sunday project, gdb, 2003-02-05

[Michael Elizabeth Chastain <mec at shout dot net>]

Highlights of this report:

. gdb HEAD has a serious regression in java support.

. gdb.mi/mi-syn-frame.exp is a new test with a lot of FAIL results.
  It needs more work.

. gdb PR count: 572 open, 16 analyzed, 8 suspended, 10 feedback,
  0 paperwork, 423 closed.  PR gdb/701 is missing from the database.
  Query executed 2003-02-05T23:51:31Z.

. I added gcc 3.2.2-20030203, a prerelease version of gcc 3.2.2.
  I have been covering gcc gcc-3_2_branch and had no trouble with it.
  gcc 3.2.2-20030203 has no regressions versus gcc 3.2.1.
  Nice branch.

. gcc gcc-3_3-branch and gcc HEAD have many -gstabs+ regressions
  compared to gcc 3.2.1.  This has been going on for months and I need to
  focus on it.

My tables are here:

  http://www.shout.net/~mec/sunday/2003-02-05/index.html

Michael C

. Old Bugs Fixed

  Michael C fixed a test script bug in gdb.base/advance.exp and
  gdb.base/until.exp.

. New Bugs Detected

  gdb has a new bug regarding setting breakpoints in Java programs.

. Libiberty Testing

  . target=native, host=i686-pc-linux-gnu, osversion=red-hat-8.0, libc=2.2.93-5-rh
      binutils HEAD                             649 tests, 0 failures
      gcc 2.95.3, binutils HEAD                 All 616 tests passed
      gcc 3.2.1, binutils HEAD                  All 648 tests passed
      gcc 3.2.2-20030203, binutils 2.13.2.1     All 648 tests passed
      gcc 3.2.2-20030203, binutils HEAD         All 648 tests passed
      gcc 3.2.2-20030203, binutils vendor       All 648 tests passed
      gcc gcc-3_2-branch, binutils 2.13.2.1     All 648 tests passed
      gcc gcc-3_2-branch, binutils HEAD         All 648 tests passed
      gcc gcc-3_2-branch, binutils vendor       All 648 tests passed
      gcc gcc-3_3-branch, binutils 2.13.2.1     649 tests, 0 failures
      gcc gcc-3_3-branch, binutils HEAD         649 tests, 0 failures
      gcc gcc-3_3-branch, binutils vendor       649 tests, 0 failures
      gcc HEAD, binutils 2.13.2.1               649 tests, 0 failures
      gcc HEAD, binutils HEAD                   649 tests, 0 failures
      gcc HEAD, binutils vendor                 649 tests, 0 failures
      gdb gdb_5_3-branch                        All 649 tests passed
      gdb HEAD                                  649 tests, 0 failures

. Gdb Testing

  My tables are at:

    http://www.shout.net/~mec/sunday/2003-02-05/index.html

  The previous report was 2003-02-01:

    http://www.shout.net/~mec/sunday/2003-02-01/Analysis.txt

  . Counts

    gdb 5.3:             0 build aborts, 0 test aborts, 377 non-PASS results
    gdb gdb_5_3_branch:  0 build aborts, 0 test aborts, 379 non-PASS results
    gdb HEAD:            0 build aborts, 0 test aborts, 324 non-PASS results

    A non-PASS result is any result except PASS.  This includes ERROR,
    WARNING, NOTE, FAIL, KPASS, KFAIL, XPASS, XFAIL, UNRESOLVED,
    UNTESTED, UNSUPPORTED, and unknown results.

  . 5.3

    . gdb.c++/annota2.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
	Fluctuation in test result from unknown cause.

    . gdb.threads/killed.exp: GDB exits after multi-threaded program exits messily

	pr gdb/568: GDB confused by messily-exiting multi-threaded programs
	http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=568

	Jim B thinks that this test may depend on a race condition:

	  http://sources.redhat.com/ml/gdb-testers/2002-q4/msg00010.html

    . gdb.threads/schedlock.exp: *

        This test script is useless in this release because of a
        signed-versus-unsigned bug.

	Daniel J has an obvious fix, which has been applied to gdb HEAD:

	  http://sources.redhat.com/ml/gdb-patches/2002-10/msg00454.html

  . gdb_5_3-branch

    checkout date is '2003-02-05 07:32:11 UTC'.

