i2c.c anomalies (IDFGH-7991)

[ammaree]

I have tried to understand the logic relating to the 'clear_bus_cnt' variable but so far have failed completely.
The lines listed below are all the references related in/directly to this variable.

90     #define I2C_ACKERR_CNT_MAX             (10)
1492   static uint8_t clear_bus_cnt = 0;
1507        clear_bus_cnt = 0;
1553                    clear_bus_cnt = 0;
1556                    clear_bus_cnt++;
 	                    if (clear_bus_cnt >= I2C_ACKERR_CNT_MAX) {
 	                        clear_bus_cnt = 0;
 	                    }
 1573            clear_bus_cnt = 0;

I have commented out all the lines, recompiled with no errors and the application has been running for months?
Am I missing something, does this variable and the related code actually do anything useful?

[o-marshmallow]

Hi @ammaree ,

You are right, that's an issue that we've found recently. Indeed, in the if (clear_bus_cnt >= I2C_ACKERR_CNT_MAX) statement, we must manually call the fsm reset function, else, the variable is useless.

An internal merge request is already on-going.

[AxelLin]

fixed in master: 75ede16
But v5.0 and v4.x still need fix.

[AxelLin]

v5.0: 4ef2ba2
v4.4: 3ec5041

[AxelLin]

@o-marshmallow
Looks like v4.3 branch needs the backport fix. (Missing i2c_hw_fsm_reset() call)
https://github.com/espressif/esp-idf/blob/release/v4.3/components/driver/i2c.c#L1245-L1247

[ammaree]

Have moved on to IDF v5.3 and the new master driver so guess this issue is not relevant any more