Extra flash when emulating 52810 (to allow debugging with Optimization=none)

[Shockel]

I compiled and debugged the PCA10040e ANCS example (ie emulating the 52810 using the 52832) from SDK 15.3 - successfully. But I wanted to turn off code optimization to facilitate debugging. That seems to bring the solution over the flash size limit of the 52810.

How can I cheat and allow emulating the 52810 which (temporarily, for debugging) taking advantage of the 52832's larger flash ?

So after adding a debug configuration in the .emProject file via

<configuration Name="Debug"
c_preprocessor_definitions="DEBUG; DEBUG_NRF"
gcc_optimization_level="None"/>

I get ".text too large to fit in FLASH memory segment" during linking. So I tried to modify the 

linker_section_placement_macros="FLASH_PH_START=0x0;FLASH_PH_SIZE=0x38000;

and bumped the FLASH_PH_SIZE a little (from 0x30000). It will link now, but the J-Link reports a "Error reported. Failed to download application. Generic error." 

Any ideas ? How could I temporarily use more flash while emulating the 52810 ?

Thanks !

[Vidar Berg]

I think the easiest way might be to just use the pca10040 (without the 'e') configuration with the same source code. That said, I think you were pretty close to achieve this with your pca10040e project. In addition to increasing the FLASH_PH_SIZE you need to change the total flash size in your "linker_section_placements_segments" macro.

The original setting from ble_app_ancs_c_pca10040e_112.emProject file

      linker_section_placements_segments="FLASH RX 0x0 0x30000;RAM RWX 0x20000000 0x6000"

must be changed to:

      linker_section_placements_segments="FLASH RX 0x0 0x38000;RAM RWX 0x20000000 0x6000"

[Shockel]

Vidar, I am still getting the "Error reported. Failed to download application. Generic error." when launching the debugger in SES.

[Vidar Berg]

It worked here, but I had to increase the address to 0x3A000 to be able to build without optimization. 

linker_section_placement_macros="FLASH_PH_START=0x0;

FLASH_PH_SIZE=0x3A000;

RAM_PH_START=0x20000000;

RAM_PH_SIZE=0x6000;

FLASH_START=0x19000;

FLASH_SIZE=0x26000;

RAM_START=0x20001c18;

RAM_SIZE=0x43e8"

linker_section_placements_segments="FLASH RX 0x0 0x3A000;RAM RWX 0x20000000 0x6000"

But the linker settings should not lead to  debugger errors. Can you try to do a chip erase with nrfjprog first and see if you get the same result? 

> nrfjprog --erase 

[Shockel]

Vidar, so first of all: Thank you for the original input to also adjust linker_section_placements_segments. Presumably that is a fix. And it sounds like you were able to flash it.

As to my specific case, it appears that it turns into a JLink / Segger issue ?!

I repeated my setup with a JLink-Lite and a different target board, and same result: I cannot flash the "emulated with extra flash" version to either target board due to the generic error.

I upgraded JLink to the latest 6.52e version and also upgraded SES to 4.22 .

Above, I assume I was to issue a "nrfjprog --eraseall". It did not help. I was able to program the softdevice and the hex file using "nrgprog --program <>.hex", if that tells you something.

* if you don't mind, could I send you my .emProject file and you give it a quick try, just to make sure there is no hidden incorrect setting.

* Otherwise, I presume, this is more a Segger issue ?

[Vidar Berg]

Attached is my modified example. I tested it with SDK 15.3.0. 

4087.ses.zip

[Shockel]

Your optimization level is set to "Level 3", which also works for me. What happens on your end if you set it to "None". It fails here.

[Vidar Berg]

Sorry for the back and forth on this, I really thought I tested it without optimization, but it's not working now. I guess the debugger reports the error because the image exceeds the address range defined for the 52810. However, it still didn't work after I changed the target device in the debug settings. I need to spend some more time on this to figure out what is going on. Can you use the "pca10040" project for debugging in the meantime?

[Shockel]

Vidar, ok. I had found the following interesting, but wouldn't know how to incorporate the results into SES, and whether this makes debugging possible now:

If I run the JLink executable in the install folder, and issue the following commands ...

>erase

>device nrf52810_xxAA

>loadfile <outputfile>.hex

Response: Unspecified error -1

>erase

>device nrf52832_xxAB

>loadfile <outputfile>.hex

Result:No errors.