[Bug rtl-optimization/57130] New: Incorrect "and --> extract" conversion in combine

[Bug rtl-optimization/57130] New: Incorrect "and --> extract" conversion in combine

[wmi at google dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57130

             Bug #: 57130
           Summary: Incorrect "and --> extract" conversion in combine
    Classification: Unclassified
           Product: gcc
           Version: 4.9.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: rtl-optimization
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: wmi@google.com


Created attachment 29986
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29986
Testcase

For the smallcase 1.ii attached.

~/workarea/gcc-r198433/build/install/bin/g++ 1.ii -o a1.out
./a1.out  --> correct output:
1
3

~/workarea/gcc-r198433/build/install/bin/g++ -fno-inline
-fno-omit-frame-pointer -O2 1.ii -o a2.out
./a2.out  --> incorrect output:
1
-1

In 1.ii.196r.ud_dce:

(insn 7 2 84 2 (set (reg:DI 64)
        (const_int -4294967296 [0xffffffff00000000])) 1.ii:26 84
{*movdi_internal}
     (nil))
...
(insn 40 36 44 2 (parallel [
            (set (reg:DI 88)
                (and:DI (reg:DI 80)
                    (reg:DI 64)))
            (clobber (reg:CC 17 flags))
        ]) 1.ii:32 386 {*anddi_1}
     (expr_list:REG_DEAD (reg:DI 80)
        (expr_list:REG_DEAD (reg:DI 64)
            (expr_list:REG_UNUSED (reg:CC 17 flags)
                (expr_list:REG_EQUAL (and:DI (reg:DI 80)
                        (const_int -4294967296 [0xffffffff00000000]))
                    (nil))))))
(insn 44 40 45 2 (set (reg:DI 92 [ mini_rect ])
        (zero_extend:DI (subreg:SI (reg:DI 88) 0))) 1.ii:33 127
{*zero_extendsidi2}
     (expr_list:REG_DEAD (reg:DI 88)
        (nil)))

The value of r92 here should always be 0.

After try_combine with params (i3==insn44, i2==insn40, i1==insn7), insn44 is
transformed to: 
(insn 44 40 45 2 (parallel [
            (set (reg:DI 92 [ mini_rect ])
                (ashiftrt:DI (reg:DI 88)
                    (const_int 63 [0x3f])))
            (clobber (reg:CC 17 flags))
        ]) 1.ii:33 528 {ashrdi3_cvt}
     (expr_list:REG_UNUSED (reg:CC 17 flags)
        (expr_list:REG_DEAD (reg:DI 88)
            (nil))))

The value of r92 now equals either 0 or -1 which depends on the highest bit of
r88. 

Try to understand what happen in try_combine:
In try_combine, after subst(PATTERN (i3), i2dest, i2src, ...), insn 44 is
transformed to the following form. This step is correct. 

(insn 44 40 45 2 (set (reg:DI 92 [ mini_rect ])
        (neg:DI (ne:DI (subreg:SI (and:DI (reg:DI 80)
                        (reg:DI 64)) 0)
                (const_int 0 [0])))) 1.ii:33 127 {*zero_extendsidi2}
     (expr_list:REG_DEAD (reg:DI 88)
        (nil)))

In subst(PATTERN (i3), i1dest, i1src, ...), insn 44 is firstly transformed to
the following in simplify_logical, which is correct:

(insn 44 40 45 2 (set (reg:DI 92 [ mini_rect ])
        (neg:DI (ne:DI (subreg:SI ((and:DI (reg:DI 80)
                        (const_int 34359738368 [0x800000000]))) 0)
                (const_int 0 [0])))) 1.ii:33 127 {*zero_extendsidi2}
     (expr_list:REG_DEAD (reg:DI 88)
        (nil)))

then it is transformed to the following in make_compound_operation, which is
incorrect:

(insn 44 40 45 2 (set (reg:DI 92 [ mini_rect ])
        (sign_extract:DI (reg:DI 80)
            (const_int 1 [0x1])
            (const_int 35 [0x23]))) 1.ii:33 127 {*zero_extendsidi2}
     (expr_list:REG_DEAD (reg:DI 88)
        (nil)))


make_compound_operation transforms

(and:DI (reg:DI 80)
    (const_int 34359738368 [0x800000000]))

to

(zero_extract:DI (reg:DI 80)
    (const_int 1 [0x1])
    (const_int 35 [0x23]))

because it thinks the "and expr" here is in a compare. But actually the "and
expr" is firstly the kid in a subreg: 

subreg:SI ((and:DI (reg:DI 80)
    (const_int 34359738368 [0x800000000]))  ==> always 0

is not identical with

subreg:SI ((zero_extract:DI (reg:DI 80)
    (const_int 1 [0x1])
    (const_int 35 [0x23]))      ==> 0 or 1

So it is the cause of the problem. The second actual of make_compound_operation
(combine.c:7701, r198433) should not be in_code.

