RFR(S) 8246019 PerfClassTraceTime slows down VM start-up

[Ioi Lam]

On 6/17/20 7:38 PM, David Holmes wrote:
> Hi Ioi,
>
> On 17/06/2020 1:19 pm, Ioi Lam wrote:
>> On 6/16/20 6:20 PM, David Holmes wrote:
>>> Hi Ioi,
>>>
>>> On 17/06/2020 6:14 am, Ioi Lam wrote:
>>>> https://bugs.openjdk.java.net/browse/JDK-8246019
>>>> http://cr.openjdk.java.net/~iklam/jdk16/8246019-avoid-PerfClassTraceTime.v01/ 
>>>>
>>>>
>>>> PerfClassTraceTime is (a rarely used feature) for measuring the 
>>>> time spent during class linking and initialization.
>>>
>>> "A special command jcmd <process id/main class> PerfCounter.print 
>>> prints all performance counters in the process."
>>>
>>> How do you know this is a "rarely used feature"?
>> Hi David,
>>
>> Sure, the counter will be dumped, but by "rarely used" -- I mean no 
>> one will find this particular counter useful, and no one will be 
>> actively looking at it.
>>
>> I changed two parts of the code -- class init and class linking.
>>
>> For class initialization, the counter may be useful for people who 
>> want to know how much time is spent in their <clinit> functions, and 
>> my patch doesn't change that. It only avoids using the counter when a 
>> class has no <clinit>, i.e., we know that the counter counts nothing 
>> (except for a logging statement).
>
> I can see where you are coming from here. Today we keep track of the 
> time taken to mark a class as initialized when there is no clinit to 
> execute, and we actually record pure timer overhead as it dwarfs the 
> simple update to the class state. With your change we now won't track 
> the time taken to mark the class as initialized. In both cases the 
> time recorded is inaccurate - in opposite senses. In that regard your 
> slight underestimation of the class initalization cost seems better 
> then the present over-estimate.
>
>> =====
>>
>> For class linking, no user code is executed, so it only measures VM 
>> code. If it's useful for anyone, that would be VM engineers like me 
>> who are trying to optimize the speed of class loading. However, due 
>> to the overhead of the counter vs what it's trying to measure, the 
>> results are pretty meaningless.
>>
>> Note that I've not disabled the counter altogether. Instead, I 
>> disable it only when linking a CDS shared class, and we know that 
>> very little is happening for this class (e.g., no verification).
>
> Yes I have little concern for this part. Linking is a multi-phase 
> process so "time for linking" is already ill-defined. And the fact you 
> only do it for CDS makes it even less interesting.
>
>> I think the class linking timer might have been useful 15 years ago 
>> when it was introduced, or it might be useful today when CDS is 
>> disabled. But with CDS enabled, we are paying a constant price that 
>> seems to benefit no one.
>>
>> I think we should short-circuit it when it seems appropriate. If this 
>> indeed causes problems for our users, it's easy to re-enable it. 
>> That's better than just keeping this forever just because we're 
>> afraid to touch anything.
>
> I'm uncomfortable with both the "keep forever as we're too afraid to 
> change" and "change it now and restore it if anyone complains" ends of 
> this spectrum. Obviously we need to make progress, but the "change it 
> now and change back later if needed" is a bit naive, as once any 
> change is made we can't change back without affecting another set of 
> users, and we don't know how long it will be before the change reaches 
> users and the problems return to us. From a CSR perspective I want to 
> see that due diligence has been applied with regard to these 
> behavioural changes, and JDK engineers are often not in a position to 
> understand how end users use this kind of information. I don't have a 
> solution for that general problem.
>
> In this particular case I think under-estimating the class 
> initialization overhead is better than the present over-estimate. 
> Though anyone tracking the trends here will be surprised when the cost 
> suddenly drops.

Hi David,

I don't have a solution, either. I am in no hurry, and the improvement 
is minor.

I could post a CSR and let it stand for a few months to see if anyone 
objects. My impression is "linking time" is such an esoteric feature 
that no one will care, but I may be wrong. Actually it would be good if 
someone tells me I am wrong -- they probably are experiencing some 
overhead in class loading that we don't currently know about.

As I mentioned earlier, if anyone is using this timer, it would be VM 
engineers who work on class loading. In fact, Yumin fixed 8178349 "Cache 
builtin class loader constraints to avoid re-initializing itable/vtable 
for shared classes" in JDK 15, which significantly reduced the amount of 
time spent during class linking. However, we didn't use this timer for 
measuring the effectiveness of that fix, as the overhead and variability 
are too high. We have reached a point in the class loading code that we 
can only make small incremental improvements, and we can only measure 
the effect of our changes with external profilers such as "perf stat -r 
200 bin/java -version" that launches the VM 200 times (and repeat that 
10 times) and averages the elapsed time.

Maybe it's time for the "class linking" timer to go away completely, or 
at least be disabled when CDS is enabled. It's pretty much useless. I 
wish we had an established deprecation process for such legacy features.


