#
…

2021-12-14

00:28jannauunrelentingtech: the ignored bytes at the beginning seem to be HID mouse reports
00:30jannauI shoud have looked at things more carefully instead assuming things when the touchpad started working in multi touch mode
00:31jannauI was more with the transport driver occupied
00:49jannauso the mouse works via hid-generic, that explains the missing multitouch
00:51jannaumarcan: https://github.com/jannau/linux/tree/dev/spi-hid-apple keyboard works via hid-apple, mouse via hid-generic
00:51jannaustill in rough WIP state
01:19slicey has quit [Quit: zzz]
01:22DarkShadow44 has quit [Quit: ZNC - https://znc.in]
01:22DarkShadow44 has joined #asahi-dev
01:35minecrell has quit [Quit: Ping timeout (120 seconds)]
01:42minecrell has joined #asahi-dev
01:47bps has quit [Ping timeout: 480 seconds]
02:01slicey has joined #asahi-dev
02:10King_InuYasha has quit [Remote host closed the connection]
02:11slicey has quit [Quit: zzz]
02:12DarkShadow44 has quit [Quit: ZNC - https://znc.in]
02:19slicey has joined #asahi-dev
02:22DarkShadow44 has joined #asahi-dev
02:29slicey has quit [Quit: zzz]
02:31slicey has joined #asahi-dev
02:59riker77_ has joined #asahi-dev
03:04riker77 has quit [Ping timeout: 480 seconds]
03:04riker77_ is now known as riker77
03:08phiologe has joined #asahi-dev
03:11PhilippvK has quit [Ping timeout: 480 seconds]
04:08Hinata[m] has joined #asahi-dev
05:50yuyichao has joined #asahi-dev
07:44the_lanetly_052___ has joined #asahi-dev
08:31povikshould be sending out NCO today, anyone wants to be CCed on that? (marcan has a default opt-in)
08:32slicey has quit [Quit: cya]
08:48svenyes. please also include kettenis for at least the device tree bindings
09:22marcanpovik: awesome! remind me again, what's the state of the audio stuff? works on the mini, have we tested the headphones jack or speakers on anything else?
09:22marcanI'm writing the progress report, so if you think the jack should trivially work everywhere I might as well test it and color that entire row green (yes I'm doing a table thing this time)
09:23marcan(well, not green, green is for upstreamed drivers, but you get the idea)
09:24marcanjannau: if it works well enough to use I might just merge it into asahi in its current state, might as well :)
09:27marcanpovik: also happy to merge in the audio work if you think that's okay
09:28marcanI basically want linux-asahi to be a testing ground for drivers headed in the right direction, even if they're rough, at least at this point in the game
09:37j`eyjannau: so hid-mouse works with multitouch?
09:37j`eyhid-generic I mean
09:38marcanI think he means hid-generic works in multitouch *mode* because it still picks up the first few bytes of the reports?
09:41_jannau_yes, I suspect that is the case without looking into it. multi touch does not work
09:43marcan_jannau_: btw, I do have access to an older Intel MacBook with SPI touchpad, so I can test that when we get there
09:44marcanI also noticed the Touch Bar multitouch is also ostensibly HID over SPI, but possibly a different protocol? unclear
09:47povikmarcan: playback through speakers and jack works on the mini. haven't been tested elsewhere
09:47povikjack should work everywhere (if the codec on t6000 macs isn't too different from the one on t8103 macs)
09:47povik^ in principle
09:52marcanawesome, I'll give it a shot after dinner :)
09:54marcansince it has the fancy table you'll probably want to run it locally; just install hugo and `hugo serve` at the repo root
09:54marcan(table will change based on what I do with the hid and audio stuff today :))
09:56povikthere's some rough edges with how the jack is presented to userspace (pulseaudio isn't happy with it, and you need to set a knob in alsamixer for jack not to be muted)
09:56povikbut this seem as minor issues to me, obviously we know how to drive the hardware
09:57povik(plug event is detected in kernel btw, just doesn't make it userspace probably)
10:02kettenismarcan: maybe mention that basic M1 support (USB type-C only) has been upstreamed in U-Boot now as well and will be part of U-Boot 2022.01?
10:02marcankettenis: ah yes, totally forgot about that, will do!
