Re: FreeBSD/Alpha Debugging Help Needed

At 15:29 31-10-2002, Matt White wrote:

However, the tests fly by so quickly that I'm not really able to figure out where they are coming from, and when I run a few of the failed tests by hand in gdb the core doesn't seem to occur.

make test 2>&1 | less will allow you to view the output on your own schedule. When the test doesn't segfault, by a simple sapi/cli/php -f path/to/tests/file.php you might be looking at shell limits, rather then actual errors. check ulimit -d especially. Add yourself to a login class, using vipw, f.e.: foo:yeah-right-sure:101:101:compile:0:0:gecos:/home/foo:/bin/bash 'compile' in this example is the login class. Modify /etc/login.conf (see login.conf(5)) and run cap_mkdb /etc/login.conf to change the it to a more appropreate value for your shell. Remember that -d (data segment size) applies to a single process, not the total ammount of vm you may use. You're discouraged to change the "default" class in /etc/login.conf.

Is there anything I can do to make the "make test" procedure much more verbose, so I can figure out which tests it's failing on?

"segmentation fault - core dumped" or something similar should occur __before__ the line with the failing test output

Is anyone else out there testing 4.3.0-pre2 on a 64-bit platform? I'd like to know if this is FreeBSD specific, or if it happens on other 64-bit platforms...

Similar issues with arrays have been noticed on Tru64 Compaq, by Sebastian Nohn. Last time I tried AIX with 64 arch, it also failed. It's about time I tested AIX again :) Met vriendelijke groeten / With kind regards, Webmaster IDG.nl Melvyn Sopacua