topic Re: Watchdog (Reset) on M5223X / M5213 in ColdFire/68K Microcontrollers and Processors

Watchdog (Reset) on M5223X / M5213

mjbcswitzerland — Tue, 31 Oct 2006 08:09:41 GMT

Hi All

I would like to know a little more about the software watchdog in the above processors - this is taking on from the previous (but now quite old) posting http://forums.freescale.com/freescale/board/message?board.id=CFCOMM&message.id=776&query.id=13063#M776

The present understanding is that it is not possible to program the core software watchdog in these devices to cause a reset in order to recover from a problem (software runaway). It is necessary to activate the corresponding interrupt routine and to command a soft reset from it.

I have just programmed this and also set the priority of this interrupt as the highest in the system.

There remains one question. What happens when a problem exists in a section of code which disables all interrupts? Wouldn't it mean that the core sw watchdog timer will timeout but its interrupt never handled? Is this not quite a risk??

Any ideas?

Regards

Mark Butcher
www.mjbc.ch / www.uTasker.com

Re: Watchdog (Reset) on M5223X / M5213

Nouchi — Tue, 31 Oct 2006 16:09:26 GMT

Hi,

Unfortunatly, you can only do a soft Reset (MCF_CIM_RCR_SOFTRST), and I if you program is blocked in a critical section, the cpu core will never be reinitialized and the program is definitely deadlocked, so you have to avoid loops in critical regions.

Emmanuel,

Re: Watchdog (Reset) on M5223X / M5213

jvantol — Thu, 07 Dec 2006 06:10:26 GMT

If you have the option, an external watchdog, or watchdog/voltage regulator IC can be a great solution to this problem. Depending on your requirements for the power supply, this can be close to cost neutral.

Re: Watchdog (Reset) on M5223X / M5213

Petter_fupp — Fri, 29 Jun 2007 15:47:11 GMT

Why not use level 7 for the watchdog interrupt? It should be non-maskable.

Re: Watchdog (Reset) on M5223X / M5213

mjbcswitzerland — Fri, 29 Jun 2007 16:24:53 GMT

Hi Petter

You are correct, the Interrupt level 7 assigned to an interrupt gives it NMI qualities and so this is a solution which has been adopted in the meantime.

There is still a small risk involved since if the interrupt vector is in SRAM and this gets corrupted by a runaway program the NMI interrupt will fire but fail itself - so there is still no 100% secure method of recovering from a fatal program error through a reset. i.e. if the runaway program writes something to the watchdog interrupt vector it will result in the board hanging forever with no automatic reset recovery (apart from when an external supervisor is utilised...)

Regards

Mark

www.uTasker.com