This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Status for simulator targets (bad, that is, but that's no news)


Oops, accidentally sent to gcc-patches.  Now you get two for the
price of one (or really, vice versa).

---------- Forwarded message ----------
Date: Fri, 27 Jul 2001 20:49:07 -0400 (EDT)
From: Hans-Peter Nilsson <hp@bitrange.com>
To: gcc-patches@gcc.gnu.org
Subject: Status for simulator targets (bad, that is, but that's no news)

I've attempted to find out, via src/dejagnu/baseboards and simulators in
src/sim, a list of simulator targets that seemed likely to be usable to
test patches against, and which seemed buildable using the instructions at
<URL:http://gcc.gnu.org/simtest-howto.html>.  The idea being that
diversity in simulator targets would lead to better patch test coverage.

Regrettably, these targets were today all broken on the trunk.  Most of the
failures (all but two) I see are due to breakage in GCC.  Most get an ICE
or equivalent when compiling libgcc2 or newlib.

As mentioned, a few days ago, I had mips-elf/mips-sim and
mn10300-elf/mn10300-sim building "all" without error and running the
testsuite.  While there was room for improvement in the test-results, they
seemed usable to test patches against.

The combinations I've been able to deduce, including those previously
known (target-nickname/simulator-name) that *should* work, modulo bugs:
arm-elf/arm-sim
d30v-elf/d30v-sim
fr30-elf/fr30-sim (needs --enable-languages=c++,java,objc to leave out fortran)
h8300-hms/h8300-sim
i960-elf/i960-sim
m32r-elf/m32r-sim  (needs libgloss patch)
mcore-elf/mcore-sim
mips-elf/mips-sim
mn10200-elf/mn10200-sim
mn10300-elf/mn10300-sim
powerpc-eabisim/powerpc-sim
sh-hms/sh-hms-sim
sparclite-elf/sparclite-sim
v850-elf/v850-sim (though simulator unbuildable seemingly due to
                  bad Makefile.in)

I'd be thankful if you tell any combinations, and any combinations left
out that are sufficiently different to those above, and which would be
buildable using simtest-howto.

I'd like to add items from that list to a table in simtest-howto as each
of them work again, assuming they unbreak for some reason.  A sample is at
<URL:http://bitrange.com/temporary/a-place-to-test-gcc-webpages--do-not-bookmark/simtest-howto2.html>

brgds, H-P


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]