RFR(S): 8140327: segfault on solaris-amd64 with "-XX:CompilerThreadStackSize=1" option
zoltan.majo at oracle.com
Mon Nov 16 08:34:12 UTC 2015
On 11/13/2015 10:38 PM, Vladimir Kozlov wrote:
> Range change is fine.
thank you for the review!
> I see that a lot of flags are removed from testing because upper limit
> testing can cause OOM.
> May be the test should be modified to at least test low limit for
> these flags. But it is an other RFE.
I'll file an RFE for that and ask Dmitry Dmitriev, the author of the
test, if he has time to look into it.
Thank you and best regards,
> On 11/13/15 8:12 AM, Zoltán Majó wrote:
>> please review the patch for 8140327.
>> Problem: The CompilerThreadStackSize flag gives the size of the
>> compiler thread's stack in KB. The VM defines the range
>> of the flag, but the upper limit of the flag is currently too high as
>> it does not account for the size being specified
>> in KB (and not in bytes).
>> Another problem is that the TestOptionsWithRanges test attempts to
>> start the VM with low/high values in the flag's
>> range. That can cause the VM to crash. It is difficult to find a
>> precise lower limit for the flag because (1) a lower
>> limit is platform-dependent and (2) it can also change over time. The
>> upper limit of the flag is also platform-dependent.
>> Solution: Update the definition of the flag's range. Disable
>> automatic testing for the flag in TestOptionWithRanges.
>> Testing: JPRT
>> Thank you and best regards,
More information about the hotspot-compiler-dev