RFR (L) 8213501 : Deploy ExceptionJniWrapper for a few tests

serguei.spitsyn at oracle.com serguei.spitsyn at oracle.com
Wed Dec 12 21:55:25 UTC 2018


Agreed.

Thanks,
Serguei


On 12/12/18 13:52, David Holmes wrote:
> FWIW I think namespaces are overkill in all of this test code and just 
> obfuscates things - the declaration is easily missed. A static 
> variable in a .cpp is clearly a global variable to the file.
>
> Cheers,
> David
>
>
>
> On 13/12/2018 5:37 am, serguei.spitsyn at oracle.com wrote:
>> Hi Jc,
>>
>>
>> On 12/11/18 21:16, JC Beyler wrote:
>>> Hi all,
>>>
>>> Here is the new webrev with the TEST.groups change. Serguei, let me 
>>> know if I convinced you with the static vs anonymous namespaces or 
>>> if you'd still rather have a "static" for now :-)
>>
>>
>> What do you think about this post? :
>> https://stackoverflow.com/questions/11623451/static-vs-non-static-variables-in-namespace 
>>
>>
>>>
>>> Webrev: http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.05/ 
>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.05/>
>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8213501
>>
>> The update looks fine.
>>
>> Thanks,
>> Serguei
>>
>>
>> Thanks,
>> Serguei
>>
>>>
>>> Thanks again for the reviews!
>>> Jc
>>>
>>> On Mon, Dec 10, 2018 at 3:10 PM JC Beyler <jcbeyler at google.com 
>>> <mailto:jcbeyler at google.com>> wrote:
>>>
>>>     Hi Serguei,
>>>
>>>     Yes basically it is equivalent :) I can put them in but they are
>>>     not required. The norm actually wanted to deprecate it but then
>>>     remembered that C compatibility would require the static key-word
>>>     for this case [1]
>>>
>>>     So, really, they are not required here and will amount to the same
>>>     thing: only that file can refer to them and you cannot get to them
>>>     without a globally available method to return a pointer to them
>>>     (ie same as a static variable in C).
>>>
>>>     I can put static if it makes it easier to see but, by being in an
>>>     anonymous namespace they are only available for the file's
>>>     translation unit. For example:
>>>
>>>     $ cat main.cpp
>>>
>>>     int totally_global;
>>>     static int explictly_static;
>>>
>>>     namespace {
>>>     int implicitly_static;
>>>     }
>>>
>>>     void foo();
>>>     int main() {
>>>       foo();
>>>     }
>>>
>>>     $ g++ -O3 main.cpp -c
>>>     $ nm main.o
>>>                      U _GLOBAL_OFFSET_TABLE_
>>>     0000000000000000 T main
>>>     0000000000000000 B totally_global
>>>                      U _Z3foov
>>>
>>>     As you can see, the static and anonymous namespace variables are
>>>     not in the file due to not being used. If you were to use them,
>>>     you'd see them show up as something like:
>>>     0000000000000008 b _ZL17explicitly_static
>>>     0000000000000004 b _ZN12_GLOBAL__N_117implicitly_staticE
>>>
>>>     Where again, it shows that it is mangling the names so that no
>>>     external usage can happen without tinkering.
>>>
>>>     Hopefully that helps :-),
>>>     Jc
>>>
>>>     [1] 
>>> http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#1012
>>>
>>>
>>>     On Mon, Dec 10, 2018 at 2:04 PM serguei.spitsyn at oracle.com
>>>     <mailto:serguei.spitsyn at oracle.com> <serguei.spitsyn at oracle.com
>>>     <mailto:serguei.spitsyn at oracle.com>> wrote:
>>>
>>>         Hi Jc,
>>>
>>>         I had little experience with the C++ namespaces.
>>>         My understanding is that static in this context should mean
>>>         internal linkage.
>>>
>>>         Thanks,
>>>         Serguei
>>>
>>>
>>>         On 12/10/18 13:57, JC Beyler wrote:
>>>>         Hi Serguei,
>>>>
>>>>         The variables and functions are in a anonymous namespace; my
>>>>         understanding of C++ is that this is equivalent to putting it
>>>>         as static.Hence, I didn't add them there. Does that make 
>>>> sense?
>>>>
>>>>         Thanks!
>>>>         Jc
>>>>
>>>>         On Mon, Dec 10, 2018 at 1:33 PM serguei.spitsyn at oracle.com
>>>>         <mailto:serguei.spitsyn at oracle.com>
>>>>         <serguei.spitsyn at oracle.com
>>>>         <mailto:serguei.spitsyn at oracle.com>> wrote:
>>>>
>>>>             Hi Jc,
>>>>
>>>>             It looks good in general.
>>>>             One question though.
>>>>
>>>> http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.03a_04/test/hotspot/jtreg/vmTestbase/nsk/share/ExceptionCheckingJniEnv/exceptionjni001/exceptionjni001.cpp.html
>>>>
>>>>             I wonder if the variables and functions have to be static.
>>>>
>>>>             Thanks,
>>>>             Serguei
>>>>
>>>>
>>>>             On 12/5/18 11:36, JC Beyler wrote:
>>>>>             Hi all,
>>>>>
>>>>>             My apologies to having to come back for another review
>>>>>             for this change: I ran into a snag when trying to pull
>>>>>             the latest changes compared to the base I was working
>>>>>             on. I basically forgot that there was an issue with
>>>>>             snprintf and that I had solved it via JDK-8213622.
>>>>>
>>>>>             Could I have a new review of this webrev:
>>>>>             Webrev:
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.04/
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.04/>
>>>>>             Bug: https://bugs.openjdk.java.net/browse/JDK-8213501
>>>>>             Incremental from the port of webrev.03 that got LGTMs:
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.03a_04/
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.03a_04/>
>>>>>
>>>>>             A few comments on this because it took me a while to get
>>>>>             things in a state I thought was good:
>>>>>               - I had to implement an itoa method, do we have
>>>>>             something like that in the test base (remember that
>>>>>             JDK-8213622 could not use sprintf due to being in the
>>>>>             test code)?
>>>>>
>>>>>               - The differences here compared to the one you all
>>>>>             reviewed are:
>>>>>                   - I found that adding to the strlen/memcpy error
>>>>>             prone and thought that I would try to make it less so.
