Hi Anson, On Tue, Dec 24, 2013 at 02:29:54AM +0000, Anson.Huang@xxxxxxxxxxxxx wrote: > Hi, Dmitry > > See below: > > Best Regards. Anson huang 黄勇才 Freescale Semiconductor Shanghai > 上海浦东新区亮景路192号A座2楼 201203 Tel:021-28937058 > > > >-----Original Message----- From: Dmitry Torokhov > >[mailto:dmitry.torokhov@xxxxxxxxx] Sent: Tuesday, December 24, 2013 > >8:31 AM To: Huang Yongcai-B20788 Cc: sachin.kamat@xxxxxxxxxx; > >linux-input@xxxxxxxxxxxxxxx Subject: Re: [PATCH] Input: gpio_keys: > >support wakeup system from freeze mode > > > >Hi Anson, > > > >On Mon, Dec 23, 2013 at 01:21:20PM -0500, Anson Huang wrote: > >> For "freeze" mode of suspend, cpu will go into idle and those > >> wakeup sources' irq should NOT be disabled during devices suspend, > >> so we need to add IRQF_NO_SUSPEND flag for those wakeup sources. > >> > >> Steps to test this patch: > >> > >> 1. echo freeze > /sys/power/state; 2. press gpio key which has > >> wakeup function, then system will resume. > > > >I do no think this is correct approach, otheriwse every driver that > >can be a wakeup source would have to use IRQF_NO_SUSPEND flag. The > >driver does use enable_irq_wake() in its suspend path, and platform > >code should perform all necessary work for this IRQ to be usable as a > >wakeup source. > > > >Thanks. > > > From the suspend flow, we can see that kernel will disable all > devices' irq unless its IRQF_NO_SUSPEND flag is set. For freeze mode, > as cpu will be in idle, SOC is not in low power mode, that means > freeze mode equals cpu in idle and devices in suspend. The wakeup > source must come from GIC. Yes, for normal standby/mem mode's suspend, > the enable_irq_wake will make everything done for waking up a system, > in gic_set_wake routine, it will call extern irq chip's wakeup > function, but for freeze mode, the wakeup process is same as normal > wakeup from cpu idle, so the wakeup source must be not masked in gic. > > Take our i.MX6 SOC for example, there is GPC module which is used to > wake up SOC from STOP mode(low power mode of SOC), normal standby/mem > mode suspend, the wakeup source's enable_irq_wake will set GPC to > monitor these irq source and wakeup system from STOP mode when there > is an wakeup event. But for freeze mode, kernel's suspend flow will > not finish the SOC's low level power management, the secondary CPUs > are even not disabled, kernel just suspend the devices and put CPU > into idle and waiting for wakeup even, in GIC's gic_set_wake routine, > it only calls our GPC's imx_gpc_irq_set_wake, here comes the problem, > as our SOC is not entering STOP mode, GPC has no use to set this > wakeup source enabled. So either we should enable GIC's wakeup in GIC > driver, or just set this flag to make kernel do NOT disable this > wakeup source's irq when executing a freeze mode suspend. > > Otherwise, we need to modify the GIC driver? Yes, I believe it is the task of platform code to make sure that enable_irq_wake() enables irq wakeup properly, be it suspend or hibernate. Thanks. -- Dmitry -- To unsubscribe from this list: send the line "unsubscribe linux-input" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