10:03marcanpovik: the userspace stuff is a whole different story anyway (there's profiles and stuff, we'll probably have to create some on top of figuring out what to do for the speaker crossovers)
10:07j`eymarcan: good report
10:08j`eymarcan: dunno if povik or jannau want their names as links..
10:09marcanI think Jannau can point to twitter, can't remember if povik has one too?
10:09marcanlet me know :)
10:10kettenisI did a bit more exploring of the battery-related SMC keys yesterday and added stuff to the wiki page
10:11kettenisfunny thing is that most values are native-endian, but the remaining capacity seems to be reverse endian
10:13kettenissome values are available both as an integer and as a float
10:13kettenisbut the float values don't seem to provide more resolution
10:14povikoh hey i do have twitter
10:14povikwith one old tweet in Czech
10:14povikmarcan: link to my github, better than nothing i guess
10:14svenmarcan: the usb quirk stuff is still pending fwiw (and i need to submit a follow up for the binding), not sure it'll make it to 5.17
10:14povikwill put up a "for hire" note there, see if anything happens :-p
10:19svenand regarding nvme... there's still at least one annoying problem. for some reason anything that uses random writes with sync (fsync, sync, etc.) is incredibly slow (<1MB/s). not sure that counts as "works well"
10:22povikmarcan: re: the report. s/supporting the various codec chips/supporting the various codec configurations/
10:22povikthe speaker amps are mostly the same, just need TDM set up to share one I2S bus
10:23povik(for speaker support elsewhere than mini)
10:24kettenissven: do those sync commands result in an NVMe flush command being issued?
10:25sveni think that's also what slows it down
10:28kettenisflushing the cache is defenitely expected to be "slower"
10:28svenit goes from 300 MB/s down to like 1MB/s. and testing the same setup with e.g. fio on macos doesn't do that
10:29kettenisOpenBSD only flushes the disk cache before shutdown/powerdown basically
10:30svenhm, i guess macos does the same. still need to verify that though
10:31_jannau_as far as I'm aware my name is globally unique so it does not need links ;)
10:33poviki need a link to not be confused with my grand grand grand father of the same name
10:33kettenis_jannau_: same here ;)
10:36_jannau_marcan: I think there are only two drivers for for the touchpad report format, hid-magicmouse will not work out of the box. but I really need to start looking at the reports in detail
10:40kettenisbtw, you're right, the "mouse" HID report is at the start of the packet
10:40kettenisand it continues to be sent, even in "raw" mode
10:41kettenis"raw" mode just extends the packet with touch information (which isn't really HID-compliant)
10:42KDDLB has quit [Quit: Ping timeout (120 seconds)]
10:42KDDLB has joined #asahi-dev
11:05marcansven: maybe send a revision to that soonish to poke things?
11:06svennvme has just been more fun ;)
11:08timokrgr has quit [Quit: User left the chat]
11:10kettenissven: don't let perfect be the enemy of good enough
11:11svenwell, the previous nvme driver would kernel panic if you just poked it enough. that's probably not good enough ;)
11:12kettenisI'd say the write cache buffer flushing slowness is something folks could live with for a while
11:13kettenisand getting usable nvme out there will actually allow other people to look into it
11:13svenright. i just happen to have a day job and it's not this.
11:14marcanpovik: https://github.com/povik/linux/commit/f5685ae510d5e9cce34c66666e9b5d335c99e848 FWIW since these are actually separate domains intended to be driven independently, you'd want "proper" PD support eventually (which involves a funny hack the PM code does for you of making up virtual devices)
11:14marcanI guess that will make mca the first driver to actually do that with pmgr
11:14marcanbut you don't have to worry about that for now
11:15kettenissven: ah yes, $DAYJOB
11:16* kettenisgoes back to answering e-mails about VLBI
11:18timokrgr has joined #asahi-dev
11:21povikmarcan: yeah, reckoned that the way out is teaching the MCA driver some PD support
11:21marcanalso, is nco_inp just a guess on the clocktree based on the frequency?
11:22povikit expresses two things: the frequency, being tied to clkref probably
11:22marcanI think we're better off with fixed frequency clocks since we don't know; that way, if something ends up being dynamic or changing, we can teach m1n1 to propagate frequencies from the ADT
11:22marcankinda harder to do that with ratios :)
11:22povikwill change that when i visit the DT again
11:23marcanif we knew the full clocktree I would absolutely do it that way... but we don't :(
11:23marcanI imagine if refclk changes (though I get the feeling it never will) they'd change all the multipliers anyway
12:08povikso, did anyone got the NCO patches? i don't see it on the lists yet
12:09j`eypovik: usually takes a bit of time
12:09povikokay, i will go do something else then
12:10povikoh there you go!
