boards/frdm: fix flashing from invalid state

[cladmi]

Contribution description

When flashing some applications the flasher sometimes gets stuck which
prevents flashing after.
It may be from a specific firmware or operation but do not have one yet.
Connect with reset asserted fix flashing from this state.

Testing procedure

I do not have an application that cannot flash anymore to really show the fix.
But I have been using a fix like this for several month and had no flashing issue in my test machine.

Flashing boards that use common/frdm so frdm-kw41z and other similar boards, frdm-k22f and frdm-k64f keep flashing correctly. (the

They now use connect_assert_srst when flashing instead of connect_deassert_srst as it was in `master

PR:
srst_only separate srst_nogate srst_open_drain connect_assert_srst

master

srst_only separate srst_nogate srst_open_drain connect_deassert_srst

Both appear with this PR as the deassert is the value in the configuration which is then changed by command line option.

frdm-k64f

BOARD=frdm-k64f RIOT_CI_BUILD=1 make -C examples/hello-world/ flash 
make: Entering directory '/home/harter/work/git/RIOT/examples/hello-world'
Building application "hello-world" for "frdm-k64f" with MCU "kinetis".

   text    data     bss     dec     hex filename
   9160     116    2544   11820    2e2c /home/harter/work/git/RIOT/examples/hello-world/bin/frdm-k64f/hello-world.elf
/home/harter/work/git/RIOT/dist/tools/openocd/openocd.sh flash /home/harter/work/git/RIOT/examples/hello-world/bin/frdm-k64f/hello-world.elf
### Flashing Target ###
/home/harter/work/git/RIOT/examples/hello-world/bin/frdm-k64f/hello-world.elf is not locked.
GNU MCU Eclipse 64-bit Open On-Chip Debugger 0.10.0+dev-00462-gdd1d90111 (2019-01-18-11:37)
Licensed under GNU GPL v2
For bug reports, read
        http://openocd.org/doc/doxygen/bugs.html
Info : auto-selecting first available session transport "swd". To override use 'transport select <transport>'.
Info : add flash_bank kinetis kx.pflash
adapter speed: 1000 kHz
none separate
cortex_m reset_config sysresetreq
srst_only separate srst_nogate srst_open_drain connect_deassert_srst
srst_only separate srst_nogate srst_open_drain connect_assert_srst
Info : CMSIS-DAP: SWD  Supported
Info : CMSIS-DAP: FW Version = 1.0
Info : CMSIS-DAP: Interface Initialised (SWD)
Info : SWCLK/TCK = 0 SWDIO/TMS = 1 TDI = 0 TDO = 0 nTRST = 0 nRESET = 1
Info : Connecting under reset
Info : CMSIS-DAP: Interface ready
Info : clock speed 1000 kHz
Info : SWD DPIDR 0x2ba01477
Info : MDM: Chip is unsecured. Continuing.
Info : kx.cpu: hardware has 6 breakpoints, 4 watchpoints
Info : Listening on port 44403 for gdb connections
    TargetName         Type       Endian TapName            State       
--  ------------------ ---------- ------ ------------------ ------------
 0* kx.cpu             cortex_m   little kx.cpu             reset
Info : MDM: Chip is unsecured. Continuing.
target halted due to debug-request, current mode: Thread 
xPSR: 0x01000000 pc: 0x0000073c msp: 0x1fff0200
auto erase enabled
Info : Kinetis MK64FN1M0xxx12 detected: 2 flash blocks
Info : 2 PFlash banks: 1024k total
wrote 12288 bytes from file /home/harter/work/git/RIOT/examples/hello-world/bin/frdm-k64f/hello-world.elf in 0.895014s (13.408 KiB/s)
Error: timed out while waiting for target halted
target halted due to debug-request, current mode: Thread 
xPSR: 0x01000000 pc: 0x00000bd2 psp: 0x1fff0408
Error: error executing cortex_m crc algorithm
verified 9276 bytes in 20.942808s (0.433 KiB/s)
Info : MDM: Chip is unsecured. Continuing.
shutdown command invoked
Done flashing
make: Leaving directory '/home/harter/work/git/RIOT/examples/hello-world'

frdm-kw41z

BOARD=frdm-kw41z RIOT_CI_BUILD=1 make -C examples/hello-world/ flash
make: Entering directory '/home/harter/work/git/RIOT/examples/hello-world'
Building application "hello-world" for "frdm-kw41z" with MCU "kinetis".

