#asahi-dev · 2024-11-08 13:33
jannauthe status output certainly breaks things but my fusb breakout setup isn't wotking with the mu uart either. tx on device side works but rx might be broken on m4#asahi-dev · 2024-11-08 13:15
davidI am unable to run hypervisor on 15.0.1 m1 m1n1 it throw an exception on CPU 0 EXCEPTION_LOWER/SYNC. does anyone face this problem#asahi-dev · 2024-11-08 13:09
mazas long as m1n1 sticks to 115200, you should be OK (baring bugs in the UART<->USB code).#asahi-dev · 2024-11-08 13:07
jannauyes, although not really interested in booting kernel but getting information from m1n1 and chainloading a new m1n1 during bringup#asahi-dev · 2024-11-08 13:06
mazor do I have the wrong end of the stick?#asahi-dev · 2024-11-08 13:06
mazand I suspect you would have to implement the UART speed change side of things, as you don't want to upload something big at 115200.#asahi-dev · 2024-11-08 13:05
mazthat's how all my devices boot.#asahi-dev · 2024-11-08 13:04
mazif so, then no, I always used the USB2.0 side of the connection.#asahi-dev · 2024-11-08 13:04
mazjannau: is that (for example) loading the kernel using the CS UART?#asahi-dev · 2024-11-08 12:51
jannaumaz: did you ever test m1n1's proxy protocol using UART over CS?#asahi-dev · 2024-11-08 09:20
nickchan(and then there's a new register that became HID11_EL1)#asahi-dev · 2024-11-08 09:20
nickchan> s3_0_c15_c13_0 used to be HID11_EL1 but now it is "BIU_TLIMIT"#asahi-dev · 2024-11-08 07:21
svenso maybe a scheduler makes sense there? dunno #asahi-dev · 2024-11-08 07:20
svenfor nvme specifically, Apple only has a single queue with very limited space for requests #asahi-dev · 2024-11-08 07:19
svenflokli: no idea about that specific scheduler but comparing Apple hardware to common x86 hw generally doesn’t work #asahi-dev · 2024-11-08 04:32
nickchans3_0_c15_c13_0 used to be HID11_EL1 but now it is "BIU_TLIMIT" #asahi-dev · 2024-11-08 04:27
nickchans3_4_c15_c2_6 used to be VMSA_LOCK_EL1 but it is VMSA_LOCK_EL12 in current chips#asahi-dev · 2024-11-08 04:27
nickchansome sysregs has changed meaning across chip generations, so apple_regs.json kind of getting tricky#asahi-dev · 2024-11-07 23:06
flokliIs there a ml thread or something to link to?#asahi-dev · 2024-11-07 23:05
floklijannau: the articles suggest not doing any scheduling on fast nvme is better than using CPU cycles trying to schedule.#asahi-dev · 2024-11-07 22:13
jannauflokli: mq-deadline is used as default by the kernel since the apple nvme controller has only a single hw queue#asahi-dev · 2024-11-07 19:55
flokli(or maybe it's some udev foo that distros ship which doesn't cover asahi yet, I was just surprised it doesn't bypass the scheduler as other ssds do)#asahi-dev · 2024-11-07 19:54
floklihmmh, I'm surprised /sys/block/nvme0n1/queue/scheduler shows mq-deadline selected for the m2air SSD. My thinkpad automatically sets this this to none, according to https://stackoverflow.com/q/27664334 this is to be expected for SSDs. Are the drivers missing some bits selecting that scheduler?#asahi-dev · 2024-11-07 13:07
chaos_princessprobably#asahi-dev · 2024-11-07 13:06
ambyjklso this is not really in the scope of this project, but would a paravirtualized driver for encode inside a linux vm on macos would be an easier task, considering there is no reverse-engineering to be done#asahi-dev · 2024-11-07 13:05
ambyjklwell i kind of had a feeling that is indeed the case#asahi-dev · 2024-11-07 13:04
chaos_princessif you are wondering - approx. nothing of that is applicable to encode#asahi-dev · 2024-11-07 13:03
chaos_princessyes#asahi-dev · 2024-11-07 13:03
ambyjklare you referring to this https://github.com/eiln/avd#asahi-dev · 2024-11-07 13:03
ambyjkli see, that's understandable#asahi-dev · 2024-11-07 13:02
chadmednot only is the RE work further along, it's more important to more people#asahi-dev · 2024-11-07 13:02
chaos_princessthat is true#asahi-dev · 2024-11-07 13:01
chadmeddecode is almost certainly going to come before encode no matter what happens tbh#asahi-dev · 2024-11-07 13:01
chaos_princessi am not saying that you can't ask devs to prioritize your specific thing, after all, everyone has a price, only that it is not very likely#asahi-dev · 2024-11-07 13:00
chadmed^#asahi-dev · 2024-11-07 12:59
chaos_princesswhat he meant is that if you throw money there, it will go to the general development, not a specific features#asahi-dev · 2024-11-07 12:58
ambyjkl@chadmed https://asahilinux.org/support/ says otherwise#asahi-dev · 2024-11-07 12:22
chadmedwherein fixing usb would break pcie and then fixing pcie would break dp alt mode etc etc#asahi-dev · 2024-11-07 12:21
chadmedmore recently the fight had turned to whack-a-mole fixes for various things#asahi-dev · 2024-11-07 12:21
chadmedbunch of very deep core code had to be completely rewritten and generalised #asahi-dev · 2024-11-07 12:21
chadmedyeah that was one of the biggies#asahi-dev · 2024-11-07 12:20
chaos_princessfrom what i remember, the problem is that linux's thunderbolt stack assumes that you are using an intel controller, and m1 is not like that#asahi-dev · 2024-11-07 12:20
chadmedwe've got to be up to a cumulative 700,000 years of release delay assuming s v e n is still keeping track :p #asahi-dev · 2024-11-07 12:18
chaos_princesslast commits 3-ish months ago if you know where to look#asahi-dev · 2024-11-07 12:14
chadmedit was being worked on actively a while ago, but not publicly. not sure what state it's in at the moment#asahi-dev · 2024-11-07 12:05
MaryHey so I was wondering about the state of thunderbolt support on M1, is anyone working on that? If not, would it be okay if I try to help a bit on that side?#asahi-dev · 2024-11-07 11:46
chadmedwe typically dont take cash for features#asahi-dev · 2024-11-07 11:44
ambyjkli see, i'm interested in this feature and I can fund its development#asahi-dev · 2024-11-07 10:05
chaos_princessi do not think anyone is actively working on it#asahi-dev · 2024-11-07 10:04
chadmedWhen It's Done