Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!news.ecn.uoknor.edu!news.wildstar.net!news.ececs.uc.edu!newsxfer.itd.umich.edu!newsxfer3.itd.umich.edu!howland.erols.net!blackbush.xlink.net!ka.sub.net!gummo.bbb.sub.org!gummo.bbb.sub.org!not-for-mail
From: bertram@gummo.bbb.sub.org (Bertram Barth)
Newsgroups: comp.unix.bsd.netbsd.misc
Subject: Re: faulting address in signal handler
Date: 25 Jan 1997 06:49:16 +0100
Organization: private site in Bruchsal, Germany
Lines: 46
Message-ID: <5cc6ss$nm7@gummo.bbb.sub.org>
References: <5bqspu$dnn@gummo.bbb.sub.org> <5c07vu$63c@innocence.interface-business.de>
NNTP-Posting-Host: gummo.bbb.sub.org
Xref: euryale.cc.adfa.oz.au comp.unix.bsd.netbsd.misc:5238
In article <5c07vu$63c@innocence.interface-business.de>,
J Wunsch <joerg_wunsch@interface-business.de> wrote:
>bertram@gummo.bbb.sub.org (Bertram Barth) wrote:
>
>> I'm trying to compile yooda-1.2 (OODBMS) for NetBSD-1.2/i386.
>> Within an signal-handler for SIGSEGV and SIGBUS I need to find
>> the faulting address but I can't find it.
>>
>> How can I access the faulting address within this signal-handler?
>
>This is fairly system-dependant. If your kernel pushes that
>information on the signal stack, you could fetch it there.
Nicely put. ;-)
It seems that NetBSD-1.2/i386 doesn't push this information on the stack.
(I've tried different ways to find this faulting address but none of
them worked.)
But there exists a patch for NetBSD (I think I found it in the mailing-
list archive in tech-kern.0022) which fixes this problem and which
never got committed into the official source tree (AFAIK).
After applying this patch I could access this address and I got this
partial problem with yooda solved. So far I have not expirienced any
bad side-effects from the patched kernel.
BTW: I can't understand why this patch (or something else addressing this
problem) was never commited to the NetBSD source-tree. OODBMS and
other persistent systems (at least yooda and texas) rely heavily on
this feature...
>The following crock seems to work on my FreeBSD system. YMMV. Note
>that it contains several DONTs: it makes an assumption about the
>parameters on the signal stack which doesn't need to be valid (hidden
>in the obscure cast to struct sigframe *), and it performs stdio
>actions inside the signal handler.
Didn't work in my environment, neither with unpatched nor with patched
kernel...
Ciao,
bertram
--
home: bertram@gummo.bbb.sub.org work: bertram@ifib.uni-karlsruhe.de
- Never ascribe to malice what can be adequately described by stupidity -