> From: Cornelia Huck [mailto:cornelia.huck@xxxxxxxxxx] > Finer granularity is certainly better here, but I'm not quite sure if > this solves our s390 problem (we don't have dma support). All those > backends should also have a non-dma version... In fact that is already there. Here is the form of async_memcpy for example: ... async_memcpy( ... ) { struct dma_chan *chan = async_tx_find_channel(depend_tx, DMA_MEMCPY); struct dma_device *device = chan ? chan->device : NULL; int int_en = callback ? 1 : 0; struct dma_async_tx_descriptor *tx = device ? device->device_prep_dma_memcpy(chan, len, int_en) : NULL; if (tx) { /* run the memcpy asynchronously */ ... } else { /* run the memcpy synchronously */ ... } } When CONFIG_DMA_ENGINE=n async_tx_find_channel takes the form: ... async_tx_find_channel( ... ) { return NULL; } So in the S390 case the entire asynchronous path will be compiled away. -- Dan - To unsubscribe from this list: send the line "unsubscribe linux-s390" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
