RFR(S): 8213014: Crash in CompileBroker::make_thread due to OOM
tobias.hartmann at oracle.com
Wed Oct 31 08:50:33 UTC 2018
okay, thanks for looking at this!
On 31.10.18 09:48, David Holmes wrote:
> Correction ...
> On 31/10/2018 6:03 PM, David Holmes wrote:
>> Tobias writes:
>>> please review the following patch:
>>> The code in compileBroker.cpp:806 dereferences 'comp' which is NULL if creation of the sweeper
>>> thread fails. I've also removed the 'compiler_thread' argument and replaced it by a "comp != NULL"
>>> check. In addition, the calls to create_thread_oop() in init_compiler_sweeper_threads() need to
>>> CHECK for exceptions.
>> This looks fine to me Tobias.
>> The only query I have (existing issue) is with the dynamic threads case: if we fail to create the
>> osThread should we not at some point call delete on the CompilerThread/SweeperThread instance?
> Sorry I forgot that smr_delete took over responsibility for this. Ignore. :)
More information about the hotspot-compiler-dev