2026-08-14
00:06hdbngr has joined #asahi-dev
00:15hdbngr has quit [Ping timeout: 480 seconds]
00:22hdbngr has joined #asahi-dev
00:30cow123 has quit [Ping timeout: 480 seconds]
00:31hdbngr has quit [Ping timeout: 480 seconds]
00:31hdbngr has joined #asahi-dev
00:41hdbngr has quit [Ping timeout: 480 seconds]
00:49hdbngr has joined #asahi-dev
00:57hdbngr has quit [Ping timeout: 480 seconds]
01:00hdbngr has joined #asahi-dev
01:08nora has quit [Ping timeout: 480 seconds]
01:08hdbngr has quit [Ping timeout: 480 seconds]
01:17hdbngr has joined #asahi-dev
01:23pb17 has quit [Ping timeout: 480 seconds]
01:25hdbngr has quit [Ping timeout: 480 seconds]
01:27nora has joined #asahi-dev
01:33hdbngr has joined #asahi-dev
01:33pb17 has joined #asahi-dev
01:45hdbngr has quit [Ping timeout: 480 seconds]
01:54hdbngr has joined #asahi-dev
01:56ZhengZhang[m] has joined #asahi-dev
02:00yurix has quit [Remote host closed the connection]
02:00yurix has joined #asahi-dev
02:03hdbngr has quit [Ping timeout: 480 seconds]
02:11hdbngr has joined #asahi-dev
02:38hdbngr has quit [Ping timeout: 480 seconds]
02:39mrkajetanp_ has joined #asahi-dev
02:40mrkajetanp has quit [Ping timeout: 480 seconds]
02:45tobhe has joined #asahi-dev
02:47hdbngr has joined #asahi-dev
02:52tobhe_ has quit [Ping timeout: 480 seconds]
02:56hdbngr has quit [Ping timeout: 480 seconds]
03:05hdbngr has joined #asahi-dev
03:14hdbngr has quit [Ping timeout: 480 seconds]
03:19nora_ has joined #asahi-dev
03:21cow123 has joined #asahi-dev
03:21hdbngr has joined #asahi-dev
03:21nora has quit [Ping timeout: 480 seconds]
03:28marvin24 has joined #asahi-dev
03:30hdbngr has quit [Ping timeout: 480 seconds]
03:39hdbngr has joined #asahi-dev
03:48hdbngr has quit [Ping timeout: 480 seconds]
03:55hdbngr has joined #asahi-dev
04:02Yushi has quit [Remote host closed the connection]
04:03Yushi has joined #asahi-dev
04:03hdbngr has quit [Ping timeout: 480 seconds]
04:12hdbngr has joined #asahi-dev
04:19mrkajetanp_ has quit [Ping timeout: 480 seconds]
04:21mrkajetanp has joined #asahi-dev
04:21hdbngr has quit [Ping timeout: 480 seconds]
04:29hdbngr has joined #asahi-dev
04:38nnym has quit [Remote host closed the connection]
04:38hdbngr has quit [Ping timeout: 480 seconds]
04:38nnym has joined #asahi-dev
04:46hdbngr has joined #asahi-dev
04:46svenbringing up dwc3 is probably very wrong but I don’t get why the full phy and dwc reset isn’t enough to fix it again
04:48nnym has quit [Remote host closed the connection]
04:48nnym has joined #asahi-dev
04:54hdbngr has quit [Ping timeout: 480 seconds]
05:00hdbngr has joined #asahi-dev
05:18hdbngr has quit [Ping timeout: 480 seconds]
05:27hdbngr has joined #asahi-dev
05:41hdbngr has quit [Ping timeout: 480 seconds]
05:46hdbngr has joined #asahi-dev
05:57hdbngr has quit [Ping timeout: 480 seconds]
06:05rany has quit [Remote host closed the connection]
06:06hdbngr has joined #asahi-dev
06:06rany has joined #asahi-dev
06:15hdbngr has quit [Ping timeout: 480 seconds]
06:23hdbngr has joined #asahi-dev
06:32hdbngr has quit [Ping timeout: 480 seconds]
06:41sofus[m] has joined #asahi-dev
06:41hdbngr has joined #asahi-dev
06:53rany has quit [Ping timeout: 480 seconds]
06:53rany has joined #asahi-dev
07:02tamar is now known as tamar-away
07:03hdbngr has quit [Ping timeout: 480 seconds]
07:10hdbngr has joined #asahi-dev
07:19hdbngr has quit [Ping timeout: 480 seconds]
07:20chadmed has quit [Quit: Konversation terminated!]
