[En-Nut-Discussion] Pending bugs - was: timer problem

[Harald Kipp]

Ole Reinhardt wrote:
> Please don't hesitate to enther those bugs into the sourceforge bug
> tracker:
> 
> http://sourceforge.net/tracker/?group_id=34079&atid=410687

This can't be stated too often. :-)

> 
>> - NutEnterCritical/NutExitCritical: not sure if solution was implemented
> 
> Indeed that's a point I'm realy interested in too. The discussion of the
> different ways to fix this issue seemd to have fizzled out a little...


Here's the ChangeLog entry

2008-07-07  Harald Kipp

         * arch/arm/os/context.c, conf/os/os.nut, include/arch/arm/atom.h,
         os/thread.c: Three optional ways of handling critical sections are
         now available for ARM targets. For fastest interrupt response,
         critical section nesting should be disabled. This is the default
         and compatible with version 4.4. Alternatively the nesting level
         is tracked by either using the stack (fast, but may crash with
         some compilers and optimization settings) or by using a global
         variable (slow, but safe).

I'm aware, that this is no optimal solution. The urgency of this problem 
required an immediate fix. IMHO, Duane's suggestion of a delayed 
interrupt handling was the most attractive, but its implementation 
doesn't look trivial to me and it may take weeks or months until some 
code shows up.

Harald