RFR 8217647: JFR: wrong recording on 32-bit systems
Boris Ulasevich
boris.ulasevich at bell-sw.com
Fri Feb 22 15:07:36 UTC 2019
Hi Markus,
We have already pushed it. Thank you!
On 21.02.2019 16:24, Markus Gronlund wrote:
> Hi again Boris,
>
> Thanks for validating and adjusting the suggestion to accommodate this proper onto a 32-bit environment.
>
> With the review from Erik, I can help sponsor this change for you.
>
> Thanks again
> Markus
>
> -----Original Message-----
> From: Boris Ulasevich <boris.ulasevich at bell-sw.com>
> Sent: den 21 februari 2019 14:18
> To: Erik Gahlin <erik.gahlin at oracle.com>
> Cc: hotspot-jfr-dev at openjdk.java.net; David Holmes <david.holmes at oracle.com>
> Subject: Re: RFR 8217647: JFR: wrong recording on 32-bit systems
>
> Thank you!
>
> On 20.02.2019 19:37, Erik Gahlin wrote:
>> Looks good.
>>
>> Erik
>>
>>> On 20 Feb 2019, at 15:15, Boris Ulasevich <boris.ulasevich at bell-sw.com> wrote:
>>>
>>> Hi all,
>>>
>>> Here is a new review for 32-bit systems JFR recording issue. Types was updated to specify variables size explicitly to avoid jfr recording format mismatch. Many thanks to Markus Gronlund who worked on this fix.
>>>
>>> http://cr.openjdk.java.net/~bulasevich/8217647/webrev.02
>>> https://bugs.openjdk.java.net/browse/JDK-8217647
>>>
>>> Tested on x64, x86, arm32.
>>>
>>> regards,
>>> Boris
>>>
>>> On 30.01.2019 8:41, David Holmes wrote:
>>>> Hi Boris,
>>>> Redirecting to hotspot-jfr-dev.
>>>> I can't tell from all the changes exactly where the conflict arises, but the JFR folk are in the best position to identify what should always be 64-bit what should follow the pointer size.
>>>> Cheers,
>>>> David
>>>> -----
>>>> On 30/01/2019 2:43 am, Boris Ulasevich wrote:
>>>>> Hi,
>>>>>
>>>>> Can I please have a review for the following FlightRecorder fix:
>>>>>
>>>>> http://cr.openjdk.java.net/~bulasevich/8217647/webrev.01
>>>>> https://bugs.openjdk.java.net/browse/JDK-8217647
>>>>>
>>>>> The issue is about intptr_t/jlong size mismatch on 32-bit systems. The essence of the fix is is to specify jlong type implicitly when storing data to jfr recording, plus minor types rearrangement was done to avoid unnecessary type conversions.
>>>>>
>>>>> Testing: JFR tests on Linux x64/x32/arm64/arm32, Mach5, Mission Control.
>>>>>
>>>>> thanks,
>>>>> Boris
>>
More information about the hotspot-jfr-dev
mailing list