07:20chadmed has joined #asahi-dev
07:26sofus[m]chadmed: i wont have access to m1n1 the next few days, would you be able to dump an avd interchange frame? like the raw pixel data. Ideally for an 128x128 red blob
07:28hdbngr has joined #asahi-dev
07:30sofus[m]yeah, im assuming dcp does not change anything? I have a suspect the rvra buffers could be the interchange things
07:35jannausofus[m]: a solid color tile will be coded in the per-tile metadata and have no pixel data
07:37jannausven: can we use 'cio@' for the "apple,t8103-usb4-acio" nodes? that would simplify dt_disable_missing_devs() in m1n1
07:39* sofus[m]uploaded an image: (19KiB) < https://matrix.org/oftc/media/v1/media/download/AU_bw93t28g7NgthP-WLmfU3vCUbWUJrb_Gn73IeFv1fODw4AUraCxrpydcf3WnZJ-r4OBLi99-4Oshq0kX3Xi1CegR09-egAG1hdHJpeC5vcmcvZ2RvT2htdmRrVHFrSE9LcVRFREdyaE5D >
07:42chadmedbut also if the whole image is one solid colour i would also expect no pixel data and just metadata
07:44sofus[m]its the 2 of the 4 rvra buffers rendered as nv12, i have the raw binary and can generate more
07:44hdbngr has quit [Ping timeout: 480 seconds]
07:46hdbngr has joined #asahi-dev
07:54hdbngr has quit [Ping timeout: 480 seconds]
08:02chadmed has quit [Quit: Konversation terminated!]
08:02hdbngr has joined #asahi-dev
08:03chadmed has joined #asahi-dev
08:11hdbngr has quit [Ping timeout: 480 seconds]
08:20jannausofus[m]: looks plausible. I apparently misremembered where the slod color is saved. the other two buffers are compression metadata
08:20hdbngr has joined #asahi-dev
08:29sofus[m]nice! if its true i assume avd uses interchange internally and converts to standard yuv when requested
08:29hdbngr has quit [Ping timeout: 480 seconds]
08:30minecrell has quit [Read error: Connection reset by peer]
08:31violette has quit [Quit: ZNC 1.10.1 - https://znc.in]
08:31violette has joined #asahi-dev
08:32minecrell has joined #asahi-dev
08:35hdbngr has joined #asahi-dev
08:36jannausven: did the "port" handling in dt_disable_missing_devs() ever do anything useful? as far as I can see it never has found a phandle match because fdt_for_each_subnode() just iterates through the the direct children
08:41WindowPain has joined #asahi-dev
08:44hdbngr has quit [Ping timeout: 480 seconds]
08:51hdbngr has joined #asahi-dev
09:00jannaufeel free to force-push, just wanted to make sure you're not confused that the remote branch changed
09:01hdbngr has quit [Ping timeout: 480 seconds]
09:10hdbngr has joined #asahi-dev
09:10___nick___ has joined #asahi-dev
09:18jannausven: have you seen https://github.com/intel/tbtools ? `tblist -A` shows useful information for host2host connections
09:18hdbngr has quit [Ping timeout: 480 seconds]
09:28hdbngr has joined #asahi-dev
09:34chaos_princessjannau: what is eudo on that cable
09:35svenhttps://f.svenpeter.dev/f9237287e3af548f79fb0d8ceaa85c49b7f8f6ed27a36d07d15d6f86d1292b21_trace_acio.py.txt
09:36sventhe weird thing i saw in chaos's trace is that linux uses "lower uuid initiates lane bonding" but both sides tried to do that which is weird
09:38svenso it should've been linux initiating the thing but i also saw macos trying to start that state machine
09:44hdbngr has quit [Ping timeout: 480 seconds]
09:44chadmed has quit [Read error: Connection reset by peer]
09:52hdbngr has joined #asahi-dev
10:01hdbngr has quit [Ping timeout: 480 seconds]
10:03hdbngr has joined #asahi-dev
10:14chadmed has joined #asahi-dev
10:15chadmedwas anyone able to test the interchange stuff with mutter? im still getting occasional dcp crashes with certain things when all three surfaces are in use (they do not happen when interchange is disabled)
10:18hdbngr has quit [Remote host closed the connection]
10:18hdbngr has joined #asahi-dev
10:27hdbngr has quit [Ping timeout: 480 seconds]
10:34hdbngr has joined #asahi-dev
10:40chadmedthis may still be something to do with mesa. the frequency of crashes can be influenced by when we allow interchange in agx_pipe
10:41chadmedon t8112 i can get it to reliably crash when glmark2-wayland is on surface 1 and kwin clears the cursor surface
10:48pb17 has quit [Ping timeout: 480 seconds]
10:48hdbngr has quit [Ping timeout: 480 seconds]
10:57hdbngr has joined #asahi-dev
11:06hdbngr has quit [Ping timeout: 480 seconds]
11:12pb17 has joined #asahi-dev
11:14hdbngr has joined #asahi-dev
11:25hdbngr has quit [Ping timeout: 480 seconds]
11:32hdbngr has joined #asahi-dev
12:00jannausven: I guess I misunderstood what you wanted to have traced. I traced both sides of a linux/macos connection
12:03chadmed has quit [Quit: Konversation terminated!]