12:11j`eypovik: you need a MAINTAINERS entry
12:11marcanpovik: MCA should probably pull in ADMAC in the kconfig, otherwise it's pretty pointless
12:11j`ey(MAINTAINERS should be in a commit of its own for v2)
12:11marcanre MAINTAINERS, you can add a new entry and volunteer yourself, or you can shove it under the apple SoC stuff; depends on how much you want to commit ;)
12:12marcan(if it's a new entry it doesn't need to be a separate commit)
12:18jnhmmm, a bit more theory-of-operation would (IMO) be nice to have in the driver, specifically an explanation of the role that the LFSR stuff plays
12:23povikah, kind of hoped MAINTAINERS would sort itself out later
12:24svenmarcan: speaking of which, we never merged the dart entry to the general apple soc one, did we?
12:25marcanah, the MAINTAINERS thing? don't think so
12:25marcanwe can do that if you want
12:25svensure. it's almost all the same people anyway
12:25povikmeaning you two? :D
12:26sven+ alyssa i think
12:26svenmarcan: also, reminder about that mailing list :D
12:27marcanah yes :)
12:27marcanI'm trying to get the audio jack to work on t6000... and getting an unhandled IRQ... on the DART fault handler IRQ?!
12:28marcanI expected breakage but not *this*
12:28sveni thought only the USB DARTs shared an interrupt
12:29sventhere's also this thing apple calls SMMU (not the ARM one) but that's only for the disp dart and i don't even know if it can create interrupts
12:29marcanonly this DART is listed under this IRQ in the ADT
12:30sventhe SIO DART also has a DAPF but i think its errors also go the normal DART error status :/
12:30povikthe MCA has an interrupt although i don't do anything to unmask it on the MCA side
12:30marcan[ 542.714722] cs42l42 1-004b: PLL failed to lock: -110
12:30marcanI'm pretty sure the codec is broken too
12:30povikno that's okay
12:30marcanalsamixer reverts any changes I make though
12:31povikit sets the PLL even if getting master clock from the SoC
12:32poviki think i see the same in my dmesg
12:32povikthat should probably be fixed in the cs42l42 driver
12:32povikanyway i forgot about one thing. i don't know what the MCA and ADMAC drivers will do on t6000
12:32povikthat's completely untested
12:32povikrestricted my attention to the codec when i said the jack has a chance of working there
12:34marcanthis is indeed forever reading the DART error registers which show no error
12:34povikis it at least the SIO DART?
12:34svenhrm, the spurious irq detection should kill it eventually. not that that's much better
12:35marcan>>> (read32(aic + 0x6800 + (1130 // 32) * 4) >> (1130 % 32)) & 1
12:35marcandefinitely firing
12:35marcan0x39b004000+000040 ERROR = 0x9000000 (FLAG=0, STREAM=0x9, CODE=0x0, NO_DAPF_MATCH=0, WRITE=0, SUBPAGE_PROT=0, PTE_READ_FAULT=0, READ_FAULT=0, WRITE_FAULT=0, NO_PTE=0, NO_PMD=0, NO_TTBR=0)
12:35marcan0x39b004000+000050 ERROR_ADDR_LO = 0x6ee306d3 ()
12:35marcan0x39b004000+000054 ERROR_ADDR_HI = 0x0 ()
12:38povikmarcan: if you didn't do it already you need at minimum to change apple,nclusters in mca and dma-channels in admac on t6000
12:39povikwell then you have done the minimum :p
12:39marcanah, not dma-channels, derp
12:40povikalthough admac driver should not touch unused channels, so that doesn't explain this mess
12:44marcan[cpu6][0xffffc000083d82c0] MMIO: R.4 0x39b409070 (admac-sio[0], offset 0x9070) = 0x12
12:44marcan[cpu6][0xffffc000083d7650] MMIO: W.4 0x39b410010 (admac-sio[0], offset 0x10010) = 0xfffa0000
12:44marcan[cpu6][0xffffc000083d7658] MMIO: W.4 0x39b410010 (admac-sio[0], offset 0x10010) = 0x0
12:44marcan[cpu6][0xffffc000083d765c] MMIO: W.4 0x39b410010 (admac-sio[0], offset 0x10010) = 0x10000
12:44marcan[DARTTracer@/arm-io/dart-sio] MMIO: R.4 ERROR = 0x1000000 (FLAG=0, STREAM=0x1, CODE=0x0, NO_DAPF_MATCH=0, WRITE=0, SUBPAGE_PROT=0, PTE_READ_FAULT=0, READ_FAULT=0, WRITE_FAULT=0, NO_PTE=0, NO_PMD=0, NO_TTBR=0)
12:44marcanthese are the last things that happened before it started looping on DART regs
12:44legarts[m] has joined #asahi-dev
12:46poviki think i know what this could be
12:46povikyou didn't change anything else in the mca config, did you?
