[LITMUS^RT] Questions about ft_trace

[Hang Su]

Hi all:

Under /kernel/sched.c, there is function with prototype void
scheduler_tick(void). Inside it, it calls:
 TS_TICK_START(current);
 litmus_tick(rq,curr) ;
 TS_TICK_END(current);

I found that litmus_tick is implemented in /litmus/sched_litmus.c. But
where is TS_TICK_START and TS_TICK_END ?

In addition, where is the definition of TS_PLUGIN_TICK_START and
TS_PLUGIN_TICK_END inside /litmus/sched_litmus.c ?

Thanks.

On Mon, Feb 13, 2012 at 11:32 PM, Hang Su <hangsu.cs at gmail.com> wrote:

> Got you. Thanks.
>
>
> On Tue, Feb 14, 2012 at 12:25 AM, Glenn Elliott <gelliott at cs.unc.edu>wrote:
>
>> CONFIG_NR_CPUS CPUs controls the maximum number of CPUs your compiled
>> Linux kernel can support.  There is no harm in setting this value to be
>> greater than the number of actual CPUs in your system-- it may just lead to
>> minor waste of memory (the Linux kernel statically allocates memory based
>> upon NR_CPUS).
>>
>> -Glenn
>>
>>
>> On Feb 14, 2012, at 12:18 AM, Hang Su wrote:
>>
>> Hi Glenn:
>>
>> Thanks for your quick reply.
>>
>> First I want to confirm the .config you mentioned is supposed to be
>> located under /litmus-rt, in which arch, litmus and other main linux
>> directory reside.  Is that right ?  If so, CONFIG_SCHED_OVERHEAD_TRACE is
>> not set in .config. It means I have to recompile Litmus kernel by marking
>> CONFIG_SCHED_OVERHEAD_TRACE item after issuing make menuconfig.
>>
>> The result of cat /proc/cpuinfo under virtual machine shows only one
>> processor in my computer, which is different from the one under my real
>> machine. Probably, I do not allow virtual machine to use two cpus.
>> CONFIG_NR_CPUS is 8 in my .config. It does not make sense. I do not have
>> so many cpus. What does    it exactly stand for ?
>>
>> Thanks
>>
>> On Mon, Feb 13, 2012 at 11:46 PM, Glenn Elliott <gelliott at cs.unc.edu>wrote:
>>
>>> Hi Hang Su,
>>>
>>> WIth regards to ft_trace, please confirm that
>>> CONFIG_SCHED_OVERHEAD_TRACE=y in your .config.
>>>
>>> With respect to the number of CPUs, is your virtual machine configured
>>> to run with two CPUs?  If so, what CONFIG_NR_CPUS set to in your .config?
>>>
>>> Please note that we recommend using Litmus in a virtual machine only for
>>> development/debug purposes.  We've found that the time accounting
>>> mechanisms in virtual machines are not accurate enough for real-time
>>> performance.
>>>
>>> Glad that you're trying Litmus out!
>>>
>>> -Glenn
>>>
>>>
>>> On Feb 13, 2012, at 11:20 PM, Hang Su wrote:
>>>
>>> > Hi all:
>>> >
>>> > LITMUS is new to me. After I installed LITMUS under Oracle Virtual Box
>>> on my Thinkpad x200, it seems that I can not find ft_trace device under
>>> /dev/litmus/. Only three devices can be seen under /dev/litmus:
>>> > ctrl log sched_trace0
>>> > How could I load a module to create ft_trace device ?
>>> >
>>> > Another question is about sched_trace buffer. As said in Tracing and
>>> Logging tutorial, sched_trace is a per-"processor" buffer. The upbound
>>> number after sched_trace is  given by num_online_cpus(). My CPU is Intel
>>> P8600 duo core processors. It is supposed to be two sched_trace buffer. But
>>> why do i have only one sched_trace buffer under /dev/litmus ?  it is
>>> because i install LITMUS on virtual machine instead of directly on real
>>> machine?
>>> >
>>> > Thanks
>>> > _______________________________________________
>>> > litmus-dev mailing list
>>> > litmus-dev at lists.litmus-rt.org
>>> > https://lists.litmus-rt.org/listinfo/litmus-dev
>>>
>>>
>>> _______________________________________________
>>> litmus-dev mailing list
>>> litmus-dev at lists.litmus-rt.org
>>> https://lists.litmus-rt.org/listinfo/litmus-dev
>>>
>>
>> _______________________________________________
>> litmus-dev mailing list
>> litmus-dev at lists.litmus-rt.org
>> https://lists.litmus-rt.org/listinfo/litmus-dev
>>
>>
>>
>> _______________________________________________
>> litmus-dev mailing list
>> litmus-dev at lists.litmus-rt.org
>> https://lists.litmus-rt.org/listinfo/litmus-dev
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.litmus-rt.org/pipermail/litmus-dev/attachments/20120215/1f6db50f/attachment.html>