RFR JDK-8193325: StackFrameInfo::getByteCodeIndex returns wrong value if bci > 32767
Mandy Chung
mandy.chung at oracle.com
Thu Aug 15 19:52:03 UTC 2019
On 8/15/19 12:43 PM, Aleksey Shipilev wrote:
> On 8/15/19 9:11 PM, Mandy Chung wrote:
>> On 8/15/19 11:59 AM, Aleksey Shipilev wrote:
>>> On 8/14/19 9:42 PM, Mandy Chung wrote:
>>>> http://cr.openjdk.java.net/~mchung/jdk14/8193325/webrev.05/
>>> Looks good.
>>>
>>> So, to reiterate, we do not need to initialize bci to -1 for StackFrameInfo, because it is not
>>> exposed anywhere?
>>>
>> No, it is only exposed via public API after the StackFrameInfo is filled by the VM.
>>
>> If the uninitialized value is observed, the MemberName may not be uninitialized either. Both bci and MemberName fields should be handled if the race becomes an issue.
> Okay then.
>
> I checked the object footprints before/after the patch, and they are the same. After-patch version has a single byte to spare for something else, if we ever needed it. (The first gap is VM-injected "version" field).
>
Thanks for checking that.
Mandy
More information about the core-libs-dev
mailing list