RFR(S) 8191437: AOT doesn't work easily after thread local handshakes
dean.long at oracle.com
dean.long at oracle.com
Wed Nov 29 20:14:19 UTC 2017
On 11/29/17 10:38 AM, dean.long at oracle.com wrote:
> On 11/29/17 3:15 AM, Andrew Dinn wrote:
>
>> Hi Dean,
>>
>> On 28/11/17 18:00, dean.long at oracle.com wrote:
>>> https://bugs.openjdk.java.net/browse/JDK-8191437
>>>
>>> http://cr.openjdk.java.net/~dlong/8191437/webrev.hs/
>>>
>>> http://cr.openjdk.java.net/~dlong/8191437/webrev.graal/
>>>
>>>
>>> This change adds support for thread local handshakes to Graal for x64
>>> and SPARC. Graal generates a different instruction encoding on x64
>>> than
>>> C1/C2, so I had to account for that difference in
>>> sharedRuntime_x86_64.cpp. I'm including the Graal changes because I
>>> would like to push them together with the HotSpot changes.
>> Is there a reason for rejecting use of thread local handshakes in the
>> Graal AArch64 code rather than just adding support for it? The changes
>> to make AArch64 implement this are targetted for 10 and in review on
>> aarch64-port-dev
>>
>> https://bugs.openjdk.java.net/browse/JDK-8189596
>>
>> http://mail.openjdk.java.net/pipermail/aarch64-port-dev/2017-November/005104.html
>>
>>
>> regards,
>
> I didn't feel like I had the time or testing setup to do the aarch64
> Graal support myself, but yes I could have asked the aarch64
> community for help. My bad.
> The x64 and SPARC changes have already gone into Graal. Does it make
> sense to file a pull request at https://github.com/graalvm/graal for
> aarch64 support?
>
Aarch64 support is building, at least:
https://github.com/dean-long/graal/tree/aarch64-tls-poll
dl
> dl
>
>>
>> Andrew Dinn
>> -----------
>> Senior Principal Software Engineer
>> Red Hat UK Ltd
>> Registered in England and Wales under Company Registration No. 03798903
>> Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander
>
More information about the hotspot-dev
mailing list