RFR (XS): 7114303: G1: assert(_g1->mark_in_progress()) failed: shouldn't be here otherwise

[Tony Printezis]

Bengt,

That methods looks indeed not to be called, which means that 
_last_satb_drain_processed_buffers is never set (I can't find any 
assignments to it) and the reported SATB processed buffers

       print_stats(2, "Processed Buffers", 
_last_satb_drain_processed_buffers);

will probably always be 0. So, if we think that this output is 
important, we should make sure we actually 
record_satb_drain_processed_buffers() it with the appropriate number. If 
we don't, we should also remove _last_satb_drain_processed_buffers and 
related code.

Tony

On 11/23/2011 02:44 AM, Bengt Rutisson wrote:
>
> Hi John,
>
> Looks good to me too.
>
> It seems like the method "void record_satb_drain_processed_buffers(int 
> processed_buffers)" in g1CollectorPolicy.hpp is dead code. I found it 
> since it contains the same assert that your CR reported.
>
> Any chance of removing that method as well as part of your change?
>
> Bengt
>
> On 2011-11-22 21:45, John Cuthbertson wrote:
>> Hi Everyone,
>>
>> Can I have a couple of volunteers review this small change? The 
>> webrev can be found at: 
>> http://cr.openjdk.java.net/~johnc/7114303/webrev.0/
>>
>> Summary:
>> There is a race between the concurrent mark thread updating the mark 
>> in progress flag and it being read by the VM thread during an 
>> evacuation pause. This is because, at the time the concurrent mark 
>> thread updates the flag, it is not part of the suspendible thread set 
>> and so is not blocked. The solution is to have the concurrent mark 
>> thread join the STS just before updating the flag.
>>
>> The race was confirmed by inserting a small delay immediately before 
>> calling G1CollectorPolicy::record_satb_drain_time(). The delay 
>> increased failure rate to 1/5 for a 20ms delay and 4/5 for a 50ms delay.
>>
>> Testing: a few thousand iterations of the failing test case with a 
>> JVM that included both fix and delay.
>>
>> Thanks,
>>
>> JohnC
>