WDT reset due to flash access lock-up
Posted: Sun Oct 07, 2018 8:07 pm
Hi,
I'm running ESP32 w/ 2 CPUs. Occasionally, I get a WDT resest:
Looking at the two addresses in gdb:
So it seems that the PRO CPU hangs waiting for the APP CPU to give up the access to the flash, but seems the task scheduler state is stuck at taskSCHEDULER_NOT_STARTED? I have at least a half-dozen tasks running - is it one of my tasks that is causing issues? Any ideas on debugging this?
Thanks
I'm running ESP32 w/ 2 CPUs. Occasionally, I get a WDT resest:
Code: Select all
W (85) boot: PRO CPU has been reset by WDT.
W (85) boot: WDT reset info: PRO CPU PC=0x40084d50
W (86) boot: WDT reset info: APP CPU PC=0x40088214
Code: Select all
(gdb) list *0x40084d50
0x40084d50 is in spi_flash_op_block_func (/proj/Firmware/esp-idf/components/spi_flash/./cache_utils.c:82).
77 // s_flash_op_complete flag is cleared on *this* CPU, otherwise the other
78 // CPU may reset the flag back to false before IPC task has a chance to check it
79 // (if it is preempted by an ISR taking non-trivial amount of time)
80 s_flash_op_complete = false;
81 s_flash_op_can_start = true;
82 while (!s_flash_op_complete) {
83 // busy loop here and wait for the other CPU to finish flash operation
84 }
85 // Flash operation is complete, re-enable cache
86 spi_flash_restore_cache(cpuid, s_flash_op_cache_state[cpuid]);
(gdb) list *0x40088214
0x40088214 is in vTaskExitCritical (/proj/Firmware/esp-idf/components/freertos/./tasks.c:4278).
4273 #ifdef CONFIG_FREERTOS_PORTMUX_DEBUG
4274 vPortCPUReleaseMutexIntsDisabled( mux, function, line );
4275 #else
4276 vPortCPUReleaseMutexIntsDisabled( mux );
4277 #endif
4278 if( xSchedulerRunning != pdFALSE )
4279 {
4280 TCB_t *tcb = pxCurrentTCB[xPortGetCoreID()];
4281 BaseType_t nesting = tcb->uxCriticalNesting;
4282 if( nesting > 0U )
Thanks