    . gdb.c++/annota2.exp: annotate-quit
      gdb.threads/killed.exp: GDB exits after multi-threaded program exits messily
      gdb.threads/schedlock.exp: *

	Same analysis as 5.3.

  . HEAD

    checkout date is '2003-02-05 07:30:23 UTC'.

    . gdb.base/advance.exp: advance function called as param
      gdb.base/advance.exp: advance function not called by current frame
      gdb.base/advance.exp: advance with no argument
      gdb.base/advance.exp: continue to call to func3 in main
      gdb.base/until.exp: continue to call to func3 in main
	FAIL -> PASS

	Michael C fixed a bug in the test script so that it accommodates
	both gcc 2.X and gcc 3.X.  gdb itself was fine.

    . gdb.c++/anon-union.exp: *

	Michael C changed the test script to handle gcc errors
	differently.  Instead of calling gdb_suppress_entire_file, the
	script calls perror and then 'continue' past all the tests.

	This results in one ERROR instead of a WARNING followed by a
	bunch of cascade problems.  The most pernicious problem was that
	the cascades continued into the next test script!

    . gdb.c++/casts.exp: cast base class pointer to derived class pointer 
	UNRESOLVED -> PASS

	This test was unresolved because of a cascade problem from the
	previous test, gdb.c++/anon-union.exp.  That was some cascade.

    . gdb.c++/local.exp: Local out of scope
	null -> PASS
	null -> FAIL
	null -> KFAIL

	Jim B wrote a new test to cover an existing bug in gdb.  The new
	test PASSed with gcc 3.X -gdwarf-2 and KFAILed with gcc 3.X
	-gstabs+.

	  pr gdb/825: gdb gets scope of class local to a function wrong
	  http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=825

	The new test FAILed with gcc 2.X, both dwarf-2 and stabs+.
	Those should also be KFAILs.  I have committed a patch.

    . gdb.c++/local.exp: ptype Local
	FAIL -> KFAIL

	Jim B improved this test and tied it to PR gdb/483.

	  pr gdb/825: gdb.c++/local.exp: ptype Local
	  http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=825

    . gdb.c++/overload.exp: print intToChar(1)
	null -> PASS

	David C wrote a new test which PASSed in all configurations.

    . gdb.c++/ovldbreak.exp: continue to bp overloaded : char
      gdb.c++/ovldbreak.exp: continue to bp overloaded : short
      gdb.c++/ovldbreak.exp: continue to bp overloaded : signed char
      gdb.c++/ovldbreak.exp: continue to bp overloaded : unsigned char
      gdb.c++/ovldbreak.exp: continue to bp overloaded : unsigned short
	FAIL -> KFAIL

	David C wrote PR gdb/1025 for an existing bug.

	  pr gdb/1025: failures in gdb.c++/ovldbreak.exp
	  http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=1025

	There are actually two problems here with common symptoms.
	One problem is that gnu binutils 2.12.1 and earlier have bugs
	(fixed in binutils 2.13).  The second problem is an un-analyzed
	problem with gcc 2.95.3 -gdwarf-2.

    . gdb.c++/pr-1023.exp: break myClass::performBlocking
	null -> KFAIL
	null -> PASS
      gdb.c++/pr-1023.exp: break myClass::performUnblocking
	null -> PASS

	Michael C wrote a new test script which reproduces a bug
	submitted by zeitler@lucent.com.  The new script produced KFAIL
	results with gcc 2.95.3 -gstabs+.

	  pr gdb/1023: setting a breakpoint on C++ member functions
	  http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=1023

    . gdb.java/jmisc2.exp: setting breakpoint at jmisc.main(java.lang.String[])
	null -> FAIL
    . gdb.java/jmisc2.exp: p *args
      gdb.java/jmisc2.exp: p args
	PASS -> FAIL

	This is a regression in gdb.  The test script is unable to set a
	breakpoint on the main function (this action does not ordinarily
	PASS, hence the null -> FAIL transition).  The other FAILs
	cascade from there.

	  pr gdb/1039: regression: cannot set breakpoint on jmisc.main(java.lang.String[])
	  http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=gdb&cmd=view&pr=1023

    . gdb.mi/mi-syn-frame.exp: *
	null -> PASS
	null -> FAIL

	Jason M wrote a new test script.  There are some PASSes and a
	lot of FAILs, particular with gcc 3.X.  This test script needs
	more work.

