PE Code Generator FAILURE

[Vagni]

I developed my K60DN512VLL10-based custom board with CodeWarrior 10.4 and MQX 4.0.2. This board has been properly working for two months.

Now I need more flash memory to make a firmware upgrade on my application, so I decided to mount a K64FN1M0VLL12 cpu on my board.

Porting my application firmware on the new cpu involves to use CodeWarrior 10.5 and to clone MQX libraries (PSP, BSP, ...) from the TWR-K64F120M demo system.

So I installed Code Warrior 10.5 along with the needed updates for 120MHz cpu (com.freescale.mcu10_5.PEX_update.v1_0_0.zip and com.freescale.mcu10_5.Kinetis_K24_K63_K64_120MHz.win.sp.v1.0.2.zip) and the MQX 4.0.2 libraries for K64F120M.

My original BSP library for the K60 cpu was built using Processor Expert LDD components, but the the BSP for TWR-K64F120M does not include a PE project file. So I cloned the TWR-K64F120M BSP library for my custom board, then I added my original PE project file. I modified the PE project file disabling the original K60DN512VLL10 processor component and adding a new K64FN1M0VLL12 processor component. I adjusted all my LDD component in the PE project file to the new cpu and finally I launched the PE Code Generator.

The generation process always ends reporting two errors:

1. Generator: FAILURE: Unexpected status of script: Drivers\Common\Static_SetConst.prg, please contact Freescale support. 
2. Generator: FAILURE: Unexpected status of script: Drivers\Kinetis\Static_KinetisK_Cpu_Config_h.prg, please contact Freescale support. 

Could you please help me to fix that issue?

I attached my PE project file.

Original Attachment has been moved to: ProcessorExpert.pe.zip

[vfilip]

Hello,

I have reproduce it. We will fix it for next release.

In mean time you can use the following workaround:

Set Initialize fast trim value to no in CPU component. and the following rows into Events.h

#define STARTUP_CLOCK_INTERNAL_FAST_TRIM_ADDRESS	0x03FEU   /* Address of non-volatile memory containing fast oscillator trim value */
#define STARTUP_CLOCK_INTERNAL_FAST_FINE_TRIM_ADDRESS	0x03FEU   /* Address of non-volatile memory containing fast oscillator fine trim value */

We are sorry for inconvenience.

Best regards

Vojtech Filip

Processor Expert Support Team

[Vagni]

Dear Vojtech Filip,

I used your workaround and got PE code generated.

But now I have some errors compiling my BSP source files:

1. The Vectors_Config.h include file is not found. This issue causes other compiler errors in the Vectors.c source file.
2. MCG_C5_PLLCLKEN0_MASK macro results undefined in the CPU_Init.c source file. But that macro is defined in MK64F12.h, which is included in CPU_Init.c.

Also, I see the PE code generation process creates four empty folders in the Generated_Code folder (IO_Map, PDD, Peripherals and System).

I attach my BSP - PSP project and source file.

Could you please help me to get a successful compilation?

Thank you.

[vfilip]

Hello,

as workaround is necessary to mark the Vectors.c file as Excluded from build.

Fix of this issue will be published within CW V10.6

Best regards

Vojtech Filip

Processor Expert Support Team

[ironsean]

Hello Vojtech,

I have experienced the same issue with Processor Expert on a FRDM K64F Board and CW 10.5. After attempting to generate code with PE I got the issue that Vectors_config.h was not found. I tried your workaround of marking Vectors.c to not be built, but now my Processor Expert CPU_Init.c code can not compile when I try to compile my project because it uses one of the vector data types from Vectors.c. Now I can no longer compile my application with "undefined reference to '__vect_table' errors.

[Vagni]

Dear Vojtech Filip,

Excluding Vector.c from build fixes some compiler's errors, but not all.

I think compiler's errors are due to the file MK64F12.h. There are two different versions of this file: one is in the <MQX_4_0>\mqx\source\psp\cortex_m\cpu folder and the other is in <CW_MCU_10_5>\MCU\ProcessorExpert\lib\Kinetis\iofiles folder.

All the MQX BSP code refers to the first one, while all the PEx generated code refers to the second one. So I cannot get my BSP code all successfully compiled.

How can I fix this issue?

[MichalP]

Dear Alessandro,

please use the same IOmap file, i.e. update the MK64F12.h in <MQX_4_0>\mqx\source\psp\cortex_m\cpu folder by the newer one that comes with the PEX update and then rebuild all MQX libs. As there could be some inconsistencies between old and new IOmap macros/strucs, please solve MQX build issues case by case when rebuilding MQX libs (renaming of some macros that addresses correct names of K64F peripheral registers like MCG_C5_PLLCLKEN_MASK -> MCG_C5_PLLCLKEN0_MASK).

FYI, there will be a new MQX 4.1.0 installer released soon (in a couple of days) as well as the new version of the MQX 4.1.0 standalone installer for the TWR-K64F120M which will come with the PEX support.

Best Regards

Michal Princ

[williamely]

I just got my K64 tower kit this morning so should I wait for MQX 4.1?

If it will be out next week then I think that would be best.

Any idea when CW 10.6 will be out?

[MichalP]

MQX 4.1.0 is already available on the FSL web. This release however does not contain support for the K64F devices. There will be the another standalone release for TWR-K64F120M based on MQX 4.1.0 code available next week (PEX support included).
As for the CW10.6 I do not know the release date.

Best Regards

MichalP

[williamely]

Hey Michal, any update on the K64 release date?

[MichalP]

Hello William,

as far as I know the standalone 4.1.0 installer for TWR-K64F120M is ready for placing to the FSL web. You can expect to be there in couple of days, probably by the end of this week.

MichalP

[williamely]

Just got it. Thanks.

[williamely]

Thanks for the update Michal. I saw that 4.1 was released just after posting. I will work on something else until the K64 version is ready.

[BlackNight]

In case this is helpful:

there is an example project (attached) with Eclipse/Keppler for the K64F with FreeRTOS, see First Steps with the Freescale TWR-K64F120M | MCU on Eclipse

[williamely]

Thanks Erich. Awesome site. I've been toying with the idea of going down the Eclipse route since I read your blog about it a month or so back. Maybe I'll give it a shot next weekend.