Performance of Scope.getSymbolsByName()

Maurizio Cimadamore maurizio.cimadamore at oracle.com
Wed Jun 12 09:54:00 UTC 2019


On 12/06/2019 07:44, Jan Lahoda wrote:
> Hi Ron,
>
> I ran the tests, I think it looks OK.
> It is this patch, right:
> http://cr.openjdk.java.net/~ronsh/8224161/webrev.00/
>
> Do you need a sponsor? Maurizio, any comments?

I already approved this simpler fix few emails ago in this thread. Looks 
good to me.

Thanks
Maurizio

>
> Thanks,
>     Jan
>
> On 11. 06. 19 20:25, Ron Shapiro wrote:
>> Is there a way I can run those? Any way I can help get this in?
>>
>> On Thu, Jun 6, 2019 at 10:36 AM Jan Lahoda <jan.lahoda at oracle.com 
>> <mailto:jan.lahoda at oracle.com>> wrote:
>>
>>     Hi Ron,
>>
>>     I think we should run it through the perf tests, just to be sure 
>> (and I
>>     didn't have time for that yet), but otherwise, I think that simple
>>     change looks useful to me.
>>
>>     Thanks,
>>           Jan
>>
>>     On 06. 06. 19 3:41, Ron Shapiro wrote:
>>      > Friendly ping - is the first webrev ready for submission?
>>      >
>>      > On Thu, May 30, 2019, 7:09 PM Ron Shapiro <ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>
>>      > <mailto:ronshapiro at google.com <mailto:ronshapiro at google.com>>> 
>> wrote:
>>      >
>>      >     Do you have any other comments on the first webrev? Is there
>>      >     anything else we need to do to submit that?
>>      >
>>      >     On Wed, May 22, 2019 at 11:00 AM Ron Shapiro
>>     <ronshapiro at google.com <mailto:ronshapiro at google.com>
>>      >     <mailto:ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>>> wrote:
>>      >
>>      >         I think there still would be benefit in that as well, 
>> as I'm
>>      >         seeing that come up in other contexts (as referenced in
>>     the bug).
>>      >
>>      >         On Wed, May 22, 2019 at 9:06 AM Jan Lahoda
>>      >         <jan.lahoda at oracle.com <mailto:jan.lahoda at oracle.com>
>>     <mailto:jan.lahoda at oracle.com <mailto:jan.lahoda at oracle.com>>> 
>> wrote:
>>      >
>>      >             Sorry, I was too busy last few days.
>>      >
>>      >             I was peeking at some possible improvements, but I
>>     think I
>>      >             like Ron's
>>      >             first (.00) patch - that caches what can be cached
>>     nicely.
>>      >
>>      >             Looking at the testcase generated by Ron's reproducer
>>     using:
>>      >             python test.py 7
>>      >
>>      >             and the (biggest) size of the outcome of:
>>      >             types.membersClosure(site,
>>     false).getSymbolsByName(sym.name <http://sym.name>
>>      >             <http://sym.name>, cf)
>>      >
>>      >             seems to be 13700 elements - which means the Scope 
>> lookup
>>      >             and iteration
>>      >             runs ~13700 times, so avoiding these additional
>>     lookup costs
>>      >             seems like
>>      >             a clear win.
>>      >
>>      >             I have an idea that might speed up the iterations 
>> through
>>      >             deeply nested
>>      >             CompoundScopes, although the effect of that in
>>     combination
>>      >             with Ron's is
>>      >             going to be fairly limited, if any, I think.
>>      >
>>      >             Jan
>>      >
>>      >             On 22. 05. 19 12:24, Maurizio Cimadamore wrote:
>>      >              > This doesn't work. You are basically relying on
>>     the order
>>      >             in which
>>      >              > symbols are entered in the members closure scope.
