Re: [OpenOCD-devel] Trouble flash-programming STM32F4xx in low-power mode

[Uwe B. <bo...@el...>]

>>>>> "Jens" == Jens Bauer <jen...@gp...> writes:

    Jens> Hi all.  On Sun, 18 Jan 2015 01:01:57 +0100, Jens Bauer wrote:
    >> If I make the following program:
    >> 
    >> int main() { while(1) { __WFI(); } }

    Jens> Note: If I change __WFI(); to __NOP(); there are no problems.

    Jens> -One thing I forgot to say is that I'll usually need to take out
    Jens> the external crystal, before I attempt to repeat a 'reset halt'
    Jens> until I get the target to halt.


I have the same problems. In principle these problems should be solved by
setting the Low Power debug bits similar like

proc stm32_default_examine_end {} {
  # Enable debug during low power modes (uses more power)
  mww 0xe0042004 0x00000007 ;# DBGMCU_CR = DBG_STANDBY | DBG_STOP |
  DBG_SLEEP

  # Stop watchdog counters during halt
  mww 0xe0042008 0x00001800 ;# DBGMCU_APB1_FZ = DBG_IWDG_STOP |
  DBG_WWDG_STOP
}

$_TARGETNAME configure -event examine-end { stm32_default_examine_end }

However when some problem with OpenOCD and a STM32 in __WFI mode happens,
it was hard to get the setup functional again. At that point I normally
switch to blackmagic
https://github.com/blacksphere/blackmagic
and things work as expected.

Probably blackmagic runs some JTAG sequence on the sleeping device in a
tight loop and can reconnect, while OpenOCD has to do Host<->debugger
transactions for that loop and probably misses some window of opportunity to
gain access to the debug target.

Bye
-- 
Uwe Bonnes                bo...@el...

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------