How to resolve a CDTDebugModelPresentation.12=signal

Technical questions regarding the xTIMEcomposer, xSOFTip Explorer and Programming with XMOS.
User avatar
aclassifier
Respected Member
Posts: 319
Joined: Wed Apr 25, 2012 8:52 pm
Contact:

How to resolve a CDTDebugModelPresentation.12=signal

Postby aclassifier » Thu Apr 05, 2018 10:59 am

I got this crash while running my code, at the same place in my debug printouts:

Code: Select all

__arduino_rfm69_on_xmos_native Default [xCORE Application]   
   xCORE Debugger (15:53 04.04.18)   
      tile[0] core[0]  (Running)CDTDebugModelPresentation.12=signal   
   xgdb (15:53 04.04.18)   
   /Users/teig/workspace/__arduino_rfm69_on_xmos_native/bin/__arduino_rfm69_on_xmos_native.xe (15:53 04.04.18)

My problem was that I was not able to find where this happened. No double-clicking got me to the source and if I stopped running with the II-button it only showed me a different process' end of while-true select block.

I have good reasons to believe that I had an array index overflow. When that was removed the code went on and I haven't seen the CDTDebugModelPresentation any more. (Since the compiler didn't err on that overflow I added my own run-time check and a fail statement with "xassert.h" and lib_xassert, and that certainly triggers the debugger so that I can double-click and it goes to the offending line).

Any help on this? What exactly is the CDTDebugModelPresentation (xTIMEcomposer 14.3.2)
User avatar
larry
Respected Member
Posts: 275
Joined: Fri Mar 12, 2010 6:03 pm

Postby larry » Fri Apr 06, 2018 10:41 am

Does the same binary run fine in xgdb alone, outside the xTIMEcomposer IDE?
User avatar
aclassifier
Respected Member
Posts: 319
Joined: Wed Apr 25, 2012 8:52 pm
Contact:

Postby aclassifier » Fri Apr 06, 2018 12:48 pm

I don't know. But I have a commit on the SW so I could always try. or just change the lines. But I do run xTIMEcomposer on Mac OS 10.11.6 El Capitan, and I haven't seen other problems on it, since I downgraded from macOS til OS X in October 2017. Now I run that old Mac OS X just because of xTIMEcomposer [1]. It this why you asked?

[1] http://www.teigfam.net/oyvind/home/technology/141-xc-is-c-plus-x/#running_xtimecomposer_functions_from_command_line
User avatar
larry
Respected Member
Posts: 275
Joined: Fri Mar 12, 2010 6:03 pm

Postby larry » Fri Apr 06, 2018 12:51 pm

I ask to establish if this could be an xgdb issue or an xTIMEcomposer (Eclipse) issue
User avatar
aclassifier
Respected Member
Posts: 319
Joined: Wed Apr 25, 2012 8:52 pm
Contact:

Postby aclassifier » Fri Apr 06, 2018 12:59 pm

With respect to the message itself "(Running)CDTDebugModelPresentation.12=signal" or with respect to (failing to ) double-clicking to get to the offending source code line? I assume the latter.
User avatar
larry
Respected Member
Posts: 275
Joined: Fri Mar 12, 2010 6:03 pm

Postby larry » Fri Apr 06, 2018 1:01 pm

The CDTDebugModelPresentation message, I guess. It looks like the double clicking happens later and might be a consequence of the CDTDebugModelPresentation message.
User avatar
aclassifier
Respected Member
Posts: 319
Joined: Wed Apr 25, 2012 8:52 pm
Contact:

Postby aclassifier » Fri Apr 06, 2018 1:54 pm

I'll be back in a day or three
User avatar
aclassifier
Respected Member
Posts: 319
Joined: Wed Apr 25, 2012 8:52 pm
Contact:

Postby aclassifier » Fri May 25, 2018 3:00 pm

Quoting myself (below): I had turned all printing off! The code was spinning as livelocked. But I still don't see why that should have caused any stops like the CDTDebugModelPresentation, that spinning was quite ok. So now I do have code where I with a print on and off get this problem. It's the code segment posted Error reporting on erroneous ":> void" in select component with DEBUG_PRINT_TEST 0 causing CDTDebugModelPresentation. Now, only as background info:
Update (after much more than three days..) I am still having problems with this. The last time there was an array index overflow that had it go. This time (it's been quiet since last time) I have no idea. I just did something very little. But I tested on startKIT and on eXplorerKIT and there was one CDTDebugModelPresentation on the startKIT but two on eXplorerKIT, even if the latter only seemed to have code on one tile. And there is no double-clicking that would take me to the offending code.

Actually it's worse than me doing something little. When I go back to the original source (a small example where I kick old and new code segments in with #ifdefs) the CDTDebugModelPresentation is haunting me. Even a printf in the first line in a task isn't seen any more.

Using 14.3.3

Update after this, again. I actually get CDTDebugModelPresentation also when it's running and printing correctly. I should probably close this case and go and take a drink.

Who is online

Users browsing this forum: bear118 and 127 guests