"Refresh While Running" in Variables view using S08PA doesn't work.

[carloscuev]

So far I've used CW v10.x with the SE8 for several proyects without mayor issues. Now I decided to jump into the PA/PT bandwagon as I was convinced in a webinar hehe.

The problem here is that I cannot use the Refresh While Running option, it displays incorrect values of variables. So far I've tried with the standard USB Multilink from P&E (USB-ML-12) and USBDM 4.10.4. In the two cases got similar issues. The difference is that Multilink displays random data and occasionally the correct values while USBDM displays zeros most of the time and also occasionally the correct values, I've tried to test using OSBDM from the TWR-S98DC-PT60 board I won on a webinar, but it stopped every 2-3 seconds telling me that user halted or something like that, which of course is not true.

So far I've read http://cache.freescale.com/files/soft_dev_tools/doc/support_info/CW_MCU_10.3_RN.pdf?fsrch=1&sr=1 but couldn't find any related issue.

Any Tips/Advice?

Thanks.

[carloscuev]

I forgot to mention that when hitting the Suspend button (Pause-like icon) the variable values are correct. The problem is only on "Refresh While Running" mode.

Regards.

[BlackNight]

I do not have that microcontroller, so I cannot try it out. But I know that the 'refresh while running' is using the 'free BDM cycles' in the hardware. The BDM (Background Debug Mode) hardware module is using free bus cycles to access the memory and pass it to the debug probe/debugger. That works for me on other S08 devices (like GT60 or GB60) I use. If it does not work for you, that looks to me like a silicon issue. Maybe the chip errata says somethign about this?

[carloscuev]

I read carefully the Release Notes of CW V10.3 and noticed something in the bug fixes for HCS08:

Enable P&E run control device to read 9S08PT memory while the device is

running

I thought that bug wasn't really fixed, to corroborate I tried reading variables using FreeMaster and got same results, random numbers and occasionally a correct value, so yes, it might be a silicon bug.

The microcontroller soldered on the prototype I'm working on has a 4N40C Mask (silicon revision?) and I could only find this:

http://cache.freescale.com/files/32bit/doc/errata/MSE9S08PT60_4N40C.pdf

Which says nothing about memory access while running.

I think this is the down side of working with the latest uc families :smileysad: However, it's not so critical as I'm about to finish the display driver and could use that as a live variable view workaround.