>> @@ -636,6 +649,9 @@ static irqreturn_t tegra_i2c_isr(int irq, void *dev_id) >> goto err; >> } >> >> + if (i2c_dev->hw->has_bus_clr_support && (status & I2C_INT_BUS_CLR_DONE)) >> + goto err; >> + > >This looks odd. If we support bus clear and the BUS_CLR_DONE interrupt is set, shouldn't we still process all other potential interrupts? Or was this meant to be inversed: > > if (!i2c_dev->hw->supports_bus_clear && (status & I2C_INT_BUS_CLR_DONE) > >? > > >Thierry Bus clear mechanism is to recover master from I2C Bus hang when I2C master loses bus arbitration due to some slave devices pulling SDA LOW continuously for some unknown reason. Master performs bus clear operation and once bus clear is done successfully, I2C_INT_BUS_CLR_DONE is set in interrupt status. During this time, no other interrupts will be received as it is already in the process of recovering from bus hang thru this bus clear operation. So transaction during the bus hang obviously is failure transaction. But master tries to recover from bus hang so transaction can be re-tried and communication with other slaves on that bus and further happen. Thanks Sowjanya