   text    data     bss     dec     hex filename
   9080     116    2544   11740    2ddc /home/harter/work/git/RIOT/examples/hello-world/bin/frdm-kw41z/hello-world.elf
/home/harter/work/git/RIOT/dist/tools/openocd/openocd.sh flash /home/harter/work/git/RIOT/examples/hello-world/bin/frdm-kw41z/hello-world.elf
### Flashing Target ###
/home/harter/work/git/RIOT/examples/hello-world/bin/frdm-kw41z/hello-world.elf is not locked.
GNU MCU Eclipse 64-bit Open On-Chip Debugger 0.10.0+dev-00462-gdd1d90111 (2019-01-18-11:37)
Licensed under GNU GPL v2
For bug reports, read
        http://openocd.org/doc/doxygen/bugs.html
Info : add flash_bank kinetis klx.pflash
adapter speed: 1000 kHz
none separate
cortex_m reset_config sysresetreq
srst_only separate srst_nogate srst_open_drain connect_deassert_srst
srst_only separate srst_nogate srst_open_drain connect_assert_srst
Info : J-Link OpenSDA 2 compiled May  6 2016 11:04:17
Info : Hardware version: 1.00
Info : VTarget = 3.300 V
Info : clock speed 1000 kHz
Info : SWD DPIDR 0x0bc11477
Info : MDM: Chip is unsecured. Continuing.
Info : klx.cpu: hardware has 2 breakpoints, 2 watchpoints
Info : Listening on port 35329 for gdb connections
    TargetName         Type       Endian TapName            State
--  ------------------ ---------- ------ ------------------ ------------
 0* klx.cpu            cortex_m   little klx.cpu            reset
Info : MDM: Chip is unsecured. Continuing.
target halted due to debug-request, current mode: Thread
xPSR: 0x01000000 pc: 0x00000758 msp: 0x1fff8200
auto erase enabled
Info : Kinetis MKW41Z512xxx4 detected: 2 flash blocks
Info : 2 PFlash banks: 512k total
Info : This device supports Program Longword execution only.
Info : Disabling Kinetis watchdog (initial SIM_COPC 0x0c)
Warn : not enough working area available(requested 2048)
Info : This device supports Program Longword execution only.
Warn : not enough working area available(requested 2048)
wrote 10240 bytes from file /home/harter/work/git/RIOT/examples/hello-world/bin/frdm-kw41z/hello-world.elf in 0.374071s (26.733 KiB/s)
verified 9196 bytes in 0.119052s (75.433 KiB/s)
Info : MDM: Chip is unsecured. Continuing.
shutdown command invoked
Done flashing
make: Leaving directory '/home/harter/work/git/RIOT/examples/hello-world'

Issues/PRs references

• Similar to boards/pba-d-01-kw2x: fix flashing from invalid state #11549
• Using the mechanism introduced in: dist/tools/openocd: add OPENOCD_EXTRA_INIT_RESET use for nucleo-f091rc dist/tools/openocd: add OPENOCD_EXTRA_INIT_RESET use for nucleo-f091rc #11976
• Found while running the test suite on the board in [WIP] Fixes to compile using DOCKER and flash boards without local toolchain #10870

[fjmolinas]

I do not see any reason why the added command could affect flashing or resetting the device, At worst it wouldn't change the base behavior. I do not have the hardware(frdm-kw41z, frdm-k64f) to test today but I will tomorrow, I will trust @cladmi restults with frdm-k22f.

