[Bug c/111818] New: GCC: 14: internal compiler error: tree check: expected tree that contains 'decl common' structure, have 'integer_cst' in tree_could_trap_p, at tree-eh.cc:2733

[Bug c/111818] New: GCC: 14: internal compiler error: tree check: expected tree that contains 'decl common' structure, have 'integer_cst' in tree_could_trap_p, at tree-eh.cc:2733

[141242068 at smail dot nju.edu.cn]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

            Bug ID: 111818
           Summary: GCC: 14: internal compiler error: tree check: expected
                    tree that contains 'decl common' structure, have
                    'integer_cst' in tree_could_trap_p, at tree-eh.cc:2733
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: 141242068 at smail dot nju.edu.cn
  Target Milestone: ---

When compile below program with GCC-14 and option `-O1`, GCC-14 crashes:
```
#define DUMMY_VALUE 0x12345678
static void foo(const volatile unsigned int x, void *p) {
  __builtin_memcpy(p, &x, sizeof x);
  __builtin_memcpy(p, &x, sizeof x);
}

void bar(int type, void *number) {
  switch (type) {
  case 1: foo(DUMMY_VALUE, number); break;
  case 7: foo(0, number); break;
  case 8: foo(0, number); break;
  case 9: foo(0, number); break;
  }
}
```

The crash reproducer: https://godbolt.org/z/7sP8zc3Kc

And the full stack dump:
```
<source>: In function 'foo':
<source>:3:23: warning: passing argument 2 of '__builtin_memcpy' discards
'volatile' qualifier from pointer target type [-Wdiscarded-qualifiers]
    3 |   __builtin_memcpy(p, &x, sizeof x);
      |                       ^~
<source>:3:23: note: expected 'const void *' but argument is of type 'const
volatile unsigned int *'
<source>:4:23: warning: passing argument 2 of '__builtin_memcpy' discards
'volatile' qualifier from pointer target type [-Wdiscarded-qualifiers]
    4 |   __builtin_memcpy(p, &x, sizeof x);
      |                       ^~
<source>:4:23: note: expected 'const void *' but argument is of type 'const
volatile unsigned int *'
during RTL pass: expand
In function 'foo',
    inlined from 'bar' at <source>:9:11:
<source>:3:3: internal compiler error: tree check: expected tree that contains
'decl common' structure, have 'integer_cst' in tree_could_trap_p, at
tree-eh.cc:2733
    3 |   __builtin_memcpy(p, &x, sizeof x);
      |   ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
0x22ff3ee internal_error(char const*, ...)
        ???:0
0x8a8243 tree_contains_struct_check_failed(tree_node const*,
tree_node_structure_enum, char const*, int, char const*)
        ???:0
0xc9ab0e set_mem_attributes_minus_bitpos(rtx_def*, tree_node*, int,
poly_int<1u, long>)
        ???:0
0xcc6c35 expand_expr_real_1(tree_node*, rtx_def*, machine_mode,
expand_modifier, rtx_def**, bool)
        ???:0
0xcd3083 store_expr(tree_node*, rtx_def*, int, bool, bool)
        ???:0
0xcd4d66 expand_assignment(tree_node*, tree_node*, bool)
        ???:0
Please submit a full bug report, with preprocessed source (by using
-freport-bug).
Please include the complete backtrace with any bug report.
See <https://gcc.gnu.org/bugs/> for instructions.
```

[Bug middle-end/111818] [11/12/13/14 Regression] ICE with __builtin_memcpy with volatile and constants

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.9.1
      Known to fail|                            |10.1.0, 5.1.0
   Target Milestone|---                         |11.5
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2023-10-15
            Summary|GCC: 14: internal compiler  |[11/12/13/14 Regression]
                   |error: tree check: expected |ICE with __builtin_memcpy
                   |tree that contains 'decl    |with volatile and constants
                   |common' structure, have     |
                   |'integer_cst' in            |
                   |tree_could_trap_p, at       |
                   |tree-eh.cc:2733             |
             Status|UNCONFIRMED                 |NEW

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Confirmed.
Even though GCC 11.x and 12.x and 13.x don't ICE without checking enabled. The
IR is still invalid:
  _9 ={v} MEM <unsigned int> [(char * {ref-all})&305419896];

That is just wrong.

[Bug middle-end/111818] [11/12/13/14 Regression] ICE with __builtin_memcpy with volatile and constants

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

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

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

--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
It's inlining of

void foo (const volatile unsigned int x, void * p)
{
  unsigned int _2;
  unsigned int _5;

  <bb 2> [local count: 1073741824]:
  _2 ={v} MEM <unsigned int> [(char * {ref-all})&x];

into

  foo (305419896, number_4(D));

Smaller testcase:

static void foo(const volatile unsigned int x, void *p) 
{ 
  __builtin_memcpy(p, &x, sizeof x);
}

void bar(void *number)
{ 
  foo(0, number);
}

[Bug middle-end/111818] [11/12/13/14 Regression] ICE with __builtin_memcpy with volatile and constants

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

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

https://gcc.gnu.org/g:ce55521bcd149fdc431f1d78e706b66d470210ae

commit r14-4678-gce55521bcd149fdc431f1d78e706b66d470210ae
Author: Richard Biener <rguenther@suse.de>
Date:   Mon Oct 16 12:50:46 2023 +0200

    middle-end/111818 - failed DECL_NOT_GIMPLE_REG_P setting of volatile

    The following addresses a missed DECL_NOT_GIMPLE_REG_P setting of
    a volatile declared parameter which causes inlining to substitute
    a constant parameter into a context where its address is required.

