[Bug c++/14755] New: miscompilation in bitfielded signed integers

[Bug c++/14755] New: miscompilation in bitfielded signed integers

[mueller at kde dot org]

Hi,  
 
this testcase is miscompiled here (with g++ -Wall -W -pedantic -v):  
 
Configured with: ../configure --prefix=/opt/gcc --enable-c99 --enable-long-long 
--enable-checking --enable-shared --enable-threads --enable-languages=c,c++ 
Thread model: posix 
gcc version 3.4.0 20040327 (prerelease) 
 
=== Cut === 
#include <stdio.h> 
#include <sys/types.h> 
#include <stdlib.h> 
 
struct QtFontStyle 
{ 
    QtFontStyle( ) 
        : count( 0 ) 
    {} 
 
    ~QtFontStyle() { 
        while ( count-- ) { 
             abort(); 
             break; 
        } 
    } 
 
    signed int count  : 31; 
}; 
 
int main() 
{ 
    QtFontStyle* qf = new QtFontStyle; 
 
    delete qf; 
} 
=== Cut ===

-- 
           Summary: miscompilation in bitfielded signed integers
           Product: gcc
           Version: 3.4.0
            Status: UNCONFIRMED
          Severity: critical
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: mueller at kde dot org
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux-gnu
  GCC host triplet: i686-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] miscompilation in bitfielded signed integers

[mueller at kde dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.4 regression] miscompilation in bitfielded signed integers

[mueller at kde dot org]

------- Additional Comments From mueller at kde dot org  2004-03-27 16:23 -------
it doesn't appear with 3.3.x. 3.5 not tested.  

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|miscompilation in bitfielded|[3.4 regression]
                   |signed integers             |miscompilation in bitfielded
                   |                            |signed integers


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug middle-end/14755] [3.4 regression] miscompilation in bitfielded signed integers

[falk at debian dot org]

------- Additional Comments From falk at debian dot org  2004-03-27 16:30 -------
Confirmed. A simple C test case is:

int main() {
    struct { int count: 31; } s = { 0 };
    while (s.count--)
	abort();
}

I see this in 3.3.3 too, but not 2.95...

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
          Component|c++                         |middle-end
     Ever Confirmed|                            |1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.4/3.5 regression] miscompilation in bitfielded signed integers

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-03-27 16:31 -------
The problem looks like fold is converting (count--)!=0 to --count != 2147483647.
Here is a simple example from Falk:
int main() {  struct { signed int count: 31; } s; s.count = 0; while (s.count--) abort(); }

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mmitchel at gcc dot gnu dot
                   |                            |org
          Component|middle-end                  |c++
      Known to fail|                            |3.4.0 3.5.0
      Known to work|                            |3.3
   Last reconfirmed|0000-00-00 00:00:00         |2004-03-27 16:31:47
               date|                            |
            Summary|[3.4 regression]            |[3.4/3.5 regression]
                   |miscompilation in bitfielded|miscompilation in bitfielded
                   |signed integers             |signed integers
   Target Milestone|---                         |3.4.1


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.4/3.5 regression] miscompilation in bitfielded signed integers

[mueller at kde dot org]

------- Additional Comments From mueller at kde dot org  2004-03-27 16:58 -------
why is this not targetted against 3.4.0 ? 
 
this is a severe miscompilation, you cannot run  
any of Qt or KDE this way :( 
 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.4/3.5 regression] miscompilation in bitfielded signed integers

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-03-27 17:03 -------
Because of the policy set by the release manager, see <http://gcc.gnu.org/ml/gcc/2004-03/
msg01259.html>:
Furthermore, please do not create any new PRs targeted for 3.4 without
my explicit permission.  If it's a regression, target it for 3.4.1.
If you think it might need to be fixed in 3.4, add me to the CC list,
and add a note asking me to move back the target.  Please do not do
this unless the PR is wrong-code, ICE-on-valid, or bootstrap for a
primary target.  New PRs referring to other categories of error are
simply not going to get fixed for 3.4.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.4/3.5 regression] miscompilation in bitfielded signed integers

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-03-27 17:07 -------
Caused by:
2004-03-21  Roger Sayle  <roger@eyesopen.com>

        * fold-const.c (fold) <EQ_EXPR>: Rewrite optimization to transform
        "foo++ == const" into "++foo == const+incr".