I do not have an application that cannot flash anymore to really show the fix.

There was one before right? Maybe you can provide a patch that reverts the change that fixed that app? If I remember correctly it was a driver test application, but I don't remember which one... If it is not reproducible I will ACK once I test it isn't breaking anything (The only reason I do not ACK right away is that in #12158 there were some issues).

[fjmolinas]

Also is this an alternative to #11549? Does this fix the issue you have in ##11549 as well?

[fjmolinas]

Also is this an alternative to #11549? Does this fix the issue you have in ##11549 as well?

I looked again and #11549 and saw no new changes have been added since the last time I tested, so this is not an alternative.

[benemorius]

You can put a kinetis board into an unflashable state by using #11789 and calling pm_set(0). This powers off the SWD hardware until reset is asserted or something else wakes the cpu.

[cladmi]

@fjmolinas no the pba-d-01-kw2x uses a different openocd configuration even if it is a kinetis board.

@benemorius oh nice I will try it. Does this PR fixes flashing for you? and which board are you using.

[benemorius]

Does this PR fixes flashing for you? and which board are you using.

I can confirm that definitely I do need connect_assert_srst or else I can't flash a kw41z-mini while it's in the lowest power mode. That's the reason for fc75cd1 in #11789.

But I don't have a board that can flash using make flash so I can't do a proper test of the PR.

[fjmolinas]

Tested on frdm-k64f using #11789, the following output are after I flashed an application that has pm_set(0) right at the start.

OPENOCD_RESET_USE_CONNECT_ASSERT_SRST=0 make -C tests/periph_pm BOARD=frdm-k64f flash -j3

make: Entering directory '/home/francisco/workspace/RIOT/tests/periph_pm'
Building application "tests_periph_pm" for "frdm-k64f" with MCU "kinetis".

"make" -C /home/francisco/workspace/RIOT/boards/frdm-k64f
"make" -C /home/francisco/workspace/RIOT/core
"make" -C /home/francisco/workspace/RIOT/cpu/kinetis
"make" -C /home/francisco/workspace/RIOT/drivers
"make" -C /home/francisco/workspace/RIOT/cpu/cortexm_common
"make" -C /home/francisco/workspace/RIOT/drivers/periph_common
"make" -C /home/francisco/workspace/RIOT/cpu/cortexm_common/periph
"make" -C /home/francisco/workspace/RIOT/cpu/kinetis/periph
"make" -C /home/francisco/workspace/RIOT/sys
"make" -C /home/francisco/workspace/RIOT/sys/auto_init
"make" -C /home/francisco/workspace/RIOT/sys/isrpipe
"make" -C /home/francisco/workspace/RIOT/sys/newlib_syscalls_default
"make" -C /home/francisco/workspace/RIOT/sys/pm_layered
"make" -C /home/francisco/workspace/RIOT/sys/shell
"make" -C /home/francisco/workspace/RIOT/sys/stdio_uart
"make" -C /home/francisco/workspace/RIOT/sys/tsrb
   text	   data	    bss	    dec	    hex	filename
  20732	    240	   2820	  23792	   5cf0	/home/francisco/workspace/RIOT/tests/periph_pm/bin/frdm-k64f/tests_periph_pm.elf
/home/francisco/workspace/RIOT/dist/tools/openocd/openocd.sh flash /home/francisco/workspace/RIOT/tests/periph_pm/bin/frdm-k64f/tests_periph_pm.elf
### Flashing Target ###
/home/francisco/workspace/RIOT/tests/periph_pm/bin/frdm-k64f/tests_periph_pm.elf is not locked.
Open On-Chip Debugger 0.10.0+dev-00703-g92bb76a4-dirty (2019-07-19-14:27)
Licensed under GNU GPL v2
For bug reports, read
	http://openocd.org/doc/doxygen/bugs.html
Info : auto-selecting first available session transport "swd". To override use 'transport select <transport>'.
Info : add flash_bank kinetis kx.pflash
adapter speed: 1000 kHz
none separate
cortex_m reset_config sysresetreq
srst_only separate srst_nogate srst_open_drain connect_deassert_srst
Info : CMSIS-DAP: SWD  Supported
Info : CMSIS-DAP: FW Version = 1.10
Info : CMSIS-DAP: Interface Initialised (SWD)
Info : SWCLK/TCK = 0 SWDIO/TMS = 1 TDI = 0 TDO = 0 nTRST = 0 nRESET = 1
Info : CMSIS-DAP: Interface ready
Info : clock speed 1000 kHz


