[Bug c/108688] New: error: ‘bit_field_ref’ of non-mode-precision operand

[Bug c/108688] New: error: ‘bit_field_ref’ of non-mode-precision operand

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

            Bug ID: 108688
           Summary: error: ‘bit_field_ref’ of non-mode-precision operand
           Product: gcc
           Version: 13.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: dcb314 at hotmail dot com
  Target Milestone: ---

For this C code:

union {
  signed f1 : 7;
  signed f2 : 2
} func_6_l_2225;
g_139, g_417, func_6___trans_tmp_7;
main() {
  for (; g_417; g_417 -= 1) {
    func_6_l_2225.f1 = g_139;
    func_6_l_2225.f2 ^= func_6___trans_tmp_7;
  }
}

Compiled by recent gcc, does this:

$ ~/gcc/results/bin/gcc -c -O2 -w bug882.c
bug882.c: In function ‘main’:
bug882.c:6:1: error: ‘bit_field_ref’ of non-mode-precision operand
    6 | main() {
      | ^~~~
_ifc__31 = BIT_FIELD_REF <_2, 2, 0>;
during GIMPLE pass: ifcvt
bug882.c:6:1: internal compiler error: verify_gimple failed

$ ~/gcc/results/bin/gcc -v 2>&1 | fgrep exp
gcc version 13.0.1 20230206 (experimental) (45e01229af33a3dc) 
$

The bug first seems to occur sometime between date 20221002
and 20221028.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|c                           |tree-optimization
   Target Milestone|---                         |13.0
           Keywords|                            |ice-on-valid-code
            Summary|error: ‘bit_field_ref’ of   |[13 Regression] error:
                   |non-mode-precision operand  |‘bit_field_ref’ of
                   |                            |non-mode-precision operand

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Most likely r13-3219-g25413fdb2ac2493321

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2023-02-06
     Ever confirmed|0                           |1

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
From:
  func_6_l_2225.f1 = _2;
  _3 = func_6_l_2225.f2;
  _4 = (signed char) _3;
  _7 = _4 ^ _6;
  _8 = (<unnamed-signed:2>) _7;
  func_6_l_2225.f2 = _8;

To:

  _ifc__34 = func_6_l_2225.D.2741;
  _ifc__35 = BIT_INSERT_EXPR <_ifc__34, _2, 0 (7 bits)>;
  _ifc__31 = BIT_FIELD_REF <_2, 2, 0>;
  _4 = (signed char) _ifc__31;
  _7 = _4 ^ _6;
  _8 = (<unnamed-signed:2>) _7;
  _ifc__33 = BIT_INSERT_EXPR <_ifc__35, _8, 0 (2 bits)>;
  func_6_l_2225.D.2742 = _ifc__33;

Yes it is trying to use the type of "func_6_l_2225.f1" as the insert point
rather than a full byte sized.

Confirmed.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |avieira at gcc dot gnu.org,
                   |                            |jakub at gcc dot gnu.org
           Keywords|needs-bisection             |

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Slightly cleaned up:

union U { signed int d : 7; signed int e : 2; } u;
int a, b;

void
foo (void)
{
  for (int i = 0; i < 64; i++)
    {
      u.d = a;
      u.e ^= b;
    }
}

Yes, indeed started with r13-3219-g25413fdb2ac2493321 .

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I think the bug is in the match.pd
/* Simplify a bit extraction from a bit insertion for the cases with
   the inserted element fully covering the extraction or the insertion
   not touching the extraction.  */
(simplify
 (BIT_FIELD_REF (bit_insert @0 @1 @ipos) @rsize @rpos)
 (with
  {
    unsigned HOST_WIDE_INT isize;
    if (INTEGRAL_TYPE_P (TREE_TYPE (@1)))
      isize = TYPE_PRECISION (TREE_TYPE (@1));
    else
      isize = tree_to_uhwi (TYPE_SIZE (TREE_TYPE (@1)));
  }
  (switch
   (if (wi::leu_p (wi::to_wide (@ipos), wi::to_wide (@rpos))
        && wi::leu_p (wi::to_wide (@rpos) + wi::to_wide (@rsize),
                      wi::to_wide (@ipos) + isize))
    (BIT_FIELD_REF @1 @rsize { wide_int_to_tree (bitsizetype,
                                                 wi::to_wide (@rpos)
                                                 - wi::to_wide (@ipos)); }))
   (if (wi::geu_p (wi::to_wide (@ipos),
                   wi::to_wide (@rpos) + wi::to_wide (@rsize))
        || wi::geu_p (wi::to_wide (@rpos),
                      wi::to_wide (@ipos) + isize))
    (BIT_FIELD_REF @0 @rsize @rpos)))))
which folds
  _ifc__35 = BIT_INSERT_EXPR <_ifc__34, _2, 0 (7 bits)>;
  _ifc__31 = BIT_FIELD_REF <_ifc__35, 2, 0>;
where 34/35 have unsigned char type, 31 signed:2 and 2 signed:7.

