RFR: [10] Backports for fixing Shenandoah JDK10 non-PCH build
Zhengyu Gu
zgu at redhat.com
Tue Mar 20 15:56:28 UTC 2018
On 03/20/2018 11:41 AM, Aleksey Shipilev wrote:
> On 03/20/2018 04:34 PM, Zhengyu Gu wrote:
>> On 03/14/2018 06:02 AM, Aleksey Shipilev wrote:
>>> On 03/14/2018 11:00 AM, Aleksey Shipilev wrote:
>>>> On 03/13/2018 09:24 PM, Zhengyu Gu wrote:
>>>>> A couple of backports to fix Shenandoah JDk10 build without precompiled headers.
>>>>>
>>>>> Both patches were applied cleanly.
>>>>>
>>>>> Webrev: http://cr.openjdk.java.net/~zgu/shenandoah/sh-jdk10-non-PCH/webrev.00/
>>>>
>>>> *) I would prefer not to pickup partial backports that fix PCH for us. Is there a way to just add
>>>> appropriate headers to make it compile?
>>
>> Okay, JDK-8193063 seems never arrived to upstream jdk10. Redid fix without it. Unfortunately, it
>> does need to touch G1 code to breakup circular dependency.
>>
>> Webrev: http://cr.openjdk.java.net/~zgu/shenandoah/sh-jdk10-non-PCH/webrev.01/
>
> *) As we have reverted the workaround in debugInfo, the additional header is not needed there?
> http://hg.openjdk.java.net/shenandoah/jdk10/rev/7e708bf03522
>
Oops, I did revert it, but forgot to regenerate webrev. It is gone.
http://cr.openjdk.java.net/~zgu/shenandoah/sh-jdk10-non-PCH/webrev.02/
> *) We need to perform the ShenandoahBarrierSet::clone_in_heap move in sh/jdk, then backport it to
> sh/jdk10, then apply this patch
Okay, will wait.
-Zhengyu
>
> Otherwise good.
>
> Thanks,
> -Aleksey
>
More information about the shenandoah-dev
mailing list