On 14/02/23 19:05, Pierre-Louis Bossart wrote: > >>>> static void amd_sdw_irq_thread(struct work_struct *work) >>>> { >>>> struct amd_sdw_manager *amd_manager = >>>> @@ -945,6 +952,9 @@ static void amd_sdw_irq_thread(struct work_struct *work) >>>> status_change_0to7 = acp_reg_readl(amd_manager->mmio + ACP_SW_STATE_CHANGE_STATUS_0TO7); >>>> dev_dbg(amd_manager->dev, "%s [SDW%d] SDW INT: 0to7=0x%x, 8to11=0x%x\n", >>>> __func__, amd_manager->instance, status_change_0to7, status_change_8to11); >>>> + if (status_change_8to11 & AMD_SDW_WAKE_STAT_MASK) >>>> + return amd_sdw_process_wake_event(amd_manager); >>>> + >>> it's not clear what 8to11 might have to do with the wake enable? >>> >>> Can't you have a wake for devices 1..7? >> It can wake any device from 1..11 . >> SoundWire Wake interrupt status bit is part of 8to11 register. > You probably want to explain what status_change_0to7 and > status_change_8to11 actually control, it's not very intuitive for me to > see device 1 interrupt being handled in status_change_8to11? Peripheral device status interrupts on manager side will be read by reading 0to7 (for devices 1..7) and 8to11 (for devices 8 ..11) registers. 8to11 register will contain field for reading Soundwire wake enable status for Soundwire manager instance.