[Bug tree-optimization/110731] New: [11/12/13/14 Regression] Wrong-code because of wide-int division since r5-424

[Bug tree-optimization/110731] New: [11/12/13/14 Regression] Wrong-code because of wide-int division since r5-424

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

            Bug ID: 110731
           Summary: [11/12/13/14 Regression] Wrong-code because of
                    wide-int division since r5-424
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: tree-optimization
          Assignee: unassigned at gcc dot gnu.org
          Reporter: jakub at gcc dot gnu.org
  Target Milestone: ---

Since r5-424-g807e902eea17f313 (merge of wide-int) we miscompile following
testcase
(noticed it first with _BitInt(119)):

__int128
foo (void)
{
  struct S { __int128 f : 119; } s = { ((__int128) -18014398509481984) << 64 };
  return s.f / 2;
}

int
main ()
{
  if (foo () != (((__int128) -9007199254740992) << 64))
    __builtin_abort ();
}

The result of dividing (-332306998946228968225951765070086143wb - 1) by 2 is
-166153499473114484112975882535043072wb but we return
166153499473114484112975882535043072wb instead (in hex that is
0, 0xffc0000000000000 signed 119 pair divided by 2 resulting in
0, 0x0020000000000000 signed 119 pair rather than
0, 0xffe0000000000000.

[Bug tree-optimization/110731] [11/12/13/14 Regression] Wrong-code because of wide-int division since r5-424

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code
   Target Milestone|---                         |11.5
           Priority|P3                          |P2
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2023-07-19
             Status|UNCONFIRMED                 |NEW

[Bug tree-optimization/110731] [11/12/13/14 Regression] Wrong-code because of wide-int division since r5-424

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
This should have done the negative (in the original patch):

1796       /* The quotient is neg if exactly one of the divisor or dividend is
1797          neg.  */
1798       if (dividend_neg != divisor_neg)
1799         quotient_len = wi::sub_large (quotient, zeros, 1, quotient,
1800                                       quotient_len, dividend_prec,
1801                                       UNSIGNED, 0);


That is you do -((332306998946228968225951765070086143wb + 1)/2)

[Bug tree-optimization/110731] [11/12/13/14 Regression] Wrong-code because of wide-int division since r5-424

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I think the problem is earlier.
If I divide the same values unsigned, while the representation of the wide_int
values
is the same, i.e.
val = {0, -18014398509481984}, len = 2, precision = 119
for x, when divmod_internal_2 is called, in the signed case it sees
x/4wx b_dividend
0x7fffffffc0b0: 0x00000000      0x00000000      0x00000000      0xffc00000
while in the unsigned one
x/4wx b_dividend
0x7fffffffc0b0: 0x00000000      0x00000000      0x00000000      0x00400000
and I think the latter is what should be used in both cases, divmod_internal_2
performs
unsigned division.

So, I think the right fix is:
--- gcc/wide-int.cc.jj  2023-06-12 15:47:22.461502821 +0200
+++ gcc/wide-int.cc     2023-07-19 09:52:40.241661869 +0200
@@ -1911,9 +1911,9 @@ wi::divmod_internal (HOST_WIDE_INT *quot
     }

   wi_unpack (b_dividend, dividend.get_val (), dividend.get_len (),
-            dividend_blocks_needed, dividend_prec, sgn);
+            dividend_blocks_needed, dividend_prec, UNSIGNED);
   wi_unpack (b_divisor, divisor.get_val (), divisor.get_len (),
-            divisor_blocks_needed, divisor_prec, sgn);
+            divisor_blocks_needed, divisor_prec, UNSIGNED);

   m = dividend_blocks_needed;
   b_dividend[m] = 0;
because at this point we are after the
  /* Make the divisor and dividend positive and remember what we
     did.  */
  if (sgn == SIGNED)
    {
      if (wi::neg_p (dividend))
        {
          neg_dividend = -dividend;
          dividend = neg_dividend;
          dividend_neg = true;
        }
      if (wi::neg_p (divisor))
        {
          neg_divisor = -divisor;
          divisor = neg_divisor;
          divisor_neg = true;
        }
    }
hunk.  Of course, this makes a difference only if either the dividend or
divisor are the smallest negative value of a type.

