Hi Andy, On 04/26/2013 08:48 PM, Andy Green wrote: > On 27/04/13 09:04, the mail apparently from Suman Anna included: > > Hi Suman - > >> Even though both the scenarios look very similar, I believe there are >> some slight differences. All the devices belonging to a controller may >> not be of the same type (meaning, intended towards the same remote or be >> used interchangeably with one another). It is definitely possible if you >> have a similar scenario to the DMA physical channels and your remote >> rx interrupt can identify the device/channel to process. This would be >> very much dependent on the architecture of a controller. The particular >> example that I have in mind is s/w clients between the same set of >> remote and host entities using the same device - the send part is anyway >> arbitrated by the controller, and the same received message can be >> delivered to the clients, with the clients making the decision whether >> the packet belongs to them or not. I agree that all remote-ends will not >> be able to cope up intermixed requests, but isn't this again a >> controller architecture dependent? > > Maybe it's helpful to describe our situation more concretely, because > the problem is not coming from "the architecture of the [mailbox] > controller".
Thanks for explaining the usecase. I do think that similar approaches will become more common (TI AM335 has something similar as well - though it is related to suspend). The right word should have been "controller functional integration", I said it as s/w architecture or usage model. In your case, it is clear that you need time-shared exclusive access, whereas I am talking about simultaneous-shared usecases. > > In the SoC we work on clock and subsystem power control registers, a > serial bus, and some other assets are not directly accessible from > Linux. We must ask a coprocessor to operate these for us, using the > mailbox. > > So at any one time, the clock driver or voltagedomain driver for the SoC > may want to own the mailbox and perform one or more operations over it > synchronously, in some cases on the remote side involving transactions > on a serial bus. We don't want other transactions to be occurring while > we wait for the serial bus to complete what the driver who started that > asked for, for example. > > We can cope with this by having an outer driver mediate access to the > mailbox. But then there are multiple sync primitives like completions > and notifiers per operation, because your core already does this. > > In short the FIFO + sync operations approach your core implements > doesn't fit our use case. That can be our problem, in which case we'll > live with the outer mediation driver on top of the mailbox, or it can be > a sign the fixed choice of FIFO + sync operations in your core did not > quite hit the nail on the head to really model all the facets of legit > mailbox usage. I agree that the current code doesn't address this usage. The changes (should have them ready in the next couple of days) I am working on actually makes this conditional. > > At least, this real scenario should be interesting to think about before > rejecting ^^ No, I didn't reject anything, we are dealing with two contrasting usecases dependent on the functional integration, and we have to find a middle ground. regards Suman -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
