[LITMUS^RT] Control Page

[Björn Brandenburg]

On 27 Jun 2014, at 04:47, Aaron Block <ablock at austincollege.edu> wrote:

> Never mind. I think I found it in my own code (Still need to compile and run, but I’m pretty certain this is the bug). Thanks for the help :)
> 
> 
> Aaron Block, Ph.D.
> Assistant Professor of Computer Science
>  
> Austin College | Math and Computer Science
> 900 North Grand Avenue, Suite 61592 | Sherman, Texas 75090
> Phone 903.813.2563 
> austincollege.edu
> 
> <image001.gif>
> 
> On Jun 26, 2014, at 5:44 PM, Aaron Block <ablock at austincollege.edu> wrote:
> 
>> Hi Glenn, 
>> 
>> it’s an x86. 
>> 
>> Thanks
>> —Aaron
>> 
>> 
>> On Jun 26, 2014, at 4:07 PM, Glenn Elliott <gelliott at cs.unc.edu> wrote:
>> 
>>> 
>>> On Jun 26, 2014, at 5:04 PM, Aaron Block <ablock at austincollege.edu> wrote:
>>> 
>>>> Howdy all, 
>>>> 
>>>> I’ve added a data field to the control page structure and it looks like its getting reset to zero every time a job has released. 
>>>> 
>>>> I haven’t dug into this behavior too much, but I thought I’d ask here before I spend some time trolling through source code to see if this is expected behavior or if anybody has any thoughts about how I can prevent this. 
>>>> 
>>>> Thanks,
>>>> 
>>>> Aaron Block
>>>> Assistant Professor of Computer Science
>>>>  
>>>> Austin College | Math and Computer Science
>>>> 900 North Grand Avenue, Suite 61592 | Sherman, Texas 75090
>>>> Phone 903.813.2563 
>>>> austincollege.edu
>>> 
>>> Hi Aaron,
>>> 
>>> Is this on x86 or ARM?  I believe Björn’s team ran into some cache-related issues on ARM that was affecting the control page.
>>> 
>>> -Glenn


I'm glad to hear you found the issue. Just for the record, there are no known problems with the control page on either x86 or ARM at the moment. The ARM-specific caching issues were resolved by Roy Spliet a couple of months ago; all relevant patches have been merged (& squashed) in master.

Thanks,
Björn