diff mbox series

c++: Fix explicit instantiation of const variable templates after earlier implicit instantation [PR113976]

Message ID ZdR2yw+J89YmCw50@tucnak
State New
Headers show
Series c++: Fix explicit instantiation of const variable templates after earlier implicit instantation [PR113976] | expand

Commit Message

Jakub Jelinek Feb. 20, 2024, 9:54 a.m. UTC
Hi!

Already previously instantiated const variable templates had
cp_apply_type_quals_to_decl called when they were instantiated,
but if they need runtime initialization, their TREE_READONLY flag
has been subsequently cleared.
Explicit variable template instantiation calls grokdeclarator which
calls cp_apply_type_quals_to_decl on them again, setting TREE_READONLY
flag again, but nothing clears it afterwards, so we emit such
instantiations into rodata sections and segfault when the dynamic
initialization attempts to initialize them.

The following patch fixes that by not calling cp_apply_type_quals_to_decl
on already instantiated variable declarations.

Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

2024-02-19  Jakub Jelinek  <jakub@redhat.com>
	    Patrick Palka  <ppalka@redhat.com>

	PR c++/113976
	* decl.cc (grokdeclarator): Don't call cp_apply_type_quals_to_decl
	on DECL_TEMPLATE_INSTANTIATED VAR_DECLs.

	* g++.dg/cpp1y/var-templ87.C: New test.


	Jakub

Comments

Patrick Palka Feb. 26, 2024, 5:10 p.m. UTC | #1
On Tue, 20 Feb 2024, Jakub Jelinek wrote:

> Hi!
> 
> Already previously instantiated const variable templates had
> cp_apply_type_quals_to_decl called when they were instantiated,
> but if they need runtime initialization, their TREE_READONLY flag
> has been subsequently cleared.
> Explicit variable template instantiation calls grokdeclarator which
> calls cp_apply_type_quals_to_decl on them again, setting TREE_READONLY
> flag again, but nothing clears it afterwards, so we emit such
> instantiations into rodata sections and segfault when the dynamic
> initialization attempts to initialize them.
> 
> The following patch fixes that by not calling cp_apply_type_quals_to_decl
> on already instantiated variable declarations.

LGTM, this seems like the safest approach for backporting.  Note
we can't check DECL_EXPLICIT_INSTANTIATION at this point because
that doesn't get set until later from do_decl_instantiation.

