This is the mail archive of the
gdb-testers@sources.redhat.com
mailing list for the GDB project.
Gdb test suite regressions, 5.2.1 to 5.2.91
- From: Michael Elizabeth Chastain <mec at shout dot net>
- To: gdb-testers at sources dot redhat dot com
- Date: Fri, 6 Dec 2002 13:30:10 -0600
- Subject: Gdb test suite regressions, 5.2.1 to 5.2.91
GDB test suite regressions from 5.2.1 to 5.2.91
These results are based on the Sunday Project tables from 2002-12-03:
http://www.shout.net/~mec/sunday/2002-12-03/index.html
(I haven't posted these tables yet, so if the link doesn't work,
check back in a few hours).
These differences are substantially identical to the differences
from the 2002-11-25 test run.
The test suite regressions break down as follows:
regressions in gdb 2
regressions in test suite 1
long-standing bugs in gdb 3
long-standing bugs in test suite 1
bugs in gcc 1
not analyzed yet 2
TOTAL 10
In my opinion, none of the 10 test suite regressions are show-stoppers.
I analyzed only the test suite results which disimproved, not any of the
improvements or the results from new tests. All test runs are native
i686-pc-linux-gnu on a red hat linux 8 system. I test with gcc v2 and v3,
several versions of binutils, and both dwarf-2 and stabs+.
Michael C
===
. regressions in gdb (works in 5.2.1, broken in 5.2.91)
. gdb.base/condbreak.exp: run until breakpoint at marker1
gdb.base/condbreak.exp: run until breakpoint at marker2
gdb.base/ena-dis-br.exp: continue to auto-disabled break marker2
gdb.base/ena-dis-br.exp: continue to break marker1
gdb.base/ena-dis-br.exp: continue to break marker1, 2nd time
pr gdb/460: Regression: gdb prints extraneous address at breakpoint
http://sources/redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=460
Happened with gcc 2.95.3, any binutils, dwarf-2.
In some cases, gdb regressed in its handling of line numbers.
This is a bug in gdb, but it is not very harmful.
. gdb.c++/templates.exp: break Garply<Garply<char> >::garply
no pr filed.
Happened with gcc 3.0.4, any binutils, dwarf-2.
gdb fails to print the source file and line number for a breakpoint
on a template function.
I am not filing a bug report for this because it happens only with
gcc 3.0.4; gcc 2.95.3 and gcc 3.1 and later all work fine. The only
reason I test with gcc 3.0.4 at all is to isolate gcc regressions:
tests that work with gcc 3.0.4 and fail with later compilers. I am
willing to ignore bugs that happen with 3.0.4 and are fixed in later
compilers.
. regressions in test suite (works in 5.2.1, broken in 5.2.91)
. gdb.base/selftest.exp: backtrace through signal handler
pr gdb/862: regression: gdb.base/selftest.exp: "backtrace through signal handler"
http://sources/redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=862
Happened with any gcc, any binutils, any gformat.
gdb does a backtrace through a signal handler. It expects to find
"in main \\(.*\\) at .*main\\.c" on the stack. But on 2002-09-26,
main() moved from main.c to gdb.c, so selftest.exp is out of sync.
. long-standing bugs in gdb (broken in 5.2.1, broken in 5.2.91)
. gdb.base/printcmds.exp: ptype &*"foo"
pr gdb/538: GDB prints type of expression incorrectly
http://sources/redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=538
Happened with any gcc, any binutils, any gformat.
gdb prints the type of &*"foo" or &*ptr as "char [4]" instead of "char *".
gdb has had this bug for a long time. gdb has not changed its
behavior, but the test script improved so that the result is now
FAIL instead of XFAIL. Actually, it should be a KFAIL!
. gdb.c++/local.exp: ptype InnerLocal::NestedInnerLocal
pr gdb/482: gdb.c++/local.exp: ptype InnerLocal::NestedInnerLocal
http://sources/redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=482
Happened with any gcc, any binutils, any gformat.
gdb fails to find the nested class "InnerLocal::NestedInnerLocal".
gdb has had this bug for a long time. gdb has not changed its
behavior, but the test script has become stricter.
. gdb.c++/local.exp: ptype Local
pr gdb/483: ptype Local
http://sources/redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=483
Happend with gcc 2.95.3, any binutils, dwarf-2
This bug is annoying but harmless. Again, gdb has remained the same,
but the test script has become stricter.
. long-standing bugs in test suite (broken in 5.2.1, broken in 5.2.91)
. gdb.cpp/local.exp: ptype NestedInnerLocal
No pr filed.
Happened with gcc HEAD, any binutils, stabs+.
gdb fails to print the type of a nested class when the class is
specified without qualification. It looks like gdb is correct and
that this is probably a test script bug.
. bugs in gcc
. gdb.c++/method.exp: print this in A::bar (missing const)
no pr filed
Happened with gcc 2.95.3 or gcc 3.0.4, any binutils, stabs+.
gdb prints "this" in a constant method. gcc versions up to 3.0.4 with
-gstabs+ do not emit "const" indicators, so the output is "A *". The
output should be "const A *" or "const A * const".
gcc 3.0.4 is certainly dead, and it is probably fruitless to file a pr
against gcc 2.95.3.
gdb has the same behavior in gdb 5.2.1 and gdb 5.2.91, but the test
script became stricter. In gdb 5.2.1, this behavior gave a PASS result;
in gdb 5.2.91, it gives an XFAIL result, which is proper.
. not analyzed yet
. gdb.c++/annota.exp: annotate-quit
pr gdb/544: gdb.c++/annota2.exp: annotate-quit test sometimes fails
http://sources/redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=544
Happened with 8 configurations out of 66, no pattern.
(Configuration #: 3 7 16 27 31 34 46 63 in my test bed).
This test has been failing randomly ever since my first test baby-crib.
It might be a symptom of a real bug in gdb. Nobody knows.
. gdb.trace/actions.exp: ERROR: couldn't load /berman/fsf/_today_/berman/test/gdb.trace/actions into gdb (timed out)
No pr filed.
Happened with gcc 3.0.4, binutils HEAD, stabs+.
I think my test bed barfed. gdb did not get as far as printing its
banner! Three other configurations with gcc 3.0.4 and stabs+ worked
fine, so there is probably no gdb bug here. I am leaving this in the
"unexplained mystery" pile.