>>      >              >
>>      >              > In simple example like these:
>>      >              >
>>      >              > class A {
>>      >              >      int m(List<String> l) {return 0;}
>>      >              > }
>>      >              >
>>      >              > class B extends A {
>>      >              >     int m(List<Integer> l) {return 0;}
>>      >              > }
>>      >              >
>>      >              >
>>      >              > The logic you proposed will not work. That's
>>     because we
>>      >             first see B::m -
>>      >              > and 'symbolByName' is empty at that stage; so we
>>     add it
>>      >             there. Then we
>>      >              > do another round and see A::m - but we don't
>>     really look
>>      >             there - given
>>      >              > that we first check to see if the symbol we're
>>      >             considering (sym) is
>>      >              > override-equivalent with B::m (the only symbol in
>>      >             symbolByName). And
>>      >              > that happens to be the case, since they are the 
>> same
>>      >             symbol. So we exit
>>      >              > the loop, w/o having found any clash.
>>      >              >
>>      >              > In other words, symbolByName would need to also
>>     contain
>>      >             A::m for the
>>      >              > code to see the clash - but that never happens; 
>> by the
>>      >             time A::m is
>>      >              > added, is already too late.
>>      >              >
>>      >              >
>>      >              > I think caching the result of
>>      >              >
>>      >              > types.membersClosure(site,
>>      >             false).getSymbolsByName(sym.name <http://sym.name>
>>     <http://sym.name>, cf)
>>      >              >
>>      >              > is a good measure.
>>      >              >
>>      >              > I'm a bit surprised that iteration is so slow
>>      >             (membersClosure is slow to
>>      >              > set up, but once you do it the results are
>>     cached). So,
>>      >             rather than
>>      >              > tweaking the algorithm, I think it'd be better to
>>      >             investigate the reason
>>      >              > was to why asking a compound scope iterator is so
>>     slow,
>>      >             which then would
>>      >              > yield dividends for the rest of the code as well.
>>      >              >
>>      >              > Maurizio
>>      >              >
>>      >              >
>>      >              > On 21/05/2019 21:21, Maurizio Cimadamore wrote:
>>      >              >>
>>      >              >> I see what you have done - I have to think 
>> about it a
>>      >             bit to see if I
>>      >              >> can come up with some counter example.
>>      >              >>
>>      >              >> Thanks
>>      >              >> Maurizio
>>      >              >>
>>      >              >> On 21/05/2019 17:39, Ron Shapiro wrote:
>>      >              >>> Are the checks of the inner loop symmetrical?
>>      >              >>>
>>      >              >>> Currently it's checking m_i against (m_0..n - 
>> m_i ).
>>      >             This second
>>      >              >>> webrev below would check it against
>>     just (m_0..i-1 ),
>>      >             which albeit
>>      >              >>> still n^2, it divides by a factor of 2.
>>      >              >>>
>>      >              >>> (sorry if the subscripting here doesn't display
>>     correctly)
>>      >              >>>
>>      >              >>> 
>> http://cr.openjdk.java.net/~ronsh/8224161/webrev.01/
>>      >              >>>
>>      >              >>> This feels conceptually logical to me, but 
>> I'm not
>>      >             seeing a
>>      >              >>> measurable change by it. It looks a little bit
>>     cleaner
>>      >             to me, but I'm
>>      >              >>> fine with either webrev given the benefits they
>>     both bring.
>>      >              >>>
>>      >              >>> I can take a look in another thread about
>>     speeding up
>>      >             CompoundScope
>>      >              >>> iteration.
