[Bug c++/114303] New: ICE with constexpr if and static_assert accessing captures across nested generic lambdas

[Bug c++/114303] New: ICE with constexpr if and static_assert accessing captures across nested generic lambdas

[benbarsdell at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

            Bug ID: 114303
           Summary: ICE with constexpr if and static_assert accessing
                    captures across nested generic lambdas
           Product: gcc
           Version: 14.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
          Assignee: unassigned at gcc dot gnu.org
          Reporter: benbarsdell at gmail dot com
  Target Milestone: ---

Similar to https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105788 but the
backtrace is different.

https://godbolt.org/z/xMPeTo633

Repros with the following versions on Godbolt:
14.0.1 ICE (gcc trunk on Godbolt)
13.2: ICE
13.1: ICE
12.3: ICE
12.2: OK
12.1: OK
11.4: ICE
11.3: OK


#include <type_traits>

int main() {
    auto foo = [](auto level1) {
        return [&](auto ) {
            return [&](auto level3) {
                if constexpr (decltype(level3)::value) {
                    static_assert(decltype(level1)::value, "");
                }
                return 0;
            }(std::integral_constant<int, 1>());
        }(0);
    }(std::integral_constant<int, 1>());
}


<source>:8:17: internal compiler error: trying to capture 'level1' in
instantiation of generic lambda
    8 |                 if constexpr (decltype(level3)::value) {
      |                 ^~
0x265eabc internal_error(char const*, ...)
        ???:0
0xb86956 add_capture(tree_node*, tree_node*, tree_node*, bool, bool, unsigned
int*)
        ???:0
0xb869d9 add_default_capture(tree_node*, tree_node*, tree_node*)
        ???:0
0x16c682c walk_tree_1(tree_node**, tree_node* (*)(tree_node**, int*, void*),
void*, hash_set<tree_node*, false, default_hash_traits<tree_node*> >*,
tree_node* (*)(tree_node**, int*, tree_node* (*)(tree_node**, int*, void*),
void*, hash_set<tree_node*, false, default_hash_traits<tree_node*> >*))
        ???:0
0x16c68a9 walk_tree_1(tree_node**, tree_node* (*)(tree_node**, int*, void*),
void*, hash_set<tree_node*, false, default_hash_traits<tree_node*> >*,
tree_node* (*)(tree_node**, int*, tree_node* (*)(tree_node**, int*, void*),
void*, hash_set<tree_node*, false, default_hash_traits<tree_node*> >*))
        ???:0
0xc75f26 build_extra_args(tree_node*, tree_node*, int)
        ???:0
0xcab766 tsubst_lambda_expr(tree_node*, tree_node*, int, tree_node*)
        ???:0
0xc919b3 instantiate_decl(tree_node*, bool, bool)
        ???:0
0xb54725 maybe_instantiate_decl(tree_node*)
        ???:0
...

[Bug c++/114303] [11/12/13/14 Regression] ICE with constexpr if and static_assert accessing captures across nested generic lambdas

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |ice-on-valid-code
      Known to fail|                            |11.4.0, 12.3.0, 13.1.0
      Known to work|                            |11.3.0, 12.1.0, 12.2.0
            Summary|ICE with constexpr if and   |[11/12/13/14 Regression]
                   |static_assert accessing     |ICE with constexpr if and
                   |captures across nested      |static_assert accessing
                   |generic lambdas             |captures across nested
                   |                            |generic lambdas
   Target Milestone|---                         |11.5
     Ever confirmed|0                           |1
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2024-03-11

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Reduced testcase:
```
struct a
{
  static constexpr int value = 1;
};

int main() {
    [](auto level1) {
        return [&](auto ) {
            return [&](auto level3) {
                if constexpr (decltype(level3)::value) {
                    static_assert(decltype(level1)::value, "");
                }
                return 0;
            }(a{});
        }(0);
    }(a{});
}
```

[Bug c++/114303] [11/12/13/14 Regression] ICE with constexpr if and accessing captures across nested generic lambdas

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

--- Comment #2 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
Note the static_assert is not needed, it could just be
`decltype(level1)::value` even.

[Bug c++/114303] [11/12/13/14 Regression] ICE with constexpr if and accessing captures across nested generic lambdas

[ppalka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

Patrick Palka <ppalka at gcc dot gnu.org> changed:

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

--- Comment #3 from Patrick Palka <ppalka at gcc dot gnu.org> ---
Started with r13-6452 (which has been backported to 11/12).

