Doug, On Wed, Nov 19, 2014 at 11:30 AM, Doug Anderson <dianders@xxxxxxxxxxxx> wrote: > Andrew, > > On Wed, Nov 19, 2014 at 11:03 AM, Andrew Bresticker > <abrestic@xxxxxxxxxxxx> wrote: >> Doug, >> >> On Wed, Nov 19, 2014 at 10:51 AM, Doug Anderson <dianders@xxxxxxxxxxxx> wrote: >>> Since the dw_mmc driver was first added to Linux it's had a TODO in it >>> that we should turn off the card clock during suspend. I have no idea >>> for sure why it wasn't done originally, but if I had to guess I'd >>> guess it was related to the lack of a common clock framework. Let's >>> do it now. >>> >>> There is no reason for the card clock to be left on during suspend and >>> most systems will eventually turn it off anyway (when whole clock >>> trees are disabled). However, it's good to be explicit that it's >>> disabled at the time that the MMC subsystem is disabled. >> >> Should the bus clock (biu) be disabled as well? > > Good question. I'm slightly hesitant to turn biu_clk off in this > patch, though. Specifically interrupts are still enabled at this > point in suspend. I guess most interrupts shouldn't be going off > right now (nobody is accessing storage, right?), but I could imagine > that a card detect or an SDIO interrupt at just the wrong time could > cause our ISR to go off _after_ this function is called. The > interrupt handling function doesn't know to turn the BIU clock back on > so you'd get a hang. Perhaps interrupts should be disabled as well across suspend/resume, like the sdhci-based hosts do? Would would happen if we tried to service an interrupt with the ciu clock disabled? -- 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
