RFR: 8266404: Fatal error report generated with -XX:+CrashOnOutOfMemoryError should not contain suggestion to submit a bug report [v2]
David Holmes
david.holmes at oracle.com
Mon May 17 02:17:32 UTC 2021
Hi Kevin,
Thanks for looking at this.
On 14/05/2021 8:18 pm, Kevin Walls wrote:
> On Thu, 13 May 2021 08:06:24 GMT, David Holmes <dholmes at openjdk.org> wrote:
>
>>> A simple fix to add a new VMErrorType (OOM_JAVA_HEAP_FATAL) to indicate we've requested a Java OOM error to be fatal. A new overload of report_fatal is provided to allow this to be passed through.
>>>
>>> VMError has an existing notion of "should_report_bug" but that encompasses more than just printing the bug submission URL, so I added a new specific check for that.
>>>
>>> Checked hs_err files before and after with the fix, from the CrashOnOutOfMemoryError test.
>>>
>>> Tested tiers 1-3 for good measure.
>>>
>>> Thanks,
>>> David
>>
>> David Holmes has updated the pull request incrementally with one additional commit since the last revision:
>>
>> Review comments from Thomas
>
> Marked as reviewed by kevinw (Committer).
>
> Hi David, looks good.
>
> I get it now I've found that should_report_bug() really means "is vm internal error". Literally, "is this NOT a native memory allocation failure".
>
> If you had time to rename should_report_bug() "is_internal_error" or similar, I think that would make vmError.cpp significantly more readable. (It's not a problem in the function you're adding, it's the old should_report_bug that seems misnamed, and more so now it has a similar related function.)
I couldn't determine exactly what "should_report_error" was really
trying to distinguish, beyond the obvious case of "report the bug submit
URL". As Thomas notes there is potential for more rework in this area,
but I don't want to increase the scope of the current change.
Thanks,
David
> -------------
>
> PR: https://git.openjdk.java.net/jdk/pull/4006
>
More information about the hotspot-runtime-dev
mailing list