[Bug tree-optimization/110731] [11/12/13/14 Regression] Wrong-code because of wide-int division since r5-424

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

--- Comment #3 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Jakub Jelinek <jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:ece799607c841676f4e00c2fea98bbec6976da3f

commit r14-2642-gece799607c841676f4e00c2fea98bbec6976da3f
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Jul 19 13:48:53 2023 +0200

    wide-int: Fix up wi::divmod_internal [PR110731]

    As the following testcase shows, wi::divmod_internal doesn't handle
    correctly signed division with precision > 64 when the dividend (and likely
    divisor as well) is the type's minimum and the precision isn't divisible
    by 64.

    A few lines above what the patch hunk changes is:
      /* Make the divisor and dividend positive and remember what we
         did.  */
      if (sgn == SIGNED)
        {
          if (wi::neg_p (dividend))
            {
              neg_dividend = -dividend;
              dividend = neg_dividend;
              dividend_neg = true;
            }
          if (wi::neg_p (divisor))
            {
              neg_divisor = -divisor;
              divisor = neg_divisor;
              divisor_neg = true;
            }
        }
    i.e. we negate negative dividend or divisor and remember those.
    But, after we do that, when unpacking those values into b_dividend and
    b_divisor we need to always treat the wide_ints as UNSIGNED,
    because divmod_internal_2 performs an unsigned division only.
    Now, if precision <= 64, we don't reach here at all, earlier code
    handles it.  If dividend or divisor aren't the most negative values,
    the negation clears their most significant bit, so it doesn't really
    matter if we unpack SIGNED or UNSIGNED.  And if precision is multiple
    of HOST_BITS_PER_WIDE_INT, there is no difference in behavior, while
    -0x80000000000000000000000000000000 negates to
    -0x80000000000000000000000000000000 the unpacking of it as SIGNED
    or UNSIGNED works the same.
    In the testcase, we have signed precision 119 and the dividend is
    val = { 0, 0xffc0000000000000 }, len = 2, precision = 119
    both before and after negation.
    Divisor is
    val = { 2 }, len = 1, precision = 119
    But we really want to divide 0x400000000000000000000000000000 by 2
    unsigned and then negate at the end.
    If it is unsigned precision 119 division
    0x400000000000000000000000000000 by 2
    dividend is
    val = { 0, 0xffc0000000000000 }, len = 2, precision = 119
    but as we unpack it UNSIGNED, it is unpacked into
    0, 0, 0, 0x00400000

    The following patch fixes it by always using UNSIGNED unpacking
    because we've already negated negative values at that point if
    sgn == SIGNED and so most negative constants should be treated as
    positive.

    2023-07-19  Jakub Jelinek  <jakub@redhat.com>

            PR tree-optimization/110731
            * wide-int.cc (wi::divmod_internal): Always unpack dividend and
            divisor as UNSIGNED regardless of sgn.

            * gcc.dg/pr110731.c: New test.

[Bug tree-optimization/110731] [11/12/13/14 Regression] Wrong-code because of wide-int division since r5-424

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

--- Comment #4 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-13 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:f76f411f2cdb62122322e709d503edd2f8e8a530

commit r13-7589-gf76f411f2cdb62122322e709d503edd2f8e8a530
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Jul 19 13:48:53 2023 +0200

    wide-int: Fix up wi::divmod_internal [PR110731]

    As the following testcase shows, wi::divmod_internal doesn't handle
    correctly signed division with precision > 64 when the dividend (and likely
    divisor as well) is the type's minimum and the precision isn't divisible
    by 64.

