Hi Dan, You're right, sorry for the noise. My v1 was written against mainline, where dma_request_slave_channel() still exists, but Mark applied Andy's conversion (b045248399d0) to for-7.4 three days before applying my patch (b4136c0d69ea), so there ->get_dma_channel() propagates error pointers and the !NULL check is ineffective.
I've just sent an incremental patch converting the check to IS_ERR()
("ASoC: fsl_asrc: fix second front-end channel check").
Best regards,
Slavin Liu
