[Bug c++/14108] New: ICE on simple testcase

[Bug c++/14108] New: ICE on simple testcase

[altacsd at yahoo dot fr]

I couldn't find a record for this bug, even if #12103 looked similar.
Here is the reduced testcase I made :

class ClassC {
public:
  ~ClassC();
};

class ClassA {
  virtual ClassC f();
};

class ClassB : public virtual ClassA {
  virtual ClassC f();
};

ClassC ClassB::f() {
  return ClassC();
}

The G++ error returned :

$ sparc-sun-solaris2.8-g++ -c bugtest.cpp
bugtest.cpp: In member function `ClassC ClassB::_ZTv0_n12_N6ClassB1fEv()':
bugtest.cpp:17: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

It seems to be related to the ClassC destructor, the 'public virtual'
inheritance of ClassA in ClassB, and the definition of ClassB::f() outside of
ClassB. They are all necessary to reproduce the error.

It concerns gcc-3.4-20040128 cross-compiled for solaris2.8 with :
configure --prefix="/cygdrive/d/work/MSYS/home/cross-builds/builds-sparc"
--target=sparc-sun-solaris2.8
--with-sysroot="/cygdrive/d/work/MSYS/home/cross-builds/arch-original/sparc-runtime"
--with-gnu-as --with-gnu-ld --without-newlib --enable-threads --disable-nls

I don't have any other gcc 3.4 to test with yet.
The testcase can be compiled without error with gcc-3.2-20020927 (cygwin target).

Regards,
Emmanuel

-- 
           Summary: ICE on simple testcase
           Product: gcc
           Version: 3.4.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: altacsd at yahoo dot fr
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-cygwin
  GCC host triplet: i686-pc-cygwin
GCC target triplet: sparc-sun-solaris2.8


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] ICE on simple testcase

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-02-11 16:41 -------
I do not see in "3.5.0 20040204".

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] ICE on simple testcase

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-02-11 16:50 -------
Can you try a newer 3.4.0 as there have been some C++ fixes there also?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] ICE on simple testcase

[altacsd at yahoo dot fr]

------- Additional Comments From altacsd at yahoo dot fr  2004-02-11 16:57 -------
Same problem with gcc-3.4-20040206.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-02-11 17:47 -------
I know what the problem is now.  The patch in <http://gcc.gnu.org/ml/gcc-patches/2004-02/
msg01037.html> should fix it.  Jan please close this PR when you commit that patch to 3.4.0.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org, hubicka at gcc dot gnu
                   |                            |dot org
         AssignedTo|unassigned at gcc dot gnu   |hubicka at gcc dot gnu dot
                   |dot org                     |org
             Status|WAITING                     |ASSIGNED
           Keywords|                            |ice-on-valid-code
      Known to fail|                            |3.4.0 3.5.0
      Known to work|                            |3.3.3
            Summary|ICE on simple testcase      |[3.4/3.5 regression] ICE on
                   |                            |simple testcase
   Target Milestone|---                         |3.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-02-11 21:10 -------
No, Jan's patch doesn't fix this problem.

I get with checking enabled on 3.4 branch and mainline, with or without Jan's patch.

[eric@localhost obj.sparc]$ gcc/xgcc -Bgcc -S pr14108.C
pr14108.C: In member function `ClassC ClassB::_ZTv0_n12_N6ClassB1fEv()':
pr14108.C:16: internal compiler error: tree check: expected tree_list, have
function_decl in friend_accessible_p, at cp/search.c:868
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-02-11 21:15 -------
Note I do not see this on either powerpc-apple-darwin or i686-pc-linux-gnu.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-02-11 22:01 -------
I guess this is again a case where the C++ front-end doesn't behave the same on
all platforms because of different macros.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-02-12 16:18 -------
Jan, the backtrace contains again use_thunk and emit_associated_thunks.  It
appears that your thunks-related patch contains implicit assumptions that are
valid on x86 but not on other platforms.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-02-12 16:22 -------
Thunks are broken on SPARC.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |critical


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-02-12 19:04 -------
Well, reverting

2004-01-28  Jan Hubicka  <jh@suse.cz>
        PR c++/12850
        * cgraph.c (cgraph_remove_node): Clear out saved/insns/arguments and
        initial pointers.
        * cgraphunit.c (cgraph_finalize_function): Clear out DECL_SAVED_INSNS
        for functions that will be only inlined.
        (cgraph_mark_function_to_output): Likewise.
        (cgraph_expand_function): Sanity check that DECL_DEFER_OUTPUT is clear;
        do not clear function body.
        * tree-optimize.c (clear_decl_rtl): Use decl_function_context.
        (tree_rest_of_compilation): Reorganize the logic releasing function
        body to use callgraph datastructure.

doesn't help.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-02-13 08:07 -------
Looks like a pure C++ front-end problem: it appears that the parser put a
FUNCTION_DECL as the DECL_BEFRIENDING_CLASSES of another FUNCTION_DECL.

