13:41 <einthecorgi2> Ive been having a couple issues with GPIO, specifically when used to acquire a threaded IRQ. I am looking at this patch set -> https://patchwork.kernel.org/project/linux-rockchip/list/?series=454377. But it seems the gpio drivers for rk3399 are not existant. However there are drivers for rockchip-pinctrl. Has it been this way for a while? The GPIO work fine from user space.

13:56 <robmur01> The GPIO driver has historically been part of the pinctrl driver - those patches are just splitting it out.

14:05 <einthecorgi2> ah okay, so then this could potentially explain my interrupt issues (before patch applied).

14:15 <einthecorgi2> I am in 5.11 now and a couple chunks from the patches get rejected, time to do it manually.

17:20 <mmind00> einthecorgi2: for rk3399 and before that series should not bring any functional changes

17:21 <mmind00> einthecorgi2: and from past experience irqs (including threaded irqs) worked fine ... so you may want to provide a bit more context for your issue ;-)

17:35 <einthecorgi2> Yeeha ive never had problems with threaded IRQs in the past. I think it is related to the irq parent in the device tree being improperly defined. As for some more context I am setting up a chain of MCP251x devices that have a single GPIO interrupt. Right now the driver fails to acquire the irq. The interesting part is that it says the irq is just 0.

17:35 <einthecorgi2> https://github.com/torvalds/linux/blob/master/drivers/net/can/spi/mcp251x.c#L1219

17:41 <robmur01> Yeah, no sign of IRQF_SHARED there, so having multiple things on the same interrupt line will be the fundamental issue.

17:42 <robmur01> the driver may need some modifications to properly support IRQ sharing (in general; I've not looked at that one in particular)

17:45 <robmur01> just blindly whacking the flag in might sometimes work, but other times might break subtle assumptions about stuff like mutual exclusion and lead to hard-to-debug misbehaviour

17:53 <robmur01> ...although the IRQ also being invalid in the first place does smack of a DT problem

18:29 <einthecorgi2> yeah, my thoughts as well. But I think it safe to say I dont need to dig around in those GPIO patches.

21:43 <einthecorgi2> Okay interrupts are sorted. Now i am working with the rkisp in mainline. I had some previous problems which got sorted out and now I am trying to build the video pipline.

21:44 <einthecorgi2> I believe I am setting up the video correctly but land with this `rkisp1 ff920000.isp1: start pipeline failed -67`

21:45 <einthecorgi2> whats a good way to dig deeper into the isp errors? I am going through the driver now.

21:45 <einthecorgi2> https://pastebin.com/BkUYRqFQ

21:46 <einthecorgi2> https://pastebin.com/FvRYrFjr