12:47marcanI'm dumb
12:47marcanI keep forgetting the DART address is reg index 1
12:47marcanoh wait, m1pro has *two*
12:47marcanm1 only had *one*
12:48poviki see you set the right dma channel :)
12:48povikmaybe you know what you are doing
12:49marcan instance = TRADDARTBULKTRADDARTLLT
12:49marcansven: is this another one of those let's use one DART for some stuff and another DART for other stuff things like USB?
12:50svencould be :/
12:50svendo you have the device tree somewhere?
12:50sven*parsed device tree
12:51svenyeah.... that smells like the USB hack
12:52povikwhat's the USB hack?
12:52svensome parts of XHCI go through dart A, other parts through dart B
12:52svenand the split makes no sense
12:53svenmarcan: if there are two DARTs like this macOS just programs them identically
12:53sventhat's why they e.g. have that sid-remap hack for the USB DARTs
12:53marcan remap = 256
12:54svenhm, yeah. maybe it was called remap
12:54svenshould be in the USB DARTs
12:54svenbecause they use sid 0 in the first but sid 1 in the second and because they just mirror register writes I think they need that
12:55sven(that's at least what the corellium driver did so i just assume that macOS does the same)
12:55marcanok, no more DART fails
12:56povikmarcan: if you post your DT i can check it wrt t6000 ADT
12:56marcanpovik: audio streaming ostensibly works now, but I don't hear anything out the jack
12:56marcando I need to poke alsamixer?
12:56povikalsamixer, MIXER knob
12:56marcando what with it?
12:57povikmake it something other than zero
12:57marcanit's maxed out and refuses to budge; pretty sure something is very broken with the codec
12:57povikyeah, could be
12:58j`eywhat does the codec hardware actually do?
12:58povikcan you post post-reset register values from the codec? haven't poked at cs42l84 yet
12:58poviki have a script somewhere
12:59marcanj`ey: the analog bits
12:59marcanADC/DAC/mixer/etc
13:00povikor not, it's easier if i read it off t6000 mac i have here
13:00j`eymarcan: I see
13:01marcani2cdump doesn't show anything for that address
13:01marcanI think it's dead
13:02marcan reg = [0x000009C40000004B, 0x0000000000000000]
13:02marcanthis *is* 4b right?
13:03marcani2cdetect finds nothing
13:04marcanwonder if we have to set a different pinconfig for this one
13:04marcanor there is some other magic gpio
13:07marcanah but wait, if I flip the reset GPIO it fails to read the device ID
13:07marcanso it works at *some* point
13:07marcanthen something breaks it
13:07poviki have seen this
13:07povikwhen i put weird frequencies on the i2s bus
13:07marcani2c breaks?
13:07povikweirdly enough
13:08marcan# 84: 1068000000 0 0xa4/0x28e038028: regval: 0x85100100
13:08marcan User: /device-tree/arm-io/nco.clk[0]
13:08marcannco clock ref is different.
13:08marcanthat's for nco0 and 1
13:09marcannco2 has a different ref though
13:09marcanor rather the 2nd input
13:09povikno, all nco's use 1st input as reference
13:10marcanalsamixer is still broken though :-)
13:11marcanhow is the mca clock -> nco mapping defined?
13:12marcanIIRC that was preconfigured by the bootloader and we don't touch those regs in the end?
13:12povikyou should be able to see from userspace if the playback progresses at the right speed
13:12povikthus confirming this all is configured right
13:13povikdon't know what's the right tool for this though
13:15marcanplayback does appear to progress at the right rate
13:16povikdo you have mclk-fs for the jack in your DT?
