RFR: 8231953: Wrong assumption in assertion in oop::register_oop

Daniel D. Daugherty daniel.daugherty at oracle.com
Wed Oct 23 13:55:38 UTC 2019

On 10/10/19 10:28 AM, christoph.goettschkes at microdoc.com wrote:
> Hi,
> please review the following changeset. This patch makes the
> CheckUnhandledOops feature work on 32-bit ARM platforms. On this platform,
> the os::current_frame() function does not return a valid frame in thumb
> mode [1]. I removed the failing assertion, since it does not hold on all
> supported platforms. I didn't want to introduce architecture specific code
> into that feature, just for an assertion.
> This fixes the hotspot tier1 test
> "runtime/CheckUnhandledOops/TestOutOfMemory.java" mentioned in the issue.
> Bug: https://bugs.openjdk.java.net/browse/JDK-8231953
> Webrev: https://cr.openjdk.java.net/~bulasevich/8231953/webrev.00/

     Looks like the unhandled oops code can tolerate a NULL pc value
     to me so this change is fine.

Thumbs up.


> Thanks,
> Christoph
> [1]
> https://hg.openjdk.java.net/jdk/jdk/file/30a9612a657d/src/hotspot/os_cpu/linux_arm/os_linux_arm.cpp#l224

More information about the hotspot-runtime-dev mailing list