    . gdb.threads/killed.exp: GDB exits after multi-threaded program exits messily
	Same analysis as 5.3.

    . gdb.threads/schedlock.exp: *

	These results are noisier than 'cp /vmlinux /dev/audio'.

	This test is still in a state where it's better to analyze the
	absolute results than to compare results from date to date.

. Test Matrix

  target     => native
  host       => i686-pc-linux-gnu
  osversion  => red-hat-8.0
  gdb        => 5.3, gdb_5_3-branch%20030205, HEAD%20030205
  gcc        => 2.95.3, 3.2-7-rh, 3.2.1, 3.2.2-20030203, gcc-3_2-branch%20030204, gcc-3_3-branch%20030204, HEAD%20030204
  binutils   => 2.13.90.0.2-rh, 2.13.2.1, HEAD%20030204
  libc       => 2.2.93-5-rh
  gformat    => dwarf-2, stabs+
  count         114 = 1 * 1 * 1 * 3 * (6*3+1*1) * 1 * 2

  'target' and 'host' are gnu configuration triples.

  'osversion' is the host operating system name, which is additional
  information beyond 'host'.

  'gdb', 'gcc', 'binutils', and 'libc' are version names.

  versions starting with a digit are official releases or snapshots.
  versions starting with a digit and ending with '-rh' are
    vendor-supplied official releases on my red hat linux host.
  versions named 'HEAD' are the cvs HEAD, also known as 'mainline' or 'trunk'.
  versions with any other name are cvs branches.
  cvs versions (head and branch) show the checkout date after a '%' delimiter.

  'gformat' is the debugging information format.

  'count' is the total number of configurations tested.
  The vendor gcc is available only with vendor binutils,
    thus the '(6*3+1*1)' term for gcc/binutils combinations.

. Baseline software

  . host=i686-pc-linux-gnu, osversion=red-hat-8.0

    make 3.79.1
    binutils 2.13.2.1
    gcc 3.2.1
    flex 2.5.4
    bison 1.35
    tcl 8.4.1
    expect 5.38.0
    dejagnu 1.4.3

    The sources.redhat.com cvs repository has its own versions of tcl,
    expect, and dejagnu.  I don't have the resources to test with both
    tcl/expect/dejagnu stacks, so I choose the stock stack for my test
    bed.
    
    The sources.redhat.com version of tcl is nearly identical to tcl
    8.4.1.  The sources.redhat.com version of expect dates from
    1998-06-15.  The sources.redhat.com version of dejagnu is nearly
    identical to dejagnu 1.4.3.

    I have packaged and published my scripts to manage the baseline
    software.  They are called Migchain (Michael's Gnu Toolchain),
    and they are licensed under the GPL.

      ftp://ftp.shout.net/pub/users/mec/migchain/migchain-0.3.tar.gz

. Test Bed Changes Since Last Report

  I implemented a suggestion from Andrew C about the way I check out
  cvs trees.  First, I run an 'ntpdate' command on my test bed host to
  adjust my clock.  Then I take the current date and subtract 180
  seconds.  Then I use this time as an argument to "cvs -D ..."
  and list the time in this report.

  In principle, anyone can reproduce my checkouts by issuing the same
  "cvs -D ...".  Or someone could tag or branch the tree at the same
  date that I checked out for testing.  This might be useful during the
  release cycle.

  The sources.redhat.com and gcc.gnu.org servers are also NTP
  synchronized, so my date argument should always be in the past
  according to the CVS server clock (the 180 seconds should be way more
  than enough to cover any residual skew between NTP'd machines).  This
  avoids a window where "cvs -D ..." might be a few seconds in the
  future on the server, and someone commits a file during that window.

  It would be nice if the CVS protocol supported "cvs time" to discover
  the current time on the server.

  I added gcc 3.2.2-20030203 as a test compiler.  I still build the test
  bed itself with gcc 3.2.1.
  
  Next report, I plan to drop gcc 3.2.1, gcc 3.2.2-20030203, and
  gcc gcc-3_2-branch in favor of gcc 3.2.2.  But for building the
  test bed itself, I will stay with gcc 3.2.1 until I get some
  spare CPU cycles.

  I changed my results cruncher so that it munges both
  'gdb.mi/mi0-*.exp' and 'gdb.mi/mi1-*.exp' to 'gdb.mi/mi[01]-*.exp'.
  This improves comparability between gdb 5.3 and gdb HEAD.