13:16marcan(using sox's play)
13:16marcanlet me paste you my diff
13:17marcans/route1/route/, that was just a test I just did
13:20povikDT looks good to me
13:22povikmclk-fs=<128> if you are up to trying random things
13:23marcan[ 19.795376] apple-mca 39b600000.mca: unsupported SERDES configuration requested (mask=0x3 slots=2 slot_width=64)
13:24marcanfwiw alsamixer fails even before I try to play any audio
13:24marcanis that expected with the i2s clock issue?
13:24povikno that's not
13:24poviklet's see
13:25marcanI feel like this codec is pretty dead
13:25povikthis could be related to the "disable regcache" hack which needs explaining
13:26marcanevery read the driver did returned 0 except the 2nd one
13:26povikso this all happens before we had a chance to upset it
13:28povikah that unsupported SERDES message you got after mclk-fs=<128>
13:28povikof course that won't fly
13:28povik(with the code as-is)
13:28povikit derives TDM slot width from mclk-fs
13:30povikcould be i2c has bad clock reference there?
13:31marcanContinue? [Y/n]
13:31marcan 0 1 2 3 4 5 6 7 8 9 a b c d e f 0123456789abcdef
13:31marcan10: 42 42 42 42 42 42 42 42 42 42 42 42 42 42 42 42 BBBBBBBBBBBBBBBB
13:31marcan00: 00 42 42 42 42 42 42 42 42 42 42 42 42 42 42 42 .BBBBBBBBBBBBBBB
13:31marcanso the first register read returns 00, the rest return 42
13:31marcanand then it sticks there
13:31povikhaha, it's cs42l42
13:31marcanit's 42 forever, even subsequent reads of 00
13:32jnas documented by Douglas Adams ;)
13:33poviki want to point out this passes the device ID check for the original driver
13:35povikah, no, almost
13:41marcanah wait, this is something weird with i2c
13:48marcan70: 42 a8 4a 01 00 00 00 00 00 00 00 00 00 00 00 00 B?J?............
13:48marcanthis looks like the device ID, but it's in the wrong place
13:48marcansomething is really messed up with the i2c interface
13:48marcanmaybe it's time for gpiola...
13:52marcanthis almost looks like the i2c addressing mechanism is different from what I expect
13:54marcanyeah, like 16-bit addresses, not the banking thing...
13:57kettenisyou mean register addresses?
13:58marcancs42l42 (aka cs42l83, apparently) uses a weird banking thing with 8 bits of page address and 8 bits of register address
13:58kettenisthere are many chips that do it like that
13:58marcancs42l83 instead seems to use 16-bits of address outright
13:58marcanyeah, I know
13:59marcancs42l92 uses *32-bit* addresses so that is precedent for cirrus doing this
13:59marcanlet me see if I can find a codec they do with 16-bit addrs...
14:14marcanyeah I get the feeling this is one of those "apple got first dibs on a new chip or got them to make something custom" things
14:26marcanpovik: yeah, this doesn't look like anything I can find, and stringsing the macos driver points in that direction too
14:26marcan84 is very different from 83
14:27povikwell that is a tricky plus-by-one
14:30marcansome register field names do match 42l42, so it's not like a total from-scratch chip, but if nothing else the addressing is different for one
14:30kettenisso it's not just a simple quirk where they basically undone the banking?
14:30marcanand there's stuff I can't find a match for in '42, though it does seem apple may be making up register names
14:30povikthe upside is the SoC side doesn't break on t6000
14:32marcanthe ID register properly says CS42L84 for this one, for one
14:36yuyichao has quit [Ping timeout: 480 seconds]
14:37povikthe 83 also says 83 in ID register
14:38marcansome of the plug detect stuff is the same as '42; but there are new concepts: "hi cap" mode, "MSM", some sample rate thing, something called "ccm"
14:38marcanapparently this one can measure headphone impedance
14:39marcandcid seems to be related to that
14:39marcan"DC impedance detect" perhaps?
14:39marcanthere's a thing about boosting the output voltage
14:40marcanit sounds like this thing got quite a bit smarter about adapting to different headphone types
14:44marcanpovik: so yeah, sounds like there's quite a bit of reversing work here to get this to work properly :/
14:44marcanwill leave it aside for now, let me test the headphone jack on the iMac or something
14:44marcanah, but I don't have a rootfs :-)
14:44marcanI should put something on a USB drive...