Mark, this is a very serious regression, can target 3.4.0.
Thanks.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |roger at eyesopen dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.3/3.4/3.5 regression] miscompilation in bitfielded signed integers

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-03-27 17:13 -------
The patch went into for 3.3.4 also so retargeting for now.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|3.4.0 3.5.0                 |3.4.0 3.5.0 3.3.4
            Summary|[3.4/3.5 regression]        |[3.3/3.4/3.5 regression]
                   |miscompilation in bitfielded|miscompilation in bitfielded
                   |signed integers             |signed integers
   Target Milestone|3.4.1                       |3.3.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.3/3.4/3.5 regression] miscompilation in bitfielded signed integers

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-03-30 17:15 -------
Yes, this should be addressed before 3.4.0.  I have retargeted the PR. 

Is there a patch available?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.3.4                       |3.4.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.3/3.4/3.5 regression] miscompilation in bitfielded signed integers

[mueller at kde dot org]

------- Additional Comments From mueller at kde dot org  2004-03-30 17:35 -------
the problem seems to be that the mask is incorrectly calculated,  
it forgets the "sign bit".  
 
does gcc only support 2-complement signed integer handling  
or does it also have support for other weird representations? 
 
otherwise its just adjusting the mask in line 7276 by another -1 
for signed integer 

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.3/3.4/3.5 regression] miscompilation in bitfielded signed integers

[jakub at gcc dot gnu dot org]

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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.3/3.4/3.5 regression] miscompilation in bitfielded signed integers

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-04-01 15:50 -------
Subject: Bug 14755

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	jakub@gcc.gnu.org	2004-04-01 15:50:12

Modified files:
	gcc            : ChangeLog fold-const.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.c-torture/execute: 20040331-1.c 
	gcc/testsuite/gcc.dg: 20040331-1.c 

Log message:
	PR c++/14755
	* fold-const.c (fold) <EQ_EXPR>: Properly compute newconst in
	"bitfld++ == const" to "++bitfld == const + incr" transformations.
	
	* gcc.c-torture/execute/20040331-1.c: New test.
	* gcc.dg/20040331-1.c: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.3323&r2=2.3324
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/fold-const.c.diff?cvsroot=gcc&r1=1.363&r2=1.364
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.3648&r2=1.3649
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/execute/20040331-1.c.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.dg/20040331-1.c.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.3/3.4/3.5 regression] miscompilation in bitfielded signed integers

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-04-01 16:09 -------
Subject: Bug 14755

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_4-branch
Changes by:	jakub@gcc.gnu.org	2004-04-01 16:09:15

Modified files:
	gcc            : ChangeLog fold-const.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.c-torture/execute: 20040331-1.c 
	gcc/testsuite/gcc.dg: 20040331-1.c 

Log message:
	PR c++/14755
	* fold-const.c (fold) <EQ_EXPR>: Properly compute newconst in
	"bitfld++ == const" to "++bitfld == const + incr" transformations.
	
	* gcc.c-torture/execute/20040331-1.c: New test.
	* gcc.dg/20040331-1.c: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=2.2326.2.374&r2=2.2326.2.375
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/fold-const.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.322.2.9&r2=1.322.2.10
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.3389.2.159&r2=1.3389.2.160
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/execute/20040331-1.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/gcc.dg/20040331-1.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=14755

[Bug c++/14755] [3.3/3.4/3.5 regression] miscompilation in bitfielded signed integers

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-04-01 16:27 -------
Subject: Bug 14755

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_3-branch
Changes by:	jakub@gcc.gnu.org	2004-04-01 16:27:07

Modified files:
	gcc            : ChangeLog fold-const.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.c-torture/execute: 20040331-1.c 

Log message:
	PR c++/14755
	* fold-const.c (fold) <EQ_EXPR>: Properly compute newconst in
	"bitfld++ == const" to "++bitfld == const + incr" transformations.
	
	* gcc.c-torture/execute/20040331-1.c: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_3-branch&r1=1.16114.2.962&r2=1.16114.2.963
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/fold-const.c.diff?cvsroot=gcc&only_with_tag=gcc-3_3-branch&r1=1.227.2.6&r2=1.227.2.7
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_3-branch&r1=1.2261.2.371&r2=1.2261.2.372
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/execute/20040331-1.c.diff?cvsroot=gcc&only_with_tag=gcc-3_3-branch&r1=NONE&r2=1.1.4.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755

[Bug c++/14755] [3.3/3.4/3.5 regression] miscompilation in bitfielded signed integers

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-04-01 18:59 -------
Fixed for 3.3.4, 3.4.0, and 3.5.0.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED
   Target Milestone|3.4.0                       |3.3.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=14755