RFR: JDK-8285675: Temporary fix for arm32 SafeFetch

Thomas Stuefe stuefe at openjdk.java.net
Wed Apr 27 11:33:44 UTC 2022


On Wed, 27 Apr 2022 10:44:31 GMT, Martin Doerr <mdoerr at openjdk.org> wrote:

>> [JDK-8283326](https://bugs.openjdk.java.net/browse/JDK-8283326) reworked the SafeFetch implementation and, on most platforms, switched to static assembly.
>> 
>> On arm (32-bit), this change led to strange crashes ([JDK-8284997](https://bugs.openjdk.java.net/browse/JDK-8284997)). These problems are time consuming to analyze (among other things they only occur when building natively on a Rasberry Pi with one particular Docker setup).
>> 
>> As a temporary workaround I'd like to switch SafeFetch to the sigjmp method for arm, until I find time to analyze these problems in more detail.
>
> Looks like a good and trivial workaround for JDK-8284997.

Thank you @TheRealMDoerr!

Also thanks to @marchof and @jgneff for testing the patch. 

I confirmed that building on my agonizingly slow Raspberry works and that safefetch gtests run through without errors.

-------------

PR: https://git.openjdk.java.net/jdk/pull/8399


More information about the hotspot-runtime-dev mailing list