12:07eConnah has joined #asahi-dev
12:08nela has joined #asahi-dev
12:09chadmed has joined #asahi-dev
12:11eConnah has quit []
12:14chadmedapple-dcp 231c00000.dcp: RTKit: Message (id=1): pcc firmware: mprintf_pcc2d_ltm.c:370 isr_inbox0_not_empty() --
12:20chadmed has quit [Read error: Connection reset by peer]
12:24bgb has quit [Remote host closed the connection]
12:26coroner21 has quit [Remote host closed the connection]
12:28chadmed has joined #asahi-dev
12:33bgb has joined #asahi-dev
13:17svenmaybe it's time to write a tracer for tipd + the VSE write and just collect a bunch of data
13:17tg479 has joined #asahi-dev
13:24tg47 has quit [Ping timeout: 480 seconds]
13:24tg479 is now known as tg47
13:30nela has quit [Quit: bye!]
13:31nela has joined #asahi-dev
13:55hdbngr has quit [Remote host closed the connection]
13:55hdbngr has joined #asahi-dev
14:10john-cabaj has joined #asahi-dev
14:25john-cabaj has quit [Quit: john-cabaj]
14:26ndufresne is now known as Guest15635
14:35hdbngr has quit [Ping timeout: 480 seconds]
14:37pb17 has quit [Remote host closed the connection]
14:38pb17 has joined #asahi-dev
15:03svenhttps://f.svpe.de/1e082eb1b38c8d840f4f1daf53f88ca5ecf68fb350381d9fe37f9fc5c1d85e99_trace_tbt_cable.py.txt okay, that should give us the cd3217 regs mac os read + what it wrote to that cable VSE, only tested on t8103
15:04svenshould give a line like this for ever plug: VSE WRITE [0x36] = 0x00000407 | hpm1: STATUS=fd820210 DATA_STATUS=03000304 USB4_STATUS=? INTEL_VID_STATUS=030000000002010100 DP_SID_STATUS=? CF_VID_STATUS=?
15:22yukathe cache_enable part seems quite easy to figure out given a trace, but the tz start/end/enable not so much. I don't understand this part: "by looking at region-id-2 and region-id-4 on a booted macos, in the /chosen/carveout-memory-map DT node"
15:24yukaalso "This can be used along with dumping the mcc reg space" I assume this does not mean dumping the mcc reg space from a booted macOS, but from m1n1? and then looking for values which would result in the carveout addresses?
15:25jannauiirc the addresses from region-id-2/region-id-4 are in the mcc mmio space around the offsets. doesn't need a booted macos just m1n1.
15:26Rafaysven: This is not part of the scope, but I noticed something really weird. MacOS Sonoma with m1n1 -> Linux is getting 1.2Gb/sec and Asahi gets 500Mb/sec but then a non-m1n1 macos Tahoe gets 19/20Gbsec? Same cable, same host on the other side. Is there a way to trace that version of mac os?
15:27svenmight also not be all that helpful, once v1 is on the list i'll look into what slows us down
15:28svenchaos' had one example where RX was >2x faster than TX so there's obviously something terribly inefficient in one of the two paths
15:30RafayOh. But macos sonoma itself is not getting those speeds. So maybe it could be a bug that asahi ends up reproducing since we can only trace sonoma?