>
>>>
>>> I find it hard to evaluate whether this short-circuiting of the time 
>>> tracing is reasonable or not. Obviously any monitoring mechanism 
>>> should impose minimal overhead compared to what is being measured, 
>>> and these timers fall short in that regard. But if these stats 
>>> become meaningless then they may as well be removed.
>>>
>>> I think the serviceability folk (cc'd) need to evaluate this in the 
>>> context of the M&M tools.
>>>
>>>> However, it's quite expensive and it needs to start and stop a 
>>>> bunch of timers. With CDS, it's quite often for the overhead of the 
>>>> timer itself to be much more than the time it's trying to measure, 
>>>> giving unreliable measurement.
>>>>
>>>> In this patch, when it's clear that the init and linking will be 
>>>> very quick, I disable the timer and count only the number of 
>>>> invocations. This shows a small improvement in start-up
>>>
>>> I'm curious if you tried to forcing EagerInitialization to be true 
>>> to see how that improves the baseline. I've always noticed 
>>> eager_init in the code, but hadn't realized it is disabled by default.
>>>
>>
>> I think it cannot be done by default, as it will violate the JLS. A 
>> class can be initialized only when it's touched by bytecodes.
>
> I'm also not sure it violates JLS as you can't directly query if a 
> class is initialized, but anyway I wasn't suggesting turning this on 
> by default, I meant only in regard to get a performance baseline to 
> compare against the changes you have implemented here.

I don't think EagerInitialization will make a difference that's related 
to the timer. Both the two instances of the timer will still be used for 
exactly the same number of times, just under different call stacks.

>
> Aside: I have to wonder whether anyone uses EagerInitialization or 
> whether we should get rid of it?
>
I wonder about the same thing. I'll ask around and file an RFE if 
appropriate.

>> It can also backfire as we may load many classes without initializing 
>> them. E.g., during bytecode verification, we load many classes and 
>> just check that one is a supertype of another.
>
> Not sure what is backfiring here ??

Here's an example that both violates the JLS and backfires as it slows 
down VM start up, because <clinit> can have observable side effects:

class Main {
     static int X;
     public void main(String args[]) {
         System.out.println(X);
     }
     void deadcode() {
         Super s = new Child();
         s.method();
     }
}

class Super {
     void method() {}
}

class Child extends Super {
     static {
         for (int i=0; i<1000000; i++) {
             Main.X ++;
         }
     }
}

When Main is linked, its bytecodes are verified, including deadcode(). 
Since deadcode() has an implicit cast of Child to Super, the verifier 
needs to load both Child and Super, and check that Child is indeed a 
subclass of Super.

If EagerInitialization is enabled, Child will be initialized as soon as 
it's entered into the system dictionary. This violates 
https://docs.oracle.com/javase/specs/jls/se14/html/jls-12.html#jls-12.4.1

    A class or interface type T will be initialized immediately before
    the first occurrence of any one of the following:

    T is a class and an instance of T is created.
    A static method declared by T is invoked.
    A static field declared by T is assigned.
    A static field declared by T is used and the field is not a constant
    variable (§4.12.4).
    ....
    A class or interface will not be initialized under any other
    circumstance.


Also, app developers will generally expect 0 to be printed. So 
EagerInitialization will probably break apps in subtle ways.

And initializing Child may recursively load in more classes during the 
verification of Child ......

Thanks
- Ioi

>
> Thanks,
> David
> -----
>
>> Thanks
>> - Ioi
>>
>>> Cheers,
>>> David
>>> -----
>>>
>>>> Results of " perf stat -r 100 bin/java -Xshare:on 
>>>> -XX:SharedArchiveFile=jdk2.jsa -Xint -version "
>>>>
>>>> 59623970 59341935 (-282035)   -----  41.774  41.591 ( -0.183) -
>>>> 59623495 59331646 (-291849)   -----  41.696  41.165 ( -0.531) --
>>>> 59627148 59329526 (-297622)   -----  41.249  41.094 ( -0.155) -
>>>> 59612439 59340760 (-271679)   ----   41.773  40.657 ( -1.116) -----
>>>> 59626438 59335681 (-290757)   -----  41.683  40.901 ( -0.782) ----
>>>> 59618436 59338953 (-279483)   -----  41.861  41.249 ( -0.612) ---
>>>> 59608782 59340173 (-268609)   ----   41.198  41.508 ( 0.310) +
>>>> 59614612 59325177 (-289435)   -----  41.397  41.738 ( 0.341) ++
>>>> 59615905 59344006 (-271899)   ----   41.921  40.969 ( -0.952) ----
>>>> 59635867 59333147 (-302720)   -----  41.491  40.836 ( -0.655) ---
>>>> ================================================
>>>> 59620708 59336100 (-284608)   -----  41.604  41.169 ( -0.434) --
>>>> instruction delta =      -284608    -0.4774%
>>>> time        delta =       -0.434 ms -1.0435%
>>>>
>>>> The number of PerfClassTraceTime's used is reduced from 564 to 116 
>>>> (so we have an overhead of about 715 instructions per use, yikes!).
>>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.java.net/pipermail/serviceability-dev/attachments/20200617/92fe8862/attachment-0001.htm>