#
…

2026-01-25

00:02jmax has quit [Quit: No Ping reply in 180 seconds.]
00:02jmax has joined #asahi-dev
00:03hdbngr has quit [Remote host closed the connection]
00:03hdbngr has joined #asahi-dev
00:04fl0_id__well at least smp seems to come up, even if init is not working...
00:07fl0_id__smp: Brought up 1 node, 16 CPUs [ 0.044822] SMP: Total of 16 processors activated.
00:13hdbngr has quit [Remote host closed the connection]
00:13hdbngr has joined #asahi-dev
00:20john-cabaj has quit [Ping timeout: 480 seconds]
00:24fl0_id__maybe once serial is working, this can be submitted as a minimal device tree
00:24fl0_id__(and baremetal)
00:27fl0_id__yay i have a shell :) turns out it somehow didn't like the /init, but just specifyig alpine /sbin/init worked
00:27fl0_id__now ideally I just need glasses or a bigger font lol
00:28hdbngr has quit [Remote host closed the connection]
00:28hdbngr has joined #asahi-dev
00:29fl0_id__(and i had specified init=/init option)
00:32fl0_id__baremetal booting works too, just no serial :/
00:32fl0_id__up to initramfs shell
00:44fl0_id__I compiled the current device tree and other changes I have here: https://gist.github.com/step21/7433ac5471d0eb8f67308fc6deca9202
00:45fl0_id__not all of those changes might be necessary, but in earlier attempts cpufreq for example was necessary. not sure if asahi-wip or device tree might mean it works without those changes, will test tomorrow probably
00:48hdbngr has quit [Remote host closed the connection]
00:48hdbngr has joined #asahi-dev
00:55mixi has quit [Ping timeout: 480 seconds]
00:58hdbngr has quit [Remote host closed the connection]
00:58hdbngr has joined #asahi-dev
01:05mixi has joined #asahi-dev
01:18midou has joined #asahi-dev
01:21fl0_id__ has quit [Ping timeout: 480 seconds]
01:26midou has quit [Ping timeout: 480 seconds]
01:48hdbngr has quit [Remote host closed the connection]
01:48hdbngr has joined #asahi-dev
02:08ansg191 has quit [Remote host closed the connection]
02:09ansg191 has joined #asahi-dev
02:28lukeleppan has quit [Quit: The Lounge - https://thelounge.chat]
02:31midou has joined #asahi-dev
02:34lukeleppan has joined #asahi-dev
02:38hdbngr has quit [Remote host closed the connection]
02:38hdbngr has joined #asahi-dev
03:06hexdump02 has joined #asahi-dev
03:08hexdump01 has quit [Ping timeout: 480 seconds]
03:10tobhe has joined #asahi-dev
03:16tobhe_ has quit [Ping timeout: 480 seconds]
03:38hdbngr has quit [Remote host closed the connection]
03:38hdbngr has joined #asahi-dev
03:44lukeleppan has quit [Quit: The Lounge - https://thelounge.chat]
03:50lukeleppan has joined #asahi-dev
04:03hdbngr has quit [Remote host closed the connection]
04:03hdbngr has joined #asahi-dev
04:05hexdump01 has joined #asahi-dev
04:06hexdump02 has quit [Ping timeout: 480 seconds]
04:25pb17 has quit [Ping timeout: 480 seconds]
04:33hdbngr has quit [Remote host closed the connection]
04:33hdbngr has joined #asahi-dev
04:38pb17 has joined #asahi-dev
04:50midou has quit [Ping timeout: 480 seconds]
04:50nora_ has joined #asahi-dev
04:52nora has quit [Ping timeout: 480 seconds]
05:07pthariensflame has joined #asahi-dev
05:08pthariensflame has quit []
05:08lukeleppan has quit [Quit: Ping timeout (120 seconds)]
05:09lukeleppan has joined #asahi-dev
05:09lukeleppan has quit []
05:14lukeleppan has joined #asahi-dev
05:17midou has joined #asahi-dev
05:23hdbngr has quit [Remote host closed the connection]
05:23hdbngr has joined #asahi-dev
06:22pb17 has quit [Ping timeout: 480 seconds]
06:33hdbngr has quit [Remote host closed the connection]
06:33hdbngr has joined #asahi-dev
06:37pb17 has joined #asahi-dev
06:53hdbngr has quit [Remote host closed the connection]
06:53hdbngr has joined #asahi-dev
07:28hdbngr has quit [Remote host closed the connection]
07:28hdbngr has joined #asahi-dev
07:32chadmed has quit [Quit: Konversation terminated!]
