On Sat, 2010-10-02 at 13:54 +0200, ext Ohad Ben-Cohen wrote: > Introduce SDIO runtime PM support: > > 1. Power to SDIO cards is kept low until one of its functions is bound > (i.e. a matching driver is successfully probed) > > 2. If the matching driver supports runtime PM, power to the card is > dropped soon after probe() returns. It is then up to the driver > to request power to its function, using runtime PM API (the get/put > variants). This is demonstrated with the wl1271 driver, in which > the power of the card power is coupled with the state of the wlan > interface (interface up -> power is up, interface down -> power is down) > > 3. If a matching driver does not support runtime PM, power to the card > is kept high during the whole lifetime of the driver > > 4. When the driver is removed, power to the card is immediately dropped > > 5. If there are multiple drivers for the same card (several SDIO functions), > power will be pulled high before the first driver probes, and dropped > down after the last driver is removed. In between, power will be > maintained accrording to the accumulated usage count of the complete > drivers group > > 6. SDIO suspend/resume semantics are unchanged. In addition, when the system > comes out of suspend, it is guaranteed that the power state of the > SDIO card will reflect its runtime PM usage count. > > 7. What was NOT changed: > - Interface: drivers can still assume that the card is powered > when probe/remove/suspend/resume are called > - Existing behavior: drivers that do not support runtime PM > are unchanged > > Changes since v1: > - Interaction with system suspend/resume > - Better commentary > > Dependencies: > - SDIO patches are against mmc-next, and have a runtime dependency on > commit "PM / Runtime: Lenient generic runtime pm callbacks" > (patch is in linux-next now) > - WLAN patches depend on recent wl1271 activity, and they are here > just to demonstrate the usage of the SDIO patchset (will be > resubmitted separately) > > The full patchset, together with all its dependencies, is also available at: > git://wizery.com/pub/mmc.git runtime-pm-v2 I have tested this on my beagle board with the wl1271 patches that Ohad provided and it seems to be working fine. I've also created a small script to stress test the implementation a little bit and didn't see any problems. Tested-by: Luciano Coelho <luciano.coelho@xxxxxxxxx> This is quite an important fix in the wl1271 point-of-view. Chris, do you think there is any chance that this could make it to 2.6.37? -- Cheers, Luca. -- 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
