[Xenomai-core] Re: XENO_OPT_DEBUG impact (was: exception 768)

[Philippe Gerum <rpm@xenomai.org>]

On Fri, 2006-11-17 at 19:41 +0100, Jan Kiszka wrote:
> Philippe Gerum wrote:
> > ...
> > Disable CONFIG_XENO_OPT_DEBUG from your kernel configuration, this is
> > the source of the PTE misses you are seeing now on ppc. This is not to
> > say that those errors are normal, and this issue still remains to be
> > fixed, but unless you want to debug the Xenomai nucleus, you don't need
> > this option on (additionally, it adds a large overhead which translates
> > in significantly augmented jitter).
> 
> CONFIG_XENO_OPT_DEBUG should not add large jitters - that's what e.g.

CONFIG_SMP + CONFIG_XENO_OPT_DEBUG instruments the nucleus lock macros
with consistency and latency tracking code. This does not come for free,
even if an order of magnitude less than checking the queues.

> CONFIG_XENO_OPT_DEBUG_QUEUES is now for.
> 
> I'm currently seeing two potential "misuses" of the common switch:
> 
>  - the posix skin (Gilles, how heavy-weighted are those checks?)
>    => CONFIG_XENO_OPT_DEBUG_POSIX
> 
>  - CONFIG_XENO_SPINLOCK_DEBUG => CONFIG_XENO_OPT_DEBUG_SPINLOCK
> 
> Both should be explicitly controllable in Kconfig.
> 

Nack for CONFIG_XENO_OPT_DEBUG_SPINLOCK. Most of the issue we tracked
with Gilles regarding the domain migration code had side-effects on the
nucleus lock. So having CONFIG_XENO_OPT_DEBUG enabled for identifying
internal state weirdnesses - like those triggered by migration bugs -
implies enabling the spinlock watchdogs too.

> Gilles, is CONFIG_XENO_OPT_DEBUG_BHEAP used in any way? Doesn't seem so.
> 
> Jan
> 
-- 
Philippe.