RFR(S): 8234108: [lworld] Call site reresolution fails for ValueBootstrapMethods::isSubstitutable0 callee

Tobias Hartmann tobias.hartmann at oracle.com
Fri Nov 15 08:38:52 UTC 2019


Thanks for the review! Pushed.

Best regards,
Tobias

On 15.11.19 09:32, David Simms wrote:
> 
> Looks good sir, especially appreciate the new test, push it
> 
> /D
> 
> On 14/11/19 4:26 PM, Tobias Hartmann wrote:
>> Hi,
>>
>> please review the following patch:
>> https://bugs.openjdk.java.net/browse/JDK-8234108
>> http://cr.openjdk.java.net/~thartmann/8234108/webrev.00/
>>
>> The C2 acmp implementation emits a runtime call to ValueBootstrapMethods::isSubstitutable0. If that
>> method got compiled but becomes non-entrant, we patch the entry to call
>> SharedRuntime::get_handle_wrong_method_stub(). The re-resolution code then looks at the current
>> bytecode of the caller and expects it to be an invoke. Roland already fixed that for static call
>> resolution [1], we should do the same for re-resolution. I've moved the check into
>> SharedRuntime::find_callee_info_helper which is used for both resolution and re-resolution.
>>
>> Thanks,
>> Tobias
>>
>> [1] http://hg.openjdk.java.net/valhalla/valhalla/rev/1d216028b57b#l8.7
> 
> 


More information about the valhalla-dev mailing list