14:52marcanpovik: the funny thing is the macOS driver has basically verbosely documented all the register reads/writes... it's like they have a macro that stringifies the arguments to write32 mask32 etc and them passes those strings too
14:52marcansoo at least there's that for understanding this mystery chip
14:52marcanOTOH; I can confirm the register addresses are different
14:53yuyichao has joined #asahi-dev
14:53marcanmight be worth extracting just the reg info from the driver (under the premise that register names aren't really copyrightable) and working off of that + HV logs
14:53marcanthat should be automatable
14:54kettenismakes sense; apple boasted about supporting high-impedance headphones on the m1 max/pro laptops in their marketing
14:55kettenisand given how they operate, that also explains why the chip isn't publically available
15:23aleasto has joined #asahi-dev
15:40mazmarcan: AICv2: is there really no way to control the affinity at all?
15:44mazmarcan: this is going to hurt with devices with multiple queues where the driver really expects a given queue to be associated with a given CPU and doesn't expect the interrupt anywhere else.
17:09ChaosPrincess has joined #asahi-dev
17:22gladiac has quit [Quit: k thx bye]
17:23gladiac has joined #asahi-dev
18:16the_lanetly_052__ has joined #asahi-dev
18:22the_lanetly_052___ has quit [Ping timeout: 480 seconds]
18:25Graypup_ has quit [Quit: meow]
18:25Graypup_ has joined #asahi-dev
18:34___nick___ has joined #asahi-dev
18:34___nick___ has quit []
18:36___nick___ has joined #asahi-dev
18:38Dcow_ has joined #asahi-dev
18:46the_lanetly_052__ has quit [Ping timeout: 480 seconds]
19:03drwhax[m]1 has joined #asahi-dev
19:04alyssa has joined #asahi-dev
19:04alyssamarcan: bluetooth I think should be nul, not ehh
19:05alyssathe corellium patches are AFAIK wifi only and I don't think anyone has looked at bluetooth?
19:05alyssaIIRC it's the same chip but the corellium patch is incomplete or something
19:05alyssaRTC should be ehh, the corellium patch works fine
19:05alyssa(and doesn't have crazy dependencies)
19:06kettenisalyssa: yes, I have seen no code for the bluetooth part of the wifi device
19:07alyssakettenis: no bluetooth support is a security feature ;-)
19:08kettenisI don't disagree ;) (we ripped out bluetooth support in OpenBSD years ago since the implementation wasn't up to our standards)
19:08alyssaThe last time I used bluetooth was in Sep 2020
19:09alyssaI have no intention of reliving that experience
19:10jni haven't installed the bluetooth firmware for my WiFi/Bluetooth module, and i don't miss it :>
19:11kettenisThat is the OpenBSD RTC driver that I write after studying the corellium code
19:13ketteniss/write/wrote/
19:14kettenisI think it is easier to read, but I may be a bit biassed ;)
19:16alyssakettenis: out of curiousity how do the cf* structures work? I don't BSD
19:17* alyssakinda wants to poke at M1 GPU
19:17alyssaI shouldn't but..
19:19alyssaOk, we'll take a vote
19:19kettenisalyssa: it's magic ;)
19:19alyssaShould I.... 1) study for finals 2) do my job 3) m1 gpu
19:19kettenisdesigned by Chris Torek IIRC
19:19alyssaI trust that #asahi-dev is an unbiased population from which to sample
19:20kettenisthe idea is that drivers form a hirarchy that matches the bus hierarchy
19:20kettenisdrivers have a match function to see whether they should attach or not
19:21kettenison arm64 this uses the device tree and the match functions basically check compatible strings
19:21kettenisbut for pci devices the match function checks vendor and device ID
19:21j`eyisnt it similar to the of_match stuff in linyx>
19:22j`ey.. *linux
19:22j`eyalyssa: 312
19:22kettenisyes, but the device tee structure in linux is very explicit
19:23kettenisis *not* very explicit I mean
20:10aleasto has quit [Remote host closed the connection]
20:11King_InuYasha has joined #asahi-dev
21:08___nick___ has quit [Ping timeout: 480 seconds]
21:52josipknezovic[m] has joined #asahi-dev
21:59Dcow_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
22:01Dcow_ has joined #asahi-dev
23:08kgarrington has joined #asahi-dev
23:09kgarrington has quit [Remote host closed the connection]
23:44Dcow_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
…