>>>>>             If you want to compare, I extended the strlen/memcpy
>>>>>             with the new format to show you if you prefer [1]
>>>>>                         - Note that the diff between the "old
>>>>>             extended way from [1]" to the webrev.04 can be found 
>>>>> in [2]
>>>>>
>>>>>                  - I added a test to test the exception wrapper in
>>>>>             tests :); I'm not sure it is deemed useful or not but
>>>>>             helped me assure myself that I was not doing things
>>>>>             wrong; you can find the base test file here [3]; should
>>>>>             we have this or not? (I know that normally we don't add
>>>>>             tests to vmTestbase but thought this might be an 
>>>>> exception)
>>>>>
>>>>>             Thanks for your help and my apologies for the snag,
>>>>>             Jc
>>>>>
>>>>>             [1]:
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.03a/test/hotspot/jtreg/vmTestbase/nsk/share/jni/ExceptionCheckingJniEnv.cpp.udiff.html
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.03a/test/hotspot/jtreg/vmTestbase/nsk/share/jni/ExceptionCheckingJniEnv.cpp.udiff.html>
>>>>>             [2]:
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.03a_04
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.03a_04>
>>>>>             [3]
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.04/test/hotspot/jtreg/vmTestbase/nsk/share/ExceptionCheckingJniEnv/exceptionjni001/exceptionjni001.cpp.html
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.04/test/hotspot/jtreg/vmTestbase/nsk/share/ExceptionCheckingJniEnv/exceptionjni001/exceptionjni001.cpp.html>
>>>>>
>>>>>             On Mon, Dec 3, 2018 at 11:29 PM David Holmes
>>>>>             <david.holmes at oracle.com
>>>>>             <mailto:david.holmes at oracle.com>> wrote:
>>>>>
>>>>>                 Looks fine to me.
>>>>>
>>>>>                 Thanks,
>>>>>                 David
>>>>>
>>>>>                 On 4/12/2018 4:04 pm, JC Beyler wrote:
>>>>>                 > Hi both,
>>>>>                 >
>>>>>                 > Thanks for the reviews! Since Serguei did not
>>>>>                 insist on get_basename, I
>>>>>                 > went for get_dirname since the method is a local
>>>>>                 static method and won't
>>>>>                 > have its name start spreading, I think it's ok too.
>>>>>                 >
>>>>>                 > For the naming of the local variable, the idea
>>>>>                 initially was to use the
>>>>>                 > same name as the local variable for JNIEnv already
>>>>>                 used to reduce the
>>>>>                 > code change. Since I'm now adding the line macro
>>>>>                 at the end anyway, this
>>>>>                 > does not matter anymore so I converged all local
>>>>>                 variables to "jni".
>>>>>                 >
>>>>>                 > So, without further ado, here is the new version:
>>>>>                 > Webrev:
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.03/
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.03/>
>>>>>                 > Bug: 
>>>>> https://bugs.openjdk.java.net/browse/JDK-8213501
>>>>>                 >
>>>>>                 > This passes the various tests changed by the
>>>>>                 webrev on my dev machine.
>>>>>                 >
>>>>>                 > Let me know what you think,
>>>>>                 > Jc
>>>>>                 >
>>>>>                 > On Mon, Dec 3, 2018 at 8:40 PM
>>>>>                 serguei.spitsyn at oracle.com
>>>>>                 <mailto:serguei.spitsyn at oracle.com>
>>>>>                 > <mailto:serguei.spitsyn at oracle.com
>>>>> <mailto:serguei.spitsyn at oracle.com>>
>>>>>                 <serguei.spitsyn at oracle.com
>>>>>                 <mailto:serguei.spitsyn at oracle.com>
>>>>>                 > <mailto:serguei.spitsyn at oracle.com
>>>>> <mailto:serguei.spitsyn at oracle.com>>> wrote:
>>>>>                 >
>>>>>                 >     On 12/3/18 20:15, Chris Plummer wrote:
>>>>>                 >      > Hi JC,
>>>>>                 >      >
>>>>>                 >      > Overall it looks good. A few naming nits
>>>>>                 thought:
>>>>>                 >      >
>>>>>                 >      > In bi01t001.cpp, why have you declared the
>>>>>                 >     ExceptionCheckingJniEnvPtr
>>>>>                 >      > using jni_env(jni). Elsewhere you use
>>>>>                 jni(jni_env) and rename the
>>>>>                 >      > method argument passed in from jni to 
>>>>> jni_env.
>>>>>                 >      >
>>>>>                 >      > Related to this, I also noticed in some
>>>>>                 files that already are using
>>>>>                 >      > ExceptionCheckingJniEnvPtr, such as
>>>>>                 CharArrayCriticalLocker.cpp, you
>>>>>                 >      > delcared it as env(jni_env). So that means
>>>>>                 there are 3 different
>>>>>                 >     names
>>>>>                 >      > you have used for the
>>>>>                 ExceptionCheckingJniEnvPtr local variable.
>>>>>                 >     They
>>>>>                 >      > should be consistent.
>>>>>                 >      >
>>>>>                 >      > Also, can you rename get_basename() to
>>>>>                 get_dirname()? I know Serguei
>>>>>                 >      > suggested get_basename() a while back, but
>>>>>                 unless "basename" is
>>>>>                 >      > commonly used for this purpose, I think
>>>>>                 "dirname" is more self
>>>>>                 >      > explanatory.
>>>>>                 >
>>>>>                 >     In general, I'm Okay with get_dirname().
>>>>>                 >     Just to mention dirname can be both short or
>>>>>                 full, so it is a little
>>>>>                 >     confusing as well.
>>>>>                 >     It is the reason why the get_basename() was
>>>>>                 suggested.
>>>>>                 >     However, I do not insist on get_basename() nor
>>>>>                 get_full_dirname(). :)
>>>>>                 >
>>>>>                 >     Thanks,
>>>>>                 >     Serguei
>>>>>                 >
>>>>>                 >
>>>>>                 >      > thanks,
>>>>>                 >      >
>>>>>                 >      > Chris
>>>>>                 >      >
>>>>>                 >      > On 12/2/18 10:29 PM, David Holmes wrote:
>>>>>                 >      >> Hi Jc,
>>>>>                 >      >>
>>>>>                 >      >> I've been lurking on this one and have had
>>>>>                 a look through. I'm okay
>>>>>                 >      >> with the FatalError approach for the tests
>>>>>                 - we don't expect
>>>>>                 >     anything
>>>>>                 >      >> to go wrong in a well written test in a
>>>>>                 correctly functioning VM.
