This is the mail archive of the
ecos-discuss@sources.redhat.com
mailing list for the eCos project.
Re: Possible bug in dbg-threads-syscall.c
- From: Robin Farine <acnrf at dial dot eunet dot ch>
- To: "Patrick Doyle" <wpd at delcomsys dot com>
- Cc: "Robin Farine" <acnrf at dial dot eunet dot ch>, "eCos" <ecos-discuss at sourceware dot cygnus dot com>
- Date: 05 Dec 2001 14:47:01 +0100
- Subject: Re: [ECOS] Possible bug in dbg-threads-syscall.c
- References: <NFBBJAJICAKJPMMKDAGBIEAICNAA.wpd@delcomsys.com>
"Patrick Doyle" <wpd@delcomsys.com> writes:
> > > That's what makes sense to me. So then what does:
> > >
> > > if (*int_fun_ptr == 0) blah()
> > >
> > > mean?
> > > (I know, we are way off topic here, but it does seem like a
> > bizarre thing to
> > > test).
> >
> > Frankly, no idea and I don't understand why the compiler eats it
> > without any
> > sign of sickness.
> >
> That's _exactly_ what caught me by surprise here. The compiler thinks that
> it is a perfectly reasonable construct. Now, I suppose it is possible that
> we have just uncovered a subtle flaw in GCC (since I have not looked at this
> on any other compilers), but I expect not., I expect that the standard
> either defines exactly what is meant by *int_fun_ptr, or states explicitly
> that it is implementation defined (more likely). It just seemed bizarre to
> me that this would have been accepted by the compiler. I was wondering if
> any C language purist gurus out there would step in and say "yep, on page
> 1234 of the spec, it says that while this is acceptable syntax, the
> implementation is free to do whatever it wants", or something like that.
Ok then ;-)
ISO/IEC 9899:1999 (E) 6.3.2.1 paragraph 4 says that a function designator,
in two special cases excepted, is converted to a pointer to a function.
Robin