#asahi-dev · 2022-04-04 20:00
svenhm. I bet this is something incredibly dumb that I’ll realize a few days from now :/#asahi-dev · 2022-04-04 18:11
svenhrm, but how do I ack that thing then? And shouldn’t the state disappear when I e.g. write to the cq doorbell or disable the irq at the nvme level?#asahi-dev · 2022-04-04 17:46
svenm1n1#asahi-dev · 2022-04-04 17:43
svenwhy does it claim the nvme irq is asserted before i even boot the processor then :(#asahi-dev · 2022-04-04 17:41
svenread32(AIC_HW_STATE + (4 * (590 / 32))) & (1 << (590 % 32)); <-- let's try that again, this looks correct now, doesn't it?#asahi-dev · 2022-04-04 17:40
svenduh#asahi-dev · 2022-04-04 17:40
svenOHHHH#asahi-dev · 2022-04-04 17:40
svenread32(AIC_HW_STATE + (4*(590/32))) & (590%32); <-- that should give me the state of interrupt 590, shouldn't it?#asahi-dev · 2022-04-04 17:40
svenhrm, i must be doing something very dumb then#asahi-dev · 2022-04-04 17:15
svenhrm, or maybe not#asahi-dev · 2022-04-04 17:08
svenlooks like AIC_HW_STATE = AIC + 0x4200 might be what i want#asahi-dev · 2022-04-04 17:06
svenmarcan: is there any register in AICv1 that tells me the current hardware state of an interrupt (even if that interrupt is disabled)#asahi-dev · 2022-04-03 18:58
svenah, nice#asahi-dev · 2022-04-03 18:56
sveni guess back when I wrote that hack we either modified the fdt in place or there just weren’t any important modifications yet#asahi-dev · 2022-04-03 18:55
svenhrm, no, it should get the modified fdt #asahi-dev · 2022-04-03 15:03
svenit’s not nice and maybe doesn’t even work anymore at this point #asahi-dev · 2022-04-03 15:03
sventhere’s some hack in linux.py to chainload uboot and a kernel together #asahi-dev · 2022-04-03 13:12
svenyeah, probably a good idea to clean it up#asahi-dev · 2022-04-03 13:08
svenlet me just get rid of those local variables#asahi-dev · 2022-04-03 13:06
svenoh, true#asahi-dev · 2022-04-03 13:01
svenrequiring three inside the binding sounds correct to me then#asahi-dev · 2022-04-03 12:57
svendo we even have different device trees for the 4T models?#asahi-dev · 2022-04-03 12:57
svenhrm, good point. i assumed it was for all of them#asahi-dev · 2022-04-03 12:54
svennote that there are some rtkit api changes#asahi-dev · 2022-04-03 12:54
svenalright, first draft for nvme v2: https://github.com/AsahiLinux/linux/commits/nvme-v2#asahi-dev · 2022-04-03 11:06
svenenabling the MMU is on of the first things it does#asahi-dev · 2022-04-02 16:07
svenah, too bad.#asahi-dev · 2022-04-02 15:03
svenif yes, can you try https://files.svenpeter.dev/15624ee8e938749519ca01a4e2ec99fe68bc95ad7280c49d716a4ce62b9262ab_nvme-intmsmc.patch?#asahi-dev · 2022-04-02 15:02
svencan you reproduce it reliably there?#asahi-dev · 2022-04-02 14:56
svenjannau: my memory is a bit fuzzy, but were you also able to reproduce that weird missed nvme interrupts bugs when there was no spinlock around command submission?#asahi-dev · 2022-04-02 14:01
svenI was thinking to already pre-fill bfr->iova and then shmem_setup() is just "do whatever is required to make bfr useable"#asahi-dev · 2022-04-02 13:54
svenalright, fingers crossed I didn't forget to reply to any of the nvme comments#asahi-dev · 2022-04-02 13:21
svendon't think that works all that well for review comments to my patches which will be sent to me directly ;)#asahi-dev · 2022-04-02 13:01
sveni need to find a better mail client and a way to separate these LKML mails from my other emails *sigh*#asahi-dev · 2022-03-31 06:12
svenstudying physics was also one of my bigger mistakes, but that’s off topic ;)#asahi-dev · 2022-03-29 17:22
svennice!#asahi-dev · 2022-03-29 05:58
svenand no idea what the tunables do#asahi-dev · 2022-03-29 05:58
svenremap allows to map an SID to another SID#asahi-dev · 2022-03-29 05:57
svenyou still map an entire 16K page but when that subpage protection is set only that subset can actually be accessed by the device#asahi-dev · 2022-03-28 17:19
svenbut nvme? lol #asahi-dev · 2022-03-28 17:19
sveni could see it breaking atcphy because it pokes some DP-phy registers #asahi-dev · 2022-03-28 17:18
sven…how??#asahi-dev · 2022-03-28 13:51
svenamarioguy: iirc it disables mostly drm keys and Apple Pay support and stuff like that #asahi-dev · 2022-03-27 15:17
sven_I_ don’t care at all fwiw #asahi-dev · 2022-03-27 15:17
svenI’m not gonna fight that fight#asahi-dev · 2022-03-27 11:05
svenI guess :(#asahi-dev · 2022-03-27 11:00
svenI’m not sure that’ll be enough for the upstream people :(#asahi-dev · 2022-03-27 10:59
svenrqou_: note that you’ll also need to sign off your commits and use a real name (Linux upstream rules)