#
…

2026-07-24

00:15nepeat has quit [Ping timeout: 480 seconds]
00:18nepeat has joined #asahi-dev
00:33Calandracas_ has joined #asahi-dev
00:40Calandracas__ has quit [Ping timeout: 480 seconds]
00:41bennofs31 has joined #asahi-dev
00:42bennofs3 has quit [Ping timeout: 480 seconds]
00:42bennofs31 is now known as bennofs3
00:58sicktriple[m] has joined #asahi-dev
01:18lion328 has quit [Remote host closed the connection]
01:33enverbalalic has quit [Ping timeout: 480 seconds]
02:00yurix has quit [Remote host closed the connection]
02:00yurix has joined #asahi-dev
02:00ciara has quit [Remote host closed the connection]
02:00jnn has joined #asahi-dev
02:03jn has quit [Ping timeout: 480 seconds]
02:13ciara has joined #asahi-dev
03:19nora_ has joined #asahi-dev
03:21nora has quit [Ping timeout: 480 seconds]
06:22picnoir1 has quit [Ping timeout: 480 seconds]
06:31picnoir has joined #asahi-dev
07:15yukaI'm slightly unsure about the t6040/t6041 pwrstate nodes
07:15yukafrom what I can tell, on t6000 vs t6002 and t6020 vs t6022 they actually had different pmgr devices in the adt
07:17yukayes, indeed it's a big diff between t6020 and t6021
07:17yukaless memory channels, less dispexts
07:18yukaI think I'm missing something in the new pmgr adt encoding
07:18yukathat marks these devices as active/inactive
07:19yukabecause: with the current interpretation of adt2dt, a t6040 adt (j614s) has all the memory channels and dispexts of t6041 still present and active (!no_ps)
07:21yukathere is literally no difference in the adt
07:21yukaDISPEXT3_SYS has all the same flags and unk* values in j614s vs j614c
07:21yukaeven though it should somehow be marked as disabled in j614s
07:58yukaAlso the fpwm* nodes have always-on which is odd
08:02jannauI suspect the pmgr* nodes for t60x0 weren't derived directly from the t60x0 ADT. The t60x1 ADT was used and we deleted the absent nodes manually
08:04yukaI also checked the t6020 vs t6021 _adt_ and they differ significantly
08:07jannaubut the difference is that the pmgr node doesn't list power-states for non-existing HW and misses one (or two) pmgr region completely
08:08yukayes, and on t6040 the adt _does_ list power states for non-existing HW
08:13yukathis is what's confusing me
08:21Pokemod9773620847128183300 has joined #asahi-dev
08:21Calandracas_ has quit [Remote host closed the connection]
08:21Calandracas__ has joined #asahi-dev
08:23Pokemod977362084712818330 has quit [Read error: Connection reset by peer]
08:26svenhrm, thinking about it some more i think i can even start with the cursed sprr/gxf hv on m1
08:28svencould even use m1n1 as the first guest to test the gxf/sprr stuff
08:28svenand then by the time that works hopefully the student who's been working on SPTM for M2 has handed in his thesis and published his code and then I can move to M3 or whatever and see about SPTM
08:38yukare samsung uart: the 1<<10 bit in UCON is set on all my devices: t6000/t6034/t8132/t8140/t6050
08:38yukaon boot
08:39yukabut only t8140 fails spectacularly if it is cleared
08:40hardy94 has joined #asahi-dev
08:50hardy94 has quit [Remote host closed the connection]
08:57yukaforeach-hpm works on t6000, after I fixed another panic in the adt code
08:58chaos_princesssven: wouldn't it require NV
08:58chadmedwhat panicked
08:58yukachadmed: adt_first_child_offset panicked on a node without children
08:58chadmedoh makes sense
08:59yukawell it should just return AdtError::NotFound as c_int
08:59svenchaos_princess: hrm... no sure, to run SPTM and secure kernel, sure. but just XNU+PPL or even a basic m1n1 payload to test the shadow paging thing?