07:32chadmed has joined #asahi-dev
07:32chadmed has quit []
07:32chadmed has joined #asahi-dev
07:33chadmed has quit []
07:33chadmed has joined #asahi-dev
07:38hdbngr has quit [Remote host closed the connection]
07:38hdbngr has joined #asahi-dev
07:44lukeleppan has quit [Quit: The Lounge - https://thelounge.chat]
07:49mattia013 has joined #asahi-dev
07:49lukeleppan has joined #asahi-dev
08:21midou has quit [Ping timeout: 480 seconds]
08:38pb17 has quit [Ping timeout: 480 seconds]
08:50pb17 has joined #asahi-dev
08:53hdbngr has quit [Remote host closed the connection]
08:53hdbngr has joined #asahi-dev
09:01midou has joined #asahi-dev
09:02lukeleppan has quit [Quit: The Lounge - https://thelounge.chat]
09:03hdbngr has quit [Remote host closed the connection]
09:03hdbngr has joined #asahi-dev
09:08lukeleppan has joined #asahi-dev
09:28hdbngr has quit [Remote host closed the connection]
09:28hdbngr has joined #asahi-dev
09:40chadmed has quit [Quit: Konversation terminated!]
09:41chadmed has joined #asahi-dev
09:48hdbngr has quit [Remote host closed the connection]
09:48hdbngr has joined #asahi-dev
10:04integralpilot has joined #asahi-dev
10:05integralpilotyay for shell, great work fl0_id! :)
10:05integralpilotjust posted updated dockchannel/MTP draft to branch
10:05integralpilotdoing SMC power stuff now and will send v4 on list shortly
10:10svendo we know if MTP does anything else that's not hid btw?
10:11svenbecause i've been looking into what dockchannel actually is (because of that new dockchannel uart) and that dockchannel as a parent node with children nodes seems rather unusual to me at this point
10:11sveni think dockchannel is actually something a bit similar to a mailbox
10:12integralpiloti don't think MTP does anything non-HID, i think it's point is to just regular internal HID events, though might be wrong
10:12integralpilotwe could change the structure if the current way doesn't look right, what would you propose?
10:12sveni'm not sure yet :D
10:13sveni'm just trying to understand/discuss what that thing actually is
10:13jannaumy first guess is mtp stands for is multi touch processor
10:13jannauand dockchannel looks like a mailbox to me as well
10:13svenhttps://github.com/apple-oss-distributions/xnu/blob/main/pexpert/pexpert/arm/dockchannel.h dockchannel itself seems to be really like some mailbox with multiple channels
10:13svenand we don't do the parent/child node thing for mailboxes either
10:15svennow i'm not sure we want to use mailbox subsystem again, but i think the issues with the asc mailbox was mostly around atomic context handling iirc
10:16integralpiloti think we could just have the HID device as a child of soc and then refer to the dockchannel within mboxes
10:17svenyeah, i wonder if a single "mtp coprocessor" that uses both the asc mailbox and the dockchannel mailbox and also does all the HID stuff is the way to go
10:17svendunno though
10:17svenwe'll also have to take care of downstream u-boot/openbsd if we change the dts representation but we might just be able to keep both variants in our downstream releases
10:18hdbngr has quit [Remote host closed the connection]
10:18hdbngr has joined #asahi-dev
10:18integralpilotwell we've already changed the dts representation so far so will need to change u-boot anyway, so might as well make it perfect
10:19svengood point
10:19jannauit wasn't just the atomic context. mailbox was overall a bad fit for asc: https://lore.kernel.org/asahi/20230328-soc-mailbox-v1-0-3953814532fd@marcan.st/
10:19svenstill not sure about the interrupt-controller part of dockchannel, will have to understand what it actually does and how it's used
10:20nickchanabout the generic part the more i look at it the more spmi looks like i2c
10:21nickchanbut maybe that's just an artifact of how spmi is used on apple silicon
10:21pb17 has quit [Ping timeout: 480 seconds]
10:23integralpilotsven: it's its own interrupt controller because it only has one interrupt line to AIC (as opposed to the many internal events that can occur) and uses mulitplexing to then indicate which even actually occured, rather than having a sepereate AIC interrupt for each of these
10:23nickchanboth contained a slave id and an address
10:23integralpilotto my understanding, at least
10:23svenintegralpilot: sure, is that used for e.g. hid though somehow?
10:24nickchanintegralpilot: do you expect multiple logically distinct hardware behind dockchannel?
