A *working* demo of Monitor Mode Debug (MMD) on SDK 15, please?

[awneil]

The zipped project attached to this blog doesn't say what SDK version it should work with - but it certainly doesn't work with v15.

https://devzone.nordicsemi.com/b/blog/posts/segger-embedded-studio-part-2-monitor-mode-debuggi

Having fixed the Include Paths (whose stupid idea was it to make them absolute?!), I can get the source to compile with SDK v15 - but the link fails:

cannot move location counter backwards (from fffffffffffffc00 to 0000000000000000)

The point at which this is flagged is:

__stack_load_start__ = __SRAM_segment_end__ - 1024;
.stack __SRAM_segment_end__ - 1024 (NOLOAD) : AT(__SRAM_segment_end__ - 1024)
{
__stack_start__ = .;
*(.stack .stack.*)
. = ALIGN(MAX(__stack_start__ + __STACKSIZE__ , .), 8);
}

What is wrong, and how to I fix it?

[haakonsh]

I've made a new tutorial for SDK14.2, and working on SDK15. 

See https://github.com/NordicPlayground/j-link-monitoring-mode-debugging and the SDK14.2 branch.

[awneil]

Thanks.

Could you clarify the question (above) about SetMonModeVTableAddr ?

(to be pedantic, note that it's usually called "monitor mode" - rather than "monitoring mode")

EDIT

The SES tutorial also doesn't mention the  mon exec SetMonModeDebug=1 command

[haakonsh]

I appreciate the input, I've raised a couple of 'Issues' in the new repo and will get to it eventually. 
Don't hesitate to ask more :)

[awneil]

haakonsh said:

In order to execute the Monitor Mode ISR the J-link driver needs to know the vector table. 

Yes, that makes sense.

So the question is: why was this command not in the original article?

https://devzone.nordicsemi.com/b/blog/posts/segger-embedded-studio-part-2-monitor-mode-debuggi 

ie, how does that one get away without it?

[haakonsh]

I believe that example does not use the SoftDevice and the application's vector table will then start at address 0x00000000. If the J-link driver's vector table entry is 0x00000000 by default then it should work as intended. 

[awneil]

Yes, that would be my guess.

But that should be explained in the article - as most Nordic projects are going to use a SoftDevice, so that option will be necessary!

[haakonsh]

It's an "old" article written by an intern from before we had any real support for SES. I guess we should mark it as deprecated and link to the new repo. 

[awneil]

That does seem to be the trouble with relying on blog posts as documentation!

I think there needs to be some way for these things to get embodied into the official documentation - with associated maintenance ...

[haakonsh]

I agree that there is a issue where blogs/tutorials on devzone are not properly maintained. So far we've relied on an "If you publish you maintain it" principle, and although there are cases where this fails, it usually works out fine as long as whoever published the works are still present. Recommendations for improvements on our method of maintainership is more than welcome :)

As to incorporating blogs/tutorials into our official set of supported works, i.e. Datasheets, SDKs, and https://github.com/NordicSemiconductor, this will likely not happen as the bureaucracy/quality required to do so will add so much to the workload of creating a blog/tutorial that there will simply be less of them. 

The demand for new supported features is ridiculously high, and we do not have the resources to support everything as well as we'd like. The devzone blog/tutorials and https://github.com/nordicplayground are our way of generating support for features that often do not have a large enough user base to warrant the extra resources required to incorporate them into our "officially supported" domain. 

[awneil]

OK - thanks for the insight.