This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: libffi and embedded mips
- From: Joel Sherrill <joel dot sherrill at OARcorp dot com>
- To: Anthony Green <green at redhat dot com>
- Cc: java at gcc dot gnu dot org
- Date: Thu, 29 Aug 2002 15:41:59 -0500
- Subject: Re: libffi and embedded mips
- Organization: OAR Corporation
- References: <3D6A5838.9BF67F60@OARcorp.com> <1030381855.1702.12.camel@dhcppc2>
Anthony Green wrote:
>
> On Mon, 2002-08-26 at 09:32, Joel Sherrill wrote:
> > It looks like the unix mips platforms go to great pains to define
> > this based upon gcc arguments.
> >
> > I have no idea how it should be set for a generic mips-elf
> > embedded target. Suggestions appreciated.
>
> I just looked at an older mips-elf-gcc build we had installed here. It
> looks like the compiler doesn't give you any hints as to what the ABI is
> for this configuration. Maybe we should just fall back on the
> assumption that we're using the MIPS GCC eabi.
Do you have suggestions on how the ifdef should look? It is pretty ugly
now.
To make progress, I pushed by this by switching target CPUs and ran into
a
more general problem. The $(LINK) command for the libffi test does not
appear to go looking for newlib in the build tree.
To continue on, I put a - in front of that line and then got to boehm-gc
and told it to use POSIX threads for RTEMS. That got me to the point
where
it wanted to know about memory layout.
Which all leads me to a more general suspicion... has anyone every used
the GNU Java libraries on an embedded target?
Looking at the configure's and target configuration files, I am not
finding many signs that another embedded target has ever been worked
on.
If I am in uncharted territory, then I will need some help if anyone
cares about this getting pushed into the embedded space.
THanks.
> AG
--joel sherrill