10:24svenmaybe i'm blind but i just see the interrupts in the dts for apple,t8112-dockchannel-hid but don't see that code that uses them
10:25svenoh wait, lol
10:26svenokay, that's weird
10:26nickchancause many hardware that's at least moderately complex has a status register that indicate which events are firing
10:26svendockchannel itself requests the interrupts
10:26nickchanthat does not make them interrupt controllers from the point of linux
10:27nickchanexample: dwc2 and dwc3
10:29jannauif you chain devices on like for dockchannel downstream an irq controller is good fit
10:30jannaubut I think those are more status flags for mailbox queues than an interrupt controller
10:30sventhe more i look at it the more dockchannel looks like a weird mbox that just needs some callbacks
10:30svenand doesn't need to be an irq controller
10:32nickchani wonder where the dockchannel "uart" go to then
10:32nickchanmaybe the other end is whatever coproc that brings up debug usb
10:33svenpretty sure that's what happens
10:34jannaumaybe aop? the dockchannel-uart node has a 'max-aop-clk' property
10:34svenall mtp and hid stuff happens in the apple,t8112-mtp-hid node, dockchannel is just a dumb mailbox
10:34sven(maybe outside the mbox system if there's a mismatch like for asc)
10:35nickchanwhatever description is used it must be compatible with a serial driver consumer
10:36svensure, anything that makes you think what i just made up isn't?
10:36integralpilotoh sorry was AFK!
10:36svenno worries :)
10:36svenirc is slow
10:36svenit's not unusual that you don't replies for hours at times
10:36nickchanit looks compatible to me i think
10:36svenit is.
10:37svenmight only be a bit strange to have a serial without a reg but oh well
10:37noopwafeldo I need to bring up all the pmgr domains before handing over to linux? or am I missing some way to get linux pm to bring up domains during probe? I stared at the docs and the api seems, uh, very suspend-centric
10:37svennoopwafel: if your device has a single domain defined in the dts it'll just work
10:38integralpilotno, you don't. if you add it to power-domains property on each node it should turn them on as needed
10:38svenif it has more than a single domain you need to do some manual work
10:38noopwafelsven: any pointers to something existing which does the manual work?
10:38svenlooking for it right now :D
10:38sventhere's code in nvme that does it, but there's a common helper function now
10:39noopwafelintegralpilot: that was what I was hoping but nop :(
10:40jannaunoopwafel: devm_pm_domain_attach_list
10:41svenyou don't want the PD_FLAG_NO_DEV_LINK flag though, i have that there because thunderbolt is special
10:41noopwafelthanks (both)!
10:41jannauit's not handled automatically for multiple domains since the ordering / timing could be device specific
10:42svenlike for thunderbolt :D
10:44jannauI don't even understand how dockchannel-uart is supposed from the ADT
10:44integralpilotsven: the way you proposed getting rid of the IRQ stuff would work for HID, but it would probably break serial as i think serial8250 (or similar) expects to be able to bind to an IRQ
10:45integralpilotwith the system of it acting as its own interrupt controller & abstracting away the multiplexing it's easier to say, oh serial, your IRQ is #3 (or whatever) on the dockchannel
10:45jannauit has disjoint regs from dockchannel-mtp and no interrupt or phandle
10:45integralpilotif you make it not a contorller anymore, you can't give an IRQ number to the serial driver, as the AIC one is for all events
10:45svensure, but the serial driver would just dockchannel_await or whatever
10:46svenjannau: oss xnu contains code to drive that dockchannel-uart fwiw
10:47lukeleppan has quit [Quit: The Lounge - https://thelounge.chat]
10:48svenmaybe i should continue figuring out that debug usb thing so that we can actually test that dockchannel uart
10:48integralpilotsven: i guess that could work, but we would need to make sure the callbacks etc. are in a threaded IRQ context
10:48integralpilotthat would probably actually work better
10:49svenat least right now i think dockchannel should really be a mailbox-like thing, possibly inside the mbox system if it doesn't run into the same problems asc mailboxes did
10:49integralpilothmm okay, i'll prototype re-writing it that way
10:49integralpilotjust finishing power v4 first
10:49svenonly if you think it's reasonable, i'm not really sure about any of this :D
10:50svenit's just that the correct parent/subnode thing and then calling dockchannel_init which requests irqs from the irq controller defined in the same file looks very weird to me
10:50integralpiloti've looked into it more (seeing how the mailbox subsystem works because i wasn't sure!) and it seems actually like a better way of handling it and more ergonomic