>>>>>                 >      >>
>>>>>                 >      >> Thanks,
>>>>>                 >      >> David
>>>>>                 >      >>
>>>>>                 >      >>
>>>>>                 >      >>
>>>>>                 >      >> On 3/12/2018 3:24 pm, JC Beyler wrote:
>>>>>                 >      >>> Hi all,
>>>>>                 >      >>>
>>>>>                 >      >>> Would someone on the GC or runtime team
>>>>>                 be motivated to give
>>>>>                 >     this a
>>>>>                 >      >>> review? :)
>>>>>                 >      >>>
>>>>>                 >      >>> It would be much appreciated!
>>>>>                 >      >>>
>>>>>                 >      >>> Webrev:
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.02/
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.02/>
>>>>>                 >      >>> Bug:
>>>>> https://bugs.openjdk.java.net/browse/JDK-8213501
>>>>>                 >      >>>
>>>>>                 >      >>> Thanks for your help,
>>>>>                 >      >>> Jc
>>>>>                 >      >>>
>>>>>                 >      >>> On Tue, Nov 27, 2018 at 4:36 PM JC Beyler
>>>>>                 <jcbeyler at google.com <mailto:jcbeyler at google.com>
>>>>>                 >     <mailto:jcbeyler at google.com
>>>>>                 <mailto:jcbeyler at google.com>>
>>>>>                 >      >>> <mailto:jcbeyler at google.com
>>>>>                 <mailto:jcbeyler at google.com>
>>>>>                 <mailto:jcbeyler at google.com
>>>>> <mailto:jcbeyler at google.com>>>> wrote:
>>>>>                 >      >>>
>>>>>                 >      >>>     Hi Chris,
>>>>>                 >      >>>
>>>>>                 >      >>>     Yes I was waiting for another review
>>>>>                 since you had explicitly
>>>>>                 >      >>> asked :)
>>>>>                 >      >>>
>>>>>                 >      >>>     And sounds good that when someone
>>>>>                 from GC or runtime gives a
>>>>>                 >      >>> review,
>>>>>                 >      >>>     I'll wait for your full review on the
>>>>>                 webrev.02!
>>>>>                 >      >>>
>>>>>                 >      >>>     Thanks again for your help,
>>>>>                 >      >>>     Jc
>>>>>                 >      >>>
>>>>>                 >      >>>
>>>>>                 >      >>>     On Tue, Nov 27, 2018 at 12:48 PM
>>>>>                 Chris Plummer
>>>>>                 >      >>> <chris.plummer at oracle.com
>>>>>                 <mailto:chris.plummer at oracle.com>
>>>>>                 <mailto:chris.plummer at oracle.com
>>>>> <mailto:chris.plummer at oracle.com>>
>>>>>                 >  <mailto:chris.plummer at oracle.com
>>>>>                 <mailto:chris.plummer at oracle.com>
>>>>>                 <mailto:chris.plummer at oracle.com
>>>>> <mailto:chris.plummer at oracle.com>>>>
>>>>>                 >     wrote:
>>>>>                 >      >>>
>>>>>                 >      >>>         Hi JC,
>>>>>                 >      >>>
>>>>>                 >      >>>         I think it would be good to get a
>>>>>                 review from the gc or
>>>>>                 >     runtime
>>>>>                 >      >>>         teams, since this also affects
>>>>>                 their tests.
>>>>>                 >      >>>
>>>>>                 >      >>>         Also, once we are settled on this
>>>>>                 FatalError approach,
>>>>>                 >     I still
>>>>>                 >      >>>         need to give your webrev-02 a
>>>>>                 full review. I only
>>>>>                 >     skimmed over
>>>>>                 >      >>>         parts of it (I did look at all
>>>>>                 the changes in webrevo-01).
>>>>>                 >      >>>
>>>>>                 >      >>>         thanks,
>>>>>                 >      >>>
>>>>>                 >      >>>         Chris
>>>>>                 >      >>>
>>>>>                 >      >>>         On 11/27/18 8:58 AM,
>>>>>                 serguei.spitsyn at oracle.com
>>>>>                 <mailto:serguei.spitsyn at oracle.com>
>>>>>                 >  <mailto:serguei.spitsyn at oracle.com
>>>>> <mailto:serguei.spitsyn at oracle.com>>
>>>>>                 >      >>>                 
>>>>> <mailto:serguei.spitsyn at oracle.com
>>>>>                 <mailto:serguei.spitsyn at oracle.com>
>>>>>                 >  <mailto:serguei.spitsyn at oracle.com
>>>>> <mailto:serguei.spitsyn at oracle.com>>> wrote:
>>>>>                 >      >>>>         Hi Jc,
>>>>>                 >      >>>>
>>>>>                 >      >>>>         I've already reviewed this too.
>>>>>                 >      >>>>
>>>>>                 >      >>>>         Thanks,
>>>>>                 >      >>>>         Serguei
>>>>>                 >      >>>>
>>>>>                 >      >>>>
>>>>>                 >      >>>>         On 11/27/18 06:56, JC Beyler 
>>>>> wrote:
>>>>>                 >      >>>>> Thanks Chris,
>>>>>                 >      >>>>>
>>>>>                 >      >>>>> Anybody else motivated to look at this
>>>>>                 and review it? :)
>>>>>                 >      >>>>>         Jc
>>>>>                 >      >>>>>
>>>>>                 >      >>>>>         On Mon, Nov 26, 2018 at 1:26 PM
>>>>>                 Chris Plummer
>>>>>                 >      >>>>> <chris.plummer at oracle.com
>>>>>                 <mailto:chris.plummer at oracle.com>
>>>>>                 >  <mailto:chris.plummer at oracle.com
>>>>> <mailto:chris.plummer at oracle.com>>
>>>>>                 <mailto:chris.plummer at oracle.com
>>>>>                 <mailto:chris.plummer at oracle.com>
>>>>>                 >  <mailto:chris.plummer at oracle.com
>>>>> <mailto:chris.plummer at oracle.com>>>>
>>>>>                 >      >>>>> wrote:
>>>>>                 >      >>>>>
>>>>>                 >      >>>>> Hi JC,
>>>>>                 >      >>>>>
>>>>>                 >      >>>>> I'm ok with the FatalError approach,
>>>>>                 but would
>>>>>                 >     like to
>>>>>                 >      >>>>> hear opinions from others also.
