RFR: JDK-8243991 Use standard -Xss argument in java command line

[David Holmes]

Hi Magnus,

On 29/04/2020 3:51 am, Magnus Ihse Bursie wrote:
> On 2020-04-28 18:33, Erik Joelsson wrote:
>>
>> On 2020-04-28 09:14, Magnus Ihse Bursie wrote:
>>> It builds without it. I tested running the tier1 test suite; but I 
>>> assume that if the build succeeds there's really all there is to it.
>>>
>>> I'm not sure where to look for performance issues. Is there some 
>>> specific thing you're worried about? If anything, I think this might 
>>> be affecting machines with different memory sizes differently, but 
>>> I'm not really in the mood to try this on a wide range of machines 
>>> just to find that out.
>>>
>>> Otherwise I'd assume you'd either get a stack overflow exception, or 
>>> everything is green. Shrinking the stack size could possibly mean 
>>> that the build will pass on low-end machines where it previously failed.
>>>
>> I did a bit of digging in the bug database and these settings are very 
>> old. Here is a quote from a comment on a build problem on Solaris 
>> Sparc 64 bit from 2001:
>>
>> "There is a comment in the VM source tree which stated that the 
>> ThreadStackSize needed to be increased to 512K in order to build the 
>> JDK. The 64 bit VM uses twice as much memory as the 32 bit VM so it 
>> might be worth increasing the ThreadStackSize for the JAVAC operations 
>> to 1024K for 64 bit builds. The jdk builds currenty use 768K 
>> ThreadStacks."
>>
>> I think this makes it clear that the intention was to increase the 
>> size of the thread stacks and at the time it was needed to even make 
>> it build. I believe the JVM has matured enough since then to safely 
>> remove these options now.

I tracked this back, way back, .... its ancient. :) Originally the first 
setting of ThreadStackSize was 768K (needed for 64-bit) in the 1.4.0 dev 
days (Jan 2001). Later (JDK-4448510 June 2001) it was determined 768K 
was too small on sparcv9 so it was split into 768K for 32-bit and 1024K 
for 64-bit. Then in August 2003 it was further increased for 64-bit IA64 
(JDK-4905568) and they just doubled the 32-bit value (though that may 
have been a workaround for a javac bug). And there you have it.

> New webrev where I remove the flag altogether:
> 
> http://cr.openjdk.java.net/~ihse/JDK-8243991-fix-Xss-in-java-args/webrev.02

I think this is perfectly reasonable. The default stack sizes should be 
more than adequate.

I have to wonder about the other VM flags that get set e.g. for heap :)

Thanks,
David
-----

> /Magnus
>>
>> /Erik
>>
>>
>