10:50svenjust have to make sure we don't run into this if we put into the mailbox system <jannau> it wasn't just the atomic context. mailbox was overall a bad fit for asc: https://lore.kernel.org/asahi/20230328-soc-mailbox-v1-0-3953814532fd@marcan.st/
10:51svenwe might also need more specialized callbacks and then maybe apple,dockchannel = <&dockchannel 0> is better, dunno
10:53pb17 has joined #asahi-dev
10:54integralpilotreading through that email many of the issues marcan brought up probably apply to dockchannel too now that i think of it
10:55svenour mailbox is in drivers/soc/apple with a customer interface now fwiw
10:55svenwe just re-use the original mbox device tree bindings
10:55sven*custom
10:57integralpilotit probably works that way, i think (i'll have to prototype and see which i'm doing rn)
10:57integralpilotideally we can probably get it working just using mboxes = <&dockchannel 1>; etc. i think
10:58integralpilotif we're doing serial as well, it also might be easier to first upstream dockchannel & serial (which is simple), THEN upstream HID
10:58svenif we're using different callbacks apple,dockchannel = ... might be better
10:58hdbngr has quit [Remote host closed the connection]
10:58hdbngr has joined #asahi-dev
10:58sveni'll see if i can figure out how to talk to debugusb so that we see the other side of that serial
11:00integralpilotdo you think in the short term you might get that working? then i'll focus more on generic dockchannel stuff and (if it's okay) include the much-simpler-than-HID serial stuff in the patchset as From: you because it will get dockchannel core approved much quicker that way. otherwise i'll focus on HID as well and send it in a patchset with dockchannel core
11:01svennot sure, when i tried to speak to debugusb yesterday i somehow got my mac mini into recovery mode :D
11:01svenwill have to find a keyboard and mouse to get it back to m1n1
11:01sveni'll bet at fosdem next weekend, so earliest probably the weekend after that
11:02sven*be at
11:02integralpilotsounds good, thank you for helping with this!
11:06integralpilotno rush with this at all though, whenever works is fine. i might try my hand at some debugusb stuff too but summer holidays are ending soon so might not have as much time
11:07nickchandockchannel uart would also need to work in earlycon
11:08nickchanshould be possible though with polling functions
11:08hdbngr has quit [Remote host closed the connection]
11:08hdbngr has joined #asahi-dev
11:11sventhere's debugusb code in libirecovery (though not serial) and there's a applekis.kext in 12.3 and 12.3.1 kernelcaches that should have support for serial
11:12nickchansending {0x5ac8012, 0x824606} in macvdmtool puts machine into debug usb mode
11:14sventhe fun part is that you can also do this from m1n1 for the device itself, so we could just do that and get something like uart without extra hardware even for early bringup
11:18hdbngr has quit [Remote host closed the connection]
11:18hdbngr has joined #asahi-dev
11:20jannauintegralpilot: I'm still puzzled why you need `nonposted-mmio;`
11:28hdbngr has quit [Remote host closed the connection]
11:28hdbngr has joined #asahi-dev
11:29nickchanit's also possible to DFU from m1n1 running on the device since A10
11:29svenls /dev/cu.kis-100000-ch-*
11:29sven ls /dev/cu.kis-100000-ch-*
11:30sven"/dev/cu.kis-100000-ch-0 /dev/cu.kis-100000-ch-1 /dev/cu.kis-100000-ch-2 /dev/cu.kis-100000-ch-3"
11:30svenso this just works with macos 26 apparently
11:30nickchanit kind of has to so iphones can restore
11:30nickchanat least from apple's perspective
11:30svenno, it doesn't
11:30sventhere's no need for serial for that to work
11:35svencat /dev/cu.kis-100000-ch-0
11:35svenHello, World!
11:35svenso that was easy
11:35svenfor c in "Hello, World!\n": write32(0x23d128000 + 0x4004, ord(c))
11:43fugi5 has quit [Quit: bye]
11:47fugi5 has joined #asahi-dev
11:48svendata is read from offset 0x401c which returns 0x0000aabb where aa is the next byte and bb the number of bytes left in the fifo
11:54lukeleppan has joined #asahi-dev
11:56tobhe_ has joined #asahi-dev
11:59nickchanhow about the other channels
12:00svenno idea what they are, also don't really care tbh
12:01tobhe has quit [Ping timeout: 480 seconds]
12:02mazit'd be awesome to be able to remove the need for extra HW (other mac or CS) for people to do early bringup, or even normal serial console.
12:03mazis this supported all the way to M1?
12:03sveni tested this on my m1
12:03pitusthow long is the queue?
12:03svenjust need to figure out an easy way to sniff usb traffic now because i can't find my usb2 sniffer anymore
12:04svenand then implement a small tool that just uses libusb to access that serial
12:04svenpitust: no idea
12:04pituste.g. what happens if you write 128k into dockchannel (from C and not python)
12:04mazI hope there's an interrupt associated with the RX side.
