test: fix stats depth variable scope

[0xc0170]

As this variable is not in scope, should not be referenced.
Fixes #4392 No, you can't do "fixes PR#" only "fixes issue#" makes sense
Fixes #4930

I am not certain this is a proper fix. The variable is not in the scope in execute() . Should it be passed somehow there, or just ignored ? This fixes it as we are not using undefined variable. The question is how it should be passed to execute function or should not and this is correct?
@theotherjimmy update: I passed stats depth correctly.

Tested with singletest.py. With this patch, it builds. Without, build does not even start, it crashes silently - FAIL build.

@MarceloSalazar @theotherjimmy You might know better, replace this patch or push to my branch.

[MarceloSalazar]

The line is needed to be able to get different levels of stats when compiling tests (default is 2).

Should it be passed somehow there, or just ignored ?
The question is how it should be passed to execute function or should not and this is correct?

IMO it should continue to be passed. I'd suggest checking whether the variable exists. If doesn't, then assign a value (e.g. 2 is the default level of depth for memap stats).

@theotherjimmy may have another suggestion.

[0xc0170]

IMO it should continue to be passed. I'd suggest checking whether the variable exists. If doesn't, then assign a value (e.g. 2 is the default level of depth for memap stats).

That is how is it with this patch, it is not passed, thus using the default value.

I am not familiar with the changes for memmap, thus not certain how it can be passed to execute_thread_slice. I had a look how to but could not see the path there.
As I see we got 2 options:

• this patch (using default value), as this has not been yet spotted outside of mbed cli (mbed 2 script broke that does not use memap at all)
• pass that parameter around to make it visible for execute_thread_slice (I'll need your help to make this happen, please either write to this branch or superceed this patch by another PR).

[theotherjimmy]

I don't think that this is the right way to go about this. Instead of allowing stats_depth to take its default, we should find where it should have been passed in and pass it appropriately.

[theotherjimmy]

@theotherjimmy note to self: help with this.

[0xc0170]

@theotherjimmy note to self: help with this.

👍 You can push to my branch or just send a new PR. It would be nice to have this fixed. I'll talk to you today

[0xc0170]

Thanks @theotherjimmy

[theotherjimmy]

I tested with singletest.py, and it seems to work.

[0xc0170]

Tested locally, working for me as well

[jeromecoutant]

👍