Re: return CERROR result in a0 too (was: More toolchain issues)

[Frank Wille <frank%phoenix.owl.de@localhost>]

Martin Husemann wrote:

> And be sure to run all regress/ and test/ regression tests with the
> change.

I have run the /usr/tests/atf testsuite yesterday and there are some
failures (more than on a 5.01/i386 system for example):

Failed test cases:
    atf-c++/t_text:to_type, atf-c++/t_fs:remove,
    test_programs/t_meta_data:require_progs_header,
    test_programs/t_meta_data:timeout, tools/t_atf_config:arch,
    tools/t_atf_cleanup:mount, tools/t_atf_cleanup:symlink,
    tools/t_atf_run:broken_tp_hdr, tools/t_atf_run:zero_tcs,
    tools/t_atf_run:no_reason, tools/t_atf_report:default,
    tools/t_atf_report:oflag, tools/t_atf_report:output_csv,
    tools/t_atf_report:output_ticker, tools/t_atf_report:output_xml,
    formats/t_parsers:atffile, formats/t_parsers:config,
    formats/t_parsers:tcs, formats/t_parsers:tps
Summary for 55 test programs:
    280 passed test cases.
    19 failed test cases.
    6 skipped test cases.

Are those expected to be normal?

Besides that the system appears to be very stable now! I'm compiling Openbox
since more 30 hours, without any segfault. nmap also works now.

So I'm considerung to commit my modification this weekend (and request a
pullup for netbsd-5).

-- 
Frank Wille