8 messages in net.sourceforge.lists.courier-maildropRe: [maildropl] Re: How to debug a br...
FromSent OnAttachments
David BorehamDec 9, 2002 12:11 pm 
Sam VarshavchikDec 9, 2002 2:39 pm 
Johannes ErdfeltDec 9, 2002 3:01 pm 
David BorehamDec 9, 2002 3:09 pm 
David BorehamDec 9, 2002 4:28 pm 
David BorehamDec 10, 2002 7:54 pm 
Sam VarshavchikDec 10, 2002 8:40 pm 
David BorehamDec 10, 2002 9:27 pm 
Actions with this message:
Paste this link in email or IM:
Paste this link in email or IM:
Atom feed for this thread
Paste this URL into your reader:
Subject:Re: [maildropl] Re: How to debug a broken maildrop configuration ?Actions...
From:David Boreham (davi@boreham.org)
Date:Dec 10, 2002 7:54:40 pm
List:net.sourceforge.lists.courier-maildrop

looks like I have screwed something up in the config. My current problem is that maildrop will not deliver mail. When run from the shell it typically exists throwing a signal (e.g. "maildrop: signal 0x0B"). I

don't

This usually happens when maildrop is compiled with non-standard, aggressive optimization flags.

More on this: the problem seems to be related to the compiler flag "-fomit-frame-pointer" in the x86 gcc code generator. Somehow C++ exception handling is screwed up when the compiler back-end produces frame pointer-less code.

I see quite a few posts in Google referencing the same problem, and some folk saying that it is a compiler bug (that -fomit-frame-pointer should work with C++ exception-generating code).

When building the RPM, the compiler flags come from a system-dependent file (/user/lib/rpm/remark in my case), which explains why I get -fomit-frame-pointer and others don't.