From: Marek Polacek <polacek@redhat.com>
To: Jason Merrill <jason@redhat.com>
Cc: GCC Patches <gcc-patches@gcc.gnu.org>
Subject: [PATCH v2] c++: ICE with template code in constexpr [PR104284]
Date: Fri, 18 Mar 2022 17:55:38 -0400 [thread overview]
Message-ID: <YjT/2vcFRIYNVWgD@redhat.com> (raw)
In-Reply-To: <edcd7956-daaa-17dc-7f82-06dfbc23e22c@redhat.com>
On Fri, Mar 11, 2022 at 06:46:42PM -0500, Jason Merrill wrote:
> On 3/10/22 18:04, Marek Polacek wrote:
> > Since r9-6073 cxx_eval_store_expression preevaluates the value to
> > be stored, and that revealed a crash where a template code (here,
> > code=IMPLICIT_CONV_EXPR) leaks into cxx_eval*.
> >
> > It happens because we're performing build_vec_init while processing
> > a template
>
> Hmm, that seems like the bug. Where's that call coming from?
From build_aggr_init. So we're handling e.g.
template<class>
constexpr void g () {
constexpr S s2[]{{'a'}};
}
cp_finish_decl (decl=s2, init={{'a'}}) sees we're in processing_template_decl,
but also that we have a constexpr var which is not dependent, nor is its
initializer:
else if (init
&& (init_const_expr_p || DECL_DECLARED_CONSTEXPR_P (decl))
&& !TYPE_REF_P (type)
&& decl_maybe_constant_var_p (decl)
&& !(dep_init = value_dependent_init_p (init)))
{
/* This variable seems to be a non-dependent constant, so process
its initializer. If check_initializer returns non-null the
initialization wasn't constant after all. */
tree init_code;
cleanups = make_tree_vector ();
init_code = check_initializer (decl, init, flags, &cleanups);
so we call check_initializer, where we go down this path:
init_code = build_aggr_init_full_exprs (decl, init, flags);
build_aggr_init sees that the type of 's2' is ARRAY_TYPE, so it calls
build_vec_init.
I now recall that we've discussed build_vec_init in a template in the
past, for example in the context of c++/93676. So I agree we ought to
make an effort to avoid calling build_vec_init in a template. Perhaps
like this: use an INIT_EXPR. With that, we should call build_vec_init
if needed while instantiating. Does that make any sense?
Bootstrapped/regtested on x86_64-pc-linux-gnu, ok for trunk?
-- >8 --
Since r9-6073 cxx_eval_store_expression preevaluates the value to
be stored, and that revealed a crash where a template code (here,
code=IMPLICIT_CONV_EXPR) leaks into cxx_eval*.
It happens because we're performing build_vec_init while processing
a template, which calls get_temp_regvar which creates an INIT_EXPR.
This INIT_EXPR's RHS contains an rvalue conversion so we create an
IMPLICIT_CONV_EXPR. Its operand is not type-dependent and the whole
INIT_EXPR is not type-dependent. So we call build_non_dependent_expr
which, with -fchecking=2, calls fold_non_dependent_expr. At this
point the expression still has an IMPLICIT_CONV_EXPR, which ought to
be handled in instantiate_non_dependent_expr_internal. However,
tsubst_copy_and_build doesn't handle INIT_EXPR; it will just call
tsubst_copy which does nothing when args is null. So we fail to
replace the IMPLICIT_CONV_EXPR and ICE.
The problem is that we call build_vec_init in a template in the
first place. It should work to create an INIT_EXPR in a template
and only perform build_vec_init when instantiating.
PR c++/104284
gcc/cp/ChangeLog:
* init.cc (build_aggr_init): Don't call build_vec_init in
a template, create an INIT_EXPR instead.
gcc/testsuite/ChangeLog:
* g++.dg/cpp1y/constexpr-104284-1.C: New test.