/home/francisco/workspace/RIOT/tests/periph_pm/../../Makefile.include:600: recipe for target 'flash' failed
make: *** [flash] Error 1
make: Leaving directory '/home/francisco/workspace/RIOT/tests/periph_pm'

OPENOCD_RESET_USE_CONNECT_ASSERT_SRST=1 make -C tests/periph_pm BOARD=frdm-k64f flash -j3

make: Entering directory '/home/francisco/workspace/RIOT/tests/periph_pm'
Building application "tests_periph_pm" for "frdm-k64f" with MCU "kinetis".

"make" -C /home/francisco/workspace/RIOT/boards/frdm-k64f
"make" -C /home/francisco/workspace/RIOT/core
"make" -C /home/francisco/workspace/RIOT/cpu/kinetis
"make" -C /home/francisco/workspace/RIOT/drivers
"make" -C /home/francisco/workspace/RIOT/cpu/cortexm_common
"make" -C /home/francisco/workspace/RIOT/drivers/periph_common
"make" -C /home/francisco/workspace/RIOT/cpu/cortexm_common/periph
"make" -C /home/francisco/workspace/RIOT/cpu/kinetis/periph
"make" -C /home/francisco/workspace/RIOT/sys
"make" -C /home/francisco/workspace/RIOT/sys/auto_init
"make" -C /home/francisco/workspace/RIOT/sys/isrpipe
"make" -C /home/francisco/workspace/RIOT/sys/newlib_syscalls_default
"make" -C /home/francisco/workspace/RIOT/sys/pm_layered
"make" -C /home/francisco/workspace/RIOT/sys/shell
"make" -C /home/francisco/workspace/RIOT/sys/stdio_uart
"make" -C /home/francisco/workspace/RIOT/sys/tsrb
   text	   data	    bss	    dec	    hex	filename
  20732	    240	   2820	  23792	   5cf0	/home/francisco/workspace/RIOT/tests/periph_pm/bin/frdm-k64f/tests_periph_pm.elf
/home/francisco/workspace/RIOT/dist/tools/openocd/openocd.sh flash /home/francisco/workspace/RIOT/tests/periph_pm/bin/frdm-k64f/tests_periph_pm.elf
### Flashing Target ###
/home/francisco/workspace/RIOT/tests/periph_pm/bin/frdm-k64f/tests_periph_pm.elf is not locked.
Open On-Chip Debugger 0.10.0+dev-00703-g92bb76a4-dirty (2019-07-19-14:27)
Licensed under GNU GPL v2
For bug reports, read
	http://openocd.org/doc/doxygen/bugs.html
Info : auto-selecting first available session transport "swd". To override use 'transport select <transport>'.
Info : add flash_bank kinetis kx.pflash
adapter speed: 1000 kHz
none separate
cortex_m reset_config sysresetreq
srst_only separate srst_nogate srst_open_drain connect_deassert_srst
srst_only separate srst_nogate srst_open_drain connect_assert_srst
Info : CMSIS-DAP: SWD  Supported
Info : CMSIS-DAP: FW Version = 1.10
Info : CMSIS-DAP: Interface Initialised (SWD)
Info : SWCLK/TCK = 0 SWDIO/TMS = 1 TDI = 0 TDO = 0 nTRST = 0 nRESET = 1
Info : Connecting under reset
Info : CMSIS-DAP: Interface ready
Info : clock speed 1000 kHz
Info : SWD DPIDR 0x2ba01477
Info : MDM: Chip is unsecured. Continuing.
Info : kx.cpu: hardware has 6 breakpoints, 4 watchpoints
Info : kx.cpu: external reset detected
Info : Listening on port 46373 for gdb connections
    TargetName         Type       Endian TapName            State       
--  ------------------ ---------- ------ ------------------ ------------
 0* kx.cpu             cortex_m   little kx.cpu             reset
Info : MDM: Chip is unsecured. Continuing.
target halted due to debug-request, current mode: Thread 
xPSR: 0x01000000 pc: 0x0000073c msp: 0x1fff0200
auto erase enabled
Info : Kinetis MK64FN1M0xxx12 detected: 2 flash blocks
Info : 2 PFlash banks: 1024k total
wrote 24576 bytes from file /home/francisco/workspace/RIOT/tests/periph_pm/bin/frdm-k64f/tests_periph_pm.elf in 1.209991s (19.835 KiB/s)
Info : kx.cpu: external reset detected
Error: Failed to write memory and, additionally, failed to find out where
Warn : target was in unknown state when halt was requested
Error: error executing cortex_m crc algorithm
Error: Failed to read memory and, additionally, failed to find out where

