Re: [WWW] simtest-howto.html clarifications, additions

[Janis Johnson <janis187 at us dot ibm dot com>]

On Wed, Jun 11, 2003 at 02:22:24PM -0700, Dan Kegel wrote:
> Janis Johnson wrote:
> > -cvs -d :pserver:anoncvs@sources.redhat.com:/cvs/src co binutils newlib dejagnu gdb
> > +cvs -d :pserver:anoncvs@sources.redhat.com:/cvs/src binutils newlib dejagnu gdb
> 
> Say, I'm not so sure about that change :-)

Eek!  I had at one time used an environment variable instead of -d xxx
and apparently removed the ever-so-important "co" when I changed it back.
 
> Also, you wrote:
> 
> > +    <p>Create a tree that consists of all of the files from the GCC and
> > +    binutils/gdb/newlib source trees (including several simulators in
> > +    <code>src/src/sim</code>), with the GCC files overriding the
> 
> By src/src/sim, I gather you're referring to
> http://sources.redhat.com/cgi-bin/cvsweb.cgi/src/sim/?cvsroot=src

That should have been "src/sim", in the src tree, which gets pulled in
with the other moduled that are checked out there.

> Are there other simulators, or are these it?

I don't know, and it looks like some of these are broken.  I'm also not
sure how to build these apart from the support in gdb; when I try to
build for fr30-elf, i960-coff, and m32r-elf I get a message that gdb
does not support the target, but it might be possible to build only
what's needed for the simulator.

I've got temporary access to an 8-processor Pentium III system with lots
of memory and disk space that I'm using to learn more about these builds,
and I'm running into lots of small problems:

  1. there's a line in the gdb sources that needs to be removed
  2. the tests won't run unless I create a symbolic link gcc/ld to
     ld/ld-new
  3. some builds fail, including arm-elf (the example, oh no!) whose
     compiler gets a segfault 
  4. some builds fail because gdb doesn't support the target
  5. builds for v850-elf and mn10300-elf fail building the simulator,
     although I noticed today that mn10300-elf is now obsolete

It would be nice to understand these so they can either be fixed or
documented.

Janis