This is the mail archive of the
mailing list for the GDB project.
Re: PATCH/testsuite: Enable two setup_kfails
- From: Michael Elizabeth Chastain <mec at shout dot net>
- To: drow at mvista dot com, gdb-patches at sources dot redhat dot com
- Date: Tue, 14 Jan 2003 01:45:07 -0600
- Subject: Re: PATCH/testsuite: Enable two setup_kfails
Daniel Jacobowitz writes:
> I left alone kfails which only show up in some configurations; we can
> come back to them after I get comments on gdb_test_multiple.
I sent you my comment last week, but I neglected to copy it to the list.
From mec Thu Jan 9 00:28:16 2003
Subject: Re: [RFA/testsuite] Re: RFC: gdb_test_multiple
Mad testing on native i686-pc-linux-gnu says that this is okay.
Recommended for approval.
The usual configuration matrix:
target => native
host => i686-pc-linux-gnu
osversion => red-hat-8.0
gdb => HEAD%20030107
gcc => 2.95.3, 3.2-7-rh, 3.2.1, gcc-3_2-branch%20030107, gcc-3_3-branch%20030107, HEAD%20030107
binutils => 18.104.22.168.2-rh, 22.214.171.124, HEAD%20030107
libc => 2.2.93-5-rh
gformat => dwarf-2, stabs+
count 32 = 1 * 1 * 1 * 2 * (5*3+1*1) * 1 * 2
I did this twice, once with TCL 8.4.1 + Expect 5.38.0 + Dejagnu 1.4.3,
and once with sourceware cvs TCL+Expect+Dejagnu ('T-E-D'). I did
before-and-after on each of those pairs, for a total of 128 test runs.
There were no significant before-and-after differences.
Good thing computers don't get bored ('Another nested for loop! Oh boy!').
There is one significant difference between FSF T+E+D and
sourceware T+E+D. With sourceware T+E+D, gdb often FAILs here:
(gdb) PASS: gdb.threads/pthreads.exp: set break at common_routine in thread 2
Cannot find thread 8192: generic error^M
(gdb) FAIL: gdb.threads/pthreads.exp: continue to bkpt at common_routine in thread 2
With FSF T+E+D, this test always PASSes. It's probably some difference
in the Expect versions and ^C handling or thread handling. In any case,
this is a separate issue from the gdb_test_multiple patch, I'm just
throwing it into the knowledge heap.