SetNextChannel can return 0 (immediate go-ahead) to ScheduleTx even if no channels enabled

[jmalangoni]

LoRaMac.c static TimerTime_t SetNextChannel( void ) returns 0 even if no channels are supported, if nextTxDelay is never assigned a value after initialization:

    if( nbEnabledChannels > 0 )
    {
        Channel = enabledChannels[randr( 0, nbEnabledChannels - 1 )];
#if defined( USE_BAND_915 ) || defined( USE_BAND_915_HYBRID )
        DisableChannelInMask( Channel, ChannelsMaskRemaining );
#endif
        return 0;
    }
    else
    {
        // Protect the function of returning the initialization value of nextTxDelay
        return ( nextTxDelay == ( TimerTime_t )( -1 ) ) ? 0 : nextTxDelay;
    }

But if .TimeOff of all bands == 0 in the following...

// Update Aggregated duty cycle
    if( AggregatedTimeOff < ( curTime - AggregatedLastTxDoneTime ) )
    {
        AggregatedTimeOff = 0;

        // Update bands Time OFF
        for( uint8_t i = 0; i < LORA_MAX_NB_BANDS; i++ )
        {
            if( DutyCycleOn == true )
            {
                if( Bands[i].TimeOff < ( curTime - Bands[i].LastTxDoneTime ) )
                {
                    Bands[i].TimeOff = 0;
                }
                if( Bands[i].TimeOff != 0 )
                {
                    nextTxDelay = MIN( Bands[i].TimeOff -
                                       ( curTime - Bands[i].LastTxDoneTime ),
                                       nextTxDelay );
                }
            }
            else
            {
                nextTxDelay = 0;
                Bands[i].TimeOff = 0;
            }
        }

...then nextTxDelay == ( TimerTime_t )( -1 ), and if nbEnabledChannels == 0, the function returns 0. This is problematic since ScheduleTx then calls SendFrameOnChannel using the current Channel value, even though no channels are currently enabled.

One quick fix would be for SetNextChannel to return ( TimerTime_t )( -1 ), and then have ScheduleTx check for this as an error. Any problems with this?

[djaeckle]

Hi jmalangoni,

thanks for the report. I see your point. We will provide an update in the next release. I'm not sure yet, which solution we apply. But I think your quick fix should work.