.
-