On 2019-05-11 12:17, Lorenzo Bianconi wrote: > Run mt76x02_edcca_init atomically in mt76_edcca_set since it runs > concurrently with calibration work and mt76x2_set_channel. > Introduce __mt76x02_edcca_init helper routine > > Signed-off-by: Lorenzo Bianconi <lorenzo@xxxxxxxxxx> I don't think this is enough. To prevent issues with calibration, we probably need to hold the mutex for the duration of the calibration anyway. Otherwise it might get enabled right in the middle of it and screw things up. Also, it probably simplifies the patch if you don't add the wrapper function that takes the mutex, and instead just explicitly take the mutex where needed. - Felix
