gcc 3.0.4 versus gcc gcc-3_1-branch%20020404 (native i686-pc-linux-gnu)

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

Here is a report comparing gcc 3.0.4 versus gcc gcc-3_1-branch%20020404.
It is based on these tables:

  http://www.shout.net/~mec/sunday/2002-04-04/Differences-compare-by-gcc.html

There are no show-stoppers here (especially since we don't have a lot of
power over the gcc 3.1 release).  I'd like someone to look at the problem
in gdb.base/funcargs.exp and maybe gdb.base/corefile.exp.  But if you
don't, that's fine, the test results will keep tracking the failures.

The configurations are:

  target   = native
  host     = i686-pc-linux-gnu%rh-7.2
  gdb      = HEAD%20020404
  gcc      = 3.0.4, gcc-3_1-branch%20020404
  binutils = HEAD%20020404
  goption  = -gdwarf-2, -gstabs+

I am looking at table #4 and table #5 (the tables for gdb HEAD),
Table #4 is -gdwarf-2; table #5 is -gstabs+.

I am looking at columns [3] and [4] (gcc 3.0.4 versus gcc-3_1-branch%20020404).

I tested languages C, C++, and assembly.  I did not test languages Java,
FORTRAN, Pascal, Ada, Objective-C, or Chill.

  result improved
    gdb.base/constvars.exp	   +5
    gdb.base/structs2.exp       +4 +5
    gdb.base/volatile.exp          +5
    gdb.trace/packetlen.exp     +4 +5

  result regressed
    gdb.base/condbreak.exp      -4
    gdb.base/corefile.exp       -4 -5
    gdb.base/ena-dis-br.exp     -4 -5
    gdb.base/funcargs.exp       -4
    gdb.c++/method.exp             -5
    gdb.c++/ovldbreak.exp       -4
    gdb.c++/templates.exp	   -5

  noise
    gdb.base/ending-run.exp     ,4 ,5
    gdb.base/gcore.exp          ,4 ,5
    gdb.c++/annota2.exp            ,5

  key

    +  result improved
    -  result regressed
    ,  noise

The individual regressions are:



  . gdb.base/condbreak.exp: run until breakpoint at marker1
    gdb.base/condbreak.exp: run until breakpoint at marker2
      -gdwarf-2 : PASS -> XFAIL

    This is an instance of PR gdb/460.



  . gdb.base/corefile.exp: print func2::coremaker_local
      -gdwarf-2 : PASS -> FAIL
      -gstabs+  : PASS -> FAIL

    Here is an excerpt from gdb.log (both versions are the same):

      print func2::coremaker_local^M
      No frame is currently executing in block func2.^M
      (gdb) FAIL: gdb.base/corefile.exp: print func2::coremaker_local

    I don't know if this is a gcc bug or a gdb bug.

    This test has been failing in both of these configurations since my
    earliest test run (2002-02-26).



  . gdb.base/ena-dis-br.exp: continue to auto-disabled break marker2
      -gdwarf-2 : PASS -> XFAIL
    gdb.base/ena-dis-br.exp: continue to break marker1
      -gdwarf-2 : PASS -> XFAIL
    gdb.base/ena-dis-br.exp: continue to break marker1, 2nd time
      -gdwarf-2 : PASS -> XFAIL
    gdb.base/ena-dis-br.exp: continue with ignore count, not stopped at bpt
      -gdwarf-2 : PASS -> XFAIL
      -gstabs+  : PASS -> XFAIL

    Not analyzed yet.



  . gdb.base/funcargs.exp: run to call5a
    gdb.base/funcargs.exp: print st
      -gdwarf-2 : PASS -> FAIL

    gdb prints the wrong value for a structure argument.  This works fine
    with gcc 3.0.4 -gdwarf-2.

    Log excerpt:

      Starting program: /berman/fsf/_today_/berman/test/gdb.base/funcargs 
      
      Breakpoint 25, 0x08048ae6 in call5a (st={s1 = 1073834212, s2 = -1073743804}) at /berman/fsf/_today_/source/gdb/HEAD/src/gdb/testsuite/gdb.base/funcargs.c:345
      345       {st.s1 = 5;}
      +(gdb) FAIL: gdb.base/funcargs.exp: run to call5a
      print st
      $31 = {s1 = 1073834212, s2 = -1073743804}
      (gdb) FAIL: gdb.base/funcargs.exp: print st



  . gdb.c++/method.exp: print this (in foo)
    gdb.c++/method.exp: print this (in bar)
    gdb.c++/method.exp: print this in getFunky
      -gstabs+ : PASS -> FAIL

    gcc improved its behavior by emitting "const" markers,
    but the gdb test suite can't handle it.



  . gdb.c++/ovldbreak.exp: continue to bp overloaded : char
    gdb.c++/ovldbreak.exp: continue to bp overloaded : double
    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
      -gdwarf-2 : PASS -> XFAIL

    This is an instance of PR gdb/460.



  . gdb.c++/templates.exp: print Garply<Garply<char> >::garply
      -gstabs+ : PASS -> FAIL

    gcc improved its behavior by emitting "const" markers,
    but the gdb test suite can't handle it.