12:04svenpitust: why would you do that?
12:05svenmaz: there is afaict, there are also status register for how much space is left i bet
12:05pitustsven: to check how long the queue is
12:05svenfeel free to try it yourself
12:05pitusti only have one mac though
12:06sventhen please don't backseat reverse engineer this
12:07mazsven: a real UART then. wish we knew that 5 years ago, would have saved me some effort! :)
12:07svenseems to work exactly the same as the downstream dockchannel thingy we have to HID on M2+
12:08svenso yeah, interrupts and status registers for how much space is left, etc
12:08sven*we have for HID
12:09sven0x800 bytes FIFO I think
12:09mazmassive!
12:10svennow what would be really great if we could also issue a reboot command through this, but that's work for later
12:11pb17 has quit [Ping timeout: 480 seconds]
12:33nickchanwill look into t8012's dockchannel
12:40slendidev has quit [Ping timeout: 480 seconds]
12:43slendidev has joined #asahi-dev
12:45integralpilot has quit [Ping timeout: 480 seconds]
12:48hdbngr has quit [Remote host closed the connection]
12:48hdbngr has joined #asahi-dev
12:57pb17 has joined #asahi-dev
13:02alxdrl has joined #asahi-dev
13:03hdbngr has quit [Remote host closed the connection]
13:03hdbngr has joined #asahi-dev
13:06integralpilot has joined #asahi-dev
13:07fl0_id__ has joined #asahi-dev
13:07integralpilotpower v4 just sent!
13:07integralpilotOMG it stuffed up sorry!
13:07maz(no cover subject)
13:07integralpilotyes just noticed that!
13:08integralpilotb4's state broke and i had to reset it again, must have broken something sorry
13:11svenapple makes sniffing usb on macos surprisingly hard :/
13:13hdbngr has quit [Remote host closed the connection]
13:13hdbngr has joined #asahi-dev
13:15nickchanon 12.3 the macos kernel would run binary signed with custom root certs without entitlement restriction
13:16nickchanwell almost every entitlement anyway
13:16integralpilotwhich ones doesn't it give?
13:17nickchanintegralpilot: "com.apple.private.cs.debugger", "dynamic-codesigning"
13:17integralpilotah, makes sense
13:17nickchanthis is just an easy to use bug
13:17integralpilotwhoever found that probably got a pretty big bounty lol
13:18integralpilotthat's quite an oversight lol, good that it works in your favour for testing though
13:18nickchanwas thinking about it since sven said macos 12.3 before but now realized hes on macos 26 not 12.3
13:20integralpilotcan confirm NVMe works on older firmware, yay!
13:20integralpilotgoing to make an ext4 parition and put a real distro there and then have alpine act as a pivot initramfs and just chroot into it and load the real distro
13:20nickchanso it's a firmware version compatibility issue
13:20nickchanusb would need that spmi mess
13:21nickchanfor wifi i think those tunables are complex
13:21nickchanpcie tunables
13:21integralpilotatp i'm going to use openSUSE jeOS, unless anyone else has any other recommendations for the "real" OS?
13:21nickchanwhy not FAR?
13:22nickchanfedora asahi remix
13:22svenoh no, i got a trace. i was just annoyed how complicated that was
13:23integralpilotwanted to try a slimmed down OS first, might try fedora minimal arm then i guess, and if it works try the full fedora asahi
13:23sventl;dr: disable SIP, ifconfig XHC0 up and then wireshark
13:23j`eyintegralpilot: far has a minimal variant
13:24integralpilotoh really, perfect, i'll try that then!
13:24nickchanif cpufreq is working, then llvmpipe shouldn't be too terrible
13:24integralpilotyeah, was going to use that
13:25nickchanthere's a clout factor in it :)
13:25integralpilotprobably just XFCe for now, i'd imagine full KDE would not perform well
13:25nickchanit actually performed ok
13:25nickchanfor the longest time in Asahi Linux under llvmpipe
13:25nickchanuntil the opengl 2.1 drivers came around
13:25integralpilotoh really, that's good. will try that then
13:25chaos_princessintegralpilot: if you don't use dcp and use raw framebuffer instead, it is surprisingly usable
13:26chaos_princesswhich, well, you will :P
13:26nickchanthere's no chance that dcp would work
13:26integralpilotthat's great then!