[Bug rtl-optimization/57130] [4.8/4.9 Regression] Incorrect "and --> extract" conversion in combine

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57130

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2013-05-01
                 CC|                            |jakub at gcc dot gnu.org
   Target Milestone|---                         |4.8.1
            Summary|Incorrect "and --> extract" |[4.8/4.9 Regression]
                   |conversion in combine       |Incorrect "and --> extract"
                   |                            |conversion in combine
     Ever Confirmed|0                           |1

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-05-01 06:41:55 UTC ---
Started with http://gcc.gnu.org/r190492

[Bug rtl-optimization/57130] [4.8/4.9 Regression] Incorrect "and --> extract" conversion in combine

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57130

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-05-02 06:58:38 UTC ---
Slightly more reduced testcase:
struct S { int a, b, c, d; } s[2] = { { 6, 8, -8, -5 }, { 0, 2, -1, 2 } };

__attribute__((noinline, noclone)) void
foo (struct S r)
{
  static int cnt;
  if (__builtin_memcmp (&r, &s[cnt++], sizeof r) != 0)
    __builtin_abort ();
}

int
main ()
{
  struct S r = { 6, 8, -8, -5 };
  foo (r);
  r = (struct S) { 0, 2, -1, 2 };
  foo (r);
  return 0;
}

[Bug rtl-optimization/57130] [4.8/4.9 Regression] Incorrect "and --> extract" conversion in combine

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57130

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-05-02 09:07:31 UTC ---
Created attachment 29994
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29994
gcc49-pr57130.patch

Untested fix, though not sure if it won't pessimize too much.  The in_code ==
COMPARE handling simply look safe to me when recursing into a SUBREG. 
Sometimes it could, e.g. for (subreg:SI (and:DI (reg:DI) (const_int
0x08000000))) with in_code == COMPARE, but we'd probably have to pass the
SUBREG mode around and special case it everywhere.

[Bug rtl-optimization/57130] [4.8/4.9 Regression] Incorrect "and --> extract" conversion in combine

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57130

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-05-02 09:23:22 UTC ---
Created attachment 29995
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29995
simplify-truncate.patch

Another thing, it is weird we don't simplify that subreg much earlier, we know
that nonzero_bits on the inner operand has all zeros in the mode that it is
truncated to, therefore it must be zero.

[Bug rtl-optimization/57130] [4.8/4.9 Regression] Incorrect "and --> extract" conversion in combine

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57130

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-05-02 10:49:08 UTC ---
Created attachment 29996
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=29996
gcc49-pr57130.patch

Alternative to the #c3 patch, using SET instead of COMPARE only when needed for
the recursive call for SUBREG.

[Bug rtl-optimization/57130] [4.8/4.9 Regression] Incorrect "and --> extract" conversion in combine

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=57130

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

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

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> 2013-05-03 13:56:59 UTC ---
Author: jakub
Date: Fri May  3 12:56:12 2013
New Revision: 198579

URL: http://gcc.gnu.org/viewcvs?rev=198579&root=gcc&view=rev
Log:
    PR rtl-optimization/57130
    * combine.c (make_compound_operation) <case SUBREG>: Pass
    SET instead of COMPARE as in_code to the recursive call
    if needed.

    * gcc.c-torture/execute/pr57130.c: New test.

Added:
    trunk/gcc/testsuite/gcc.c-torture/execute/pr57130.c
Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/combine.c
    trunk/gcc/testsuite/ChangeLog

Author: jakub
Date: Fri May  3 13:19:51 2013
New Revision: 198581

URL: http://gcc.gnu.org/viewcvs?rev=198581&root=gcc&view=rev
Log:
    PR rtl-optimization/57130
    * combine.c (make_compound_operation) <case SUBREG>: Pass
    SET instead of COMPARE as in_code to the recursive call
    if needed.

    * gcc.c-torture/execute/pr57130.c: New test.

Added:
    branches/gcc-4_8-branch/gcc/testsuite/gcc.c-torture/execute/pr57130.c
Modified:
    branches/gcc-4_8-branch/gcc/ChangeLog
    branches/gcc-4_8-branch/gcc/combine.c
    branches/gcc-4_8-branch/gcc/testsuite/ChangeLog

[Bug rtl-optimization/57130] [4.8/4.9 Regression] Incorrect "and --> extract" conversion in combine

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=57130

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |yahemore at gmail dot com

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
*** Bug 50291 has been marked as a duplicate of this bug. ***