Hi all,

I've been working on a library which will provide a standard execution
trace output format across each of the simulation models.

It pulls in the libopcodes and libbfd which are generated when
building our binutils port.

https://github.com/juliusbaxter/or1ktrace

I have a number of problems, though and was wondering if anyone can
suggest how to solve them.

First is linking with the libopcodes and libbfd from the binutils
build directory. At present, I pass the directory where the binutils
output is stored via an environment variable to the or1ktrace
configure scripts, like so:

./configure BINUTILS_BLD_DIR=$HOME/svn/opencores/openrisc/trunk/gnu-src/bd-elf

The library appears to compile fine, but when the autotools-generated
Makefile tries to compile the little test program in there it passes
libor1ktrace.la last and thus can't find the things it relies on in
libbfd and libopcodes because "-lbfd -lopcodes etc." is before the
direct reference to the libor1ktrace library. I can compile it
manually, though, and everything works fine (see the README for
details on that.)

I'm not really sure what the best solution here is, though. Ideally
libbfd and libopcodes would get installed somewhere near the toolchain
and when libor1ktrace is built, we can install it there too.

This would, handily, result in just a single place to point the
models' compile scripts at when we build them.

As these libraries are not cross-compiled we probably need to keep
them separate from the actual toolchain stuff, similar to what we have
to do for libor1ksim.

So if this is the way forward, can anyone suggest how I get the libbfd
and libopcodes we compile with our binutils installed somewhere? (I've
tried make install-libbfd and install-libopcodes in the build
directory and saw it only installed a bunch of .po files get installed
and no .a's or .so's.) And does anyone know how to get those library
switches in after the one we've built for the libor1ktrace test
executable?

Thanks,

    Julius
_______________________________________________
OpenRISC mailing list
[email protected]
http://lists.openrisc.net/listinfo/openrisc

Reply via email to