13:26integralpilotyes, haven't even tried DCP
13:26integralpilotjust using framebuffer for now
13:26sveneven dcp and llvmpipe was surprisingly usable at ccc
13:27integralpiloti'd imagine a lot changed in 13.5 -> 14.latest given how much they change DCP
13:27nickchanwell, maybe it could work if you use 13.5/13.6.2 specifically
13:27sventook me almost a day to realize i built the kernel without rust support :D
13:27nickchansince base m3 does have those
13:27svenonly noticed because there was occasional lag and because all cores were somehow busy at ~30% all the time
13:27integralpilothappened to me why i was testing SEP, i was like why is it not probing???
13:27nickchansep *must* be stable
13:28integralpilotthe problem was it's a rust driver and i hadn't enabled rust support lol
13:28integralpilotbut weirdly, the driver did not work on M3. neither does the way m1n1 talks to sep to try to get a KASLR RNG
13:28integralpiloti think they changed the way BootROM talks to give early randomness and boot firmware
13:29integralpilot*SEP BootROM
13:29nickchanwould be the first change since it's added in a10
13:30integralpilotoh weird then! will look into SEP more when i have time
13:30nickchannote the mailbox hardware changed in a11, on a10 messages are 32+32 and on a11 to m2 it's 64+32
13:30nickchanbut sep uses only 64 bit anyways so the high level commands are identical
13:30svenhuh, especially the KASLR RNG breaking is weird
13:31integralpilot has quit [Quit: integralpilot]
13:32nickchanhigh level seprom communication is the same since A7 even
13:32nickchanjust the trng command is added in a10
13:33nickchanintegralpilot: do you have the unexpected getrand reply?
13:34nickchanalas of course this trng is also the source of misery with racing seprom firmware load on a10
13:38hdbngr has quit [Remote host closed the connection]
13:38hdbngr has joined #asahi-dev
13:52fl0_id__integralpilot - could I take a look at a more complete device tree of yours?
13:53slendidev has quit [Read error: Connection reset by peer]
13:54slendidev has joined #asahi-dev
13:54svenso yeah, looking at dockchannel this thing definitely wants to be something like a mailbox and not that weird parent/subnode things with an interrupt controller
13:55svenanyone got a m2 device tree for me? i suspect that the uart dockchannel fifo is separate from the mtp dockchannel fifo
13:55sveni think the uart dockchannel one is directly connected to whatever that "kanzi in system" co-proc is, and the mtp one to, well, MTP
13:58jannausven: since I already prepated that: https://paste.debian.net/hidden/c8d41c53
13:58hdbngr has quit [Remote host closed the connection]
13:58hdbngr has joined #asahi-dev
13:58sventhanks, that matches what i suspected
13:59svenfrom what i can tell the uart implementation is directly hacked into xnu to allow something similar to earlycon
14:00nickchanthat's just how xnu originally is
14:00nickchanbefore more and more functions are moved into kext over time
14:00svennot my point
14:00svenmy point is that the dockchannel kext isn't used for that early uart
14:00svenit's all open-coded in xnu
14:01svenso this all really looks like two instantiations of the same HW block
14:01nickchanthere's seem to be code for apple uart and bcm2837 uart (since removed) in there
14:01svenjust that one is hooked up to KIS on the other end and the other to MTP
14:02nickchanand also pl011 uart used in virtualmac2,1
14:04nickchanadt seem to be missing dockchannel uart interrupt
14:09svenyeah, but we can figure that one out
14:09sveni assume it exists
14:31jannauj493 has an interrupt in dockchannel-uart
14:34jannaunot sure why the one I paste has none. looked at a few others and they have all interrups
14:37nickchanj105a has interrupt in dockchannel-uart but no j680
14:39nickchanthe ones that don't have interrupts all have enable-sw-drain = 1
14:41jannauj274 has interrupts and enable-sw-drain = 1
14:44jannauthe same for j473, j314c, j375d
14:48hdbngr has quit [Remote host closed the connection]
14:48hdbngr has joined #asahi-dev
14:50alxdrl has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
14:56fl0_idj514 has one too
14:58fl0_iddoes asahi-wip have dockchannel-mtp code that keyboard on m3 might work? I found some code about it, but not sure if that's enough or if integralpilot added anything not on that branch yet
14:59lederstr- has quit [Quit: ZNC 1.8.2 - https://znc.in]
15:00lederstrumpf has joined #asahi-dev
15:22fl0_id__ has quit [Ping timeout: 480 seconds]
15:44svenbah, this usb trace isn't complete afaict, missing all the host->kis data. time to get a hardware usb sniffer again :D
15:44sventhe protocol looks simple enough but i think i'm missing some "actually enable serial" packet
16:44pb17 has quit [Ping timeout: 480 seconds]
16:58hdbngr has quit [Remote host closed the connection]
16:58hdbngr has joined #asahi-dev
17:13pb17 has joined #asahi-dev
17:23hdbngr has quit [Remote host closed the connection]
17:23hdbngr has joined #asahi-dev
17:43hdbngr has quit [Remote host closed the connection]
17:43hdbngr has joined #asahi-dev
18:23hdbngr has quit [Remote host closed the connection]
18:23hdbngr has joined #asahi-dev
18:44___nick___ has joined #asahi-dev
18:45___nick___ has quit []
18:47___nick___ has joined #asahi-dev
18:53hdbngr has quit [Remote host closed the connection]
18:53hdbngr has joined #asahi-dev
19:34catred has joined #asahi-dev
19:37nela has quit [Ping timeout: 480 seconds]
19:45catred has quit [Quit: Leaving]
19:53hdbngr has quit [Remote host closed the connection]
19:53hdbngr has joined #asahi-dev
19:55axel_fl has joined #asahi-dev
20:00axel_fl has quit [Quit: Konversation terminated!]