>>>>>                 >      >>>>>
>>>>>                 >      >>>>> thanks,
>>>>>                 >      >>>>>
>>>>>                 >      >>>>> Chris
>>>>>                 >      >>>>>
>>>>>                 >      >>>>> On 11/21/18 8:19 AM, JC Beyler wrote:
>>>>>                 >      >>>>>>             Hi Chris,
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             Thanks for taking the time
>>>>>                 to look at it and yes you
>>>>>                 >      >>>>>>             have raised exactly why
>>>>>                 the webrev is between two
>>>>>                 >      >>>>>>             worlds: in cases where a
>>>>>                 fatal error on failure is
>>>>>                 >      >>>>>>             wanted, should we simplify
>>>>>                 the code to remove
>>>>>                 >     the return
>>>>>                 >      >>>>>>             tests since we do them
>>>>>                 internally? Now that I've
>>>>>                 >     looked
>>>>>                 >      >>>>>>             around for non-fatal
>>>>>                 cases, I think the answer
>>>>>                 >     is yes,
>>>>>                 >      >>>>>>             it simplifies the code
>>>>>                 while maintaining the checks.
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             I looked a bit and it
>>>>>                 seems that I can't find
>>>>>                 >     easily a
>>>>>                 >      >>>>>>             case where the test
>>>>>                 accepts a JNI failure to
>>>>>                 >     then move
>>>>>                 >      >>>>>>             on. Therefore, perhaps,
>>>>>                 for now, the fail with a
>>>>>                 >     Fatal
>>>>>                 >      >>>>>>             is enough and we can work
>>>>>                 on the tests to clean
>>>>>                 >     them up?
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             That means that this is
>>>>>                 the new webrev with only
>>>>>                 >     Fatal
>>>>>                 >      >>>>>>             and cleans up the tests so
>>>>>                 that it is no longer in
>>>>>                 >      >>>>>>             between two worlds:
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             Webrev:
>>>>>                 >      >>>>>>
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.02/
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.02/>
>>>>>                 >      >>>>>>
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.02/>
>>>>>                 >      >>>>>>             Bug:
>>>>>                 > https://bugs.openjdk.java.net/browse/JDK-8213501
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             (This passes testing on my
>>>>>                 dev machine for all the
>>>>>                 >      >>>>>>             modified tests)
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             with the example you
>>>>>                 provided, it now looks like:
>>>>>                 >      >>>>>>
>>>>>                 >
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.02/test/hotspot/jtreg/vmTestbase/nsk/jvmti/scenarios/allocation/AP04/ap04t003/ap04t003.cpp.frames.html
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.02/test/hotspot/jtreg/vmTestbase/nsk/jvmti/scenarios/allocation/AP04/ap04t003/ap04t003.cpp.frames.html>
>>>>>                 >
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 > 
>>>>>  <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.02/test/hotspot/jtreg/vmTestbase/nsk/jvmti/scenarios/allocation/AP04/ap04t003/ap04t003.cpp.frames.html>
>>>>>                 >
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             Where it does, to me at
>>>>>                 least, seem cleaner and less
>>>>>                 >      >>>>>>             "noisy".
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             Let me know what you think,
>>>>>                 >      >>>>>>             Jc
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             On Tue, Nov 20, 2018 at
>>>>>                 9:33 PM Chris Plummer
>>>>>                 >      >>>>>>             <chris.plummer at oracle.com
>>>>>                 <mailto:chris.plummer at oracle.com>
>>>>>                 >  <mailto:chris.plummer at oracle.com
>>>>> <mailto:chris.plummer at oracle.com>>
>>>>>                 > >>>>>> <mailto:chris.plummer at oracle.com
>>>>>                 <mailto:chris.plummer at oracle.com>
>>>>>                 >  <mailto:chris.plummer at oracle.com
>>>>> <mailto:chris.plummer at oracle.com>>>> wrote:
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 Hi JC,
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 Sorry about the delay.
>>>>>                 I had to go back an
>>>>>                 >     look at
>>>>>                 >      >>>>>>                 the initial 8210842
>>>>>                 webrev and RFR thread to see
>>>>>                 >      >>>>>>                 what this was
>>>>>                 initially all about.
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 In general the changes
>>>>>                 look good.
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 I don't have a good
>>>>>                 answer to your
>>>>>                 > >>>>>> FatalError/NonFatalError question. It makes
>>>>>                 >     the code
>>>>>                 >      >>>>>>                 a lot cleaner to use
>>>>>                 FatalError, but then it
>>>>>                 >     is a
>>>>>                 >      >>>>>>                 behavior change, and
>>>>>                 you also need to deal with
>>>>>                 >      >>>>>>                 tests that
>>>>>                 intentionally induce errors (do
>>>>>                 >     you have
>>>>>                 >      >>>>>>                 an example of that).
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 In any case, right now
>>>>>                 your webrev seems to be
>>>>>                 >      >>>>>>                 between two worlds.
>>>>>                 You are producing
>>>>>                 >     FatalError,
>>>>>                 >      >>>>>>                 but still checking
>>>>>                 results. Here's a good
>>>>>                 >     example:
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 >
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.01/test/hotspot/jtreg/vmTestbase/nsk/jvmti/scenarios/allocation/AP04/ap04t003/ap04t003.cpp.frames.html
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.01/test/hotspot/jtreg/vmTestbase/nsk/jvmti/scenarios/allocation/AP04/ap04t003/ap04t003.cpp.frames.html>
>>>>>                 >
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 > 
>>>>>  <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.01/test/hotspot/jtreg/vmTestbase/nsk/jvmti/scenarios/allocation/AP04/ap04t003/ap04t003.cpp.frames.html>
>>>>>                 >
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 I'm not sure if this
>>>>>                 is just a temporary
>>>>>                 >     state until
>>>>>                 >      >>>>>>                 it was decided which
>>>>>                 approach to take.