09:00chaos_princessbasic payload may work
09:00sveni'm gonna get some M3 or whatever soonish anyway, finally figured out the tax issues
09:00chadmedyuka: yeah it should, C version never did and I overlooked "fixing" it as i was more concerned about the load-bearing UB in another part of the offset stuff :P
09:00chadmedthanks for fixing
09:01girona[m] has joined #asahi-dev
09:01girona[m]You're welcome
09:29kevans91 has quit [Ping timeout: 480 seconds]
09:40hardy94 has joined #asahi-dev
09:49enverbalalic has joined #asahi-dev
10:08ansg191_ has joined #asahi-dev
10:19ansg191 has quit [Remote host closed the connection]
10:23snek has quit [Remote host closed the connection]
10:23snek has joined #asahi-dev
10:24ansg191_ is now known as ansg191
10:32hardy94 has quit [Remote host closed the connection]
10:41os has quit [Ping timeout: 480 seconds]
11:02kitlith_ has joined #asahi-dev
11:05kitlith has quit [Ping timeout: 480 seconds]
11:06kitlith_ is now known as kitlith
11:33enverbalalic has quit [Ping timeout: 480 seconds]
11:35enverbalalic has joined #asahi-dev
11:38mrkajetanp has joined #asahi-dev
11:43enverbalalic has quit [Ping timeout: 480 seconds]
11:50enverbalalic has joined #asahi-dev
12:11enverbalalic has quit [Ping timeout: 480 seconds]
12:12kevans91 has joined #asahi-dev
12:15lion328 has joined #asahi-dev
12:26enverbalalic has joined #asahi-dev
12:42hdbngr has joined #asahi-dev
12:45madmatt0677 has joined #asahi-dev
12:50madmatt067 has quit [Ping timeout: 480 seconds]
12:50madmatt0677 is now known as madmatt067
12:54qyliss has quit [Quit: bye]
12:56qyliss has joined #asahi-dev
13:01hdbngr has quit [Ping timeout: 480 seconds]
13:04hdbngr has joined #asahi-dev
13:17nickchan1<<10 in UCON is set on s5l8960x
13:17nickchanit's probably safe to set it everywhere
13:18yukaQuestion is still what it does exactly
13:19yukaOr if "UNK" will fly with upstream
13:24svenUNK + some explanation why it's required is probably fine
13:29nela has quit [Quit: bye!]
13:30nela has joined #asahi-dev
13:32hdbngr has quit [Ping timeout: 480 seconds]
13:34hdbngr has joined #asahi-dev
13:57hdbngr has quit [Ping timeout: 480 seconds]
13:58hdbngr has joined #asahi-dev
14:07hdbngr has quit [Ping timeout: 480 seconds]
14:15hdbngr has joined #asahi-dev
14:24hdbngr has quit [Ping timeout: 480 seconds]
14:32hdbngr has joined #asahi-dev
14:40SG has joined #asahi-dev
14:46hdbngr has quit [Ping timeout: 480 seconds]
14:48rodmac has joined #asahi-dev
14:53hdbngr has joined #asahi-dev
14:58rodmac has quit [Quit: leaving]
15:05hdbngr has quit [Ping timeout: 480 seconds]
15:06hdbngr has joined #asahi-dev
15:15hdbngr has quit [Ping timeout: 480 seconds]
15:31svenalright, so genter with an imm values works kinda like hvc, the imm ends up in ESR_GL1
15:32svenGL0 also works essentially as i expected. i can eret there from GL1 and then svc back up to GL1
15:40enverbalalic has quit [Ping timeout: 480 seconds]
15:42enverbalalic has joined #asahi-dev
15:43hdbngr has joined #asahi-dev
15:52hdbngr has quit [Ping timeout: 480 seconds]
15:53hdbngr has joined #asahi-dev
15:57yukaHas anyone gotten M4+ nvme to work? a few weeks back someone with the nick starcatmeow posted something on it but haven't seen them around
15:57yukaI think I'm close, but this is not a particularly helpful error message: "assert failed: [13075]:undefined error. interrupt status 0x0, status valid 0x0"
15:58chaos_princesshave you fixed the 15 firmware differences?
15:58yukayou mean that APPLE_ANS_UNKNOWN_CTRL moved?
16:00yukaI am not convinced this is the case here
16:01yukathe old addresses contain sensible values, and the trace I am working with also has the old ones
16:04yukab'EFI PART...'