20:07michaelrostom has joined #asahi-dev
20:08nela has joined #asahi-dev
20:08hdbngr has quit [Remote host closed the connection]
20:08hdbngr has joined #asahi-dev
20:09ethantwardy has quit [Ping timeout: 480 seconds]
20:28lukeleppan has quit [Quit: The Lounge - https://thelounge.chat]
20:33lukeleppan has joined #asahi-dev
20:41axel_fl has joined #asahi-dev
20:41axel_fl has quit [Remote host closed the connection]
20:42alxdrl has joined #asahi-dev
20:42noopwafelTTY> pcie: Port 0 max speed = 2
20:42noopwafelTTY> pcie: Initialized controller 0
20:43fl0_idnoopwafel nice
20:44noopwafelt8122 pcie seems basically t602x-but-less-cursed
20:48hdbngr has quit [Remote host closed the connection]
20:48hdbngr has joined #asahi-dev
20:51nela7 has joined #asahi-dev
20:56nela has quit [Ping timeout: 480 seconds]
20:56nela7 is now known as nela
20:58hdbngr has quit [Remote host closed the connection]
20:58hdbngr has joined #asahi-dev
21:02___nick___ has quit [Remote host closed the connection]
21:04john-cabaj has joined #asahi-dev
21:05MikeM64 has quit [Read error: Connection reset by peer]
21:06MikeM64 has joined #asahi-dev
21:12john-cabaj has quit [Ping timeout: 480 seconds]
21:13chadmed has quit [Quit: Konversation terminated!]
21:13john-cabaj has joined #asahi-dev
21:14john-cabaj has quit []
21:14john-cabaj has joined #asahi-dev
21:22john-cabaj has quit [Ping timeout: 480 seconds]
21:23ipatch_ has quit [Read error: Connection reset by peer]
21:28hdbngr has quit [Remote host closed the connection]
21:28hdbngr has joined #asahi-dev
21:31ipatch has joined #asahi-dev
21:32svennow how about usb and then thunderbolt? 😇
21:41michaelrostom has quit [Ping timeout: 480 seconds]
21:45Shizwe're still debugging pcie stuff for now :p
21:48hdbngr has quit [Remote host closed the connection]
21:48hdbngr has joined #asahi-dev
21:49michaelrostom has joined #asahi-dev
21:50fl0_idsven would you look over a first draft for a minimal device tree to submit upstream, sometime this week?
21:51svencan't really send that initial dt upstream before the AIC bindings are at least reviewed
21:51sventhe merge window will also open next monday, submitting after that probably makes more sense
21:52svenbut sure, feel free to post a tree (i.e. not just the device tree, the entire branch based on the latest -rc1)
21:55michaelrostom has quit [Quit: Konversation terminated!]
21:57alxdrl has quit [Quit: Textual IRC Client: www.textualapp.com]
21:58alessaba has joined #asahi-dev
21:58hdbngr has quit [Remote host closed the connection]
21:58hdbngr has joined #asahi-dev
22:18hdbngr has quit [Remote host closed the connection]
22:18hdbngr has joined #asahi-dev
22:33integralpilot has joined #asahi-dev
22:34integralpilotomg yay, great job on pcie! :)
22:34fl0_idsven right aic would need to be there. it will take me some time anyway, so might as well start
22:35integralpilotintegralpilot: sure - would you prefer t8112 (which is also based upon jannau's work) or t6030 (which has less stuff as i mainly use t8112 for testing)
22:35integralpilotwait meant to ping fl0_id, sorry
22:35integralpilotwe will probably in the future need to merge our t6030 & t6031 work together into a t603x DT like M2 has
22:36fl0_idintegralpilot I don't really mind, t6030 might be closer. for now I found the gist of t8122.dtsi you posted on github which might help already...