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 thanks,
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 Chris
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>                 On 11/20/18 2:14 PM,
>>>>>                 JC Beyler wrote:
>>>>>                 >      >>>>>>>                 Hi all,
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Chris thought it made
>>>>>                 sense to have more
>>>>>                 >     eyes on
>>>>>                 >      >>>>>>>                 this change than just
>>>>>                 serviceability as it will
>>>>>                 >      >>>>>>>                 modify to tests that
>>>>>                 are not only
>>>>>                 >     serviceability
>>>>>                 >      >>>>>>>                 tests so I've moved
>>>>>                 this to conversation
>>>>>                 >     here :)
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 For convenience, I've
>>>>>                 copy-pasted the
>>>>>                 >     initial RFR:
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Could I have a review
>>>>>                 for the extension and
>>>>>                 >     usage
>>>>>                 >      >>>>>>>                 of the
>>>>>                 ExceptionJniWrapper. This adds lines and
>>>>>                 >      >>>>>>>                 filenames to the end
>>>>>                 of the wrapper JNI
>>>>>                 >     methods,
>>>>>                 >      >>>>>>>                 adds tracing, and
>>>>>                 throws an error if need
>>>>>                 >     be. I've
>>>>>                 >      >>>>>>>                 ported the gc/lock
>>>>>                 files to use the new
>>>>>                 >      >>>>>>>                 TRACE_JNI_CALL add-on
>>>>>                 and I've ported a few
>>>>>                 >     of the
>>>>>                 >      >>>>>>>                 tests that were
>>>>>                 already changed for the
>>>>>                 >     assignment
>>>>>                 >      >>>>>>>                 webrev for 
>>>>> JDK-8212884.
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Webrev:
>>>>>                 >      >>>>>>>
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.01
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.01>
>>>>>                 >      >>>>>>>
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.01>
>>>>>                 >      >>>>>>>                 Bug:
>>>>>                 >      >>>>>>>
>>>>> https://bugs.openjdk.java.net/browse/JDK-8213501
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 For illustration, if
>>>>>                 I force an error to the
>>>>>                 >      >>>>>>>                 AP04/ap04t03 test and
>>>>>                 set the verbosity on,
>>>>>                 >     I get
>>>>>                 >      >>>>>>>                 something like:
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 >> Calling JNI method
>>>>>                 FindClass from
>>>>>                 >      >>>>>>> ap04t003.cpp:343
>>>>>                 >      >>>>>>>                 >> Calling with these
>>>>>                 parameter(s):
>>>>>                 >      >>>>>>>                         
>>>>> java/lang/Threadd
>>>>>                 >      >>>>>>>                 Wait for thread to 
>>>>> finish
>>>>>                 >      >>>>>>>                 << Called JNI method
>>>>>                 FindClass from
>>>>>                 >      >>>>>>> ap04t003.cpp:343
>>>>>                 >      >>>>>>>                 Exception in thread
>>>>>                 "Thread-0"
>>>>>                 >      >>>>>>> java.lang.NoClassDefFoundError:
>>>>>                 >     java/lang/Threadd
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003.runIterateOverHeap(Native 
>>>>>
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Method)
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003HeapIterator.runIteration(ap04t003.java:140)
>>>>>                 >
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003Thread.run(ap04t003.java:201)
>>>>>                 >
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Caused by:
>>>>>                 java.lang.ClassNotFoundException:
>>>>>                 >      >>>>>>>                 java.lang.Threadd
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>>                 >
>>>>>  java.base/jdk.internal.loader.BuiltinClassLoader.loadClass(BuiltinClassLoader.java:583)
>>>>>                 >
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>>                 >
>>>>>  java.base/jdk.internal.loader.ClassLoaders$AppClassLoader.loadClass(ClassLoaders.java:178)
>>>>>                 >
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>> java.base/java.lang.ClassLoader.loadClass(ClassLoader.java:521)
>>>>>                 >      >>>>>>>                         ... 3 more
>>>>>                 >      >>>>>>>                 FATAL ERROR in native
>>>>>                 method: JNI method
>>>>>                 >     FindClass
>>>>>                 >      >>>>>>>                 : internal error from
>>>>>                 ap04t003.cpp:343
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003.runIterateOverHeap(Native 
>>>>>
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Method)
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003HeapIterator.runIteration(ap04t003.java:140)
>>>>>                 >
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                         at
>>>>>                 >      >>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003Thread.run(ap04t003.java:201)
>>>>>                 >
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Questions/comments I
>>>>>                 have about this are:
>>>>>                 >      >>>>>>>                   - Do we want to
>>>>>                 force fatal errors when a JNI
>>>>>                 >      >>>>>>>                 call fails in
>>>>>                 general? Most of these tests
>>>>>                 >     do the
>>>>>                 >      >>>>>>>                 right thing and test
>>>>>                 the return of the JNI
>>>>>                 >     calls,
>>>>>                 >      >>>>>>>                 for example:
>>>>>                 >      >>>>>>>                     thrClass =
>>>>>                 >  jni->FindClass("java/lang/Threadd",
>>>>>                 >      >>>>>>>                 TRACE_JNI_CALL);
>>>>>                 >      >>>>>>>                     if (thrClass ==
>>>>>                 NULL) {
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 but now the wrapper
>>>>>                 actually would do a
>>>>>                 >     fatal if
>>>>>                 >      >>>>>>>                 the FindClass call
>>>>>                 would return a nullptr,
>>>>>                 >     so we
>>>>>                 >      >>>>>>>                 could remove that
>>>>>                 test altogether. What do you
>>>>>                 >      >>>>>>> think?
>>>>>                 >      >>>>>>>                     - I prefer to
>>>>>                 leave them as the tests then
>>>>>                 >      >>>>>>>                 become closer to what
>>>>>                 real users would have in
>>>>>                 >      >>>>>>>                 their code and is the
>>>>>                 "recommended" way of
>>>>>                 >     doing it
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                    - The alternative
>>>>>                 is to use the
>>>>>                 >     NonFatalError I
>>>>>                 >      >>>>>>>                 added which then just
>>>>>                 prints out that something
>>>>>                 >      >>>>>>>                 went wrong, letting
>>>>>                 the test continue. Question
>>>>>                 >      >>>>>>>                 will be what should
>>>>>                 be the default? The
>>>>>                 >     fatal or
>>>>>                 >      >>>>>>>                 the non-fatal error
>>>>>                 handling?