> 
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
> 
> 2024-02-19  Jakub Jelinek  <jakub@redhat.com>
> 	    Patrick Palka  <ppalka@redhat.com>
> 
> 	PR c++/113976
> 	* decl.cc (grokdeclarator): Don't call cp_apply_type_quals_to_decl
> 	on DECL_TEMPLATE_INSTANTIATED VAR_DECLs.
> 
> 	* g++.dg/cpp1y/var-templ87.C: New test.
> 
> --- gcc/cp/decl.cc.jj	2024-02-15 09:51:34.460065992 +0100
> +++ gcc/cp/decl.cc	2024-02-19 19:18:09.839188137 +0100
> @@ -15263,7 +15263,12 @@ grokdeclarator (const cp_declarator *dec
>      /* Record constancy and volatility on the DECL itself .  There's
>         no need to do this when processing a template; we'll do this
>         for the instantiated declaration based on the type of DECL.  */
> -    if (!processing_template_decl)
> +    if (!processing_template_decl
> +	/* Don't do it for instantiated variable templates either,
> +	   cp_apply_type_quals_to_decl should have been called on it
> +	   already and might have been overridden in cp_finish_decl
> +	   if initializer needs runtime initialization.  */
> +	&& (!VAR_P (decl) || !DECL_TEMPLATE_INSTANTIATED (decl)))
>        cp_apply_type_quals_to_decl (type_quals, decl);
>  
>      return decl;
> --- gcc/testsuite/g++.dg/cpp1y/var-templ87.C.jj	2024-02-19 19:21:49.668129195 +0100
> +++ gcc/testsuite/g++.dg/cpp1y/var-templ87.C	2024-02-19 19:21:42.218232862 +0100
> @@ -0,0 +1,43 @@
> +// PR c++/113976
> +// { dg-do run { target c++14 } }
> +
> +int
> +foo ()
> +{
> +  return 42;
> +}
> +
> +template <int N>
> +const int a = foo ();
> +const int *b = &a <0>;
> +template <int N>
> +const int c = foo ();
> +template const int c <0>;
> +template <int N>
> +const int d = foo ();
> +const int *e = &d <0>;
> +template const int d <0>;
> +template <int N>
> +const int f = foo ();
> +template const int f <0>;
> +const int *g = &f <0>;
> +struct S { int a, b; };
> +template <int N>
> +const S h = { 42, foo () };
> +const S *i = &h <0>;
> +template <int N>
> +const S j =  { 42, foo () };
> +template const S j <0>;
> +template <int N>
> +const S k =  { 42, foo () };
> +const S *l = &k <0>;
> +template const S k <0>;
> +template <int N>
> +const S m =  { 42, foo () };
> +template const S m <0>;
> +const S *n = &m <0>;
> +
> +int
> +main ()
> +{
> +}
> 
> 	Jakub
> 
>
Jason Merrill Feb. 28, 2024, 9:49 p.m. UTC | #2
On 2/26/24 12:10, Patrick Palka wrote:
> On Tue, 20 Feb 2024, Jakub Jelinek wrote:
> 
>> Hi!
>>
>> Already previously instantiated const variable templates had
>> cp_apply_type_quals_to_decl called when they were instantiated,
>> but if they need runtime initialization, their TREE_READONLY flag
>> has been subsequently cleared.
>> Explicit variable template instantiation calls grokdeclarator which
>> calls cp_apply_type_quals_to_decl on them again, setting TREE_READONLY
>> flag again, but nothing clears it afterwards, so we emit such
>> instantiations into rodata sections and segfault when the dynamic
>> initialization attempts to initialize them.
>>
>> The following patch fixes that by not calling cp_apply_type_quals_to_decl
>> on already instantiated variable declarations.
> 
> LGTM, this seems like the safest approach for backporting.  Note
> we can't check DECL_EXPLICIT_INSTANTIATION at this point because
> that doesn't get set until later from do_decl_instantiation.

Agreed, OK.
diff mbox series

Patch

--- gcc/cp/decl.cc.jj	2024-02-15 09:51:34.460065992 +0100
+++ gcc/cp/decl.cc	2024-02-19 19:18:09.839188137 +0100
@@ -15263,7 +15263,12 @@  grokdeclarator (const cp_declarator *dec
     /* Record constancy and volatility on the DECL itself .  There's
        no need to do this when processing a template; we'll do this
        for the instantiated declaration based on the type of DECL.  */
-    if (!processing_template_decl)
+    if (!processing_template_decl
+	/* Don't do it for instantiated variable templates either,
+	   cp_apply_type_quals_to_decl should have been called on it
+	   already and might have been overridden in cp_finish_decl
+	   if initializer needs runtime initialization.  */
+	&& (!VAR_P (decl) || !DECL_TEMPLATE_INSTANTIATED (decl)))
       cp_apply_type_quals_to_decl (type_quals, decl);
 
     return decl;
--- gcc/testsuite/g++.dg/cpp1y/var-templ87.C.jj	2024-02-19 19:21:49.668129195 +0100
+++ gcc/testsuite/g++.dg/cpp1y/var-templ87.C	2024-02-19 19:21:42.218232862 +0100
@@ -0,0 +1,43 @@ 
+// PR c++/113976
+// { dg-do run { target c++14 } }
+
+int
+foo ()
+{
+  return 42;
+}
+
+template <int N>
+const int a = foo ();
+const int *b = &a <0>;
+template <int N>
+const int c = foo ();
+template const int c <0>;
+template <int N>
+const int d = foo ();
+const int *e = &d <0>;
+template const int d <0>;
+template <int N>
+const int f = foo ();
+template const int f <0>;
+const int *g = &f <0>;
+struct S { int a, b; };
+template <int N>
+const S h = { 42, foo () };
+const S *i = &h <0>;
+template <int N>
+const S j =  { 42, foo () };
+template const S j <0>;
+template <int N>
+const S k =  { 42, foo () };
+const S *l = &k <0>;
+template const S k <0>;
+template <int N>
+const S m =  { 42, foo () };
+template const S m <0>;
+const S *n = &m <0>;
+
+int
+main ()
+{
+}