FRDM-K64F flashing issue

[jjp]

Hello *,

I try to port a project on frdm-k64f (700-28163 REVA/SCH-28163 REVC) and I am facing flashing issue.

I am using :

  - a modified version of linker script (MK64FN1M0xxx12_flash.ld) with added sections

  - a modified version of assembler file (startup_MK64F12.S) that just disable IRQ and jump to C code

(Flash configuration is not modified)

The original files come from installation of KSDK 1.3.0 and I use GNU toolchain on a GNU/Linux

host.

The sections .interrupts, .flash_config and .text has same addresses of ELF file generated

from hello_world example located in $KSDK/examples/frdmk64f/demo_apps/hello_world/kds/debug.

When I try to flash binary with openOCD coming from KDS (3.2.0-1), I lose control from

debugging interface and get following messages from openOCD

8<---

[..]

Error: Failed to read memory at 0xe000ed00

Examination failed, GDB will be halted. Polling again in 100ms

Polling target k64.cpu failed, trying to reexamine

Info : SWD DPIDR 0x2ba01477

Info : SWD DPIDR 0x2ba01477

Error: Failed to read memory at 0xe000ed00

Examination failed, GDB will be halted. Polling again in 300ms

Polling target k64.cpu failed, trying to reexamine

Info : SWD DPIDR 0x2ba01477

Info : SWD DPIDR 0x2ba01477

Error: Failed to read memory at 0xe000ed00

Examination failed, GDB will be halted. Polling again in 700ms

Polling target k64.cpu failed, trying to reexamine

--->8

Debugging LED stays in RED and I can do nothing with target...

Some topics from NXP forum look like my issue :

  - https://community.nxp.com/thread/375103#comment-566948

  - https://community.nxp.com/message/443763?

I try them but they don't solve my issue.

I can only "unbrick" the board with *upstream* openOCD (0.10.0-dev-00321-gd4b7cbf (2016-05-27-16:48))

with command *kinetis mdm mass_erase* from telnet interface.

So .flash_config with values

8<---

Hex dump of section '.flash_config':

  0x00000400 ffffffff ffffffff ffffffff feffffff ................

--->8

OR

8<---

Hex dump of section '.flash_config':

  0x00000400 ffffffff ffffffff ffffffff fefdffff ................

--->8

don't change anything..

In my use case it is not possible to debug *one* line of loaded binary..

I implement NMI exception handler to see if i can help but nothing change..

Moreover I can observe when openOCD lose control :

8<---

Debug: 101105 45154 command.c:143 script_debug(): command - ocd_k64.cpu ocd_k64.cpu invoke-event reset-end

Debug: 101106 45154 cmsis_dap_usb.c:522 cmsis_dap_swd_run_queue(): Executing 3 queued transactions

Debug: 101107 45154 cmsis_dap_usb.c:545 cmsis_dap_swd_run_queue(): DP write reg 8 10

Debug: 101108 45154 cmsis_dap_usb.c:545 cmsis_dap_swd_run_queue(): AP read reg 0 0

Debug: 101109 45154 cmsis_dap_usb.c:545 cmsis_dap_swd_run_queue(): DP read reg c 0

Debug: 101110 45156 cmsis_dap_usb.c:583 cmsis_dap_swd_run_queue(): SWD ack not OK: 2 FAULT

--->8

Does anyone face this issue?

Did I miss something from vectors initialization, linker script or compiler option?

[ZhangJennie]

Hi,

first we need to know the problem is on open OCD interface or project. if you create a new project with wizard, can you download this project vis OpenOCD to target?

Can you download the default project C:\Freescale\KSDK_1.3.0\examples\frdmk64f\demo_apps\hello_world\kds to board without problem?

if the problem is only on your specific project, please send your project to me, I will check it on my side.

Have a great day,
Jennie Zhang

-----------------------------------------------------------------------------------------------------------------------
Note: If this post answers your question, please click the Correct Answer button. Thank you!
-----------------------------------------------------------------------------------------------------------------------

[jjp]

Hello,

Thanks for your quick answer (:

Yes I can download the binary hello_world_frdmk64f.elf just after mass erasing board with openOCD and debug it

with no issue.

The project is confidential and I can only send you binary (ELF) in a private channel..

It will be possible to explain problem solution in this forum without publishing binary and I hope we can catch this issue..

We already exchange with NXP through *encompass extranet*.

Is it an appropriate channel to send you this binary?

Let me know if there is a better solution.

[jjp]

Hello *,

Finally find the problem.

It concerns semihosting option of OpenOCD in gdb script file.

So add :

monitor arm semihosting enable

solve the issue.

Have a nice day.

[ZhangJennie]

thanks for sharing the solutions.

so I may understand your code includes semihosting functions, which need enable semihosting in debugger to support it. Correct?