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 On a different 
>>>>> subject:
>>>>>                 >      >>>>>>>                   - On the new tests,
>>>>>                 I've removed the
>>>>>                 >      >>>>>>>                 NSK_JNI_VERIFY since
>>>>>                 the JNI wrapper
>>>>>                 >     handles the
>>>>>                 >      >>>>>>>                 tracing and the
>>>>>                 verify in almost the same
>>>>>                 >     way; only
>>>>>                 >      >>>>>>>                 difference I can
>>>>>                 really tell is that the
>>>>>                 >     complain
>>>>>                 >      >>>>>>>                 method from NSK has a
>>>>>                 max complain before
>>>>>                 >     stopping
>>>>>                 >      >>>>>>>                 to "complain"; I have
>>>>>                 not added that part
>>>>>                 >     of the
>>>>>                 >      >>>>>>>                 code in this webrev
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Once we decide on
>>>>>                 these, I can continue on the
>>>>>                 >      >>>>>>>                 files from
>>>>>                 JDK-8212884 and then do both the
>>>>>                 >      >>>>>>>                 assignment in an if
>>>>>                 extraction followed-by this
>>>>>                 >      >>>>>>>                 type of webrev in an
>>>>>                 easier fashion.
>>>>>                 >     Depending on
>>>>>                 >      >>>>>>>                 decisions here,
>>>>>                 NSK*VERIFY can be deprecated as
>>>>>                 >      >>>>>>>                 well as we go forward.
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 Thanks!
>>>>>                 >      >>>>>>>                 Jc
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 On Mon, Nov 19, 2018
>>>>>                 at 11:34 AM Chris Plummer
>>>>>                 > >>>>>>>                 <chris.plummer at oracle.com
>>>>>                 <mailto:chris.plummer at oracle.com>
>>>>>                 >  <mailto:chris.plummer at oracle.com
>>>>> <mailto:chris.plummer at oracle.com>>
>>>>>                 >      >>>>>>> <mailto:chris.plummer at oracle.com
>>>>>                 <mailto:chris.plummer at oracle.com>
>>>>>                 >  <mailto:chris.plummer at oracle.com
>>>>> <mailto:chris.plummer at oracle.com>>>> wrote:
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                     On 11/19/18 10:07
>>>>>                 AM, JC Beyler wrote:
>>>>>                 >      >>>>>>>>                     Hi all,
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                     @David/Chris:
>>>>>                 should I then push this
>>>>>                 >     RFR to
>>>>>                 >      >>>>>>>>                     the hotspot
>>>>>                 mailing or the runtime
>>>>>                 >     one? For
>>>>>                 >      >>>>>>>>                     what it's worth,
>>>>>                 a lot of the tests
>>>>>                 >     under the
>>>>>                 >      >>>>>>>>                     vmTestbase are
>>>>>                 jvmti so the review also
>>>>>                 >      >>>>>>>>                     affects
>>>>>                 serviceability; it just turns
>>>>>                 >     out I
>>>>>                 >      >>>>>>>>                     started with the
>>>>>                 GC originally and
>>>>>                 >     then hit
>>>>>                 >      >>>>>>>>                     some other tests
>>>>>                 I had touched via the
>>>>>                 >      >>>>>>>>                     assignment
>>>>>                 extraction.
>>>>>                 >      >>>>>>>                     I think hotspot
>>>>>                 would be best.
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                     Chris
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                     @Serguei: Done
>>>>>                 for the method
>>>>>                 >     renaming, for
>>>>>                 >      >>>>>>>>                     the indent, are
>>>>>                 you talking about
>>>>>                 >     going from
>>>>>                 >      >>>>>>>>                     the 8-indent to
>>>>>                 4-indent? If so, would
>>>>>                 >     it not
>>>>>                 >      >>>>>>>>                     just be better
>>>>>                 to do a new JBS bug and
>>>>>                 >     do the
>>>>>                 >      >>>>>>>>                     whole files in
>>>>>                 one go? I ask because
>>>>>                 >      >>>>>>>>                     otherwise, it
>>>>>                 will look a bit weird to
>>>>>                 >     have
>>>>>                 >      >>>>>>>>                     parts of the
>>>>>                 file as 8-indent and others
>>>>>                 >      >>>>>>>> 4-indent?
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                     Thanks for
>>>>>                 looking at it!
>>>>>                 >      >>>>>>>>                     Jc
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                     On Mon, Nov 19,
>>>>>                 2018 at 1:25 AM
>>>>>                 >      >>>>>>>> serguei.spitsyn at oracle.com
>>>>>                 <mailto:serguei.spitsyn at oracle.com>
>>>>>                 <mailto:serguei.spitsyn at oracle.com
>>>>> <mailto:serguei.spitsyn at oracle.com>>
>>>>>                 >      >>>>>>>> <mailto:serguei.spitsyn at oracle.com
>>>>>                 <mailto:serguei.spitsyn at oracle.com>
>>>>>                 >  <mailto:serguei.spitsyn at oracle.com
>>>>> <mailto:serguei.spitsyn at oracle.com>>>
>>>>>                 > >>>>>>>>                 
>>>>> <serguei.spitsyn at oracle.com
>>>>>                 <mailto:serguei.spitsyn at oracle.com>
>>>>>                 >  <mailto:serguei.spitsyn at oracle.com
>>>>> <mailto:serguei.spitsyn at oracle.com>>
>>>>>                 >      >>>>>>>> <mailto:serguei.spitsyn at oracle.com
>>>>>                 <mailto:serguei.spitsyn at oracle.com>
>>>>>                 >  <mailto:serguei.spitsyn at oracle.com
>>>>> <mailto:serguei.spitsyn at oracle.com>>>> wrote:
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                         Hi Jc,
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                         We have to
>>>>>                 start this review
>>>>>                 >     anyway. :)
>>>>>                 >      >>>>>>>>                         It looks
>>>>>                 good to me in general.
>>>>>                 >      >>>>>>>>                         Thank you
>>>>>                 for your consistency in this
>>>>>                 >      >>>>>>>>                         refactoring!
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                         Some minor
>>>>>                 comments.
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>
>>>>>                 >
>>>>> http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.00/test/hotspot/jtreg/vmTestbase/nsk/share/jni/ExceptionCheckingJniEnv.cpp.udiff.html
>>>>>                 >
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                         +static
>>>>>                 const char*
>>>>>                 >     remove_folders(const
>>>>>                 >      >>>>>>>>                         char*
>>>>>                 fullname) { I'd suggest to
>>>>>                 >     rename
>>>>>                 >      >>>>>>>>                         the function
>>>>>                 name to something
>>>>>                 >     traditional
>>>>>                 >      >>>>>>>>                         like
>>>>>                 get_basename. Otherwise, it
>>>>>                 >     sounds
>>>>>                 >      >>>>>>>>                         like this
>>>>>                 function has to really
>>>>>                 >     remove
>>>>>                 >      >>>>>>>>                         folders. :)
>>>>>                 Also, all *Locker.cpp have
>>>>>                 >      >>>>>>>>                         wrong indent
>>>>>                 in the bodies of if
>>>>>                 >     and while
>>>>>                 >      >>>>>>>>                         statements.