15:30sveni don't think there's anything in the trace that will help speed at this point, except for lane bonding
15:31svenjannau: he had 20 Gbps for mac->linux but 7.something for linux->mac on the same connection
15:33jannauwith the fixed router uuid tblist now reports Speed (Rx/Tx): 40/40 Gb/s for linux as well
15:36jannausven: I could reproduce that. additionally linux->linux had the same 7.x gpbs speed (in both directions)
15:37lynndotpy72 has quit [Quit: bye bye]
15:38lynndotpy72 has joined #asahi-dev
15:38chaos_princessin case it matters, i also tried playing with iperf configs, and neither setting the zero copy receive nor doing 9000 mtu helped
15:42svenmy suspicion is something dumb in usb4net that's only triggered on apple platforms but works fine on intel now
15:46jannauno, it wasn't particularly spammy. log file is 2.1M large and contains 2 or 3 connections
16:03chaos_princesswhats up with rust libraries and using "smol" in their names
16:16Rafayjannau: I have an intel controller as well and I have to power off, power on and then pci rescan to get it working.
16:24elvishjerricco5 has joined #asahi-dev
16:31elvishjerricco has quit [Ping timeout: 480 seconds]
16:31elvishjerricco5 is now known as elvishjerricco
16:48svenaaand while focusing on xdomain i somehow broke device so bad that acio doesn't shut down anymore once i disconnect it again :D
16:52rhysmdnz has quit [Quit: Bridge terminating on SIGTERM]
16:52Guest15587 has quit [Quit: Bridge terminating on SIGTERM]
17:01___nick___ has quit [Ping timeout: 480 seconds]
17:03___nick___ has joined #asahi-dev
17:10rhysmdnz has joined #asahi-dev
17:11Guest15641 has joined #asahi-dev
17:46Rafaysven: It was almost entirely m1n1 overhead. After not tracing acio like jannau said, I get 15Gbps which is the max the other host supports.
17:55leah has quit [Remote host closed the connection]
17:56leah has joined #asahi-dev
18:00svenyup, it's hanging at wait_for_completion(&anhi->nhi.domain_released) and something still has a reference to that domain
18:15pb17 has quit [Ping timeout: 480 seconds]
18:16pb17 has joined #asahi-dev
18:19sveni'm willing to bet it's d6d458d42e1e ("thunderbolt: Handle DisplayPort tunnel activation asynchronously")
18:19sventhat was introduced after the previous tbt branch that worked and it already had a bunch of subtle fixes
18:49yukathe mcc regs on t8132 are funny: mcc[7]+0x704 = 0x103e6da8, mcc[7]+0x708 = 0x103ffda7, which would be start and end of carveout-memory-map region-id-2
18:51Felixonvacationtill26ofAugust[ has joined #asahi-dev
18:52yukawell, if I assume stride = 0x14, then the entries above and below all have start = end = 0x10000000
18:57jannauthose entries are not enabled though. looks the same on t8122. I wouldn't be surpised if those entries even match region-id-2/4 (1 indexed)
19:04roxfan2 has joined #asahi-dev
19:09roxfan has quit [Ping timeout: 480 seconds]
19:13yukajannau: that means if I got a match for region-id-2, then its start/end value minus 1*stride should be the ones in the tz_regs?
19:19jannauthe slower side direction has tcp retransmits (50 - 300) while the faster variant has none. not sure if that is enough to explain the difference
19:21bgb has quit [Remote host closed the connection]
19:21bgb has joined #asahi-dev
19:33pb17 has quit [Ping timeout: 480 seconds]
19:42pb17 has joined #asahi-dev
20:04bgb has quit [Remote host closed the connection]
20:04bgb has joined #asahi-dev
20:07___nick___ has quit [Remote host closed the connection]
20:08bgb has quit [Remote host closed the connection]
20:09bgb has joined #asahi-dev
20:49millefy14 has quit [Quit: Ping timeout (120 seconds)]
20:50LuisHenriques[m] has joined #asahi-dev
20:58millefy14 has joined #asahi-dev
21:27___nick___ has joined #asahi-dev
21:44svenand if the RX direction constantly churns iommu mappings and invalidates DART all the time that might just be very expensive
21:53nela has quit [Quit: Ping timeout (120 seconds)]
21:53nela has joined #asahi-dev
22:21linuxgemini is now known as Guest15656
22:22Guest15656 has quit [Remote host closed the connection]
22:22linuxgemini has joined #asahi-dev
23:58pb17 has quit [Ping timeout: 480 seconds]