    A few lines above what the patch hunk changes is:
      /* Make the divisor and dividend positive and remember what we
         did.  */
      if (sgn == SIGNED)
        {
          if (wi::neg_p (dividend))
            {
              neg_dividend = -dividend;
              dividend = neg_dividend;
              dividend_neg = true;
            }
          if (wi::neg_p (divisor))
            {
              neg_divisor = -divisor;
              divisor = neg_divisor;
              divisor_neg = true;
            }
        }
    i.e. we negate negative dividend or divisor and remember those.
    But, after we do that, when unpacking those values into b_dividend and
    b_divisor we need to always treat the wide_ints as UNSIGNED,
    because divmod_internal_2 performs an unsigned division only.
    Now, if precision <= 64, we don't reach here at all, earlier code
    handles it.  If dividend or divisor aren't the most negative values,
    the negation clears their most significant bit, so it doesn't really
    matter if we unpack SIGNED or UNSIGNED.  And if precision is multiple
    of HOST_BITS_PER_WIDE_INT, there is no difference in behavior, while
    -0x80000000000000000000000000000000 negates to
    -0x80000000000000000000000000000000 the unpacking of it as SIGNED
    or UNSIGNED works the same.
    In the testcase, we have signed precision 119 and the dividend is
    val = { 0, 0xffc0000000000000 }, len = 2, precision = 119
    both before and after negation.
    Divisor is
    val = { 2 }, len = 1, precision = 119
    But we really want to divide 0x400000000000000000000000000000 by 2
    unsigned and then negate at the end.
    If it is unsigned precision 119 division
    0x400000000000000000000000000000 by 2
    dividend is
    val = { 0, 0xffc0000000000000 }, len = 2, precision = 119
    but as we unpack it UNSIGNED, it is unpacked into
    0, 0, 0, 0x00400000

    The following patch fixes it by always using UNSIGNED unpacking
    because we've already negated negative values at that point if
    sgn == SIGNED and so most negative constants should be treated as
    positive.

    2023-07-19  Jakub Jelinek  <jakub@redhat.com>

            PR tree-optimization/110731
            * wide-int.cc (wi::divmod_internal): Always unpack dividend and
            divisor as UNSIGNED regardless of sgn.

            * gcc.dg/pr110731.c: New test.

    (cherry picked from commit ece799607c841676f4e00c2fea98bbec6976da3f)

[Bug tree-optimization/110731] [11/12 Regression] Wrong-code because of wide-int division since r5-424

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[11/12/13/14 Regression]    |[11/12 Regression]
                   |Wrong-code because of       |Wrong-code because of
                   |wide-int division since     |wide-int division since
                   |r5-424                      |r5-424
           Assignee|unassigned at gcc dot gnu.org      |jakub at gcc dot gnu.org
             Status|NEW                         |ASSIGNED

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Fixed for 13.2+ so far.

[Bug tree-optimization/110731] [11/12 Regression] Wrong-code because of wide-int division since r5-424

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

--- Comment #6 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-12 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:9a93c334865af18ce2fd50cd77a9e90939f3c231

commit r12-10046-g9a93c334865af18ce2fd50cd77a9e90939f3c231
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Jul 19 13:48:53 2023 +0200

    wide-int: Fix up wi::divmod_internal [PR110731]

    As the following testcase shows, wi::divmod_internal doesn't handle
    correctly signed division with precision > 64 when the dividend (and likely
    divisor as well) is the type's minimum and the precision isn't divisible
    by 64.

    A few lines above what the patch hunk changes is:
      /* Make the divisor and dividend positive and remember what we
         did.  */
      if (sgn == SIGNED)
        {
          if (wi::neg_p (dividend))
            {
              neg_dividend = -dividend;
              dividend = neg_dividend;
              dividend_neg = true;
            }
          if (wi::neg_p (divisor))
            {
              neg_divisor = -divisor;
              divisor = neg_divisor;
              divisor_neg = true;
            }
        }
    i.e. we negate negative dividend or divisor and remember those.
    But, after we do that, when unpacking those values into b_dividend and
    b_divisor we need to always treat the wide_ints as UNSIGNED,
    because divmod_internal_2 performs an unsigned division only.
    Now, if precision <= 64, we don't reach here at all, earlier code
    handles it.  If dividend or divisor aren't the most negative values,
    the negation clears their most significant bit, so it doesn't really
    matter if we unpack SIGNED or UNSIGNED.  And if precision is multiple
    of HOST_BITS_PER_WIDE_INT, there is no difference in behavior, while
    -0x80000000000000000000000000000000 negates to
    -0x80000000000000000000000000000000 the unpacking of it as SIGNED
    or UNSIGNED works the same.
    In the testcase, we have signed precision 119 and the dividend is
    val = { 0, 0xffc0000000000000 }, len = 2, precision = 119
    both before and after negation.
    Divisor is
    val = { 2 }, len = 1, precision = 119
    But we really want to divide 0x400000000000000000000000000000 by 2
    unsigned and then negate at the end.
    If it is unsigned precision 119 division
    0x400000000000000000000000000000 by 2
    dividend is
    val = { 0, 0xffc0000000000000 }, len = 2, precision = 119
    but as we unpack it UNSIGNED, it is unpacked into
    0, 0, 0, 0x00400000

