Re: CVS commit: src/regress/lib/libc/siginfo/sigfpe

[christos%zoulas.com@localhost (Christos Zoulas)]

On Feb 18,  2:55pm, M.Drochner%fz-juelich.de@localhost (Matthias Drochner) 
wrote:
-- Subject: Re: CVS commit: src/regress/lib/libc/siginfo/sigfpe

| christos%NetBSD.org@localhost said:
| > Module Name:        src
| > Committed By:       christos
| > Date:               Tue Feb 17 19:22:43 UTC 2009
| > Modified Files:
| >     src/regress/lib/libc/siginfo/sigfpe: sigfpe.c 
| 
| This is basically what regress/lib/libc/ieeefp/except
| does (one of four testcases in the "unmasked" case).

Actually not. It siglongjmps out of the signal handler which is illegal
according to:

http://www.opengroup.org/onlinepubs/000095399/functions/xsh_chap02_04.html#tag_02_04
http://opengroup.org/onlinepubs/007908799/xsh/siglongjmp.html

The question is what to do with SIGFPE?
Do we advance to the next instruction so returning from a signal handler
can proceed? Currently most platforms don't (except the vax, which I changed
a day or so ago). I think it is useful to have the option to advance to
the next instruction at a fault. Perhaps this should be set with a signal
flag? SA_NOADVANCE or something?

christos