On 02/28/2014 04:30 PM, Ulf Hansson wrote: > On 28 February 2014 03:37, Aaron Lu <aaron.lu@xxxxxxxxx> wrote: >> On 02/27/2014 09:05 PM, Ulf Hansson wrote: >>> I think it's more a matter of having a controlled suspend sequence. >>> The mmc core are not able to serve any new SDIO requests while it is >>> suspended, therefore it tells the sdio func driver about when it safe >>> to send request - using it's PM callbacks. >> >> Does this mean after the function device is suspended from PM core's >> pespective, the mmc core will still send some requests to the function >> device? I did see one such request, disable_width, get sent after the >> function driver's suspend callback is invoked, don't know if there are >> others. > > Nope. The mmc core will send request to the "card device" during > suspend. Those are part's of a sequence to put the card in a proper > "suspend mode". I suppose this request is sent after calling function driver's suspend callback? BTW, what request is it? > >> >> From the mmc_sdio_suspend function I can see two things are done: >> 1 function device driver's suspend callback is called; >> 2 optionally disable_width and power off the card according to some flags. >> >> So once we fixed the -ENOSYS problem, can we have the function device >> run its own suspend callback and have the mmc_sdio_suspend just did the >> disable_width and power off thing? > > That requires that all sdio func devices get suspended before the card > devices. (And resumed after, but that will of course be implicit.) Yes, and this is guaranteed by PM core as the function device is the child of the card device. > > I guess the guarantee for that is already present, since we have added > the func device to the driver model after we have added the card > device at mmc_sdio_attach(). So, I suppose this could work. Exactly. > > Do you want me to create a patch that you can test? That would be good. I do not have any SDIO card at hand, but I would be glad to review the patch. Thanks, Aaron -- To unsubscribe from this list: send the line "unsubscribe linux-mmc" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