>>      >              >>>
>>      >              >>> On Tue, May 21, 2019 at 8:05 AM Maurizio 
>> Cimadamore
>>      >              >>> <maurizio.cimadamore at oracle.com
>>     <mailto:maurizio.cimadamore at oracle.com>
>>      >             <mailto:maurizio.cimadamore at oracle.com
>>     <mailto:maurizio.cimadamore at oracle.com>>
>>      >              >>> <mailto:maurizio.cimadamore at oracle.com
>>     <mailto:maurizio.cimadamore at oracle.com>
>>      >             <mailto:maurizio.cimadamore at oracle.com
>>     <mailto:maurizio.cimadamore at oracle.com>>>> wrote:
>>      >              >>>
>>      >              >>>
>>      >              >>>     On 21/05/2019 12:16, Ron Shapiro wrote:
>>      >              >>>>     I still think that something to optimize 
>> the
>>      >             actual ScopeImpl
>>      >              >>>>     Iterable is a worthwhile endeavor, as that
>>     would
>>      >             alleviate the
>>      >              >>>>     need to materialize here (and solve
>>     hopefully the
>>      >             other issues
>>      >              >>>>     I'm seeing), but I was having trouble
>>     figuring out
>>      >             how to do
>>      >              >>>>     that. This may be a good interim option 
>> without
>>      >             much cost.
>>      >              >>>
>>      >              >>>     Sure - I'm not opposed to optimizing the
>>     iteration
>>      >             process - I
>>      >              >>>     was expressing my skepticism w.r.t. making
>>      >             checkOverrideClash
>>      >              >>>     simpler/non quadratic.
>>      >              >>>
>>      >              >>>     Maurizio
>>      >              >>>
>>      >              >>>>
>>      >              >>>>
>>      >              >>>>
>>      >              >>>>     On Tue, May 21, 2019, 5:59 AM Maurizio
>>     Cimadamore
>>      >              >>>>  <maurizio.cimadamore at oracle.com
>>     <mailto:maurizio.cimadamore at oracle.com>
>>      >             <mailto:maurizio.cimadamore at oracle.com
>>     <mailto:maurizio.cimadamore at oracle.com>>
>>      >              >>>>  <mailto:maurizio.cimadamore at oracle.com
>>     <mailto:maurizio.cimadamore at oracle.com>
>>      >             <mailto:maurizio.cimadamore at oracle.com
>>     <mailto:maurizio.cimadamore at oracle.com>>>> wrote:
>>      >              >>>>
>>      >              >>>>         I think your fix is a good one. We
>>     spent some
>>      >             cycles
>>      >              >>>>         optimizing this, a bit odd we have
>>     missed this :-)
>>      >              >>>>
>>      >              >>>>         I'm very skeptical you can collapse 
>> into a
>>      >             single loop, as
>>      >              >>>>         this implement the logic in JLS 
>> 8.4.8.3 [1]
>>      >             which, as you
>>      >              >>>>         can see, is inherently quadratic 
>> (for each
>>      >             method, we have
>>      >              >>>>         to scan all methods with same name in
>>      >             supertypes to see if
>>      >              >>>>         there is an override clash). The 
>> algorithm
>>      >             that was there
>>      >              >>>>         before wasn't - and it lead to the 
>> wrong
>>      >             answers in tricky
>>      >              >>>>         cases - so while you can get 80% there
>>     with a
>>      >             non-quadratic
>>      >              >>>>         algorithm, you will miss issues by
>>     doing so.
>>      >              >>>>
>>      >              >>>>         One thing that would help would be,
>>     instead,
>>      >             to limit the
>>      >              >>>>         analysis only in cases where it adds
>>     value -
>>      >             for instance,
>>      >              >>>>         if your hierarchy is just non-generic
>>     classes
>>      >             (as in your
>>      >              >>>>         example), then there's no way for 
>> you to
>>      >             accidentally
>>      >              >>>>         override a 'bridge' method, since no
>>     bridges
>>      >             will be
>>      >              >>>>         generated! But when looking at this, I
>>      >             couldn't find great
>>      >              >>>>         ways to detect these conditions w/o
>>     spending
>>      >             more time than
>>      >              >>>>         the check itself.
