[ECOS] How to Disable FIQ on AT91 (eg EB40A)

[ECOS] How to Disable FIQ on AT91 (eg EB40A)

[Keith Ross (E-Lands)]

Hi,

I have noise on my unused FIQ pin and no handler, so my stack keep full, 
etc when I run my app but ok when single step. The question is how do I 
disable FIQ or what do I do to intall a safe handler.

Keith

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

RE: [ECOS] How to Disable FIQ on AT91 (eg EB40A)

[Paul D. DeRocco]

> From: Keith Ross (E-Lands)
>
> I have noise on my unused FIQ pin and no handler, so my stack keep full,
> etc when I run my app but ok when single step. The question is how do I
> disable FIQ or what do I do to intall a safe handler.

Set bit 6 in the CPSR. (This cannot be done in user mode.)

I don't understand why you'd get stack overflow, though. When an FIQ is
handled, it automatically masks further FIQ interrupts until the handler
returns, so the interrupts can't nest unless you explicitly reenable FIQ
inside the handler.

In the long run, you should really tie off the FIQ pin. Floating CMOS inputs
are not a Good Thing.

--

Ciao,               Paul D. DeRocco
Paul                mailto:pderocco@ix.netcom.com


-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss

[ECOS] "!!!-- Configuration broken - no scheduler selected --!!!"

[Erik Christiansen]

That straighforward error message occurs several times, deep in my ecos.ecc,
but is CYGSEM_KERNEL_SCHED_MLQUEUE not enabled here?:

cdl_component CYGSEM_KERNEL_SCHED_MLQUEUE {
    # Flavor: bool
    # No user value, uncomment the following line to provide one.
    # user_value 1
    # value_source default 
    # Default value: 1

    # The following properties are affected by this value
    # option CYGPRI_KERNEL_SCHED_IMPL_HXX
    #     Calculated:   CYGSEM_KERNEL_SCHED_BITMAP  ? "<cyg/kernel/bitmap.hxx>"  : CYGSEM_KERNEL_SCHED_MLQUEUE ? "<cyg/kernel/mlqueue.hxx>" : CYGSEM_KERNEL_SCHED_LOTTERY ? "<cyg/kernel/lottery.hxx>" : "!!!-- Configuration broken - no scheduler selected --!!!"
    # component CYGSEM_KERNEL_SYNCH_MUTEX_PRIORITY_INVERSION_PROTOCOL
    #     ActiveIf:  CYGSEM_KERNEL_SCHED_MLQUEUE
    # option CYGIMP_UITRON_STRICT_CONFORMANCE
    #     Requires: CYGSEM_KERNEL_SCHED_MLQUEUE
};

cdl_option CYGSEM_KERNEL_SCHED_BITMAP {
    # Flavor: bool
    # No user value, uncomment the following line to provide one.
    # user_value 0
    # value_source default
    # Default value: 0
    # Requires: !CYGPKG_KERNEL_SMP_SUPPORT
    #     CYGPKG_KERNEL_SMP_SUPPORT == 0
    #   --> 1

    # The following properties are affected by this value
    # option CYGPRI_KERNEL_SCHED_IMPL_HXX
    #     Calculated:   CYGSEM_KERNEL_SCHED_BITMAP  ? "<cyg/kernel/bitmap.hxx>"  : CYGSEM_KERNEL_SCHED_MLQUEUE ? "<cyg/kernel/mlqueue.hxx>" : CYGSEM_KERNEL_SCHED_LOTTERY ? "<cyg/kernel/lottery.hxx>" : "!!!-- Configuration broken - no scheduler selected --!!!"
    # component CYGSEM_KERNEL_SCHED_TIMESLICE
    #     Requires: !CYGSEM_KERNEL_SCHED_BITMAP
};

Since ecosconfig thinks that the configuration is OK;

$ ecosconfig check
Adder
Template: kernel
Added:
 CYGPKG_UITRON
Removed:
 CYGPKG_IO_SERIAL
 CYGPKG_HAL_QUICC
 CYGPKG_IO_SERIAL_POWERPC_QUICC_SMC
No conflicts

and the code built from this does schedule threads, I'm led to infer
that CYGSEM_KERNEL_SCHED_MLQUEUE is active by default. But then the
configuration is not broken!?

No benefit is obtained from trying:
   a) Uncommenting the "# user_value 1" in CYGSEM_KERNEL_SCHED_MLQUEUE

   b) ecosconfig add CYGSEM_KERNEL_SCHED_MLQUEUE

   c) ecosconfig remove CYGPKG_UITRON

Is there some other measure that can make this situation less scary?

Erik

-- 
Before posting, please read the FAQ: http://ecos.sourceware.org/fom/ecos
and search the list archive: http://ecos.sourceware.org/ml/ecos-discuss