From mboxrd@z Thu Jan 1 00:00:00 1970 References: From: Philippe Gerum Subject: Re: Realtime interrupt handling in driver Date: Wed, 23 Feb 2022 09:12:17 +0100 In-reply-to: Message-ID: <87h78qf0qe.fsf@xenomai.org> MIME-Version: 1.0 Content-Type: text/plain List-Id: Discussions about the Xenomai project List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Russell Johnson Cc: xenomai@xenomai.org Russell Johnson via Xenomai writes: > Hello, > > In my driver that I have been converting to support OOB functionality > with the EVL core, I have an interrupt handler that handles all > interrupts from a device sitting on the PCIe bus. I was reading > through the information on your wiki about interrupt pipelining in > Dovetail. I was confused a bit on if there is work that I need to do > in order to get interrupts & interrupt handlers to run out-of-band > (and not be preempted by in-band linux processing), or if the EVL core > takes care of that for me? > If there is work I need to do with > interrupt config & interrupt handlers in order to make them able to > run out-of-band, where do I need to start? > > Thanks! > > Russell [1] describes the general flow, this is a bit outdated starting from the upcoming kernel v5.17 series [2], but the logic remains the same though. [3] explains why and how to fix up the irqchip handlers for IRQ pipelining. [1] https://evlproject.org/dovetail/porting/irqflow/ [2] https://source.denx.de/Xenomai/xenomai4/linux-evl/-/tree/v5.17-evl-rebase [3] https://evlproject.org/dovetail/pipeline/irq_handling/ -- Philippe.