[Xenomai-help] xnpod_schedule 2.4.6.1

[Xenomai-help] xnpod_schedule 2.4.6.1

[Steven Seeger]

[-- Attachment #1: Type: text/plain, Size: 531 bytes --]

Forgive me if this has been asked before.

 

In xnpod_schedule I notice there is a call to
xnlock_get_irqsave(&nklock, s);

 

Further down, this call is made again followed by
xnlock_put_irqrestore(&nklock, s);

 

The comments say that the lock is not held, but from what I can tell it
previously was. I was just pointing out this observation in case it was
a bug. It looks like this function has been changed a lot in trunk, so
perhaps anything relating to this was fixed since 2.4.6.1.

 

Steven

 


[-- Attachment #2: Type: text/html, Size: 2624 bytes --]

Re: [Xenomai-help] xnpod_schedule 2.4.6.1

[Philippe Gerum]

Steven Seeger wrote:
> Forgive me if this has been asked before.
> 
>  
> 
> In xnpod_schedule I notice there is a call to
> xnlock_get_irqsave(&nklock, s);
> 
>  
> 
> Further down, this call is made again followed by
> xnlock_put_irqrestore(&nklock, s);
> 
>  
> 
> The comments say that the lock is not held, but from what I can tell it
> previously was. I was just pointing out this observation in case it was
> a bug. It looks like this function has been changed a lot in trunk, so
> perhaps anything relating to this was fixed since 2.4.6.1.
>

There is a Xenomai to Linux context switch in between, and the nklock was indeed
released in that case. This code is correct.

>  
> 
> Steven
> 
>  
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Xenomai-help mailing list
> Xenomai-help@domain.hid
> https://mail.gna.org/listinfo/xenomai-help


-- 
Philippe.

Re: [Xenomai-help] xnpod_schedule 2.4.6.1

[Steven Seeger]

> There is a Xenomai to Linux context switch in between, and the nklock
> was indeed
> released in that case. This code is correct.

Thanks for letting me know. I am inspecting things and trying to
understand what happens in Xenomai so I can help debug the FPU issue we
are still having. I noticed on the RTAI mailing list there was another
user with my board who had the same problem with LXRT. It looks like
geodes do something weird.

Steven