On Mon, Feb 09, 2015 at 11:07:13PM +1100, Ben Elliston wrote:
> Can you please send a reproducible test case (or script to generate
> one)?
Sure. I've just tested the following procedure with the latest
upstream code from the Gcc trunk on an x86_64 box (I usually work
with s390x). The behaviour is slightly different, but it should
be "good" enough to reproduce the problem.
1) Download the latest Gcc code.
2) Put the file attached to the message starting this thread into
the directory gcc/testsuite/gcc.target/x86_64/abi and rename it
to test_foo.C.
3) Edit the file gcc/testsuite/gcc.target/x86_64/abi/abi-x86_64.exp
and replace the pattern "test_*.c" in the line beginning with
"foreach" with the pattern "test_*.C" (lower case c to upper
case C).
4) After building Gcc in the usual way, cd into the build
directory and run
$ make -k check-gcc RUNTESTFLAGS="abi-x86_64.exp=test_foo.C"
When that is complete, take a look at the file
<builddir>/gcc/testsuite/gcc/gcc.log. You should see that only
the error messages up to source line 107 appear in the log. The
messages for line 108 and 109 are missing (if they are not, try
duplicating the last line of the function a couple of times). (I
don't get any PASS or FAIL messages on x86_64 though, but that's
not the root of the problem.)
I also have this script as a replacement for expect in my PATH
before /usr/bin to add the match_max command to the expect script:
-- BEGIN: $HOME/bin/expect --
#!/bin/sh -x
/home/vogt/bin/expect.bin -c "match_max 1000000" -c "match_max -d 1000000" $@
-- END --
Finally the version information from the x86_64 box:
Expect version is 5.45
Tcl version is 8.5
Framework version is 1.5
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
_______________________________________________
DejaGnu mailing list
DejaGnu@gnu.org
https://lists.gnu.org/mailman/listinfo/dejagnu
