[PATCH] c++: Fix resolving the address of overloaded pmf [PR96647]

[Jason Merrill]

On 8/28/20 12:45 PM, Patrick Palka wrote:
> (Removing libstdc++@gcc.gnu.org from CC list)
> 
> On Fri, 28 Aug 2020, Patrick Palka wrote:
>> In resolve_address_of_overloaded_function, currently only the second
>> pass over the overload set (which considers just the function templates
>> in the overload set) checks constraints and performs return type
>> deduction when necessary.  But as the testcases below show, we need to
>> do this when considering non-template functions during the first pass,
>> too.
>>
>> Tested on x86_64-pc-linux-gnu, does this look OK for trunk?
>>
>> gcc/cp/ChangeLog:
>>
>> 	PR c++/96647
>> 	* class.c (resolve_address_of_overloaded_function): Also check
>> 	constraints and perform return type deduction when considering
>> 	non-template functions in the overload set.
>>
>> gcc/testsuite/ChangeLog:
>>
>> 	PR c++/96647
>> 	* g++.dg/cpp0x/auto-96647.C: New test.
>> 	* g++.dg/cpp2a/concepts-fn6.C: New test.
>> ---
>>   gcc/cp/class.c                            | 16 ++++++++++++++++
>>   gcc/testsuite/g++.dg/cpp0x/auto-96647.C   | 10 ++++++++++
>>   gcc/testsuite/g++.dg/cpp2a/concepts-fn6.C | 10 ++++++++++
>>   3 files changed, 36 insertions(+)
>>   create mode 100644 gcc/testsuite/g++.dg/cpp0x/auto-96647.C
>>   create mode 100644 gcc/testsuite/g++.dg/cpp2a/concepts-fn6.C

>> +	if (undeduced_auto_decl (fn))
>> +	  {
>> +	    /* Force instantiation to do return type deduction.  */
>> +	    ++function_depth;
>> +	    instantiate_decl (fn, /*defer*/false, /*class*/false);
>> +	    --function_depth;

How about maybe_instantiate_decl instead of this hunk?  This looks like 
it could call instantiate_decl for a non-template function, which is wrong.

Jason