On 10/24/2014 10:59 AM, Marc Zyngier wrote: > Hi Bjorn, > > On 24/10/14 18:31, Bjorn Andersson wrote: >> >> Stephen Boyd talked about the need to be able to mask/unmask interrupts from >> client code in the Qualcomm platform as well - most likely to block wakeup >> sources(?) > What's wrong with irq_disable? The problem is irq_disable() is lazy and doesn't actually disable the interrupt. This is the scenario. During idle we want to communicate with another processor in the SoC and tell it to turn off something after we go to idle. The communication mechanism uses some shared memory and an interrupt to trigger the other processor to go look at what we told it to do. When the other processor is done handling the message it will send us an ack interrupt. We want to temporarily ignore that ack interrupt because 1) we're going to execute a wfi to trigger a deep idle and any pending interrupt will abort that instruction and 2) if the interrupt comes after we execute the wfi it will wake us up out of deep idle unnecessarily and waste power. There isn't anything in the communication protocol that says we don't want the interrupt to be sent and it would probably make things more complicated anyway. The solution is to mask the interrupt before we send the message, and then execute the wfi. Once we wake up, we unmask the interrupt and then handle the ack. -- Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project -- To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
