[tree-ssa] Regressions vs mainline (2004-05-07)

[Diego Novillo]

There have been several fixes in the last couple of days.  This is the
current state of regressions vs mainline:

New regressions in gcc.sum.gz:
	 FAIL: gcc.dg/debug/dwarf2/dwarf-die7.c scan-assembler 1.*DW_AT_inline
	 FAIL: 209: expected branch percentages not found: 25
	 FAIL: gcc.misc-tests/gcov-4b.c gcov: 0 failures in line counts

        The dwarf2 regression may be related to one of the PRs opened by
        mec (PR 14048, PR 14049).  I do not know yet whether it's
        reasonable to wait after the merge to fix this one (rth?).
        
        The two others require that we build a flowgraph at -O0.  I
        would like to leave these until after the merge.
        


New regressions in g++.sum.gz:
	 FAIL: g++.dg/parse/stack1.C (test for excess errors)

        This one requires an extensive modification to the gimplifier to
        avoid deep recursion.  This will have to be fixed after the
        merge.
        
	 FAIL: 258: expected branch percentages not found: 25
	 FAIL: g++.dg/gcov/gcov-1.C gcov: 0 failures in line counts
	 FAIL: 23:is 2:should be 1
	 FAIL: g++.dg/gcov/gcov-2.C gcov: 1 failures in line counts

        Similar to the GCC tests.  We need the flowgraph at -O0.  Post
        merge.
        
        
	 FAIL: g++.old-deja/g++.brendan/crash13.C  (test for errors
	 FAIL: g++.old-deja/g++.bugs/900205_03.C  (test for errors
	 FAIL: g++.old-deja/g++.bugs/900205_03.C  (test for errors
	 FAIL: g++.old-deja/g++.bugs/900205_03.C  (test for errors
	 FAIL: g++.old-deja/g++.bugs/900205_03.C  (test for errors
	 FAIL: g++.old-deja/g++.law/friend5.C  (test for errors
	 FAIL: g++.old-deja/g++.law/friend5.C  (test for errors
	 FAIL: g++.old-deja/g++.mike/p646.C  (test for warnings
	 FAIL: g++.old-deja/g++.mike/p646.C  (test for warnings
	 FAIL: g++.old-deja/g++.mike/p646.C  (test for warnings
	 FAIL: g++.old-deja/g++.mike/p646.C  (test for warnings
	 FAIL: g++.old-deja/g++.mike/p646.C  (test for warnings

        These are diagnostics that are either not emitted or at
        different lines.  Jason, are they fixable pre-merge?
        


New regressions in libffi.sum.gz:
        NONE



New regressions in libjava.sum.gz:
	 FAIL: Array_3 execution - gij test
	 FAIL: Array_3 execution - gij test

        In [link] aph says that it's a C++ bug.  We are generating bad
        EH code.  rth?
        
        
	 FAIL: PR4766 -O3 compilation from source
	 FAIL: String_overflow -O3 compilation from source
	 FAIL: err3 output - source compiled test
	 FAIL: err3 -O3 output - source compiled test

        Andrew, what should we do with these?
        

Fixed regressions in libjava.sum.gz:
	 FAIL: Thread_Monitor execution - gij test

        At least one test where the branch does better.
        


New regressions in libstdc++-abi.sum.gz:
	NONE


New regressions in libstdc++.sum.gz:
	NONE


New regressions in objc.sum.gz:
	NONE


Diego.