>>      >              >>>>
>>      >              >>>>         [1] -
>>      >              >>>>
>>      >
>> https://docs.oracle.com/javase/specs/jls/se12/html/jls-8.html#jls-8.4.8.3
>>      >              >>>>
>>      >              >>>>         Maurizio
>>      >              >>>>
>>      >              >>>>         On 20/05/2019 21:58, Ron Shapiro wrote:
>>      >              >>>>>         In the real world example, I'm seeing
>>     the 40s
>>      >             that was
>>      >              >>>>>         previously spent in
>>      >             Check.checkOverrideClashes drop to to
>>      >              >>>>>         9.5s when using this patch. Of that
>>     9.5, 8.9
>>      >             is spent in
>>      >              >>>>>         iterating through the 
>> CompoundIterator and
>>      >             calling
>>      >              >>>>>  getSymbolsByName.
>>      >              >>>>>
>>      >              >>>>>         On Mon, May 20, 2019 at 4:34 PM Ron
>>     Shapiro
>>      >              >>>>>  <ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>
>>      >             <mailto:ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>> <mailto:ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>
>>      >             <mailto:ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>>>> wrote:
>>      >              >>>>>
>>      >              >>>>>             This patch, which materializes the
>>      >             duplicate outer and
>>      >              >>>>>             inner Iterables first into a 
>> list. It
>>      >             removes the
>>      >              >>>>>             entire section of the 
>> CompoundIterator
>>      >             iteration from
>>      >              >>>>>             the profile.
>>      >              >>>>>
>>      >              >>>>>             webrev:
>>      >              >>>>>
>>      >
>> http://cr.openjdk.java.net/~ronsh/8224161/webrev.00/src/jdk.compiler/share/classes/com/sun/tools/javac/comp/Check.java.sdiff.html
>>      >              >>>>>
>>      >              >>>>>             I'm not sure it's the absolutely
>>     correct
>>      >             solution as it
>>      >              >>>>>             possibly masks an underlying 
>> issue.
>>      >              >>>>>
>>      >              >>>>>             I'm still seeing some time 
>> spent in
>>      >              >>>>>  MethodSymbol.overrides,
>>      >             Types.isSubSignature, and
>>      >              >>>>>  Types.memberType, all of which
>>     happen in
>>      >             the inner
>>      >              >>>>>             loop. If we can remove those and
>>     collapse
>>      >             the nested
>>      >              >>>>>             loops into one, then this solution
>>     isn't
>>      >             necessary and
>>      >              >>>>>             it would also solve that
>>     performance issue.
>>      >              >>>>>
>>      >              >>>>>             On Fri, May 17, 2019 at 5:55 PM
>>     Ron Shapiro
>>      >              >>>>>  <ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>
>>      >             <mailto:ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>> <mailto:ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>
>>      >             <mailto:ronshapiro at google.com
>>     <mailto:ronshapiro at google.com>>>>
>>      >              >>>>>             wrote:
>>      >              >>>>>
>>      >              >>>>>                 I still have more to
>>     investigate to
>>      >             fully wrap my
>>      >              >>>>>                 head around it, but I finally
>>     found a
>>      >             sample
>>      >              >>>>>  program that exhibits this.
>>     Filed a
>>      >             bug here:
>>      >              >>>>> 
>> https://bugs.openjdk.java.net/browse/JDK-8224161
>>      >              >>>>>
>>      >              >>>>>                 On Fri, May 17, 2019 at 11:21
>>     AM Jan
>>      >             Lahoda
>>      >              >>>>>  <jan.lahoda at oracle.com
>>     <mailto:jan.lahoda at oracle.com>
>>      >             <mailto:jan.lahoda at oracle.com
>>     <mailto:jan.lahoda at oracle.com>>
>>      >              >>>>>  <mailto:jan.lahoda at oracle.com
>>     <mailto:jan.lahoda at oracle.com>
>>      >             <mailto:jan.lahoda at oracle.com
>>     <mailto:jan.lahoda at oracle.com>>>> wrote:
>>      >              >>>>>
>>      >              >>>>>  Hi Ron,
>>      >              >>>>>
>>      >              >>>>>  I am afraid it is hard to
>>     guess
>>      >             what is the
>>      >              >>>>>  problem without some
>>      >              >>>>>  testcase. So, at least to me,
>>      >             having a sample
>>      >              >>>>>  would be helpful.