>>>>>                 Could this be fixed
>>>>>                 >     with the
>>>>>                 >      >>>>>>>>                         refactoring?
>>>>>                 I did not look on how
>>>>>                 >     this
>>>>>                 >      >>>>>>>>                         impacts the
>>>>>                 tests other than
>>>>>                 > >>>>>>>>                 serviceability. Thanks, 
>>>>> Serguei
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                         On 11/16/18
>>>>>                 19:43, JC Beyler wrote:
>>>>>                 > >>>>>>>>>                         Hi all,
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                         Anybody
>>>>>                 motivated to review this? :)
>>>>>                 > >>>>>>>>>                         Jc
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                         On Wed, Nov 7,
>>>>>                 2018 at 9:53 PM JC
>>>>>                 >     Beyler
>>>>>                 > >>>>>>>>>                 <jcbeyler at google.com 
>>>>> <mailto:jcbeyler at google.com>
>>>>>                 >     <mailto:jcbeyler at google.com
>>>>>                 <mailto:jcbeyler at google.com>>
>>>>>                 > >>>>>>>>> <mailto:jcbeyler at google.com
>>>>>                 <mailto:jcbeyler at google.com>
>>>>>                 >     <mailto:jcbeyler at google.com
>>>>> <mailto:jcbeyler at google.com>>>> wrote:
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             Hi all,
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             Could I have
>>>>>                 a review for the
>>>>>                 > >>>>>>>>>                             extension
>>>>>                 and usage of the
>>>>>                 > >>>>>>>>> ExceptionJniWrapper. This
>>>>>                 >     adds lines
>>>>>                 > >>>>>>>>>                             and
>>>>>                 filenames to the end of the
>>>>>                 > >>>>>>>>>                             wrapper JNI
>>>>>                 methods, adds
>>>>>                 >     tracing,
>>>>>                 > >>>>>>>>>                             and throws
>>>>>                 an error if need
>>>>>                 >     be. I've
>>>>>                 > >>>>>>>>>                             ported the
>>>>>                 gc/lock files to
>>>>>                 >     use the
>>>>>                 > >>>>>>>>>                             new
>>>>>                 TRACE_JNI_CALL add-on and
>>>>>                 >     I've
>>>>>                 > >>>>>>>>>                             ported a few
>>>>>                 of the tests
>>>>>                 >     that were
>>>>>                 > >>>>>>>>>                             already
>>>>>                 changed for the
>>>>>                 >     assignment
>>>>>                 > >>>>>>>>>                             webrev for
>>>>>                 JDK-8212884.
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             Webrev:
>>>>>                 > >>>>>>>>>
>>>>> http://cr.openjdk.java.net/~jcbeyler/8213501/webrev.00/
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.00/>
>>>>>                 > >>>>>>>>>
>>>>> <http://cr.openjdk.java.net/%7Ejcbeyler/8213501/webrev.00/>
>>>>>                 > >>>>>>>>>                             Bug:
>>>>>                 > >>>>>>>>>
>>>>> https://bugs.openjdk.java.net/browse/JDK-8213501
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             For
>>>>>                 illustration, if I force
>>>>>                 >     an error
>>>>>                 > >>>>>>>>>                             to the
>>>>>                 AP04/ap04t03 test and
>>>>>                 >     set the
>>>>>                 > >>>>>>>>>                             verbosity
>>>>>                 on, I get something
>>>>>                 >     like:
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             >> Calling
>>>>>                 JNI method
>>>>>                 >     FindClass from
>>>>>                 > >>>>>>>>> ap04t003.cpp:343
>>>>>                 > >>>>>>>>>                             >> Calling
>>>>>                 with these
>>>>>                 >     parameter(s):
>>>>>                 > >>>>>>>>> java/lang/Threadd
>>>>>                 > >>>>>>>>>                             Wait for
>>>>>                 thread to finish
>>>>>                 > >>>>>>>>>                             << Called
>>>>>                 JNI method
>>>>>                 >     FindClass from
>>>>>                 > >>>>>>>>> ap04t003.cpp:343
>>>>>                 > >>>>>>>>>                             Exception in
>>>>>                 thread "Thread-0"
>>>>>                 > >>>>>>>>> java.lang.NoClassDefFoundError:
>>>>>                 > >>>>>>>>> java/lang/Threadd
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003.runIterateOverHeap(Native 
>>>>>
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             Method)
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003HeapIterator.runIteration(ap04t003.java:140)
>>>>>                 >
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003Thread.run(ap04t003.java:201)
>>>>>                 >
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             Caused by:
>>>>>                 > >>>>>>>>> java.lang.ClassNotFoundException:
>>>>>                 > >>>>>>>>> java.lang.Threadd
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  java.base/jdk.internal.loader.BuiltinClassLoader.loadClass(BuiltinClassLoader.java:583)
>>>>>                 >
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  java.base/jdk.internal.loader.ClassLoaders$AppClassLoader.loadClass(ClassLoaders.java:178)
>>>>>                 >
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  java.base/java.lang.ClassLoader.loadClass(ClassLoader.java:521)
>>>>>                 > >>>>>>>>>                             ... 3 more
>>>>>                 > >>>>>>>>>                             FATAL ERROR
>>>>>                 in native method: JNI
>>>>>                 > >>>>>>>>>                             method
>>>>>                 FindClass : internal error
>>>>>                 > >>>>>>>>>                             from
>>>>>                 ap04t003.cpp:343
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003.runIterateOverHeap(Native 
>>>>>
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             Method)
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003HeapIterator.runIteration(ap04t003.java:140)
>>>>>                 >
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             at
>>>>>                 > >>>>>>>>>
>>>>>                 >
>>>>>  nsk.jvmti.scenarios.allocation.AP04.ap04t003Thread.run(ap04t003.java:201)
>>>>>                 >
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>> Questions/comments I have about
>>>>>                 > >>>>>>>>> this are:
>>>>>                 > >>>>>>>>>                               - Do we
>>>>>                 want to force fatal
>>>>>                 >     errors
>>>>>                 > >>>>>>>>>                             when a JNI
>>>>>                 call fails in general?
