[Bug target/43671] New: -fsched2-use-superblocks -m32 causes FAIL: gcc.dg/pr32912-1.c execution test

[Bug target/43671] New: -fsched2-use-superblocks -m32 causes FAIL: gcc.dg/pr32912-1.c execution test

[zsojka at seznam dot cz]

Command line:
gcc -O1 -fschedule-insns2 -fsched2-use-superblocks -m32 testcase.c && ./a.out
or
gcc -O2 -fsched2-use-superblocks -m32 testcase.c && ./a.out

Tested revisions:
r157965 - fail
r157877 - fail
r153685 - OK
4.4.3, 4.3.4, 4.2.4, 4.1.2, 3.4.6 - OK

Output:
$ /mnt/svn/gcc-trunk/binary-157965-lto/bin/gcc -O1 -fschedule-insns2
-fsched2-use-superblocks -m32 testcase.c && ./a.out
Aborted

The problem is the value is read before stored:
(reduced testcase, -O1 version)
...
        pxor    %xmm0, %xmm0    # tmp63
...
        cmpl    $0, 4(%esp)     #, u.i
        movdqa  %xmm0, (%esp)   # tmp63, u.v
        jne     .L2     #,
...
.L2:
        call    abort   #


-- 
           Summary: -fsched2-use-superblocks -m32 causes FAIL:
                    gcc.dg/pr32912-1.c execution test
           Product: gcc
           Version: 4.5.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: target
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: zsojka at seznam dot cz
  GCC host triplet: x86_64-pc-linux-gnu
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] -fsched2-use-superblocks -m32 causes FAIL: gcc.dg/pr32912-1.c execution test

[zsojka at seznam dot cz]

------- Comment #1 from zsojka at seznam dot cz  2010-04-07 01:19 -------
Created an attachment (id=20325)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20325&action=view)
reduced executable testcase

$ gcc -O1 -fschedule-insns2 -fsched2-use-superblocks -m32 pr43671.c && ./a.out
Aborted


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] [4.5 Regression] -fsched2-use-superblocks -m32 -finline-functions causes FAIL: gcc.dg/pr32912-1.c execution test

[zsojka at seznam dot cz]

------- Comment #2 from zsojka at seznam dot cz  2010-04-07 02:07 -------
Created an attachment (id=20326)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=20326&action=view)
another testcase, probably better

copied from gcc/testsuite/gcc.dg/tree-ssa/gen-vect-32.c (gen-vect-2.c fails the
same way), but removed headers to prevent confusion

Issue is the same, reading value before store:
...
        cmpb    $5, 1(%esp)     #, s.ca
        movdqa  .LC0, %xmm0     #, tmp78
        movdqa  %xmm0, (%esp)   # tmp78,
        jne     .L2     #,
...

Command line:
gcc -O3 -fsched2-use-superblocks -m32 && ./a.out

flags can be changed to:
-O1 -fpeel-loops -fschedule-insns2 -ftree-vectorize -fsched2-use-superblocks
-m32

This testcase fails in 4.4.3 too (but 4.3.4 works)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[rguenth at gcc dot gnu dot org]

------- Comment #3 from rguenth at gcc dot gnu dot org  2010-04-07 09:12 -------
Confirmed.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2010-04-07 09:12:13
               date|                            |
   Target Milestone|---                         |4.4.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[hjl dot tools at gmail dot com]

------- Comment #4 from hjl dot tools at gmail dot com  2010-04-07 18:36 -------
It is caused by revision 138953:

http://gcc.gnu.org/ml/gcc-cvs/2008-08/msg00512.html


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |matz at suse dot de


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[matz at gcc dot gnu dot org]

------- Comment #5 from matz at gcc dot gnu dot org  2010-04-08 13:40 -------
Um, how can r138953 be the reason when (as the original report says) it's
still okay with r153685?  In any case, my patch just shuffled around the
activation/non-activation of the scheduler, so if at all it exposed a latent
bug in it, which given that -fsched2-use-superblocks isn't used by default
is very likely.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[zsojka at seznam dot cz]

------- Comment #6 from zsojka at seznam dot cz  2010-04-08 13:48 -------
I am sorry for the confusion. The original testcase from comment #1 doesn't
fail in r153685 and 4.4.3. Later I found different testcase (comment #2) that
fails there.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[jakub at gcc dot gnu dot org]

-- 

jakub at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|4.4.4                       |4.4.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug target/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[hjl dot tools at gmail dot com]

------- Comment #7 from hjl dot tools at gmail dot com  2010-05-02 20:09 -------
It is a bug in true_dependence. We get