* g++.dg/cpp1y/constexpr-104284-2.C: New test.
* g++.dg/cpp1y/constexpr-104284-3.C: New test.
* g++.dg/cpp1y/constexpr-104284-4.C: New test.
---
gcc/cp/init.cc | 11 +++---
.../g++.dg/cpp1y/constexpr-104284-1.C | 34 ++++++++++++++++++
.../g++.dg/cpp1y/constexpr-104284-2.C | 33 +++++++++++++++++
.../g++.dg/cpp1y/constexpr-104284-3.C | 33 +++++++++++++++++
.../g++.dg/cpp1y/constexpr-104284-4.C | 35 +++++++++++++++++++
5 files changed, 142 insertions(+), 4 deletions(-)
create mode 100644 gcc/testsuite/g++.dg/cpp1y/constexpr-104284-1.C
create mode 100644 gcc/testsuite/g++.dg/cpp1y/constexpr-104284-2.C
create mode 100644 gcc/testsuite/g++.dg/cpp1y/constexpr-104284-3.C
create mode 100644 gcc/testsuite/g++.dg/cpp1y/constexpr-104284-4.C
diff --git a/gcc/cp/init.cc b/gcc/cp/init.cc
index 7575597c8fd..58e66adbfe1 100644
--- a/gcc/cp/init.cc
+++ b/gcc/cp/init.cc
@@ -2006,10 +2006,13 @@ build_aggr_init (tree exp, tree init, int flags, tsubst_flags_t complain)
}
}
- stmt_expr = build_vec_init (exp, NULL_TREE, init,
- /*explicit_value_init_p=*/false,
- from_array,
- complain);
+ /* build_vec_init is not meant to be used in templates. */
+ if (processing_template_decl)
+ stmt_expr = build2 (INIT_EXPR, itype, exp, init);
+ else
+ stmt_expr = build_vec_init (exp, NULL_TREE, init,
+ /*explicit_value_init_p=*/false,
+ from_array, complain);
TREE_READONLY (exp) = was_const;
TREE_THIS_VOLATILE (exp) = was_volatile;
TREE_TYPE (exp) = type;
diff --git a/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-1.C b/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-1.C
new file mode 100644
index 00000000000..809c26a6161
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-1.C
@@ -0,0 +1,34 @@
+// PR c++/104284
+// { dg-do compile { target c++14 } }
+// { dg-additional-options "-fchecking=2" }
+
+struct S {
+ char c{};
+};
+
+auto x1 = [](auto) { constexpr S s[]{{}}; };
+auto x2 = [](auto) { constexpr S s[]{{'a'}}; };
+#if __cpp_designated_initializers >= 201707L
+auto x3 = [](auto) { constexpr S s[]{{.c = 'a'}}; };
+#endif
+auto x4 = [](auto) { constexpr S s[]{'a'}; };
+auto x5 = [](auto) { constexpr S s[]{{{}}}; };
+
+template<class>
+constexpr void g ()
+{
+ constexpr S s1[]{{}};
+ static_assert(s1[0].c == '\0', "");
+ constexpr S s2[]{{'a'}};
+ static_assert(s2[0].c == 'a', "");
+#if __cpp_designated_initializers >= 201707L
+ constexpr S s3[]{{.c = 'a'}};
+ static_assert(s3[0].c == 'a', "");
+#endif
+ constexpr S s4[]{'a'};
+ static_assert(s4[0].c == 'a', "");
+ constexpr S s5[]{{{}}};
+ static_assert(s5[0].c == '\0', "");
+}
+
+static_assert ((g<int>(), true), "");
diff --git a/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-2.C b/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-2.C
new file mode 100644
index 00000000000..704d37de129
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-2.C
@@ -0,0 +1,33 @@
+// PR c++/104284
+// { dg-do compile { target c++14 } }
+// { dg-additional-options "-fchecking=2" }
+
+struct S {
+ char a;
+ constexpr S() : a{'a'} { }
+ constexpr S(char a_) : a{a_} { }
