Messages in this thread | | | Date | Thu, 28 Jul 2011 11:14:00 -0700 | Subject | Re: [PATCHv2] DMAEngine: Let dmac drivers to set chan_id | From | "Williams, Dan J" <> |
| |
On Thu, Jul 28, 2011 at 10:54 AM, Jaswinder Singh <jaswinder.singh@linaro.org> wrote: > On 28 July 2011 02:07, Russell King <rmk@arm.linux.org.uk> wrote: >> >>> On a serious note, my proposal, and the reply, shows the possibility >>> of having :- >>> a) Client drivers that are truly platform agnostic -- no platform_data >>> poking for >>> channel selection >> >> I really doubt that's even possible. Take this setup: > > I don't want to suggest anything wrong just because I didn't understand > your h/w. > I hope you will be kind enough to help me better understand your setup, > so that I have a fair chance to present my proposal.
I don't understand all the slave usage models but the proposal seems fragile. It wants to encode a whole bunch of arch specific details into a 32-bit code, but it still seems to me that there will always be platform specific details that don't fit the code. Especially when we are still striving to ensure common behavior across drivers. It seems the first step should be getting a few instances of a client driver interfacing with multiple dma drivers (via platform specific machinery) and see if anything common falls out of those implementations that can be pulled into dmaengine. The proposal seems to be addressing a problem that we don't yet have (several clients associating with multiple channels in disparate and un-maintainable ways).
-- Dan -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |