This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] to java::lang::ConcreteProcess::destroy
- From: Ronald Landheer-Cieslak <ronald at landheer dot com>
- To: Andrew Haley <aph at redhat dot com>
- Cc: Ronald Landheer-Cieslak <ronald at landheer dot com>, Tom Tromey <tromey at redhat dot com>, <gcc-patches at gcc dot gnu dot org>, <java-patches at gcc dot gnu dot org>
- Date: Mon, 17 Feb 2003 17:58:43 +0100 (CET)
- Subject: Re: [PATCH] to java::lang::ConcreteProcess::destroy
On Sat, 15 Feb 2003, it was written:
> Ronald Landheer-Cieslak writes:
> > The Makefile defines NO_SIGNALS when there is no signal support for the
> > current target. In fact, it tried to compile the file with -DNO_SIGNALS as
> > an option, and the compile broke on the kill() (which is rather logical if
> > you don't have signals) so I patched it up like this and it worked like a
> > charm :)
> > Might leave a couple of zombies around if there are no signals, though..
> I think we're confused about what system has no signals but has
> processes. Wel, I certainly am.
Hmm.. Of course there is no doubt that any POSIX system must have signals,
but how, then, is it possible that gcc is compiling itself without signal
support, and is still giving me POSIX threads -especially when I've asked
him to disable threads. Configure options for the cross-compiler are:
--prefix=$HOME/tmp --target=i386-unknown-freebsd4.7 --with-newlib
(these are the options for the full compiler, of course; the bootstrap
compiler doesn't get into trouble).
FWIW, I was building a linux-x-freebsd cross compiler w/ newlib as CRTL -
latest versions of everything, of course.
IMHO, there is a bug somewhere around here:
1. the file I patched should enever have been compiler (because I disabled
threads on the command-line
2. if compiled without signal support (as done here) kill() should not be
called (as fixed by my patch)
--> IMHO, this is true regardless of whether the platform is supposedly
POSIX-compliant: AFAIK, no platform is completely POSIX compliant to date,
and it *is* possible (though not very likely) to run into a platform with
threads but without signals (though I have no idea how *that* would be
IMHO, the patch, though it doesn't fix all of the problem, fixes at least
part of it. The other part of the problem is in the Makefile.in and
configure files, but I really can't help you there :\
There's another problem with the same cross-compiler, BTW, (wrong headers
being included) but I'll try to whip up something decent for that later.