[13] RFR (S): 8219513: [TESTBUG] : compiler/codegen/aes/TestCipherBlockChainingEncrypt.java timeout on Solaris-sparc

Vladimir Kozlov vladimir.kozlov at oracle.com
Wed Feb 27 17:47:03 UTC 2019


Hi Fairoz,

I am fine with reduced number of iterations but did you verified that intrinsics are still used/generated in execute() 
method?

And I think the test is missing intrinsic availability check on testing system - see 8207153 changes for 
TestBase64.java, for example [1].

Thanks,
Vladimir

[1] http://hg.openjdk.java.net/jdk/jdk/rev/ae001a1deb74#l2.1

On 2/26/19 8:36 PM, Fairoz Matte wrote:
> Hi Igor,
> 
> Thanks for the review, below is the webrev with suggested changes.
> http://cr.openjdk.java.net/~fmatte/8219513/webrev.01/
> 
> Testing Mach5 hs-tier1 - 3 (Observed that test case passes in product and debug builds)
> 
> Thanks,
> Fairoz
> 
>> -----Original Message-----
>> From: Igor Ignatyev
>> Sent: Monday, February 25, 2019 10:46 PM
>> To: Fairoz Matte <fairoz.matte at oracle.com>
>> Cc: hotspot compiler <hotspot-compiler-dev at openjdk.java.net>
>> Subject: Re: [13] RFR (S): 8219513: [TESTBUG] :
>> compiler/codegen/aes/TestCipherBlockChainingEncrypt.java timeout on
>> Solaris-sparc
>>
>> Hi Fairoz,
>>
>> reduction loop iterations to 2k is good.
>>
>> adjustment of timeout for slow debug build (or any other type of vm
>> configuration) should not be done at test-level, and should be done via
>> TIMEOUT_FACTOR make var. instead.
>>
>> Thanks,
>> -- Igor
>>
>>> On Feb 25, 2019, at 8:29 AM, Fairoz Matte <fairoz.matte at oracle.com>
>> wrote:
>>>
>>> Hi,
>>>
>>> http://cr.openjdk.java.net/~fmatte/8219513/webrev.00/
>>> https://bugs.openjdk.java.net/browse/JDK-8219513
>>>
>>> Please review this tiny change to adjust the timeout observed on debug
>> builds.
>>> Single test iteration exercises the aes intrinsic several times.
>>> There are encrypt and decrypt calls in the loop that all use aes.
>>> Reducing count to 2_000 iterations still triggers the compilation.
>>>
>>> Test case uses timeout of 300 seconds is not sufficient for slow debug
>> builds.
>>> That has been adjusted to 600 seconds.
>>>
>>> Thanks,
>>> Fairoz
>>


More information about the hotspot-compiler-dev mailing list