>>      >              >>>>>
>>      >              >>>>>  Thanks,
>>      >              >>>>>       Jan
>>      >              >>>>>
>>      >              >>>>>  On 17. 05. 19 0:41, Ron
>>     Shapiro
>>      >             wrote:
>>      >              >>>>>  > Hi,
>>      >              >>>>>  >
>>      >              >>>>>  > I'm observing a particularly
>>      >             bizarre
>>      >              >>>>>  compilation. It's a single
>>     file
>>      >              >>>>>  > with annotation
>>     processing, and
>>      >             the type that
>>      >              >>>>>  is being compiled and
>>      >              >>>>>  > processed has ~1000 declared
>>      >             and inherited
>>      >              >>>>>  methods combined. The total
>>      >              >>>>>  > compilation is 3
>>     minutes, but
>>      >             65% of the
>>      >              >>>>>  entire compilation is spent in
>>      >              >>>>>  > 3 methods:
>>      >              >>>>>  >
>>      >              >>>>>
>>      >  Check.checkOverrideClashes(), Resolve.findInheritedMemberType(),
>>      >              >>>>>
>>      >              >>>>>  > and Resolve.findField().
>>      >              >>>>>  >
>>      >              >>>>>  > Looking at profiles, it
>>     looks like
>>      >              >>>>>  getSymbolsByName() is the
>>     major
>>      >              >>>>>  > culprit here. I initially
>>      >             thought the reason
>>      >              >>>>>  was that there were far too
>>      >              >>>>>  > many overloads (this
>>     type had >600
>>      >              >>>>>  overloads...) and that
>>     that was
>>      >              >>>>>  > causing a bad regression
>>     for the
>>      >              >>>>>  pseudo-hashmap that
>>     ScopeImpl uses.
>>      >              >>>>>  > However, renaming the
>>     methods
>>      >             did not
>>      >              >>>>>  alleviate the build pain
>>     and these
>>      >              >>>>>  > methods continue to be
>>     taking
>>      >             long amounts of
>>      >              >>>>>  time.
>>      >              >>>>>  >
>>      >              >>>>>  > I was wondering what
>>     could be
>>      >             done to improve
>>      >              >>>>>  the performance of this
>>      >              >>>>>  > code. It seemed to me that
>>      >             something like a
>>      >              >>>>>  Map<Name, List<Symbol>>
>>      >              >>>>>  > could be a reasonable+modern
>>      >             replacement for
>>      >              >>>>>  this table, which would
>>      >              >>>>>  > naturally have a fast
>>      >             getSymbolsForName()
>>      >              >>>>>  implementation. I'm having
>>      >              >>>>>  > some trouble implementing it
>>      >             correctly, and I
>>      >              >>>>>  believe it's partially
>>      >              >>>>>  > related to not fully
>>      >             understanding some of
>>      >              >>>>>  the semantics of the class.
>>      >              >>>>>  >
>>      >              >>>>>  > Does what I wrote make
>>     sense to
>>      >             anyone, and
>>      >              >>>>>  maybe spark a lightbulb?
>>      >              >>>>>  >
>>      >              >>>>>  > I'm trying to put together a
>>      >             repro in case
>>      >              >>>>>  that helps, but I'm not 100%
>>      >              >>>>>  > sure I even understand
>>     what the
>>      >             regression
>>      >              >>>>>  case is.
>>      >              >>>>>  >
>>      >              >>>>>  > Thanks for you help!
>>      >              >>>>>
>>      >
>>


More information about the compiler-dev mailing list