On Fri, Apr 10, 2015 at 5:06 PM, H.J. Lu <[email protected]> wrote:
> On Thu, Apr 9, 2015 at 11:28 AM, Marek Polacek <[email protected]> wrote:
>> This test was failing at -O due to non-trivial conversion at assignment.
>> The reason is that gimple_fold_builtin_memory_op created invalid GIMPLE,
>> because fold_ctor_reference called via fold_const_aggregate_ref was
>> discarding a cast, so it returned a different type than it got. Fixed by
>> dropping useless type conversions only.
>>
>> It also fixes a stale comment that I noticed.
>>
>> Bootstrapped/regtested on x86_64-linux. Preapproved by Jakub on IRC,
>> applying to trunk.
>>
>> 2015-04-09 Marek Polacek <[email protected]>
>> Jakub Jelinek <[email protected]>
>>
>> PR middle-end/65554
>> * gimple-fold.c (gimple_fold_builtin_memory_op): Update comment.
>> (fold_ctor_reference): Use STRIP_USELESS_TYPE_CONVERSION instead
>> of STRIP_NOPS.
>>
>> * g++.dg/opt/pr65554.C: New test.
>>
>
> The testcase failed for 32-bit target:
>
> /export/gnu/import/git/gcc-test-ia32/src-trunk/gcc/testsuite/g++.dg/opt/pr65554.C:47:30:
> fatal error: definition of std::initializer_list does not match
> #include <initializer_list>^M
> compilation terminated.^M
> compiler exited with status 1
> output is:
> /export/gnu/import/git/gcc-test-ia32/src-trunk/gcc/testsuite/g++.dg/opt/pr65554.C:47:30:
> fatal error: definition of std::initializer_list does not match
> #include <initializer_list>^M
> compilation terminated.^M
>
> FAIL: g++.dg/opt/pr65554.C -std=gnu++11 (test for excess errors)
This patch fixes it.
--
H.J.
---
diff --git a/gcc/testsuite/g++.dg/opt/pr65554.C
b/gcc/testsuite/g++.dg/opt/pr65554.C
index 70e43a4..1519964 100644
--- a/gcc/testsuite/g++.dg/opt/pr65554.C
+++ b/gcc/testsuite/g++.dg/opt/pr65554.C
@@ -48,7 +48,7 @@ namespace std
{
typedef _E *const_iterator;
_E *_M_array;
- unsigned long _M_len;
+ __SIZE_TYPE__ _M_len;
const_iterator begin () { return _M_array; }
const_iterator end () { return begin () + 1; }
};