>>>>>                 > >>>>>>>>>                             Most of
>>>>>                 these tests do the right
>>>>>                 > >>>>>>>>>                             thing and
>>>>>                 test the return of
>>>>>                 >     the JNI
>>>>>                 > >>>>>>>>>                             calls, for
>>>>>                 example:
>>>>>                 > >>>>>>>>>                             thrClass =
>>>>>                 > >>>>>>>>> jni->FindClass("java/lang/Threadd",
>>>>>                 > >>>>>>>>> TRACE_JNI_CALL);
>>>>>                 > >>>>>>>>>                                 if
>>>>>                 (thrClass == NULL) {
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             but now the
>>>>>                 wrapper actually
>>>>>                 >     would do
>>>>>                 > >>>>>>>>>                             a fatal if
>>>>>                 the FindClass call
>>>>>                 >     would
>>>>>                 > >>>>>>>>>                             return a
>>>>>                 nullptr, so we could
>>>>>                 >     remove
>>>>>                 > >>>>>>>>>                             that test
>>>>>                 altogether. What do
>>>>>                 >     you
>>>>>                 > >>>>>>>>> think?
>>>>>                 > >>>>>>>>>                                 - I
>>>>>                 prefer to leave them
>>>>>                 >     as the
>>>>>                 > >>>>>>>>>                             tests then
>>>>>                 become closer to
>>>>>                 >     what real
>>>>>                 > >>>>>>>>>                             users would
>>>>>                 have in their
>>>>>                 >     code and is
>>>>>                 > >>>>>>>>>                             the
>>>>>                 "recommended" way of doing it
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                                - The
>>>>>                 alternative is to
>>>>>                 >     use the
>>>>>                 > >>>>>>>>>                 NonFatalError I added 
>>>>> which
>>>>>                 >     then just
>>>>>                 > >>>>>>>>>                             prints out
>>>>>                 that something
>>>>>                 >     went wrong,
>>>>>                 > >>>>>>>>>                             letting the
>>>>>                 test continue.
>>>>>                 >     Question
>>>>>                 > >>>>>>>>>                             will be what
>>>>>                 should be the
>>>>>                 >     default?
>>>>>                 > >>>>>>>>>                             The fatal or
>>>>>                 the non-fatal error
>>>>>                 > >>>>>>>>>                             handling?
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             On a
>>>>>                 different subject:
>>>>>                 > >>>>>>>>>                               - On the
>>>>>                 new tests, I've
>>>>>                 >     removed
>>>>>                 > >>>>>>>>>                             the
>>>>>                 NSK_JNI_VERIFY since the JNI
>>>>>                 > >>>>>>>>>                             wrapper
>>>>>                 handles the tracing
>>>>>                 >     and the
>>>>>                 > >>>>>>>>>                             verify in
>>>>>                 almost the same
>>>>>                 >     way; only
>>>>>                 > >>>>>>>>>                             difference I
>>>>>                 can really tell
>>>>>                 >     is that
>>>>>                 > >>>>>>>>>                             the complain
>>>>>                 method from NSK
>>>>>                 >     has a
>>>>>                 > >>>>>>>>>                             max complain
>>>>>                 before stopping to
>>>>>                 > >>>>>>>>>                             "complain";
>>>>>                 I have not added that
>>>>>                 > >>>>>>>>>                             part of the
>>>>>                 code in this webrev
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             Once we
>>>>>                 decide on these, I can
>>>>>                 > >>>>>>>>>                             continue on
>>>>>                 the files from
>>>>>                 > >>>>>>>>>                             JDK-8212884
>>>>>                 and then do both the
>>>>>                 > >>>>>>>>>                             assignment
>>>>>                 in an if extraction
>>>>>                 > >>>>>>>>>                             followed-by
>>>>>                 this type of
>>>>>                 >     webrev in an
>>>>>                 > >>>>>>>>>                             easier
>>>>>                 fashion. Depending on
>>>>>                 > >>>>>>>>>                             decisions
>>>>>                 here, NSK*VERIFY can be
>>>>>                 > >>>>>>>>>                             deprecated
>>>>>                 as well as we go
>>>>>                 >     forward.
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                             Thank you
>>>>>                 for the
>>>>>                 >     reviews/comments :)
>>>>>                 > >>>>>>>>>                             Jc
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>
>>>>>                 > >>>>>>>>>                         --
>>>>>                 > >>>>>>>>>                         Thanks,
>>>>>                 > >>>>>>>>>                         Jc
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>
>>>>>                 >      >>>>>>>>                     --
>>>>>                 >      >>>>>>>>                     Thanks,
>>>>>                 >      >>>>>>>>                     Jc
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>
>>>>>                 >      >>>>>>>                 --
>>>>>                 >      >>>>>>>                 Thanks,
>>>>>                 >      >>>>>>>                 Jc
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>
>>>>>                 >      >>>>>>             --
>>>>>                 >      >>>>>>             Thanks,
>>>>>                 >      >>>>>>             Jc
>>>>>                 >      >>>>>
>>>>>                 >      >>>>>
>>>>>                 >      >>>>>
>>>>>                 >      >>>>>
>>>>>                 >      >>>>>         --
>>>>>                 >      >>>>> Thanks,
>>>>>                 >      >>>>>         Jc
>>>>>                 >      >>>>
>>>>>                 >      >>>
>>>>>                 >      >>>
>>>>>                 >      >>>
>>>>>                 >      >>>     --
>>>>>                 >      >>>     Thanks,
>>>>>                 >      >>>     Jc
>>>>>                 >      >>>
>>>>>                 >      >>>
>>>>>                 >      >>>
>>>>>                 >      >>> --
>>>>>                 >      >>>
>>>>>                 >      >>> Thanks,
>>>>>                 >      >>> Jc
>>>>>                 >      >
>>>>>                 >      >
>>>>>                 >
>>>>>                 >
>>>>>                 >
>>>>>                 > --
>>>>>                 >
>>>>>                 > Thanks,
>>>>>                 > Jc
>>>>>
>>>>>
>>>>>
>>>>>             --
>>>>>             Thanks,
>>>>>             Jc
>>>>
>>>>
>>>>
>>>>         --
>>>>         Thanks,
>>>>         Jc
>>>
>>>
>>>
>>>     --
>>>     Thanks,
>>>     Jc
>>>
>>>
>>>
>>> -- 
>>>
>>> Thanks,
>>> Jc
>>



More information about the hotspot-dev mailing list