22:36fl0_idintegralpilot yeah for sure at some point. I also saw that t62xx has different subparts
22:37integralpilotthat gist is ONLY minimum required for boot & MTP as its a minimum reproducer for the issue requiring nonposted-mmio
22:37integralpilotits not the full DT and most stuff is removed
22:38integralpilotoh also, i've learnt my lesson about sending patches late at night when i'm tired
22:38hdbngr has quit [Remote host closed the connection]
22:38hdbngr has joined #asahi-dev
22:38integralpilothave to send v6 now cause i forgot the fs header (it did work fully locally for me though which was weird, might have included it as a byproduct of some configuration-specific header) :(
22:40noopwafelintegralpilot: well the linux side doesn't work yet but..
22:41noopwafelit seems almost(?) there
22:43noopwafelfor my t8122 devicetree changes, is "give patch to jannau" the best approach?
22:46jannauI'm wondering whether we should start adding the t8122 dt to the downstream kernel
22:46svenideally we just send an initial series with all the m3 variants. can just be aic, serial and pmgr and similar very basic stuff
22:47sventhen once we have that base structure we can start extending them with all the other hw as it’s brought up
22:47jannauit seems to be a little cumbersome to have these distributed trees
22:47noopwafelit would be great to have it in downstream.
22:48hdbngr has quit [Remote host closed the connection]
22:48jannauI'm not sure that upstream first is a practical idea for device trees
22:48hdbngr has joined #asahi-dev
22:49jannauI think it's acceptable if we require upstreamable commits
22:51jannauand ensure that we send out patches quickly
22:52integralpilot has quit [Ping timeout: 480 seconds]
22:56integralpilot has joined #asahi-dev
22:58sventhe only thing we should definitely avoid is adding un-reviewed device tree bindings and nodes if it's more than a simple compatible
22:58svenwe've had enough pain with that and we're still not through with everything
22:59integralpilotnickchan: "SEP: couldn't get enough random bytes for KASLR seed"
23:00integralpilotno message about unexpected reply
23:01integralpiloti'll put some more logging in and see why its failing
23:01jannauack, I was thinking of just adding the new soc compatibles required to make make dtbs_check pass. we could even require to those simple dt binding changes as commits and only merge dt changes if make dtbs_check passes for m3+
23:01john-cabaj has joined #asahi-dev
23:02svensounds good to me
23:02jannaunew dt-bindings (or significant changes) should be send upstream first
23:03hdbngr has quit [Remote host closed the connection]
23:03hdbngr has joined #asahi-dev
23:08integralpilot has quit [Remote host closed the connection]
23:08integralpilot has joined #asahi-dev
23:12fl0_idintegralpilot yeah I know it's minimal, but the mtp stuff was still informative
23:16fl0_idintegralpilot isn't that SEP / KASLR msg normal? I got that always, but afterwards KASLR said it was initiliazed
23:17svenno, that's not normal
23:17svenit means something changed on M3 with SEP that we have to figure out
23:17sventhere's a fallback to re-use some random data from apple's device tree iirc
23:18hdbngr has quit [Remote host closed the connection]
23:18hdbngr has joined #asahi-dev
23:24fl0_idsven ok, thx
23:25integralpilot has quit [Remote host closed the connection]
23:25jannauhuh, the base m3 has 4 SoC USB4 ports and 2 dcpext if installed in an imac
23:25integralpilot has joined #asahi-dev
23:26svenhuh, but no when installed in anything else?
23:28jannauno, but to be fair the other devices are laptops
23:32jannauhttps://support.apple.com/en-us/117734 just mentions 1 external display and 2 usb4 and 2 usb3.1 ports
23:36jannaubut at least the template adt has 4 complete usb4 ports including all including acio, apciec, dpphy, dpin[01] and dpxbar
23:40integralpilotis the ubuntu version very behind on kernel releases? someone online is reporting an error using the latest ubuntu asahi that was fixed in asahi kernel 6.14.
23:43integralpilotoh just realised they have their own channel, will post something there.
23:44integralpilot has quit [Quit: integralpilot]
23:45integralpilot has joined #asahi-dev
23:48hdbngr has quit [Remote host closed the connection]
23:48hdbngr has joined #asahi-dev
23:57integralpilotokay so after adding logs, the SEP issue is that it times out waiting for a reply to the getrand
23:57integralpilotthe issue for the linux driver is also a timeout
…