home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!auspex-gw!guy
- From: guy@Auspex.COM (Guy Harris)
- Newsgroups: comp.unix.sys5.r4
- Subject: Re: method for detecting NULL pointer dereferences (incl. source)
- Message-ID: <15362@auspex-gw.auspex.com>
- Date: 5 Nov 92 23:47:44 GMT
- References: <1992Oct13.085006.110429@ilx018.intel.com> <ckl.719753912@uwdesk> <PINKAS.92Oct23151837@skywalker.intel.com>
- Sender: news@auspex-gw.auspex.com
- Organization: Auspex Systems, Santa Clara
- Lines: 24
- Nntp-Posting-Host: auspex.auspex.com
-
- >Where we found it, we fixed it, but we discovered too many instances of it
- >to try and release the system in the forseeable future. In addition, there
- >were hundreds of millions of lines of application code that was not clean,
- >and the last straw was that SVR4 was specified to run Xenix, SVR3.x, SCO,
- >and ISC binaries as is. Since most (or all) of these allowed dereferencing
- >a NULL pointer, there was a lot of code that would break.
-
- It might be nice to provide a "no page 0" option in the OS, to turn that
- behavior off for the benefit of e.g. application developers (especially
- those who want to put SVR4 applications on other architectures as well;
- for example, Solaris 2.x will, I suspect, not give you an all-zeroes
- page 0 by default).
-
- >Somebody should string up the HW designer at DEC that hardwired a 0 word at
- >address 0 in the VAX.
-
- Impossible - there's no such hardware designer. In fact, VMS doesn't
- map page 0 into the virtual address space.
-
- >Someone else should string up the Unix VM writer that didn't map page
- >zero elsewhere.
-
- Yup. After all, Sun probably won't even notice losing one vice
- president. :-)
-