[Bug middle-end/16968] New: RTX_UNCHANGING_P related miscompilation

[Bug middle-end/16968] New: RTX_UNCHANGING_P related miscompilation

[jakub at gcc dot gnu dot org]

On ppc (and ppc64 -m32) following testcase is miscompiled with -O2
in at least 3.4.2 20040810 (prerelease) and 3.4.1 20040714 (Red Hat 3.4.1-7).

In RTL dumps the problem is immediately obvious, t initialization is done with
MEM/s (note missing /u), while later on reads from it are MEM/s/u.

-- 
           Summary: RTX_UNCHANGING_P related miscompilation
           Product: gcc
           Version: 3.4.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: middle-end
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: jakub at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org
GCC target triplet: powerpc-redhat-linux


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug middle-end/16968] RTX_UNCHANGING_P related miscompilation

[jakub at gcc dot gnu dot org]

------- Additional Comments From jakub at gcc dot gnu dot org  2004-08-10 15:44 -------
Created an attachment (id=6911)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=6911&action=view)
ppc-const.c testcase


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] RTX_UNCHANGING_P related miscompilation

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-08-10 15:45 -------
I really wish RTX_UNCHANGING_P would die now.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] RTX_UNCHANGING_P related miscompilation

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|middle-end                  |rtl-optimization
           Keywords|                            |wrong-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] RTX_UNCHANGING_P related miscompilation

[jakub at gcc dot gnu dot org]

------- Additional Comments From jakub at gcc dot gnu dot org  2004-08-10 15:46 -------
*** Bug 16969 has been marked as a duplicate of this bug. ***

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] RTX_UNCHANGING_P related miscompilation

[jakub at gcc dot gnu dot org]

------- Additional Comments From jakub at gcc dot gnu dot org  2004-08-10 15:48 -------
On mainline the initialization seems to be done with MEM/s/u, so it looks like
it is 3.4.x only (not sure if it is tree-ssa which cures this or there were
RTX_UNCHANGING_P changes I did not know about).

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] RTX_UNCHANGING_P related miscompilation

[jakub at gcc dot gnu dot org]

------- Additional Comments From jakub at gcc dot gnu dot org  2004-08-10 16:02 -------
On closer look, this is because loop optimizer doesn't care about asm ("")
being an optimization barrier and happily moves instructions accross it.
It seems at least on this testcase no other optimization attempts to do so
and the testcase works just fine with -O2 -m32 -fno-loop-optimize.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 only] RTX_UNCHANGING_P related miscompilation

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|RTX_UNCHANGING_P related    |[3.4 only] RTX_UNCHANGING_P
                   |miscompilation              |related miscompilation
   Target Milestone|---                         |3.4.2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[giovannibajo at libero dot it]

------- Additional Comments From giovannibajo at libero dot it  2004-08-16 10:24 -------
Andrew, can you please confirm this on PPC?

Zdenek, this looks like a miscompilation in the RTL loop optimizer. Can you 
please have a look?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org, rakdver at atrey dot
                   |                            |karlin dot mff dot cuni dot
                   |                            |cz
      Known to fail|                            |3.5.0
      Known to work|                            |3.4.0 3.4.1
            Summary|[3.4 only] RTX_UNCHANGING_P |[3.4 Regression] loop
                   |related miscompilation      |optimizer miscompilation


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[jakub at gcc dot gnu dot org]

------- Additional Comments From jakub at gcc dot gnu dot org  2004-08-16 10:31 -------
It is miscompiled in 3.4.x, so setting known to work to 3.4.0 and 3.4.1 seems
wrong to me.
GCC 3.5 doesn't miscompile this (it doesn't use asm_input ("") magic),
but the loop bug which is the cause of this problem is most probably there
too (i.e. I think it will happily move asm volatile or asm without output
around while it shouldn't).

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-08-29 19:02 -------
Postponed until GCC 3.4.3.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.2                       |3.4.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[pinskia at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
      Known to fail|4.0                         |4.0.0
   Last reconfirmed|0000-00-00 00:00:00         |2004-10-22 17:38:25
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[mmitchel at gcc dot gnu dot org]

------- Additional Comments From mmitchel at gcc dot gnu dot org  2004-11-01 00:46 -------
Postponed until GCC 3.4.4.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4.3                       |3.4.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-12-10 10:29 -------
Investigating, RTX_UNCHANGING_P was my preferred RTL flag. ;-)


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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-12-13 11:02 -------
Hum... the problem lies in the old loop optimizer, which indeed doesn't care
about optimization barriers.  This apparently had had very limited consequences
until we started to emit them more frequently as part of the stopgap solution
for the RTX_UNCHANGING_P.  We might need to resort to disabling all
optimizations if we detect such a barrier in a loop.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-12-17 17:11 -------
Present on x86 and x86-64 too as of today on 3.4 branch.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
 GCC target triplet|powerpc-redhat-linux        |
   Last reconfirmed|2004-10-22 17:38:25         |2004-12-17 17:11:29
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-18 07:55 -------
Subject: Bug 16968

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	ebotcazou@gcc.gnu.org	2004-12-18 07:55:42

Modified files:
	gcc            : ChangeLog loop.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.c-torture/execute: 20041218-1.c 

Log message:
	PR rtl-optimization/16968
	* loop.c (scan_loop): Stop scanning the loop for movable
	insns as soon as an optimization barrier is encountered.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.6872&r2=2.6873
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/loop.c.diff?cvsroot=gcc&r1=1.517&r2=1.518
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.4779&r2=1.4780
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/execute/20041218-1.c.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-12-18 07:58 -------
Subject: Bug 16968

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_4-branch
Changes by:	ebotcazou@gcc.gnu.org	2004-12-18 07:58:12

Modified files:
	gcc            : ChangeLog loop.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.c-torture/execute: 20041218-1.c 

Log message:
	PR rtl-optimization/16968
	* loop.c (scan_loop): Stop scanning the loop for movable
	insns as soon as an optimization barrier is encountered.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=2.2326.2.746&r2=2.2326.2.747
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/loop.c.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.488.2.5&r2=1.488.2.6
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_4-branch&r1=1.3389.2.331&r2=1.3389.2.332
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/execute/20041218-1.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=16968

[Bug rtl-optimization/16968] [3.4 Regression] loop optimizer miscompilation

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-12-18 08:00 -------
See http://gcc.gnu.org/ml/gcc-patches/2004-12/msg01315.html


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


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16968