Request for reviews (XS): 7017240: C2: native memory leak in nsk/regression/b4675027 on windows-x86 in comp mode with G1

[Vladimir Kozlov]

Thank you, Tom, for catching _dom_stk. An other one is _dom_lca_tags(C->comp_arena()). I used arena() set by base class PhaseTransform() to Thread::resource_area.

Updated webrev:

http://cr.openjdk.java.net/~kvn/7017240/webrev

thanks,
Vladimir

Tom Rodriguez wrote:
> On Feb 4, 2011, at 11:48 AM, Vladimir Kozlov wrote:
> 
>> http://cr.openjdk.java.net/~kvn/7017240/webrev
>>
>> Fixed 7017240: C2: native memory leak in nsk/regression/b4675027 on windows-x86 in comp mode with G1
>>
>> C2 adds additional code for G1 barriers so ideal graph
>> may become larger then with other collectors.
>> Also the compiled method itself is huge: 7102 bytes.
>> Main part of the problem is memory resources usage during
>> escape analysis and it will be resolved with 7013538 fix.
>> But I also noticed that PhaseIdealLoop::build_and_optimize()
>> allocates next arrays without freeing resources on exit
>> (there is no ResourceMark):
>>
>>  _idom      = NEW_RESOURCE_ARRAY( Node*, _idom_size );
>>  _dom_depth = NEW_RESOURCE_ARRAY( uint,  _idom_size );
>>
>> Add ResourceMark.
>>
> 
> That looks good.  Should _dom_stk be resource allocated as well?
> 
>   if (_dom_stk == NULL) {
>     uint init_size = C->unique() / 100; // Guess that 1/100 is a reasonable initial size.
>     if (init_size < 10) init_size = 10;
>     _dom_stk = new (C->node_arena()) GrowableArray<uint>(C->node_arena(), init_size, 0, 0);
>   }
> 
> tom
>