Ok, I guess I didn't update all the patch descriptions properly...
Haavard Skinnemoen <email@example.com> wrote:
This isn't correct anymore, as the dma_slave_descriptor struct turned
out not to be needed after all.
This isn't entirely correct either -- regular dma_async_tx_descriptors
are created, but one such descriptor can represent a whole scatterlist.
I'll update the changelog for the next round...