marcan changed the topic of #asahi-dev to: Asahi Linux: porting Linux to Apple Silicon macs | Non-development talk: #asahi | General development | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Logs: https://alx.sh/l/asahi-dev
jole has joined #asahi-dev
DarkShadow44 has quit [Quit: ZNC - https://znc.in]
DarkShadow44 has joined #asahi-dev
CME has quit [Read error: No route to host]
CME has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
nsklaus has quit [Ping timeout: 480 seconds]
nsklaus has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
jacobcrypusa[m] has left #asahi-dev [#asahi-dev]
thelounge7571340 has joined #asahi-dev
thelounge7571340 has quit [Read error: Connection reset by peer]
thelounge7571340 has joined #asahi-dev
skipwich has quit [Quit: DISCONNECT]
kov has quit [Quit: Coyote finally caught me]
chadmed has quit [Ping timeout: 480 seconds]
nicolas17 has quit [Quit: Konversation terminated!]
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
OfirRubenov[m] has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
SSJ_GZ has joined #asahi-dev
thelounge7571340 has quit [Read error: Connection reset by peer]
<jannau> still no update from Joerg
thelounge7571340 has joined #asahi-dev
MajorBiscuit has joined #asahi-dev
<sven> huh, if I bring up atcphy for thunderbolt and then enable acio the pciec mmio becomes accessible... so why doesn't that happen inside linux as well?!
<sven> jannau: :(
<sven> maybe he's on vacation or sick or something like that
<_jannau_> probably, he was planning to look at the t600x dart series last week
<_jannau_> maybe a power-domain that gets disabled in linux?
thelounge7571340 has quit [Read error: Connection reset by peer]
<j`ey> I wonder if there is a power domain equivalent to clk_ignore_unused
<sven> hm... let me check that. maybe i'm just missing some domain in my hacked up device tree
thelounge7571340 has joined #asahi-dev
<sven> j`ey: I guess I could mark all of them as always on :D
<sven> looks like there's pd_ignore_unused though
<j`ey> give it a try!
<steev> yeah pd_ignore_unused is power domains
<sven> same result, I guess there's a subtle bug in my atcphy.c then :(
<j`ey> can you trace it?
<sven> yeah
jluthra has quit [Remote host closed the connection]
<sven> still going to be annoying to figure out the difference to my python code
jluthra has joined #asahi-dev
<sven> I guess I could run m1n1 inside the hv and then run atcphy.py against that el1-m1n1 and trace it with el2-m1n1 and look for differences
<j`ey> I guess double check the register offsets/values is a good place to start too
<sven> yeah, i bet it some BIT() #define that's wrong
<sven> *it's
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
roxfan has quit [Remote host closed the connection]
roxfan has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
duban6 has quit []
duban6 has joined #asahi-dev
thelounge7571340 has joined #asahi-dev
chengsun_ has joined #asahi-dev
chengsun has quit [Ping timeout: 480 seconds]
nickchan has joined #asahi-dev
nickchan has quit []
Major_Biscuit has joined #asahi-dev
Major_Biscuit has quit []
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
gilutsu_test03[m] has joined #asahi-dev
gilutsu_test03[m] has left #asahi-dev [#asahi-dev]
tanty has quit []
qdot has quit [Ping timeout: 480 seconds]
tanty has joined #asahi-dev
akemin_dayo has quit [Ping timeout: 480 seconds]
mattgirv has quit [Ping timeout: 480 seconds]
irth` has quit [Ping timeout: 480 seconds]
maz has quit [Ping timeout: 480 seconds]
maz has joined #asahi-dev
mattgirv has joined #asahi-dev
akemin_dayo has joined #asahi-dev
maz has quit [Quit: ZNC 1.8.2+deb2+b5 - https://znc.in]
tanty has quit []
qdot has joined #asahi-dev
maz has joined #asahi-dev
kov has joined #asahi-dev
tanty has joined #asahi-dev
irth` has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
nehsou^ has joined #asahi-dev
thelounge7571340 has quit [Ping timeout: 480 seconds]
chipxxx has joined #asahi-dev
caligulam[m] has joined #asahi-dev
<sven> [syslog] * [splayInterface.cpp:1107]operator(): WARNING: no color or timing modes from the link.Faking, since we have non-null VI.
<sven> got that once with dp-altmode but can't reproduce it anymore now :(
<sven> I guess that means it didn't manage to negotiate modes?
<sven> actually.. looks like I get that everytime now. progress I guess?
<_jannau_> "... from the link." sounds if dcp can't read the EDID
<sven> if anyone speaks DCP and can make sense of that.
