#asahi-dev · 2026-08-16 10:37
jannauyes, we want shutdown-gpios and not reset-gpios. I don't think the reset controler will work for this#asahi-dev · 2026-08-16 10:20
jannauthe tas27xx dt-bindings example even uses the gpio pin for reset and shutdown#asahi-dev · 2026-08-16 10:14
jannauhuh, both drivers also use shutdown-gpios and map this to sdz#asahi-dev · 2026-08-16 10:10
jannauI think we just have to change the drivers from using the reset gpio pin directly via gpiod to the reset controller framework#asahi-dev · 2026-08-16 10:04
jannaulooking at the tas2764/tas2770 data sheets I don't see a reset pin, just sdz and power supply oins#asahi-dev · 2026-08-16 09:42
jannaubut why would using the shared reset controller require binding changes? I thought it was designed to work with the existing "reset-gpios" without changes#asahi-dev · 2026-08-16 09:33
jannauchadmed: I wouldn't be sure of that given that both bindings are written by TI#asahi-dev · 2026-08-16 09:32
jannauyuka: https://www.youtube.com/watch?v=GUSFOP_lFmA or the "Bringing up Asahi Linux on the M1 Pro" series might be relevant. Those are just before the initial src/mcc.c commit#asahi-dev · 2026-08-16 09:18
jannauT6031_DCS_STRIDE is 0x200000. are you testing on T6000? maybe it's just wrong since it is never used#asahi-dev · 2026-08-16 09:07
jannauwe probably can switch to parsing segment-ranges#asahi-dev · 2026-08-16 09:05
jannauchaos_princess: re dcp carveouts, matching segment-ranges against carveout-memory-map/region-id-*. we started with matching pre-existing mapped regions in dart-{dcp,disp}* against carveouts but if I'm not mistaken it's just the annotated segments#asahi-dev · 2026-08-15 10:45
jannaulooks like a factor 10 error is mine. tracing counts 220k thunderbolt_net:tbnet_rx_ip_frame events per second and not 25k I calculated with#asahi-dev · 2026-08-15 10:32
jannau~20 us per map or unmap page (~1 gigabyte per second and 4k payload per package) feels at least a order of magnitude too slow#asahi-dev · 2026-08-15 10:07
jannauerr, I missed that it unmaps and reallocates buffers in tbnet_poll (the receive path)#asahi-dev · 2026-08-15 09:53
jannauit allocs a fixed number of buffers for rx and tx#asahi-dev · 2026-08-15 09:40
jannauyes#asahi-dev · 2026-08-15 09:38
jannauthe perf profile shifted towards dart_map_pages https://paste.centos.org/view/9576827b#asahi-dev · 2026-08-15 09:36
jannaufurther increase to 10.5 Gbits/sec#asahi-dev · 2026-08-15 09:32
jannauincreased throughput from 7.2 to 8.8 Gbits/sec by forcing the irq to performance core and switching cpufreq gov to performance#asahi-dev · 2026-08-15 09:26
jannaunot sure that perf samples correctly (I see less samples than I would expect) but https://paste.centos.org/view/cc49f0f1#asahi-dev · 2026-08-15 09:22
jannauwe can on t8103#asahi-dev · 2026-08-15 09:21
jannauwhy are all AIC interrupts served on core 0 (t8103)#asahi-dev · 2026-08-15 09:18
jannaupossibly with https://github.com/AsahiLinux/m1n1/pull/563#asahi-dev · 2026-08-15 09:17
jannausmells like irq though. on the receiving side a single core is at ~90% in soft irq state. for some reason the first (efficiency) core#asahi-dev · 2026-08-15 09:06
jannauI'd expect that the intel laptop uses an iommu as well so it could only be an issue how well iommu and driver handle the usage pattern#asahi-dev · 2026-08-15 09:03
jannausven: when forcing usb4_dart to identity tbt breaks with "thunderbolt-apple-nhi 501f00000.nhi: error -ENODEV: Failed to init software connection manager"#asahi-dev · 2026-08-14 20:33
jannauusb4stream is faster, 13.5 gbps instead of 7.2#asahi-dev · 2026-08-14 19:20
jannauthe faster variant is then the linux side transmits data#asahi-dev · 2026-08-14 19:19
jannauthe slower side direction has tcp retransmits (50 - 300) while the faster variant has none. not sure if that is enough to explain the difference#asahi-dev · 2026-08-14 19:03
jannaut8103/macos <> intel/linux does 30 / 24 Gbit/s in iperf3 over tbt net#asahi-dev · 2026-08-14 18:57
jannauthose 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)#asahi-dev · 2026-08-14 16:22
jannauyes#asahi-dev · 2026-08-14 16:21
jannauno working tbt_net, and only announced as 20gbps#asahi-dev · 2026-08-14 16:15
jannauhuh, after waiting 1400 seconds the thunderbolt connection to the intel notebook is up#asahi-dev · 2026-08-14 15:50
jannauno, just usb / xhci messages#asahi-dev · 2026-08-14 15:48
jannauucsi_acpi reports helpfully "unknown error 0" and "UCSI_GET_PDOS failed (-5)"#asahi-dev · 2026-08-14 15:47
jannaustill doesn't connect with my intel work notebook#asahi-dev · 2026-08-14 15:46
jannauno, it wasn't particularly spammy. log file is 2.1M large and contains 2 or 3 connections#asahi-dev · 2026-08-14 15:43
jannauwithout tracing back to 7.x Gbits/sec#asahi-dev · 2026-08-14 15:42
jannauthunderbolt0 mac address and ipv4 link local address differ now#asahi-dev · 2026-08-14 15:41
jannauerr, with acio tracing#asahi-dev · 2026-08-14 15:41
jannauI'm now down to 12.8 Mbits/sec for linux2linux#asahi-dev · 2026-08-14 15:36
jannausven: I could reproduce that. additionally linux->linux had the same 7.x gpbs speed (in both directions)#asahi-dev · 2026-08-14 15:33
jannauwith the fixed router uuid tblist now reports Speed (Rx/Tx): 40/40 Gb/s for linux as well#asahi-dev · 2026-08-14 15:30
jannauthe factor 2 might be lane bonding though#asahi-dev · 2026-08-14 15:26
jannauit's helpful to look what happens on an already supported device#asahi-dev · 2026-08-14 15:25
jannauiirc 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.#asahi-dev · 2026-08-14 13:52
jannauit's not used. I supposed it's derived for newer gens becasue it was in the adt for m1#asahi-dev · 2026-08-14 12:01
jannauI'll trace t8103 linux to linux later and add it to the gist#asahi-dev · 2026-08-14 12:01
jannauso the 40gbps connection