    The following patch fixes it by always using UNSIGNED unpacking
    because we've already negated negative values at that point if
    sgn == SIGNED and so most negative constants should be treated as
    positive.

    2023-07-19  Jakub Jelinek  <jakub@redhat.com>

            PR tree-optimization/110731
            * wide-int.cc (wi::divmod_internal): Always unpack dividend and
            divisor as UNSIGNED regardless of sgn.

            * gcc.dg/pr110731.c: New test.

    (cherry picked from commit ece799607c841676f4e00c2fea98bbec6976da3f)

[Bug tree-optimization/110731] [11/12 Regression] Wrong-code because of wide-int division since r5-424

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110731

--- Comment #7 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-11 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:9e30ca09c01124333dc97e8c2cb704b74d3c7b60

commit r11-11148-g9e30ca09c01124333dc97e8c2cb704b74d3c7b60
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Wed Jul 19 13:48:53 2023 +0200

    wide-int: Fix up wi::divmod_internal [PR110731]

    As the following testcase shows, wi::divmod_internal doesn't handle
    correctly signed division with precision > 64 when the dividend (and likely
    divisor as well) is the type's minimum and the precision isn't divisible
    by 64.

    A few lines above what the patch hunk changes is:
      /* Make the divisor and dividend positive and remember what we
         did.  */
      if (sgn == SIGNED)
        {
          if (wi::neg_p (dividend))
            {
              neg_dividend = -dividend;
              dividend = neg_dividend;
              dividend_neg = true;
            }
          if (wi::neg_p (divisor))
            {
              neg_divisor = -divisor;
              divisor = neg_divisor;
              divisor_neg = true;
            }
        }
    i.e. we negate negative dividend or divisor and remember those.
    But, after we do that, when unpacking those values into b_dividend and
    b_divisor we need to always treat the wide_ints as UNSIGNED,
    because divmod_internal_2 performs an unsigned division only.
    Now, if precision <= 64, we don't reach here at all, earlier code
    handles it.  If dividend or divisor aren't the most negative values,
    the negation clears their most significant bit, so it doesn't really
    matter if we unpack SIGNED or UNSIGNED.  And if precision is multiple
    of HOST_BITS_PER_WIDE_INT, there is no difference in behavior, while
    -0x80000000000000000000000000000000 negates to
    -0x80000000000000000000000000000000 the unpacking of it as SIGNED
    or UNSIGNED works the same.
    In the testcase, we have signed precision 119 and the dividend is
    val = { 0, 0xffc0000000000000 }, len = 2, precision = 119
    both before and after negation.
    Divisor is
    val = { 2 }, len = 1, precision = 119
    But we really want to divide 0x400000000000000000000000000000 by 2
    unsigned and then negate at the end.
    If it is unsigned precision 119 division
    0x400000000000000000000000000000 by 2
    dividend is
    val = { 0, 0xffc0000000000000 }, len = 2, precision = 119
    but as we unpack it UNSIGNED, it is unpacked into
    0, 0, 0, 0x00400000

    The following patch fixes it by always using UNSIGNED unpacking
    because we've already negated negative values at that point if
    sgn == SIGNED and so most negative constants should be treated as
    positive.

    2023-07-19  Jakub Jelinek  <jakub@redhat.com>

            PR tree-optimization/110731
            * wide-int.cc (wi::divmod_internal): Always unpack dividend and
            divisor as UNSIGNED regardless of sgn.

            * gcc.dg/pr110731.c: New test.

    (cherry picked from commit ece799607c841676f4e00c2fea98bbec6976da3f)