[Bug c++/114303] [11/12/13/14 Regression] ICE with constexpr if and accessing captures across nested generic lambdas

[law at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

Jeffrey A. Law <law at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at gcc dot gnu.org
           Priority|P3                          |P1

[Bug c++/114303] [11/12/13/14 Regression] ICE with constexpr if and accessing captures across nested generic lambdas

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P1                          |P2

--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
P2, we releases with this bug.

[Bug c++/114303] [11/12/13/14 Regression] ICE with constexpr if and accessing captures across nested generic lambdas

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

--- Comment #5 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Patrick Palka <ppalka@gcc.gnu.org>:

https://gcc.gnu.org/g:b262b17636e47ae969a74f16e86ccb00678d5e88

commit r14-9916-gb262b17636e47ae969a74f16e86ccb00678d5e88
Author: Patrick Palka <ppalka@redhat.com>
Date:   Thu Apr 11 10:16:41 2024 -0400

    c++: build_extra_args recapturing local specs [PR114303]

    r13-6452-g341e6cd8d603a3 made build_extra_args walk evaluated contexts
    first so that we prefer processing a local specialization in an evaluated
    context even if its first use is in an unevaluated context.  But this
    means we need to avoid walking a tree that already has extra args/specs
    saved because the list of saved specs appears to be an evaluated
    context which we'll now walk first.  It seems then that we should be
    calculating the saved specs from scratch each time, rather than
    potentially walking the saved specs list from an earlier partial
    instantiation when calling build_extra_args a second time around.

            PR c++/114303

    gcc/cp/ChangeLog:

            * constraint.cc (tsubst_requires_expr): Clear
            REQUIRES_EXPR_EXTRA_ARGS before calling build_extra_args.
            * pt.cc (tree_extra_args): Define.
            (extract_locals_r): Assert *_EXTRA_ARGS is empty.
            (tsubst_stmt) <case IF_STMT>: Clear IF_SCOPE on the new
            IF_STMT.  Call build_extra_args on the new IF_STMT instead
            of t which might already have IF_STMT_EXTRA_ARGS.

    gcc/testsuite/ChangeLog:

            * g++.dg/cpp1z/constexpr-if-lambda6.C: New test.

    Reviewed-by: Jason Merrill <jason@redhat.com>

[Bug c++/114303] [11/12/13 Regression] ICE with constexpr if and accessing captures across nested generic lambdas

[ppalka at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

Patrick Palka <ppalka at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[11/12/13/14 Regression]    |[11/12/13 Regression] ICE
                   |ICE with constexpr if and   |with constexpr if and
                   |accessing captures across   |accessing captures across
                   |nested generic lambdas      |nested generic lambdas

--- Comment #6 from Patrick Palka <ppalka at gcc dot gnu.org> ---
Fixed for GCC 14 so far

[Bug c++/114303] [11/12/13 Regression] ICE with constexpr if and accessing captures across nested generic lambdas

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=114303

--- Comment #7 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The releases/gcc-13 branch has been updated by Patrick Palka
<ppalka@gcc.gnu.org>:

https://gcc.gnu.org/g:47cac09307874ff1d640392e3d986453f34f7bcb

commit r13-8767-g47cac09307874ff1d640392e3d986453f34f7bcb
Author: Patrick Palka <ppalka@redhat.com>
Date:   Thu Apr 11 10:16:41 2024 -0400

    c++: build_extra_args recapturing local specs [PR114303]

    r13-6452-g341e6cd8d603a3 made build_extra_args walk evaluated contexts
    first so that we prefer processing a local specialization in an evaluated
    context even if its first use is in an unevaluated context.  But this
    means we need to avoid walking a tree that already has extra args/specs
    saved because the list of saved specs appears to be an evaluated
    context which we'll now walk first.  It seems then that we should be
    calculating the saved specs from scratch each time, rather than
    potentially walking the saved specs list from an earlier partial
    instantiation when calling build_extra_args a second time around.

            PR c++/114303

    gcc/cp/ChangeLog:

            * constraint.cc (tsubst_requires_expr): Clear
            REQUIRES_EXPR_EXTRA_ARGS before calling build_extra_args.
            * pt.cc (tree_extra_args): Define.
            (extract_locals_r): Assert *_EXTRA_ARGS is empty.
            (tsubst_stmt) <case IF_STMT>: Clear IF_SCOPE on the new
            IF_STMT.  Call build_extra_args on the new IF_STMT instead
            of t which might already have IF_STMT_EXTRA_ARGS.

    gcc/testsuite/ChangeLog:

            * g++.dg/cpp1z/constexpr-if-lambda6.C: New test.

    Reviewed-by: Jason Merrill <jason@redhat.com>
    (cherry picked from commit b262b17636e47ae969a74f16e86ccb00678d5e88)