So, in this case it wants to extract low 2 bits out of the 7 bits _2 which is
invalid
because _2 doesn't have mode precision.  If it tried to extract 7 bits out of
it, we have:
(simplify
 (BIT_FIELD_REF @0 @1 integer_zerop)
 (if (tree_int_cst_equal (@1, TYPE_SIZE (TREE_TYPE (@0))))
  (view_convert @0)))
simplification but that one wouldn't trigger, because those signed:2 types have
TYPE_SIZE 8 rather than 7, only TYPE_PRECISION of 7.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|unassigned at gcc dot gnu.org      |pinskia at gcc dot gnu.org
             Status|NEW                         |ASSIGNED

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Oh I might have a fix for this already just didn't have a testcase for the
upstream compiler at the time.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #5)
> Oh I might have a fix for this already just didn't have a testcase for the
> upstream compiler at the time.

Note, it is just the first case from the above because @0 already ought to be
guaranteed to have mode precision because BIT_INSERT_EXPR requires it too.
On the other side, we should for the first case differentiate between @1 has
mode precision, then we can do what we do right now, or when isize == rsize and
rpos == ipos (then we can just simplify to possibly casted @1) and finally
otherwise, IMHO we should just punt.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=88739

--- Comment #7 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I did mention about this exact match.pd pattern issue in bug 88739 comment #63,
I have a slightly simplified patch from that one referenced there too.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #8 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Created attachment 54433
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=54433&action=edit
Patch which I had around and will be testing

I had this patch around while I was working on full bitfield lowering.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                URL|                            |https://gcc.gnu.org/piperma
                   |                            |il/gcc-patches/2023-Februar
                   |                            |y/611594.html
           Keywords|                            |patch

--- Comment #9 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Patch submitted:
https://gcc.gnu.org/pipermail/gcc-patches/2023-February/611594.html

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #10 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:44f308e59bfa0f93ae05b17e257d8563c12399fd

commit r13-5757-g44f308e59bfa0f93ae05b17e257d8563c12399fd
Author: Andrew Pinski <apinski@marvell.com>
Date:   Thu Feb 9 16:03:54 2023 +0100

    match.pd: When simplifying BFR of an insert, require a mode precision
integral type [PR108688]

    The same problem as PR 88739 has crept in but
    this time in match.pd when simplifying bit_field_ref of
    an bit_insert. That is we are generating a BIT_FIELD_REF
    of a non-mode-precision integral type.

            PR tree-optimization/108688
            * match.pd (bit_field_ref [bit_insert]): Avoid generating
            BIT_FIELD_REFs of non-mode-precision integral operands.

            * gcc.c-torture/compile/pr108688-1.c: New test.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #11 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:bcca64d70ce91e29717fb70cff252639df6902be

commit r13-5758-gbcca64d70ce91e29717fb70cff252639df6902be
Author: Jakub Jelinek <jakub@redhat.com>
Date:   Thu Feb 9 16:07:40 2023 +0100

    match.pd: Simplify BFR of insert when extracting exactly all inserted bits
[PR108688]

    On Thu, Feb 09, 2023 at 09:16:17AM +0100, Richard Biener via Gcc-patches
wrote:
    > Hmm.  Can we handle the case of the extraction exactly covering the
    > insertion separately then and simplify to plain @1?

    I was suggesting that in the PR.  Here it is as an incremental patch
    on top of Andrew's patch.

    On the newly added testcase the ifcvt-folding difference without/with the
    incremental patch is:
    --- pr108688.c.171t.ifcvt_      2023-02-09 10:47:30.169916845 +0100
    +++ pr108688.c.171t.ifcvt       2023-02-09 10:48:44.942793453 +0100
    @@ -25,6 +25,8 @@ Number of blocks in CFG: 11
     Number of blocks to update: 5 ( 45%)

