RFR(xs): 8176442: [aix] assert(_thr_current == 0L) failed: Thread::current already initialized

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

RFR(xs): 8176442: [aix] assert(_thr_current == 0L) failed: Thread::current already initialized

Thomas Stüfe-2
Please review this small AIX change.

Issue: https://bugs.openjdk.java.net/browse/JDK-8176442
webrev:
http://cr.openjdk.java.net/~stuefe/webrevs/8176442-switch-off-compiler-level-tls-on-aix/webrev.00/webrev/


This change switches the implementation of Thread::current() from compiler
level TLS back to pthread library TLS. We see errors on AIX 5.3 machines
where compiler level TLS seems not to work correctly - new TLS variable
instances are sporadically left uninitialized.

For more details, please see the description and comments in the bug report.

Thank you, Thomas
Reply | Threaded
Open this post in threaded view
|

Re: RFR(xs): 8176442: [aix] assert(_thr_current == 0L) failed: Thread::current already initialized

David Holmes
Looks good to me.

David

On 11/03/2017 7:02 PM, Thomas Stüfe wrote:

> Please review this small AIX change.
>
> Issue: https://bugs.openjdk.java.net/browse/JDK-8176442
> webrev:
> http://cr.openjdk.java.net/~stuefe/webrevs/8176442-switch-off-compiler-level-tls-on-aix/webrev.00/webrev/
>
>
> This change switches the implementation of Thread::current() from compiler
> level TLS back to pthread library TLS. We see errors on AIX 5.3 machines
> where compiler level TLS seems not to work correctly - new TLS variable
> instances are sporadically left uninitialized.
>
> For more details, please see the description and comments in the bug report.
>
> Thank you, Thomas
>
Reply | Threaded
Open this post in threaded view
|

Re: RFR(xs): 8176442: [aix] assert(_thr_current == 0L) failed: Thread::current already initialized

Thomas Stüfe-2
May I get a second review please?`

Thank you, Thomas

On Sun, Mar 12, 2017 at 12:34 AM, David Holmes <[hidden email]>
wrote:

> Looks good to me.
>
> David
>
>
> On 11/03/2017 7:02 PM, Thomas Stüfe wrote:
>
>> Please review this small AIX change.
>>
>> Issue: https://bugs.openjdk.java.net/browse/JDK-8176442
>> webrev:
>> http://cr.openjdk.java.net/~stuefe/webrevs/8176442-switch-of
>> f-compiler-level-tls-on-aix/webrev.00/webrev/
>>
>>
>> This change switches the implementation of Thread::current() from compiler
>> level TLS back to pthread library TLS. We see errors on AIX 5.3 machines
>> where compiler level TLS seems not to work correctly - new TLS variable
>> instances are sporadically left uninitialized.
>>
>> For more details, please see the description and comments in the bug
>> report.
>>
>> Thank you, Thomas
>>
>>
Reply | Threaded
Open this post in threaded view
|

Re: RFR(xs): 8176442: [aix] assert(_thr_current == 0L) failed: Thread::current already initialized

David Holmes
Hi Thomas,

As this only affects AIX and is a very simple change I am claiming this
falls under the category of a "trivial" fix and will push it for you. I
re-targeted the bug to 9 but this must get in before RDP2 to avoid
additional approval processes.

David

On 13/03/2017 9:19 PM, Thomas Stüfe wrote:

> May I get a second review please?`
>
> Thank you, Thomas
>
> On Sun, Mar 12, 2017 at 12:34 AM, David Holmes <[hidden email]
> <mailto:[hidden email]>> wrote:
>
>     Looks good to me.
>
>     David
>
>
>     On 11/03/2017 7:02 PM, Thomas Stüfe wrote:
>
>         Please review this small AIX change.
>
>         Issue: https://bugs.openjdk.java.net/browse/JDK-8176442
>         <https://bugs.openjdk.java.net/browse/JDK-8176442>
>         webrev:
>         http://cr.openjdk.java.net/~stuefe/webrevs/8176442-switch-off-compiler-level-tls-on-aix/webrev.00/webrev/
>         <http://cr.openjdk.java.net/~stuefe/webrevs/8176442-switch-off-compiler-level-tls-on-aix/webrev.00/webrev/>
>
>
>         This change switches the implementation of Thread::current()
>         from compiler
>         level TLS back to pthread library TLS. We see errors on AIX 5.3
>         machines
>         where compiler level TLS seems not to work correctly - new TLS
>         variable
>         instances are sporadically left uninitialized.
>
>         For more details, please see the description and comments in the
>         bug report.
>
>         Thank you, Thomas
>
>
Reply | Threaded
Open this post in threaded view
|

Re: RFR(xs): 8176442: [aix] assert(_thr_current == 0L) failed: Thread::current already initialized

Thomas Stüfe-2
Thank you David!

Kind regards, Thomas

On Tue, Mar 14, 2017 at 2:41 AM, David Holmes <[hidden email]>
wrote:

> Hi Thomas,
>
> As this only affects AIX and is a very simple change I am claiming this
> falls under the category of a "trivial" fix and will push it for you. I
> re-targeted the bug to 9 but this must get in before RDP2 to avoid
> additional approval processes.
>
> David
>
> On 13/03/2017 9:19 PM, Thomas Stüfe wrote:
>
>> May I get a second review please?`
>>
>> Thank you, Thomas
>>
>> On Sun, Mar 12, 2017 at 12:34 AM, David Holmes <[hidden email]
>> <mailto:[hidden email]>> wrote:
>>
>>     Looks good to me.
>>
>>     David
>>
>>
>>     On 11/03/2017 7:02 PM, Thomas Stüfe wrote:
>>
>>         Please review this small AIX change.
>>
>>         Issue: https://bugs.openjdk.java.net/browse/JDK-8176442
>>         <https://bugs.openjdk.java.net/browse/JDK-8176442>
>>         webrev:
>>         http://cr.openjdk.java.net/~stuefe/webrevs/8176442-switch-of
>> f-compiler-level-tls-on-aix/webrev.00/webrev/
>>         <http://cr.openjdk.java.net/~stuefe/webrevs/8176442-switch-o
>> ff-compiler-level-tls-on-aix/webrev.00/webrev/>
>>
>>
>>         This change switches the implementation of Thread::current()
>>         from compiler
>>         level TLS back to pthread library TLS. We see errors on AIX 5.3
>>         machines
>>         where compiler level TLS seems not to work correctly - new TLS
>>         variable
>>         instances are sporadically left uninitialized.
>>
>>         For more details, please see the description and comments in the
>>         bug report.
>>
>>         Thank you, Thomas
>>
>>
>>