    The main issue is in update_address_taken which clears
    DECL_NOT_GIMPLE_REG_P from the parameter but fails to rewrite it
    because is_gimple_reg returns false for volatiles.  The following
    changes maybe_optimize_var to make the 1:1 correspondence between
    clearing DECL_NOT_GIMPLE_REG_P of a register typed decl and
    actually rewriting it to SSA.

            PR middle-end/111818
            * tree-ssa.cc (maybe_optimize_var): When clearing
            DECL_NOT_GIMPLE_REG_P always rewrite into SSA.

            * gcc.dg/torture/pr111818.c: New testcase.

[Bug middle-end/111818] [11/12/13 Regression] ICE with __builtin_memcpy with volatile and constants

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P2
           Keywords|                            |wrong-code
            Summary|[11/12/13/14 Regression]    |[11/12/13 Regression] ICE
                   |ICE with __builtin_memcpy   |with __builtin_memcpy with
                   |with volatile and constants |volatile and constants
      Known to fail|                            |13.2.1
      Known to work|                            |14.0

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed on trunk sofar.

[Bug middle-end/111818] [11/12/13 Regression] ICE with __builtin_memcpy with volatile and constants

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

--- Comment #5 from CVS Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-13 branch has been updated by Richard Biener
<rguenth@gcc.gnu.org>:

https://gcc.gnu.org/g:ffb0615cf1aa628ca8aa10bda31fe1bc554eccf3

commit r13-8044-gffb0615cf1aa628ca8aa10bda31fe1bc554eccf3
Author: Richard Biener <rguenther@suse.de>
Date:   Mon Oct 16 12:50:46 2023 +0200

    middle-end/111818 - failed DECL_NOT_GIMPLE_REG_P setting of volatile

    The following addresses a missed DECL_NOT_GIMPLE_REG_P setting of
    a volatile declared parameter which causes inlining to substitute
    a constant parameter into a context where its address is required.

    The main issue is in update_address_taken which clears
    DECL_NOT_GIMPLE_REG_P from the parameter but fails to rewrite it
    because is_gimple_reg returns false for volatiles.  The following
    changes maybe_optimize_var to make the 1:1 correspondence between
    clearing DECL_NOT_GIMPLE_REG_P of a register typed decl and
    actually rewriting it to SSA.

            PR middle-end/111818
            * tree-ssa.cc (maybe_optimize_var): When clearing
            DECL_NOT_GIMPLE_REG_P always rewrite into SSA.

            * gcc.dg/torture/pr111818.c: New testcase.

    (cherry picked from commit ce55521bcd149fdc431f1d78e706b66d470210ae)

[Bug middle-end/111818] [11/12 Regression] ICE with __builtin_memcpy with volatile and constants

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

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

https://gcc.gnu.org/g:2cc700fec5c72d9c1385ec83621771bb9db58d96

commit r12-10014-g2cc700fec5c72d9c1385ec83621771bb9db58d96
Author: Richard Biener <rguenther@suse.de>
Date:   Mon Oct 16 12:50:46 2023 +0200

    middle-end/111818 - failed DECL_NOT_GIMPLE_REG_P setting of volatile

    The following addresses a missed DECL_NOT_GIMPLE_REG_P setting of
    a volatile declared parameter which causes inlining to substitute
    a constant parameter into a context where its address is required.

    The main issue is in update_address_taken which clears
    DECL_NOT_GIMPLE_REG_P from the parameter but fails to rewrite it
    because is_gimple_reg returns false for volatiles.  The following
    changes maybe_optimize_var to make the 1:1 correspondence between
    clearing DECL_NOT_GIMPLE_REG_P of a register typed decl and
    actually rewriting it to SSA.

            PR middle-end/111818
            * tree-ssa.cc (maybe_optimize_var): When clearing
            DECL_NOT_GIMPLE_REG_P always rewrite into SSA.

            * gcc.dg/torture/pr111818.c: New testcase.

    (cherry picked from commit ce55521bcd149fdc431f1d78e706b66d470210ae)

[Bug middle-end/111818] [11 Regression] ICE with __builtin_memcpy with volatile and constants

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

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

https://gcc.gnu.org/g:45c296f2d81f6259e268e07707b4aa87dffba20a

commit r11-11142-g45c296f2d81f6259e268e07707b4aa87dffba20a
Author: Richard Biener <rguenther@suse.de>
Date:   Mon Oct 16 12:50:46 2023 +0200

    middle-end/111818 - failed DECL_NOT_GIMPLE_REG_P setting of volatile

    The following addresses a missed DECL_NOT_GIMPLE_REG_P setting of
    a volatile declared parameter which causes inlining to substitute
    a constant parameter into a context where its address is required.

    The main issue is in update_address_taken which clears
    DECL_NOT_GIMPLE_REG_P from the parameter but fails to rewrite it
    because is_gimple_reg returns false for volatiles.  The following
    changes maybe_optimize_var to make the 1:1 correspondence between
    clearing DECL_NOT_GIMPLE_REG_P of a register typed decl and
    actually rewriting it to SSA.

            PR middle-end/111818
            * tree-ssa.c (maybe_optimize_var): When clearing
            DECL_NOT_GIMPLE_REG_P always rewrite into SSA.

            * gcc.dg/torture/pr111818.c: New testcase.

    (cherry picked from commit ce55521bcd149fdc431f1d78e706b66d470210ae)

[Bug middle-end/111818] [11 Regression] ICE with __builtin_memcpy with volatile and constants

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111818

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|                            |11.4.0
         Resolution|---                         |FIXED
             Status|ASSIGNED                    |RESOLVED
      Known to work|                            |11.4.1

--- Comment #8 from Richard Biener <rguenth at gcc dot gnu.org> ---
Fixed.