+};
+
+auto x1 = [](auto) { constexpr S s[]{{}}; };
+auto x2 = [](auto) { constexpr S s[]{{'a'}}; };
+auto x3 = [](auto) { constexpr S s[]{'a'}; };
+auto x4 = [](auto) { constexpr S s[]{{{}}}; };
+
+template<typename>
+constexpr void g()
+{
+ constexpr S s1[]{{}};
+ static_assert(s1[0].a == 'a', "");
+ constexpr S s2[]{{'a'}};
+ static_assert(s2[0].a == 'a', "");
+ constexpr S s3[]{'a'};
+ static_assert(s3[0].a == 'a', "");
+ constexpr S s4[]{{{}}};
+ static_assert(s4[0].a == '\0', "");
+}
+
+void
+f ()
+{
+ g<int>();
+}
diff --git a/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-3.C b/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-3.C
new file mode 100644
index 00000000000..6f23b255f9c
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-3.C
@@ -0,0 +1,33 @@
+// PR c++/104284
+// { dg-do compile { target c++14 } }
+// { dg-additional-options "-fchecking=2" }
+// Like constexpr-104284.C, but the function template is not
+// constexpr. In that case, we were still calling build_vec_init
+// in a template, just not crashing.
+
+struct S {
+ char c{};
+};
+
+template<class>
+void g ()
+{
+ constexpr S s1[]{{}};
+ static_assert(s1[0].c == '\0', "");
+ constexpr S s2[]{{'a'}};
+ static_assert(s2[0].c == 'a', "");
+#if __cpp_designated_initializers >= 201707L
+ constexpr S s3[]{{.c = 'a'}};
+ static_assert(s3[0].c == 'a', "");
+#endif
+ constexpr S s4[]{'a'};
+ static_assert(s4[0].c == 'a', "");
+ constexpr S s5[]{{{}}};
+ static_assert(s5[0].c == '\0', "");
+}
+
+void
+f ()
+{
+ g<int>();
+}
diff --git a/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-4.C b/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-4.C
new file mode 100644
index 00000000000..a99d3255a47
--- /dev/null
+++ b/gcc/testsuite/g++.dg/cpp1y/constexpr-104284-4.C
@@ -0,0 +1,35 @@
+// PR c++/104284
+// { dg-do run { target c++14 } }
+// { dg-additional-options "-fchecking=2" }
+
+struct S {
+ char c{};
+};
+
+template<class>
+constexpr void g ()
+{
+ S s1[]{{}};
+ if (s1[0].c != '\0')
+ __builtin_abort ();
+ S s2[]{{'a'}};
+ if (s2[0].c != 'a')
+ __builtin_abort ();
+#if __cpp_designated_initializers >= 201707L
+ S s3[]{{.c = 'a'}};
+ if (s3[0].c != 'a')
+ __builtin_abort ();
+#endif
+ S s4[]{'a'};
+ if (s4[0].c != 'a')
+ __builtin_abort ();
+ S s5[]{{{}}};
+ if (s5[0].c != '\0')
+ __builtin_abort ();
+}
+
+int
+main ()
+{
+ g<int>();
+}
base-commit: 0c016888ffd569c4b70722cf7df2efcc003f397b
--
2.35.1
next prev parent reply other threads:[~2022-03-18 21:55 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-03-10 22:04 [PATCH] " Marek Polacek
2022-03-10 22:27 ` Marek Polacek
2022-03-11 23:46 ` Jason Merrill
2022-03-18 21:55 ` Marek Polacek [this message]
2022-03-24 15:40 ` [PATCH v2] " Jason Merrill
2022-03-24 21:53 ` [PATCH v3] " Marek Polacek
2022-03-24 22:14 ` Jason Merrill
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=YjT/2vcFRIYNVWgD@redhat.com \
--to=polacek@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jason@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).