On Sat, Apr 09, 2016 at 10:14:31PM -0700, Peter Hurley wrote: > Hi Greg, et. al. > > This series aims to keep 8250 dma and the omap8250 variant from > diverging further, in the hopes that at some point the implementations > will converge. > > The first 2 patches add an overlooked requirement of using the > dmaengine interface; namely ensuring the actual dmaengine driver is > capable of the operations required by the 8250 client. > > The 8250 driver requires the dmaengine driver to support dma pause > with terminate for the rx channel, and further requires better-than- > descriptor residue accuracy (so that it is possible for dma to be > paused without completing followed by pio read of remaining rx data > _without losing data_. > > Note however that the dmaengine driver query is known to be > insufficient, as at least one driver (omap-dma) does not support > the requirements for this use case but reports that it does. > My hope is to extend slave caps reporting to allow dmaengine > clients to distinguish incomplete pause/resume support. > > A similar check is added for the tx chanenl. Soon I expect to add pause > check to tx channel as well, as I believe it will be required to > implement XON/XOFF transmit *which is completely missing* - with dma > enabled the x_char is never transmitted. > > The remaining patches cleanup the rx dma interface; the rx dma flush > is extracted as a separate operation, rather than as a sub-operation > of dma flow. > > REQUIRES: 4-patch series "serial: 8250: misc cleanups" > > Regards, > > Peter Hurley (6): > serial: 8250: Validate dmaengine rx chan meets requirements > serial: 8250: Validate dmaengine tx chan meets requirements > serial: omap8250: Drop rx buffer sync > serial: omap8250: Simplify rx dma completion interface > serial: 8250: Extract IIR logic steering from rx dma > serial: omap8250: Terminate rx dma only for flushes For the whole series, FWIW: Reviewed-by: Heikki Krogerus <heikki.krogerus@xxxxxxxxxxxxxxx> Thanks Peter, -- heikki -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