16:04yukathanks for reminding me that this gist on the ans3 changes exists
16:04yukait works
16:05yukahowever, I just skipped the unknown_ctrl and linear_sq_ctrl writes completely
16:05yukainstead what I needed to change was cmd.cdw12 1 -> 0
16:05yukanoted in the same gist
16:05svenyeah, that's a bug in m1n1 because i misread the spec iirc
16:06sveni guess it just defaults to linear sq now and what i guessed what a chicken bit in unknown_ctrl just isn't required anymore then
16:06yukaI wonder where the person who wrote the gist got the new addresses from
16:06yukayes it seems like it
16:06sventhrowing kexts at ghidra iirc
16:06svenso they wrote that gist instead of implementing anything themselves
16:07hdbngr has quit [Ping timeout: 480 seconds]
16:07chaos_princesst600x is still sptm-less on 26, right?
16:07chaos_princessso that can be used for future tracing
16:07yukaSo... the main insight I got from looking at the t8140/t8132 adts is that there is more ans regs
16:07hdbngr has joined #asahi-dev
16:08yukathe t8140 trace someone gave me was helpful in figuring out what they are used for: nvmmu and nvme are now two separate bases (reg 3 = nvmmu, reg 9 = nvme)
16:08yukat8140 does not throw serrors when trying to write to the nvme_base + NVMMU_*
16:08yukabut t8132 does
16:09yukaanyways I will prepare a PR
16:09chaos_princesssven: is atc.c doing writes in the same order as xnu does?
16:10svenalmost iirc
16:10chaos_princesst603x does a lot of the same things but in different order
16:11chaos_princessso, hopefully it does not matter?
16:18svenonly one way to figure out :D
16:20chaos_princessyea, i am filling out the rest of the changes
16:24hdbngr has quit [Ping timeout: 480 seconds]
16:38hdbngr has joined #asahi-dev
16:56hdbngr has quit [Ping timeout: 480 seconds]
16:57hdbngr has joined #asahi-dev
17:05hdbngr has quit [Ping timeout: 480 seconds]
17:13yukaOkay, I'm testing nvme things with 26.5 on my M3 Max
17:14yuka0x24008 is indeed abad1dea
17:14yuka(the old offset for unknown_ctrl)
17:14yukahowever I'm pretty sure it's just no longer needed
17:14yukathe bit that we'd clear is already cleared at 0x28130 (the new offset according to the gist)
17:16hdbngr has joined #asahi-dev
17:22yukaI guess I should test M1 with new firmware too just to be sure
17:23yukathe conditionals for this will be a bit of a mess since some of the things depend on the firmware versions, others on the hardware, but the hardware-dependend things don't have any apparent indicator in the adt (compatible is same, only that there's more regs)
17:29svenfor linux we probably want some apple,firmware-abi compatible then :(
17:30chaos_princessdidn't we decide to do sth like t8103-nvme-fw-15 for compatible override in m1n1, and then add firmware-abi?
17:31sveni know we discussed that at some point
17:32hdbngr has quit [Remote host closed the connection]
17:32hdbngr has joined #asahi-dev
17:37jannauwe need a new compatible anyway because the the macOS 15 fw is very not compatible. it even breaks boot if it is not used
17:38yukaYes, I'm currently trying to figure out how to make nvme in linux happy
17:39yukaon M3 26.x, but it's the same errors as M4 26.x
17:41jannauwhether it's a good idea to encode the fw version in the compatible itself or in a property is more of a style question
17:42jannauI was planning to bring this up with M3* nvme support as the M3 Ultra requires macOS 15.x fw support
17:44elvishjerricco5 has joined #asahi-dev
17:50elvishjerricco has quit [Ping timeout: 480 seconds]
17:50elvishjerricco5 is now known as elvishjerricco
17:52enverbalalic has quit [Remote host closed the connection]
18:18jannauv3 M3 Pro/Max/Ultra intial devicetrees sent
18:22svennice! i'll pick them up sometime next week unless there are more comments
18:34hdbngr has quit [Ping timeout: 480 seconds]
18:49hdbngr has joined #asahi-dev
19:00svenhttps://f.svpe.de/847eb6e8e302dfe3d136f2bbde5106b4ddc23def82c4f613fe42157912c1df55_cursed_sprr.txt looks like basic genter/exit and _GL sysreg emulation works. that was surprisingly easy
19:00svenprobably breaks in various ways with XNU though, but my small test guest works
19:00svennow for the truly cursed part: the pagetables
19:02svenhrm, though maybe i should test gl0 there as well
19:02hdbngr has quit [Ping timeout: 480 seconds]
19:04pb17 has quit [Ping timeout: 480 seconds]
19:04hdbngr has joined #asahi-dev
19:13pb17 has joined #asahi-dev
19:13hdbngr has quit [Ping timeout: 480 seconds]
19:28yukahmmmmm
19:28yukathe nvme with new fw behaves like old fw without clearing the bit in unknown_ctrl
19:29yukain some ways
19:51hdbngr has joined #asahi-dev
19:58barusunickchan: it doesn't book up with 100w charger either
19:58barusus/book/boot/
20:00hdbngr has quit [Ping timeout: 480 seconds]
20:01svenalright, gl0 works as well.
