This is the mail archive of the
gdb-testers@sources.redhat.com
mailing list for the GDB project.
sunday project report, gdb, 2002-05-18
- From: Michael Elizabeth Chastain <mec at shout dot net>
- To: gdb-testers at sources dot redhat dot com
- Date: Sun, 19 May 2002 23:25:06 -0500
- Subject: sunday project report, gdb, 2002-05-18
There are two new bugs in this report:
pr gdb/553 is a change in binutils HEAD which damages gdb in with -gdwarf-2.
Someone who understands line number records (Daniel J?) needs to fix this
in gdb or push it upstream to binutils people.
pr gdb/554 is a gdb crash in gdb.c++/virtfunc with gdb 5.2, gcc 3.1, and
binutils 2.11.2. It's always bad when gdb crashes, but we might leave this
one alone until I upgrade my testbed to binutils 2.12. I've been postponing
that upgrade because binutils 2.12 introduces some regression failures
compared to binutils 2.11.2, so I have to set up a giant before-and-after
environment and then report the failures.
My tables are here:
http://www.shout.net/~mec/sunday/2002-05-18/index.html
gdb/553 does not appear in the tables at all, because I fell back to an
earlier binutils after I reported the problem. gdb/554 appears in the
attention tables, but does not appear in the difference tables by-date
because this is the first report with gcc 3.1.
Michael C
. Summary
. Test Matrix
. Matrix
target: native
host: i686-pc-linux-gnu%rh-7.2
gdb: 5.2, gdb_5_2-branch%20020518, HEAD%20020518
gcc: 2.95.3, 2.96-rh, 3.0.4, 3.1, gcc-3_1-branch%20020518, HEAD%20020518
glibc: vendor
goption: -gdwarf-2, -gstabs+
count: 36 = *(1 1 3 6 1 2)
. Notes
target and host names are gnu triples with extra information after
a '%' delimiter.
versions which start with a number are official releases or snapshots.
versions named "HEAD" are the cvs HEAD.
versions with any other name are cvs branches.
cvs versions show the pull date after a '%' delimiter.
gcc releases and snapshots are built with binutils 2.11.2
gcc cvs versions are built with binutils HEAD from 2002-05-13.
gas has a change on 2002-05-14 which breaks gdb. (pr gdb/553).
. libiberty
All tests passed.
. gdb
. Tables
. http://www.shout.net/~mec/sunday/2002-05-18/index.html
. Overview
gdb 5.2: 0 build aborts, 0 test aborts, 495 attention lines
gdb gdb_5_2-branch: 0 build aborts, 0 test aborts, 495 attention lines
gdb HEAD: 0 build aborts, 0 test aborts, 304 attention lines
An attention line is a line with one or more attention results.
An attention result is any result except PASS. This includes
ERROR, WARNING, FAIL, KPASS, KFAIL, XPASS, XFAIL, UNRESOLVED,
UNTESTED, UNSUPPORTED, and unknown results.
. Old bugs fixed
http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=277
Support for qualified types is lacking common information
. New bugs detected
http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=553
gdb with -gdwarf2 fails with binutils HEAD%20020514
http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=554
gdb 5.2 crashes on gdb.c++/virtfunc with gcc 3.1, binutils 2.11.2, -gstabs+
. Test Protocol Changes Since Last Report
I added gcc 3.1.
The tables now handle the NOTE result.
. Baseline Software
. host i686-pc-linux-gnu%rh-7.2
make 3.79.1
binutils 2.11.2
gcc 3.0.4
flex 2.5.4
bison 1.35
tcl 8.3.4
expect 5.33.0
dejagnu 1.4.2 + FernandoN kfail patch.
. Analysis
. libiberty
. results
. target native
. host i686-pc-linux-gnu%rh-7.2
binutils HEAD All 648 tests passed
gcc gcc-3_1-branch All 648 tests passed
gcc HEAD All 648 tests passed
gdb gdb_5_2-branch All 648 tests passed
gdb HEAD All 648 tests passed
. gdb
The last report was 2002-05-10.
http://www.shout.net/~mec/sunday/2002-05-10/Analysis.txt
Configuration numbers for the difference grids are in:
http://www.shout.net/~mec/sunday/2002-05-10/Differences-compare-by-date-two.html
Difference grid key:
+ gdb improved
- gdb regressed
$ gcc improved
= gcc regressed
@ test suite improved
~ test suite regressed
* new test with good result
_ new test with bad result
, noise: result changed but gdb behavior did not
. 5.2
No significant differences.
. gdb_5_2-branch
No significant differences.
. HEAD
gdb.base/break.exp *20 *21 *22 *23 *24 *25 *26 *27 *28 *29
gdb.base/ending-run.exp ,26 ,27 ,28 ,29
gdb.base/gcore.exp ,26 ,27 ,28 ,29
gdb.c++/annota2.exp ,24 ,25 ,27
gdb.c++/inherit.exp ~27 ~29
gdb.c++/method.exp +27 +29
gdb.c++/templates.exp +25 +27 +29
. Gdb Improved
. gdb.c++/method.exp: print this in A::bar
This was pr gdb/277. This has been fixed.
. gdb.c++/templates.exp: constructor breakpoint
Gdb improved its behavior when completing a template constructor name
in gcc v3 -gstabs+ configurations.
. New Test with Good Result
. gdb.base/break.exp: breakpoint quoted function
gdb.base/break.exp: run until quoted breakpoint
These new tests PASS in all configurations tested.
. Test Suite Regressed
. gdb.c++/inherit.exp: print g_vE
Last report, this test printed:
print g_vE
$94 = {<vD> = {<vB> = {<vA> = {va = 0, vx = 0}, _vptr.vB = 0x804f0f8, vb = 0, vx = 0}, <vC> = {_vptr.vC = 0x804f104, vc = 0, vx = 0}, _vptr.vD = 0x804f0ec, vd = 0, vx = 0}, _vptr.vE = 0x804f0d8, ve = 27, vx = 28}
(gdb) PASS: gdb.c++/inherit.exp: print g_vE (FIXME v3 vtbl ptr)
This report, this test prints:
print g_vE
$94 = {<vD> = {<vB> = {<vA> = {va = 0, vx = 0}, _vptr.vB = 0x804f0d8, vb = 0, vx = 0}, <vC> = {_vptr.vC = 0x804f0e4 <VTT for vD>, vc = 0, vx = 0}, _vptr.vD = 0x804f0cc, vd = 0, vx = 0}, _vptr.vE = 0x804f0b8, ve = 27, vx = 28}
(gdb) FAIL: gdb.c++/inherit.exp: print g_vE
Gdb has added "<VTT for vD>" to the output. This output is okay;
the test script needs an enhancement.
. Noise
. gdb.base/ending-run.exp: b *0x80483ce
gdb.base/gcore.exp: heap_string = 0x8049638 "I'm a little teapot, short and stou ..."
These tests have a machine address in their names, so the test names
fluctuate.
. gdb.c++/annota2.exp: annotate-quit
This test is unstable.