<sven> and ofc there's also timing issues/some race involved *sigh*
<sven> ah, no, i'm just an idiot and disconnected the cable :D
<_jannau_> fps_frac = 0 doesn't look good
<_jannau_> message sounds if the dp aux channel is not enabled/working
nehsou^ has quit [Remote host closed the connection]
chipxxx has quit [Read error: No route to host]
<_jannau_> dp aux is using sbu1/2, that could require routing in tps6598x
thelounge7571340 has joined #asahi-dev
<sven> ohhhh... that's a good point
<_jannau_> if you have something someone else can test it could make sense to try it on atc3/hdmi on macbook pro 14"/16"
<sven> this code is horribly cursed fwiw
<sven> i'll have to clean it up a little bit first
<sven> there are some strings related to displayport in the tps6598x kext
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
<_jannau_> tps6598x has a multiplexer for SBU and AUX_P/N pins for displayport aux
<_jannau_> but it sounds as if tps6598x handles that itself without configuration
yamii has quit [Quit: WeeChat 3.6]
yamii has joined #asahi-dev
EricCurtin[m] has joined #asahi-dev
<sven> yeah
<sven> but maybe apple did something special for their variant
chipxxx has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
gladiac has joined #asahi-dev
MajorBiscuit has quit [Ping timeout: 480 seconds]
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
<sven> lol, running it again gets a bit further
<sven> jannau:, marcan: ^--
<sven> still doesn't work though
<sven> [syslog] * [PmgrService.cpp:319][AFK]_pmgrConfigVideoClock Failed to configure video clock freq= 74250000
<sven> hrm
<sven> selecting different modes doesn’t seem to help either :(
<jannau> "VideoInterfaceIOAV::open_ioav_gated(): IOAVVideoInterface open failed" sounds broken as well
<jannau> can you test dcpep instead of dcp_iboot?
<sven> not easily, I’ll have to untangle some of this mess first
<jannau> which mode did you set? 1920x1080@30 or 1280x720@60
<sven> in that trace 1920x1080@60, then later I tried @30 and some other random lower resolution mode
<jannau> 74.25 MHz would be CEA-861 pixel clock for 1920x1080@30 or 1280x720@60
<sven> huh, weird
<sven> Chosen timing mode claims fps_int = 60 though :/
<sven> (At 1920x1080)
<sven> [syslog] * [PmgrService.cpp:319][AFK]_pmgrConfigVideoClock Failed to configure video clock freq= 13500000 <-- that's with 1440x480@29
<sven> [syslog] * [PmgrService.cpp:319][AFK]_pmgrConfigVideoClock Failed to configure video clock freq= 37087912 <-- and that with 1920x1080@29
<jannau> so it's ~2 off, strange
<jannau> +factor
<sven> yeah. i'll double check my atcphy configuration
<sven> but maybe the iboot protocol just doesn't work with dcpext for some reason
<sven> it also reliably oscillates between "no color modes" and "valid colors modes" so there's still something that's obviously broken
thelounge7571340 has quit [Remote host closed the connection]
<jannau> not sure how the pixel clock calculation should be dependent on the atc phy config. it should just depend on the selected mode
<sven> knowing how entangled that mess is it wouldn't have surprised me
<sven> there's some PCLK (which I assume means pixel clock) configuration inside atcphy and something else that depends on the link rate
<sven> but it doesn't look like there's an additional diver or multipler to be configured
thelounge7571340 has joined #asahi-dev
<jannau> let me check what the dcp does with the pixel clock. I remember that it was not what I expected but since it works I didn't pay attention what was off
min[m] has joined #asahi-dev
<jannau> no, I misremembered. pixel clock is as expected for h/v total pixels and refresh rate
<sven> real dcp protocol seems to first print [syslog] * [nifiedPipeline.cpp:6553]set_digital_out_mode returned 80000104
<sven> and then
<sven> while mgr.iomfb_prop['DPTimingModeId'] != timing_mode:
<sven> KeyError: 'DPTimingModeId'
<sven> ahh.. nvm
<sven> also fails with [syslog] * [PmgrService.cpp:319][AFK]_pmgrConfigVideoClock Failed to configure video clock freq= 74250000
<jannau> it logs however a vid0 value which is half rate of the pixel clock: "syslog message: ock_Config_H13P.cpp:886: program_frame_size: pixel_clock 265250000, vid0 132625000, 000084a0 000001ef"
<sven> i think it currently fails before that
<sven> real dcp seems to always get the correct mode list though fwiw
<sven> so maybe the iboot code has a weird race somewhere
<sven> so this is either some subtle DCP issue or some subtle ATCPHY issue. i actually don't know which would be worse :(
<jannau> full dcp seems to have "IOMobileFramebufferAP::setProperty_int(key='VideoClock', value=74250000)" in a log which should be 1920x1080@60. resolution is certain, but I can't easily see the display refresh rate
<sven> so that would match [syslog] * [PmgrService.cpp:319][AFK]_pmgrConfigVideoClock Failed to configure video clock freq= 74250000 at the same resolution here
<sven> same thing happens when I just replay the hv trace instead of using my atcphy code
<jannau> yes. so there's a video clock which runs at half-rate of the pixel clock
<sven> why doesn't it just work then =(
gladiac has quit [Quit: k thx bye]
<jannau> have you checked pmgr with tools/dump_pmgr.py?