I guess we need a specialist of the C++ front-end at this point.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|hubicka at gcc dot gnu dot  |unassigned at gcc dot gnu
                   |org                         |dot org
             Status|ASSIGNED                    |NEW


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-02-13 19:01 -------
The problem is that the THUNK_TARGET macro and DECL_BEFRIENDING_CLASSES macro
reference the same field in the nested structure.  Now the former is set to the
thunked-to function in make_thunk by

  THUNK_TARGET (thunk) = function;

so ends up holding a DECL, while the latter expects a TREE_LIST.


There seems to be a contradiction in cp-tree.h between

      struct full_lang_decl 
      {
	/* For a non-thunk function decl, this is a tree list of
  	   friendly classes. For a thunk function decl, it is the
  	   thunked to function decl.  */
	tree befriending_classes;

and

   A thunk is an alternate entry point for an ordinary FUNCTION_DECL.
   The address of the ordinary FUNCTION_DECL is given by the
   DECL_INITIAL, which is always an ADDR_EXPR whose operand is a
   FUNCTION_DECL.


The following ad-hoc patch fixes the ICE:

Index: search.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/cp/search.c,v
retrieving revision 1.284
diff -u -p -r1.284 search.c
--- search.c    19 Dec 2003 23:28:10 -0000      1.284
+++ search.c    13 Feb 2004 19:00:39 -0000
@@ -856,7 +856,7 @@ friend_accessible_p (tree scope, tree de
   if (!scope)
     return 0;

-  if (TREE_CODE (scope) == FUNCTION_DECL
+  if ((TREE_CODE (scope) == FUNCTION_DECL && !DECL_THUNK_P (scope))
       || DECL_FUNCTION_TEMPLATE_P (scope))
     befriending_classes = DECL_BEFRIENDING_CLASSES (scope);
   else if (TYPE_P (scope))


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[mmitchel at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |mark at codesourcery dot com
                   |dot org                     |
             Status|NEW                         |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-02-14 00:49 -------
Subject: Bug 14108

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	mmitchel@gcc.gnu.org	2004-02-14 00:49:15

Modified files:
	gcc/cp         : ChangeLog cp-tree.h decl2.c search.c typeck.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/g++.dg/inherit: thunk2.C 
	gcc/testsuite/g++.dg/template: array4.C 

Log message:
	PR c++/14122
	* cp-tree.h (delete_sanity): Change prototype.
	* decl2.c (delete_sanity): Make doing_vec a bool, not an int.
	Remove dead code.  Adjust code to warn about deleting an array.
	* typekc.c (decay_conversion): Use build_address and build_nop.
	
	PR c++/14108
	* search.c (accessible_p): Do not check access in thunks.
	
	PR c++/14122
	* g++.dg/template/array4.C: New test.
	
	PR c++/14108
	* g++.dg/inherit/thunk2.C: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/ChangeLog.diff?cvsroot=gcc&r1=1.3953&r2=1.3954
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/cp-tree.h.diff?cvsroot=gcc&r1=1.952&r2=1.953
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/decl2.c.diff?cvsroot=gcc&r1=1.698&r2=1.699
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/search.c.diff?cvsroot=gcc&r1=1.286&r2=1.287
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/typeck.c.diff?cvsroot=gcc&r1=1.527&r2=1.528
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.3496&r2=1.3497
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/inherit/thunk2.C.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/array4.C.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-02-14 02:33 -------
Subject: Bug 14108

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_4-branch
Changes by:	mmitchel@gcc.gnu.org	2004-02-14 02:32:54

Modified files:
	gcc/cp         : ChangeLog cp-tree.h decl2.c search.c typeck.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/g++.dg/inherit: thunk2.C 
	gcc/testsuite/g++.dg/template: array4.C 

Log message:
	PR c++/14122
	* cp-tree.h (delete_sanity): Change prototype.
	* decl2.c (delete_sanity): Make doing_vec a bool, not an int.
	Remove dead code.  Adjust code to warn about deleting an array.
	* typekc.c (decay_conversion): Use build_address and build_nop.
	
	PR c++/14108
	* search.c (accessible_p): Do not check access in thunks.
	
	PR c++/14122
	* g++.dg/template/array4.C: New test.
	
	PR c++/14108
	* g++.dg/inherit/thunk2.C: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.3892.2.50&r2=1.3892.2.51
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/cp-tree.h.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.946.4.4&r2=1.946.4.5
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/decl2.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.695.4.2&r2=1.695.4.3
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/search.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.284&r2=1.284.4.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/cp/typeck.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.519.2.7&r2=1.519.2.8
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.3389.2.74&r2=1.3389.2.75
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/inherit/thunk2.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=NONE&r2=1.1.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/g++.dg/template/array4.C.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=NONE&r2=1.1.2.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108

[Bug c++/14108] [3.4/3.5 regression] ICE on simple testcase

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-02-14 02:36 -------
Fixed in GCC 3.4 and GCC 3.5.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14108