RFR(S): 8196626: [Graal] runtime/appcds/UseAppCDS.java crashes with "VM thread using lock Heap_lock (not allowed to block on)"

Calvin Cheung calvin.cheung at oracle.com
Fri Mar 9 20:49:55 UTC 2018


Thanks, Misha.
I've got 2 reviews and will push the change soon.

Calvin

On 3/9/18, 12:31 PM, Mikhailo Seledtsov wrote:
> Test looks good,
>
> Misha
>
> On 3/8/18, 10:32 AM, Calvin Cheung wrote:
>>
>>
>> On 3/7/18, 10:42 PM, Ioi Lam wrote:
>>> Hi Calvin,
>>>
>>> The VM changes look good.
>> Thanks for your review.
>>>
>>> For the test, instead of invoking the jartool directly, maybe you 
>>> can do something like this?
>>>
>>> * @run driver ClassFileInstaller -jar test.jar UseAppCDS_Test
>> Done.
>>>
>>> Also, the "String[] unexpectedClasses" is always empty. Maybe this 
>>> code should be removed?
>> I've removed it and some unused import and variables.
>>
>> updated webrev:
>>     http://cr.openjdk.java.net/~ccheung/8196626/webrev.01/
>>
>> thanks,
>> Calvin
>>>
>>> Thanks
>>>
>>> - Ioi
>>>
>>>
>>> On 3/7/18 11:20 AM, Calvin Cheung wrote:
>>>>
>>>> bug: https://bugs.openjdk.java.net/browse/JDK-8196626
>>>>
>>>> webrev: http://cr.openjdk.java.net/~ccheung/8196626/webrev.00/
>>>>
>>>> This change is to print a message and abort CDS dumping when the 
>>>> allocation of metaspace has failed.
>>>>
>>>> The message will look like the following:
>>>>
>>>>     Failed allocating metaspace object type TypeArrayU8 of size 24. 
>>>> CDS dump aborted.
>>>>
>>>> The crash can be reproduced reliably with the new test case.
>>>>
>>>> Testing:
>>>>     The new test case on linux-x64, macosx, windows-x64.
>>>>     hs-tier[1|2|3] on the above platforms plus sparcv9.
>>>>
>>>> thanks,
>>>> Calvin
>>>>
>>>


More information about the hotspot-runtime-dev mailing list