diff mbox series

c++: bad 'this' conversion for nullary memfn [PR106760]

Message ID 20230602155519.2300026-1-ppalka@redhat.com
State New
Headers
Series c++: bad 'this' conversion for nullary memfn [PR106760] |

Checks

Context Check Description
linaro-tcwg-bot/tcwg_gcc_build--master-arm success Testing passed
linaro-tcwg-bot/tcwg_gcc_build--master-aarch64 success Testing passed
linaro-tcwg-bot/tcwg_gcc_check--master-aarch64 success Testing passed
linaro-tcwg-bot/tcwg_gcc_check--master-arm success Testing passed

Commit Message

Patrick Palka June 2, 2023, 3:55 p.m. UTC 
  Bootstrapped and regtested on x86_64-pc-linu-xgnu, does this look OK for trunk?

-- >8 --

Here we notice the 'this' conversion for the call f<void>() is bad, so
we correctly defer instantiating it, but we end up never adding it to
'bad_fns' since missing_conversion_p for it returns false (its only
argument is the already computed 'this' argument).  This is not a huge
deal, but it causes us to longer accept the call with -fpermissive.

So if we have a non-strictly viable template candidate that has not been
instantiated, then we need to add it to 'bad_fns' even if it doesn't
have any missing conversions.

	PR c++/106760

gcc/cp/ChangeLog:

	* call.cc (add_candidates): Relax test for adding a candidate
	to 'bad_fns' to accept an uninstantiated template candidate that
	has no missing conversions.

gcc/testsuite/ChangeLog:

	* g++.dg/ext/conv3.C: New test.
---
 gcc/cp/call.cc                   |  3 ++-
 gcc/testsuite/g++.dg/ext/conv3.C | 16 ++++++++++++++++
 2 files changed, 18 insertions(+), 1 deletion(-)
 create mode 100644 gcc/testsuite/g++.dg/ext/conv3.C

Comments

Jason Merrill June 2, 2023, 4:28 p.m. UTC | #1 
On 6/2/23 11:55, Patrick Palka wrote:
> Bootstrapped and regtested on x86_64-pc-linu-xgnu, does this look OK for trunk?
> 
> -- >8 --
> 
> Here we notice the 'this' conversion for the call f<void>() is bad, so
> we correctly defer instantiating it, but we end up never adding it to
> 'bad_fns' since missing_conversion_p for it returns false (its only
> argument is the already computed 'this' argument).  This is not a huge
> deal, but it causes us to longer accept the call with -fpermissive.
> 
> So if we have a non-strictly viable template candidate that has not been
> instantiated, then we need to add it to 'bad_fns' even if it doesn't
> have any missing conversions.

I wonder about using ck_deferred_bad in add_template_candidate for this 
case rather than changing the test here?  OK either way.

> 	PR c++/106760
> 
> gcc/cp/ChangeLog:
> 
> 	* call.cc (add_candidates): Relax test for adding a candidate
> 	to 'bad_fns' to accept an uninstantiated template candidate that
> 	has no missing conversions.
> 
> gcc/testsuite/ChangeLog:
> 
> 	* g++.dg/ext/conv3.C: New test.
> ---
>   gcc/cp/call.cc                   |  3 ++-
>   gcc/testsuite/g++.dg/ext/conv3.C | 16 ++++++++++++++++
>   2 files changed, 18 insertions(+), 1 deletion(-)
>   create mode 100644 gcc/testsuite/g++.dg/ext/conv3.C
> 
> diff --git a/gcc/cp/call.cc b/gcc/cp/call.cc
> index 2736f55f229..dbf42567cc9 100644
> --- a/gcc/cp/call.cc
> +++ b/gcc/cp/call.cc
> @@ -6632,7 +6632,8 @@ add_candidates (tree fns, tree first_arg, const vec<tree, va_gc> *args,
>   
>         if (cand->viable == -1
>   	  && shortcut_bad_convs
> -	  && missing_conversion_p (cand))
> +	  && (TREE_CODE (cand->fn) == TEMPLATE_DECL
> +	      || missing_conversion_p (cand)))
>   	{
>   	  /* This candidate has been tentatively marked non-strictly viable,
>   	     and we didn't compute all argument conversions for it (having
> diff --git a/gcc/testsuite/g++.dg/ext/conv3.C b/gcc/testsuite/g++.dg/ext/conv3.C
> new file mode 100644
> index 00000000000..5f4b4d4cc50
> --- /dev/null
> +++ b/gcc/testsuite/g++.dg/ext/conv3.C
> @@ -0,0 +1,16 @@
> +// PR c++/106760
> +// { dg-additional-options "-fpermissive" }
> +
> +struct S {
> +  template<class>
> +  int f(...);
> +
> +  int g() const {
> +    return f<void>(); // { dg-warning "discards qualifiers" }
> +  }
> +};
> +
> +int main() {
> +  S s;
> +  s.g();
> +}
diff mbox series

Patch

diff --git a/gcc/cp/call.cc b/gcc/cp/call.cc
index 2736f55f229..dbf42567cc9 100644
--- a/gcc/cp/call.cc
+++ b/gcc/cp/call.cc
@@ -6632,7 +6632,8 @@  add_candidates (tree fns, tree first_arg, const vec<tree, va_gc> *args,
 
       if (cand->viable == -1
 	  && shortcut_bad_convs
-	  && missing_conversion_p (cand))
+	  && (TREE_CODE (cand->fn) == TEMPLATE_DECL
+	      || missing_conversion_p (cand)))
 	{
 	  /* This candidate has been tentatively marked non-strictly viable,
 	     and we didn't compute all argument conversions for it (having
diff --git a/gcc/testsuite/g++.dg/ext/conv3.C b/gcc/testsuite/g++.dg/ext/conv3.C
new file mode 100644
index 00000000000..5f4b4d4cc50
--- /dev/null
+++ b/gcc/testsuite/g++.dg/ext/conv3.C
@@ -0,0 +1,16 @@ 
+// PR c++/106760
+// { dg-additional-options "-fpermissive" }
+
+struct S {
+  template<class>
+  int f(...);
+
+  int g() const {
+    return f<void>(); // { dg-warning "discards qualifiers" }
+  }
+};
+
+int main() {
+  S s;
+  s.g();
+}