06:11 <nickchan> okay a8x cpmu is getting even more interesting

06:11 <nickchan> PMI mode generates a sync exception with a ESR of 0x94000200

06:11 <nickchan> AIC mode can be set but does nothing

06:12 <nickchan> FIQ mode is RES0 so probably not supported

06:12 <nickchan> and HALT mode halts the CPU as expected

06:13 <nickchan> 16-bit instruction trapped and ISV is 0

06:13 <nickchan> am pretty sure that's illegal

06:13 <nickchan> data abort exception too

06:15 <nickchan> EA, bit [9]

06:15 <nickchan> External abort type. This bit can provide an IMPLEMENTATION DEFINED classification of External aborts.

06:16 <nickchan> ffs i don't want to add impdef aborts into linux

06:19 <nickchan> by generating a sync exception I mean it is generated at the expected time, where the counter reaches 48 bit

06:26 <nickchan> even as IMPDEF delivering a data abort doesn't sound right to me however

07:11 <nickchan> all of a7-a11 can deliever the sync exception

07:11 <nickchan> but only a11 can deliver the fiq

07:12 <LarstiQ> nickchan: do you have a whole testbench of devices?

07:12 <nickchan> LarstiQ: i have a lot

07:14 <nickchan> these devices are lot cheaper than macs :)

07:16 <nickchan> there is no hypervisor however so a lot more blind probing

07:18 <LarstiQ> just fun poking around or is this yak shaving for some other goal?

07:19 <nickchan> my experiments https://github.com/asdfugil/m1n1-idevice/blob/cpmu/proxyclient/experiments/pmi.py

07:19 <nickchan> LarstiQ: to get performance counters working properly in Linux

07:20 <LarstiQ> nickchan: for those devices themselves, or as inspiration for Asahi targeted macs?

07:21 <nickchan> LarstiQ: for those device themselves

07:21 <nickchan> there's quite a bit of hw differences

07:22 <nickchan> I really hope I can get AIC mode to work because otherwise it's to insert an IMP DEF hack inside Linux's sync exceptio handler

07:23 <nickchan> This also means the current AIC FIQ handler is wrong for A7-A10

07:24 <nickchan> because the CPMU on these can't deliever a FIQ, the fiq handler should not touch the PMC registers

07:26 <nickchan> actually

07:26 <nickchan> it's technically wrong for every CPU

07:26 <nickchan> because I believe the CPMU has other modes even on M1

07:27 <nickchan> the FIQ handler probably should only touch the PMCR0 register when PMCR0_IMODE is set to FIQ

07:34 <nickchan> this is actually m1n1's behavior

07:34 <nickchan> so it should be fine

07:34 <nickchan> for linux too

07:39 <LarstiQ> nickchan: I'll applaude from the sidelines. I like reading your progress but can't help much. Unless you need some rubber duck debugging!

07:49 <nickchan> or maybe no hack but also give sync exception the fiq treatment when adding m1 support except the default handler is the normal sync exception handler

07:50 <nickchan> and then allow the custom sync exception handler to call the default sync exception handler

08:16 <nickchan> in the AIC mode that ostensibly is supposed to deliver an IRQ not only no IRQs are received but PMCR0_IACT never gets set as if it's in OFF mode

08:16 <nickchan> so it probably won't work

10:12 <nickchan> actually xnu uses aic mode on a7-a10

12:17 <anuragrao> Hey, I'm new here and would like to try to help with WiFi P2P support. If this is supported, we can partially solve the external display problem on Macbook Air devices. I tried setting it up with gnome-network-displays but then I can see errors in journalctl and dmesg. I looked through the driver code for brcmfmac and saw there is a p2p.c file. I'm new to kernel development and want to understand the scope of this problem an

12:19 <anuragrao> what is the state of wifi p2p at the moment, what layer is it supported until and what layer has to be implemented? the output of iw phy lists p2p-client p2p-go and p2p-device as supported interface modes

12:32 <anuragrao> https://github.com/AsahiLinux/linux/issues/363

22:27 <jannau> I've tagged asahi-6.12.9-1 not much interesting compared to 6.12.8-2. fix for SError on boot for t602x and a small dcp audio fix for some devices (still not enabled)