RFR(S): 8005572: fatal error: acquiring lock JfrBuffer_lock/19 out of order with lock MethodData_lock/19 -- possible deadlock

Staffan Larsen staffan.larsen at oracle.com
Thu Mar 7 03:28:21 PST 2013


Thanks David!

On 7 mar 2013, at 11:53, David Holmes <David.Holmes at oracle.com> wrote:

> On 7/03/2013 6:27 PM, Staffan Larsen wrote:
>> Can I have a Review for this small change, please.
> 
> I could have sworn I'd already seen this one and reviewed it :)
> 
> If it fixes the problem - looks fine.
> 
> David
> 
>> Thanks,
>> /Staffan
>> 
>> On 18 feb 2013, at 11:15, Staffan Larsen <staffan.larsen at oracle.com> wrote:
>> 
>>> Please review this change for hs24 / jdku.
>>> 
>>> webrev: http://cr.openjdk.java.net/~sla/8005572/webrev.00/
>>> 
>>> The problem lies with the order of the newly introduced JfrBuffer_lock and JfrStream_lock compared to the MethodData_lock. There are cases where we need to take the Jfr* locks while holding the MethodData_lock. Ths solution is simply to increase the lock order for the Jfr* locks. This change has baked for a while in internal testing and showed no regressions.
>>> 
>>> Thanks,
>>> /Staffan
>> 



More information about the hotspot-runtime-dev mailing list