/home/francisco/workspace/RIOT/tests/periph_pm/../../Makefile.include:600: recipe for target 'flash' failed
make: *** [flash] Error 1
make: Leaving directory '/home/francisco/workspace/RIOT/tests/periph_pm'

NOTE: my diff

diff --git a/boards/common/frdm/dist/openocd-klx.cfg b/boards/common/frdm/dist/openocd-klx.cfg
index 5d7f29cb1..7970abc14 100644
--- a/boards/common/frdm/dist/openocd-klx.cfg
+++ b/boards/common/frdm/dist/openocd-klx.cfg
@@ -19,7 +19,7 @@ source [find target/klx.cfg]
 # The debug module is stopped in low leakage modes, we use connect_assert_srst
 # to hold the device in reset while connecting with OpenOCD to make sure that we
 # always get a connection even when the application uses low power modes.
-reset_config srst_only connect_assert_srst
+reset_config srst_only
 
 $_TARGETNAME configure -event gdb-attach {
   halt
diff --git a/boards/common/frdm/dist/openocd-kx.cfg b/boards/common/frdm/dist/openocd-kx.cfg
index 2e128dcb9..84bd6243e 100644
--- a/boards/common/frdm/dist/openocd-kx.cfg
+++ b/boards/common/frdm/dist/openocd-kx.cfg
@@ -16,7 +16,7 @@ source [find target/kx.cfg]
 # The debug module is stopped in low leakage modes, we use connect_assert_srst
 # to hold the device in reset while connecting with OpenOCD to make sure that we
 # always get a connection even when the application uses low power modes.
-reset_config srst_only connect_assert_srst
+reset_config srst_only
 
 $_TARGETNAME configure -event gdb-attach {
   halt
diff --git a/tests/periph_pm/main.c b/tests/periph_pm/main.c
index 4fe74b3bf..3f5705956 100644
--- a/tests/periph_pm/main.c
+++ b/tests/periph_pm/main.c
@@ -241,6 +241,8 @@ int main(void)
          "needed.");
 #endif
 
+    pm_set(0);
+
     /* run the shell and wait for the user to enter a mode */
     shell_run(shell_commands, line_buf, SHELL_DEFAULT_BUFSIZE);

[fjmolinas]

I was able to test this fixes the issue for frdm-kw41z and frdm-k64f. Changes are good ACK.

[cladmi]

Thanks @benemorius for your feedback and giving the perfect test for this.
Thanks @fjmolinas for testing and reviewing.

[cladmi]

I also cross-checked locally with frdm-kw41z and frdm-k64f.