RE: [RFC PATCH 0/2] dmaengine: at_hdmac: fix locking according toslave DMA requirements

[Jouko Haapaluoma]

Hi

Thanks for the patches. We got the DMA working for now but the tasklet_disable() and tasklet_enable() had to be removed from atc_control().

If the device driver calls dmaengine_terminate_all() from the callback (like in our previous deadlock example), the tasklet_disable() will cause
another deadlock because the tasklet will then wait for itself to close.

The tasklet_disable() seems to be used to ensure that no tasklet is running when terminating the DMA transfers. This prevents the terminate_all from
happening in between the critical sections in the tasklet which are locked with atchan->lock. Maybe the tasklet needs to be implemented so that it can
detect and recover if the terminate_all was called in between the critical sections in the tasklet?

BR,
Jouko Haapaluoma
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/