[gcc(refs/vendors/ARM/heads/morello)] cp: Use same_type_p in cp_build_binary_op for intcap check

[gcc(refs/vendors/ARM/heads/morello)] cp: Use same_type_p in cp_build_binary_op for intcap check

[Alex Coplan]

https://gcc.gnu.org/g:b479d8363d3f0387747de0d039ad9870d3908624

commit b479d8363d3f0387747de0d039ad9870d3908624
Author: Alex Coplan <alex.coplan@arm.com>
Date:   Sat May 7 18:38:44 2022 +0100

    cp: Use same_type_p in cp_build_binary_op for intcap check
    
    This fixes an ICE seen with the following testcase (reduced from building
    purecap libstdc++):
    
    typedef long T;
    T sz;
    __intcap f() {
      return sz + (__intcap)f;
    }
    
    the issue here is that we have two different non-capability types: T and
    long are both really the same type, but are represented by different
    nodes in the IR, so pointer equality fails us, and we ICE in
    cp_build_binary_op. Adjust the check to use same_type_p instead.
    
    gcc/cp/ChangeLog:
    
            * typeck.c (cp_build_binary_op): Use same_type_p to check the
            non-capability type of the final intcap type is the same type as
            the type we did the arithmetic in.

Diff:
---
 gcc/cp/typeck.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gcc/cp/typeck.c b/gcc/cp/typeck.c
index 84fecd410cf..5bab85e412b 100644
--- a/gcc/cp/typeck.c
+++ b/gcc/cp/typeck.c
@@ -5879,7 +5879,7 @@ cp_build_binary_op (const op_location_t &location,
 	{
 	  tree ic_type = cp_common_type (TREE_TYPE (intcap),
 					 TREE_TYPE (result));
-	  gcc_assert (TREE_TYPE (result) == noncapability_type (ic_type));
+	  gcc_assert (same_type_p (TREE_TYPE (result), noncapability_type (ic_type)));
 	  intcap = cp_convert (ic_type, intcap, complain);
 	  result = fold_build_replace_address_value_loc (location,
 							 intcap,