RFR (XS) 8223589: Build failures after JDK-8223534 (add back fixed test_markOop.cpp)

[Aleksey Shipilev]

Thanks, I am going to skip jdk-submit on this trivial one, okay?

-Aleksey

On 5/8/19 11:47 PM, coleen.phillimore at oracle.com wrote:
> Yes, that looks like a good fix, and trivial.  Thank you for fixing it.
> Coleen
> 
> On 5/8/19 5:44 PM, Aleksey Shipilev wrote:
>> Bug:
>>    https://bugs.openjdk.java.net/browse/JDK-8223589
>>
>> The only use of assert_not_test_pattern is protected by _LP64, so default x86_32 build config fails
>> with "unused function". We could ifdef the definition too, but better yet, let's just use it on
>> !_LP64 path. Since this is a negative assert, it would not fail if we get the expected string wrong.
>> I had to look into markOopDesc::print_on and guess what the mark word printout on 32-bit would look
>> like. I mean not look like. Err...
>>
>> Fix:
>>
>> diff -r 62b6e7587b1f test/hotspot/gtest/oops/test_markOop.cpp
>> --- a/test/hotspot/gtest/oops/test_markOop.cpp  Tue Apr 16 13:06:23 2019 -0700
>> +++ b/test/hotspot/gtest/oops/test_markOop.cpp  Wed May 08 23:37:51 2019 +0200
>> @@ -102,7 +102,9 @@
>>     obj->set_mark(mark);
>>     ObjectSynchronizer::fast_enter(h_obj, lock.lock(), true, THREAD);
>> +  // Look for the biased_locker in markOop, not prototype_header.
>>   #ifdef _LP64
>> -  // Look for the biased_locker in markOop, not prototype_header.
>>     assert_not_test_pattern(h_obj, "mark(is_biased biased_locker=0x0000000000000000");
>> +#else
>> +  assert_not_test_pattern(h_obj, "mark(is_biased biased_locker=0x00000000");
>>   #endif
>>
>>
>> Testing: Linux (x86_64, x86_32) build, gtest:markOopDesc
>>