Hello, Jack
> Can you post you results from contrib/test_summary for a make
> check-fortran of llvm 2.4's gcc 4.2 compilers on linux? I'll
> remove the offending flags from CC1_SPECS in darwin.h on
> llvm 2.4 and post the same results for i686-apple-darwin9.
I tried to run testsuite several times, but there is a bug in the
frontend itself, which causes memory corruption and stall of the
testsuite run :(
> Are there any major issues left with the functionality of gfortran
> under llvm?
As I can see, I'd separate all problems into 2 groups:
1. Errors in lowering. There is some amount of this problems, since
fortran backend sometimes produces bogus tree. We're doing much better
typechecking than gcc and thus such stuff either asserts or leads to
invalid code.
I'd say, that in most cases frontend passes incorrect number of
arguments to the lowering code.
2. Invalid codegen.
3. Unhandled tree types.
There are some examples in standard testsuite for all types of mentioned
problems.
--
With best regards, Anton Korobeynikov.
Faculty of Mathematics & Mechanics, Saint Petersburg State University.