Breakpoint 7, true_dependence (mem=0x7ffff19246d8, mem_mode=SImode, 
    x=0x7ffff1924918, varies=0x93bbea <rtx_varies_p>)
    at /export/gnu/import/git/gcc/gcc/alias.c:2360
2360      x_addr = get_addr (XEXP (x, 0));
(mem/s/j:QI (plus:SI (value:SI 7:7854 @0x17f5800/0x17f56a0)
        (const_int 1 [0x1])) [0 s.ca+1 S1 A8])
(mem:SI (value:SI 7:7854 @0x17f5800/0x17f56a0) [0 *vect_ps.7_1+0 S4 A32])
(gdb) 

(gdb) call debug_rtx (x_addr)
(plus:SI (value:SI 7:7854 @0x17f5800/0x17f56a0)
    (const_int 1 [0x1]))
(gdb) call debug_rtx (mem_addr)
(plus:SI (value:SI 5:3935 @0x17f57d0/0x17f5640)
    (const_int -16 [0xfffffffffffffff0]))
(gdb) 

We think x and mem don't overlap. It is clearly wrong.


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rguenth at gcc dot gnu dot
                   |                            |org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug middle-end/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[hjl dot tools at gmail dot com]

------- Comment #8 from hjl dot tools at gmail dot com  2010-05-02 22:22 -------
A patch is posted at

http://gcc.gnu.org/ml/gcc-patches/2010-05/msg00089.html


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |http://gcc.gnu.org/ml/gcc-
                   |                            |patches/2010-
                   |                            |05/msg00089.html
          Component|target                      |middle-end


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug middle-end/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[hjl at gcc dot gnu dot org]

------- Comment #9 from hjl at gcc dot gnu dot org  2010-05-04 15:47 -------
Subject: Bug 43671

Author: hjl
Date: Tue May  4 15:47:25 2010
New Revision: 159033

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=159033
Log:
Handle the same VALUE for true dependence.

gcc/

2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        PR middle-end/43671
        * alias.c (true_dependence): Handle the same VALUE in x and mem.
        (canon_true_dependence): Likewise.
        (write_dependence_p): Likewise.

gcc/testsuite/

2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        PR middle-end/43671
        * gcc.target/i386/pr43671.c: New.

Added:
    trunk/gcc/testsuite/gcc.target/i386/pr43671.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/alias.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug middle-end/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[hjl at gcc dot gnu dot org]

------- Comment #10 from hjl at gcc dot gnu dot org  2010-05-04 15:49 -------
Subject: Bug 43671

Author: hjl
Date: Tue May  4 15:49:24 2010
New Revision: 159034

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=159034
Log:
Handle the same VALUE for true dependence.

gcc/

2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        Backport from mainline
        2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        PR middle-end/43671
        * alias.c (true_dependence): Handle the same VALUE in x and mem.
        (canon_true_dependence): Likewise.
        (write_dependence_p): Likewise.

gcc/testsuite/

2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        Backport from mainline
        2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        PR middle-end/43671
        * gcc.target/i386/pr43671.c: New.

Added:
    branches/gcc-4_5-branch/gcc/testsuite/gcc.target/i386/pr43671.c
Modified:
    branches/gcc-4_5-branch/gcc/ChangeLog
    branches/gcc-4_5-branch/gcc/alias.c
    branches/gcc-4_5-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug middle-end/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[hjl at gcc dot gnu dot org]

------- Comment #11 from hjl at gcc dot gnu dot org  2010-05-04 15:51 -------
Subject: Bug 43671

Author: hjl
Date: Tue May  4 15:50:53 2010
New Revision: 159035

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=159035
Log:
Handle the same VALUE for true dependence.

gcc/

2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        Backport from mainline
        2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        PR middle-end/43671
        * alias.c (true_dependence): Handle the same VALUE in x and mem.
        (canon_true_dependence): Likewise.
        (write_dependence_p): Likewise.

gcc/testsuite/

2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        Backport from mainline
        2010-05-04  H.J. Lu  <hongjiu.lu@intel.com>

        PR middle-end/43671
        * gcc.target/i386/pr43671.c: New.

Added:
    branches/gcc-4_4-branch/gcc/testsuite/gcc.target/i386/pr43671.c
Modified:
    branches/gcc-4_4-branch/gcc/ChangeLog
    branches/gcc-4_4-branch/gcc/alias.c
    branches/gcc-4_4-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671

[Bug middle-end/43671] [4.4/4.5/4.6 Regression] -fsched2-use-superblocks -m32 produces wrong code with vectorization

[hjl dot tools at gmail dot com]

------- Comment #12 from hjl dot tools at gmail dot com  2010-05-04 15:53 -------
Fixed.


-- 

hjl dot tools at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
      Known to fail|4.4.3 4.5.0                 |4.4.4 4.5.0
      Known to work|3.4.6 4.1.2 4.2.4 4.3.4     |3.4.6 4.1.2 4.2.4 4.3.4
                   |                            |4.4.5 4.5.1 4.6.0
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43671