Re: [RFC] improve the SDHCI wakeup support.

[Peppe CAVALLARO <peppe.cavallaro@st.com>]

Hello

On 11/10/2010 4:28 PM, Giuseppe CAVALLARO wrote:
>
> Hello.
>
> The following patches, currently built against an ST Kernel
> "2.6.32", show how to improve the wakeup support in the SDHCI
> device driver.
>
> Note:
> I'm going to rework them, for example, against the mmc-next
> after performing the review process and, obviously, if you
> think that they can be actually useful.
>
> Indeed, on a STB, it can be nice to have the capability to
> wake-up the system when a card is inserted ;-).
>
> This work allows the SDHCI to wake up the system
> on the following events:
>  1) Card Interrupts.
>  2) Card Insertion.
>  3) Card Removal.
>
> To do that the sdhci has to perform some operations
> described in the patch named:
>   "mmc_sdhci: improve the wake-up support"
>
> A new parameter has been also added to select which wakeup
> event has to be used. At any rate, a device based on
> the sdchi, e.g. sdhci-pci, can use a default mode (that
> can be modified at run-time as well).
> In case of the sdhci-pci the "Card Interrupt event" has not
> been modified according to the logic behind the recent commit:
>   5f619704d18b93869d045abc49e09cdba109b04b
>
> The define MMC_PM_KEEP_POWER has been used to notify that
> a device driver (e.g. sdhci-pltfm) want to wakeup the system.
> In any case, the logic for programming the HC register is
> embedded in the suspend and it's self contained.
>
> In the sdhci-pltfm, it has been also introduced another fix:
> see patch named:
>  "mmc: sdhci-pltfm calls the sdhci_alloc_host with pdev->dev"
> The driver calls the sdhci_alloc_host passing as device pointer
> the parent.
> Note: parent name is "platform" and dev name is "sdhci.0".
> IMO it makes sense to pass the pdev->dev pointer instead of the
> parent. This also helps when invoke the device_set_wakeup_capable etc.
>
> This is a piece of output on our ST platforms:
>
>  bash-3.00# echo 2 > /sys/module/sdhci/parameters/wakeup
>  bash-3.00# echo standby > /sys/power/state
>
>  PM: Syncing filesystems ... done.
>  Freezing user space processes ... (elapsed 0.00 seconds) done.
>  Freezing remaining freezable tasks ... (elapsed 0.00 seconds) done.
>  [STM][PM] Analyzing the wakeup devices
>  [STM][PM] -> device sdhci.0 can wakeup
>  [STM][PM] -> device stm-asc.0 can wakeup
>
> [snip]
>
> As shown above, to modify at run time the wakeup event we can do:
>  bash-3.00 echo X > /sys/module/sdhci/parameters/wakeup
>
> where X can be:
>  0: no wakeup
>  1: Card Interrupts
>  2: Card Insertion
>  3: Card Removal
>
> Hmm, maybe it's worth having more parameters instead of the
> wakeup. I mean, something like this (welcome feedback):
>  wake_on_card_int
>  wake_on_card_ins
>  wake_on_card_rem
>
> Welcome advice and feedback as usual.
>

I've reworked the patches, I sent the mailing list last week, now built
against the mmc-next Git repository.
I've removed some piece of the code commented by yourself (in this first
thread of emails) and re-generated a single path (marked as V2).

Many thanks for your advice and welcome feedback on the new patch.

Regards
Peppe