How to set optimisation level in SES 5.34a (Nordic version)

[skajam66]

Hi,

As in the subject, I cannot find where to change the optimisation level.  I am working with the Zephyr bluetooth sample project - peripheral_hr.

I have found a previous post on this topic (Case ID 213802) that says to right click on the project->Options-Code Generation->Optimisation Level.

In my setup, the only project where the Options menu item is not greyed out is this one:

and the displayed window does not have a Code Generation and Optimisation selection.

So how do I set the optimisation?

Regards,

AC

[Amanda Hsieh]

Hi Andrew, 

In NCS, You can edit prj.conf and add "NO_OPTIMIZATIONS=Y" to set compiler optimizations as -O0 or "DEBUG_OPTIMIZATIONS=Y" to set the compiler optimizations as -Og. Please see <choice: Optimization level>.

-Amanda H.

[skajam66]

Thanks for this.  I didn't say but I'm using the nrf5340 which has two core. Changing the settings in prj.conf seems to impact optimisation for both the network and the application cores.  I don't want to change the optimisation for the network core (the bluetooth stack) - only the application core.  Where do I adjust the settings for just the app core?

Regards,

AC

[Amanda Hsieh]

If you change the optimization configuration in the samples prj.conf, then that will not affect the network core.

-Amanda H.

[skajam66]

OK. So I've changed the optimisation in prj.conf to (and rebuilt the system):

CONFIG_NO_OPTIMIZATIONS=y

This change causes this crash:

Next, I changed the main stack size up in increments of 256 bytes. At 1536 bytes it still crashes:

Increasing it to 2048 bytes cases a hard fault:

When the stack size is set to 1536 bytes, the map file (.../build/zephyr/zephyr.map) shows:

 - a logging stack of 1024
 - an interrupt stack of 2048
 - a main stack of 1536
 - a work queue stack of 1024

FYI - The faulting instruction in the Usage Fault images (0x0002b202) is somewhere in a kernel mutex.

Any recommendations on how to turn optimisation off and still get the app to run?

REgards,

ac

[Amanda Hsieh]

Hi AC, 

Sorry for the dealy. 

Could you try the following config in the prj.conf?

CONFIG_NO_OPTIMIZATIONS=y
CONFIG_MAIN_STACK_SIZE=4096
CONFIG_BT_RX_STACK_SIZE=4096

-Amanda H.

[skajam66]

Hi Amanda,

Many thanks for your reply. Apologies for the slow response but I had to step away for a few days.

I have tried your suggestion but, unfortunately, there is no change in behaviour.  I updated prj.conf as you suggested:

CONFIG_NO_OPTIMIZATIONS=y
CONFIG_MAIN_STACK_SIZE=4096
CONFIG_BT_RX_STACK_SIZE=4096

Then I quit SES, deleted the build folder to force a project rebuild then re-built the project for debug and ran it on the DK.  As you can see from the screenshot, I'm still getting a hard fault after a few seconds of running.

There isn't much useful on the stack trace:

Kind regards,

AC

[Amanda Hsieh]

Hi AC, 

What NCS version are you using?

Which does 0x20008820 this address match with the one in the build-folder/zephyr/zephyr.map?

-Amanda H.

[skajam66]

Hi Amanda,

NCS is v1.4.2

The address 0x20008820 is in the zephyr logging thread

Regards,

AC

[Amanda Hsieh]

Hi AC, 

Try adjusting "CONFIG_LOG_PROCESS_THREAD_STACK_SIZE".

-Amanda H.

[skajam66]

Hi Amanda,

CONFIG_LOG_PROCESS_THREAD_STACK_SIZE=4096 works.

Thank you for your support but where are all these configuration options documented? I could figure this out myself if I could find the docs...

Kind regards,

AC