code review for JVM/TI DynamicCodeGenerated event fix (7039447)

Daniel D. Daugherty daniel.daugherty at oracle.com
Thu Apr 28 13:42:50 PDT 2011


Thanks for confirmation on the second webrev.

Dan


On 4/28/2011 1:12 PM, Dmitry Samersoff wrote:
> Dan,
>
> Thumbs up!
>
> -Dmitry
>
> On 2011-04-28 20:05, Daniel D. Daugherty wrote:
>> Updated webrev to close the loop on this review:
>>
>> http://cr.openjdk.java.net/~dcubed/7039447-webrev/1/
>>
>> This webrev matches the e-mail discussions and diffs that
>> we've been throwing around...
>>
>> Dan
>>
>>
>> On 4/27/2011 1:46 PM, Daniel D. Daugherty wrote:
>>> Greetings,
>>>
>>> I have a fix for a bug in JVM/TI DynamicCodeGenerated event posting:
>>>
>>> 7039447 2/1 java profiling is broken in build 139 (garbage in
>>> function name)
>>>
>>> The bug has actually been tracked back to HSX-21-B04/JDK7-B134, but
>>> the failure mode can be intermittent. Here is the webrev URL:
>>>
>>> http://cr.openjdk.java.net/~dcubed/7039447-webrev/0/
>>>
>>> Here is my proposed commit message:
>>>
>>> 7039447: 2/1 java profiling is broken in build 139 (garbage in
>>> function name)
>>> Summary: The name in a deferred JVM/TI DynamicCodeGenerated event
>>> needs to be explicitly saved.
>>> Reviewed-by:
>>>
>>>
>>> I'm targeting this fix at HSX-21-B12/JDK7-B142; I just missed the
>>> RT_Baseline cutoff for HSX-21-B11/JDK7-B141. Since we are getting
>>> down to the wire on JDK7, I would like at least three reviewers.
>>>
>>> I'm in the process of running the following test suites (the
>>> JPDA parts of the Serviceability stack):
>>>
>>> nsk.jvmti, nsk.jvmti_unit,
>>> nsk.jdwp,
>>> nsk.jdi, SDK_JDI, SDK_JLI,
>>> nsk.hprof, nsk.jdb,
>>> nsk.sajdi,
>>> vm.heapdump, SDK_MISC_ATTACH, SDK_MISC_JVMSTAT, SDK_MISC_TOOLS
>>>
>>> on the following configs:
>>>
>>> {Solaris X86, WinXP} x {Client VM} x {product, fastdebug} x {Xmixed,
>>> Xcomp}
>>>
>>> So far preliminary results show no regressions and no new failures.
>>>
>>> I've also provided JPRT test bits to the Analyzer team and they
>>> have confirmed that the issue is resolved.
>>>
>>> Thanks, in advance, for any reviews.
>>>
>>> Dan
>>>
>>>
>>>
>
>


More information about the hotspot-runtime-dev mailing list