20:10hdbngr has joined #asahi-dev
20:26hdbngr has quit [Ping timeout: 480 seconds]
20:34chaos_princessBus 002 Device 002: ID 152d:0562 JMicron Technology Corp. / JMicron USA Technology Corp. JMS567 SATA 6Gb/s bridge
20:35svennice!!
20:35chaos_princessthough 1. hello, "pipehandler lock not acked" my old friend. 2. i needed to remove a wait that macos does, and i am very confused as to why
20:36svenis that a wait for some bit in the BIST dance?
20:36chaos_princessbist dance is quite different on t603x
20:36svenah. 'caused i used to have a bug where some bit in there was never set whenever i got pipehandler lock not acked
20:37chaos_princessret = readl_poll_timeout(atcphy->regs.core + AUS_COMMON_DIG_RCAL1, reg, (reg & AUS_COMMON_DIG_RCAL1_ALL_CODES_DONE), 10, 100000);
20:37chaos_princessthis is the one macos does but i had to remove
20:38chaos_princessAUS_COMMON_DIG_RCAL1 is 0x804 AUS_COMMON_DIG_RCAL1_ALL_CODES_DONE is 0x1
20:38chaos_princessalso bist dance involves resetting ACIOPHY_BIST_CFG1_USB_EN 3 times, cause why not
20:39chaos_princessi guess the wait can be because my trace is from the usb hub that has a dp port, and i am now using a usb3-only ssd enclosure?
20:41svenhrm, doesn't it do usb3+dp even when usb3-only is requested? the dp part just has additional crap like dp aux enable
20:42chaos_princessi straight up commented out most of dp init
20:48chaos_princesshow much should i be concerned about pipehandler lock if it is only on teardown?
20:48svenoh, when unplugging? not much
20:48Mrmaxmeier1 has quit [Quit: The Lounge - https://thelounge.chat]
20:48sveniirc it's only critical around the BIST dance
20:49Mrmaxmeier17 has joined #asahi-dev
20:49Mrmaxmeier17 has quit []
20:50chaos_princessare there even docs around what the moving parts are supposed to be? or is everything apple/snps specific, and can just be treated as "black box thingy i guess"
20:50Mrmaxmeier17 has joined #asahi-dev
20:51hdbngr has joined #asahi-dev
20:52svenblack box thingy :(
20:53sventhe register names allows some guessing, and e.g. PIPE is the interface between the PHY and dwc3
20:53svenbut it's mostly guesswork
20:57pb17 has quit [Ping timeout: 480 seconds]
21:00hdbngr has quit [Ping timeout: 480 seconds]
21:00pb17 has joined #asahi-dev
21:11Yushi has quit [Remote host closed the connection]
21:12Yushi has joined #asahi-dev
21:14hdbngr has joined #asahi-dev
21:22hdbngr has quit [Ping timeout: 480 seconds]
21:31chaos_princessok, the weird wait was because macos already takes the phy out of reset at this point, and we only do it later, moving reset to macos position makes it no longer timeout
21:45JamesCalligeros[m]github is having issues again
21:46JamesCalligeros[m]Im pretty sure ive got better uptime and availability at this point
21:52hdbngr has joined #asahi-dev
22:02hdbngr has quit [Ping timeout: 480 seconds]
22:15hdbngr has joined #asahi-dev
22:24hdbngr has quit [Ping timeout: 480 seconds]
23:05hdbngr has joined #asahi-dev
23:19crabbedhaloablut has quit []
23:20crabbedhaloablut has joined #asahi-dev
23:46hdbngr has quit [Ping timeout: 480 seconds]
…