<sven> User: /device-tree/arm-io/disp0.clk[0] User: /device-tree/arm-io/disp0.clk[0]
<sven> [...]
<sven> #101: 24000000 0 0xa0/0x23b0400a4: regval: 0x80100000 |#101: 24000000 0 0xa0/0x23b0400a4: regval: 0x88100000
<sven> #102: 0 0 0xa4/0x23b0400a8: regval: 0xa000000c |#102: 0 0 0xa4/0x23b0400a8: regval: 0xa0000002
<sven> that... sounds wrong.. shouldn't it change the dispext clk?
<jannau> maybe, for disp0 it none of the 2 clocks is the pixel or video clock though
<sven> ah.. could be just mislabled ADT or something then
<jannau> or the video clock is not in the adt since dcp manages it directly
<sven> true
thelounge7571340 has quit [Remote host closed the connection]
<sven> 0x23b738018 is passed instead of 0x23b738014 in rt_bandwidth_setup_ap but that doesn't seem to make a difference
<jannau> is the DISPEXT_CPU0 power-domain enabled? I doubt you would get that far if it wasn't
<sven> yeah, it's on
thelounge7571340 has joined #asahi-dev
<sven> also bit=3 instead of 2 there but that also doesn't help
<jannau> are you sure that the values in rt_bandwidth_setup_ap don't make a difference? that are offsets from the pmgr reg in dcp(ext)'s regs
<sven> I changed them to what I observed in my macOS trace and nothing changed
<jannau> i.e. wrong values should try to set the wrong clock
<sven> 00000000 00 ff ff ff 01 00 52 47 18 80 73 3b 02 00 00 00 |......RG..s;....|
<sven> 00000010 00 c0 c3 3b 02 00 00 00 72 61 63 74 03 00 00 00 |...;....ract....|
<sven> 00000030 42 4c 61 79 65 72 5f 4d 00 00 00 00 |BLayer_M.... |
<sven> 00000020 6e 02 00 00 00 00 00 00 00 73 11 00 04 00 00 00 |n........s......|
<sven> so that's from my macos trace
<sven> from that I got bit=3, 0x23b738018 and 0x23bc3c000 for the regs
<sven> disp0 is also hardcoded in sr_mapDeviceMemoryWithIndex but changing that to dispext0 also doesn't make a difference
<jannau> 0x23bc3c000 is pmp and 0x23b738018 is inside pmgr, the last 3 regs of disp0 and dispext0 are identical
chipxxx has quit [Read error: Connection reset by peer]
<sven> ah, that explains it. it just requested one of those 3 regs
<jannau> have you already found trace_dcp dump? uncomment/copy the last line in hv/trace_dcp.py
<jannau> the dump can be analysed with python -m m1n1.fw.dcp.parse_log dcp.log
<jannau> that's easier to compare than the full output. it's only dcpep though
<sven> hah, didn’t know about that
<sven> ill compare DCP and dcpext tomorrow then and see if there’s any difference
<jannau> I was more thinking of comparing dcpext macos / your m1n1 scripts
<sven> I think there the more interesting stuff happens to the other endpoints
<sven> but I’ve wanted to run m1n1 in el1 to check the atcphy bringup anyway
dingodoppelt has quit [Quit: ZNC 1.9.x-git-170-9be0cae1 - https://znc.in]
dingodoppelt has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev
Phorous has joined #asahi-dev
thelounge7571340 has quit [Ping timeout: 480 seconds]
SSJ_GZ has quit [Ping timeout: 480 seconds]
mps has quit [Ping timeout: 480 seconds]
thelounge7571340 has joined #asahi-dev
loki_val has joined #asahi-dev
crabbedhaloablut has quit [Ping timeout: 480 seconds]
compassion1 has joined #asahi-dev
compassion has quit [Ping timeout: 480 seconds]
compassion1 is now known as compassion
thelounge7571340 has quit [Remote host closed the connection]
thelounge7571340 has joined #asahi-dev