Watchdog failed to reset esp properly

[gailu]

Hi Experts,

We are doing regression testing of our software running on ESP-12 with SDK 2.0 (ESP8266_NONOS_SDK_V2.0.0_16_07_19). When generating lots of interrupts and TCP communication we see that Watchdog is kicked in which is expected. However we are facing an issue approx once out of 6-7 attempts where Watchdog is kicked-in but board is not reset properly and remains in hanged position.

When Watchdog resets the board properly wee see following logs at console.

Code: Select all

ets Jan  8 2013,rst cause:4, boot mode:(3,7)

wdt reset
load 0x40100000, len 2408, room 16
tail 8
chksum 0xe5
load 0x3ffe8000, len 776, room 0
tail 8
chksum 0x84
load 0x3ffe8310, len 632, room 0
tail 8
chksum 0xd8
csum 0xd8

Occasionally when watchdog is kicked in but board is not reset properly we see only following prints on console.

Code: Select all

ets Jan  8 2013,rst cause:4, boot mode:(2,7)

wdt reset
;[1E]ts_main.c

Can you please let me know what could be the reason where watchdog is not able to reset the board as expected.

Can you please let me know what could be the reason where watchdog is not able to reset the board as expected.

Can someone please explain the difference between boot mode (3,7) in successful case and (2,7) in failure case. Any documentation on boot mode please?

Also please let me know the meaning of "[1E]ts_main.c" print

[pratik]

Please make sure that the boot mode GPIOs are in a valid state when the reset occurs!
No external circuitry shoulld be driving those pins to value = 2, which may cause this issue.