    +Applying pattern match.pd:7487, gimple-match.cc:243200
    +Applying pattern match.pd:3987, gimple-match.cc:75423
     Matching expression match.pd:1677, gimple-match.cc:209
     Applying pattern match.pd:1733, gimple-match.cc:109481
     Matching expression match.pd:2393, gimple-match.cc:852
    @@ -70,7 +72,6 @@ void foo ()
       signed char _29;
       <unnamed-signed:7> _30;
       unsigned int ivtmp_33;
    -  <unnamed-signed:7> _ifc__35;
       unsigned char _ifc__37;
       unsigned char _ifc__38;
       unsigned char _ifc__39;
    @@ -91,8 +92,7 @@ void foo ()
       _2 = (<unnamed-signed:7>) a.0_1;
       _ifc__38 = u.D.2741;
       _ifc__39 = BIT_INSERT_EXPR <_ifc__38, _2, 0 (7 bits)>;
    -  _ifc__35 = BIT_FIELD_REF <_ifc__39, 7, 0>;
    -  _4 = (signed char) _ifc__35;
    +  _4 = (signed char) _2;
       b.1_5 = b;
       _6 = (signed char) b.1_5;
       _7 = _4 ^ _6;

    2023-02-09  Jakub Jelinek  <jakub@redhat.com>

            PR tree-optimization/108688
            * match.pd (bit_field_ref [bit_insert]): Simplify BIT_FIELD_REF
            of BIT_INSERT_EXPR extracting exactly all inserted bits even
            when without mode precision.  Formatting fixes.

            * gcc.c-torture/compile/pr108688-1.c: Add PR number as comment.
            * gcc.dg/pr108688.c: New test.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

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

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

--- Comment #12 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Should be fixed now.

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #13 from CVS 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:a5453c659bce698d80f5aebdab5ab0eba39b1ed6

commit r12-9153-ga5453c659bce698d80f5aebdab5ab0eba39b1ed6
Author: Andrew Pinski <apinski@marvell.com>
Date:   Thu Feb 9 16:03:54 2023 +0100

    match.pd: When simplifying BFR of an insert, require a mode precision
integral type [PR108688]

    The same problem as PR 88739 has crept in but
    this time in match.pd when simplifying bit_field_ref of
    an bit_insert. That is we are generating a BIT_FIELD_REF
    of a non-mode-precision integral type.

            PR tree-optimization/108688
            * match.pd (bit_field_ref [bit_insert]): Avoid generating
            BIT_FIELD_REFs of non-mode-precision integral operands.

            * gcc.c-torture/compile/pr108688-1.c: New test.

    (cherry picked from commit 44f308e59bfa0f93ae05b17e257d8563c12399fd)

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #14 from CVS 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:d87adfe1a6455025eb3ae787a6da245a62cdb446

commit r11-10710-gd87adfe1a6455025eb3ae787a6da245a62cdb446
Author: Andrew Pinski <apinski@marvell.com>
Date:   Thu Feb 9 16:03:54 2023 +0100

    match.pd: When simplifying BFR of an insert, require a mode precision
integral type [PR108688]

    The same problem as PR 88739 has crept in but
    this time in match.pd when simplifying bit_field_ref of
    an bit_insert. That is we are generating a BIT_FIELD_REF
    of a non-mode-precision integral type.

            PR tree-optimization/108688
            * match.pd (bit_field_ref [bit_insert]): Avoid generating
            BIT_FIELD_REFs of non-mode-precision integral operands.

            * gcc.c-torture/compile/pr108688-1.c: New test.

    (cherry picked from commit 44f308e59bfa0f93ae05b17e257d8563c12399fd)

[Bug tree-optimization/108688] [13 Regression] error: ‘bit_field_ref’ of non-mode-precision operand

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=108688

--- Comment #15 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-10 branch has been updated by Jakub Jelinek
<jakub@gcc.gnu.org>:

https://gcc.gnu.org/g:0200ac49e6281fd86ad391b539a4f564ca44e397

commit r10-11364-g0200ac49e6281fd86ad391b539a4f564ca44e397
Author: Andrew Pinski <apinski@marvell.com>
Date:   Thu Feb 9 16:03:54 2023 +0100

    match.pd: When simplifying BFR of an insert, require a mode precision
integral type [PR108688]

    The same problem as PR 88739 has crept in but
    this time in match.pd when simplifying bit_field_ref of
    an bit_insert. That is we are generating a BIT_FIELD_REF
    of a non-mode-precision integral type.

            PR tree-optimization/108688
            * match.pd (bit_field_ref [bit_insert]): Avoid generating
            BIT_FIELD_REFs of non-mode-precision integral operands.

            * gcc.c-torture/compile/pr108688-1.c: New test.

    (cherry picked from commit 44f308e59bfa0f93ae05b17e257d8563c12399fd)