#
✕ nick:jannau
#asahi-dev · 2026-03-21 13:33
jannauand marking the twp power-doimains always on
#asahi-dev · 2026-03-21 13:32
jannauchaos_princess: it was working on m1 max with "val |= 0x60000000;
#asahi-dev · 2026-03-21 13:17
jannauyour pmp device tree misses pio-ranges for t8112
#asahi-dev · 2026-03-21 13:11
jannaunot sure. I got pmp start but it crashes, likely after a pmp_report
#asahi-dev · 2026-03-21 12:56
jannauyes, either there is a race or macos doesn't set one. the iova in the ack is 0 and that it's not mapped in dart
#asahi-dev · 2026-03-21 12:36
jannauno, dump looks correct. might be my fault by setting DCAP to 8, while macos doesn't touch it on M2
#asahi-dev · 2026-03-21 11:46
jannauI guess I need to trace pmp on M2, bootargs offset seems to be wrong, keys don't look like fourccs
#asahi-dev · 2026-03-21 09:34
jannauI think we should prepare a m1n1 release soon (after merging PMP)
#asahi-dev · 2026-03-21 09:32
jannauI was waiting / trying to look into PMP
#asahi-dev · 2026-03-19 12:25
jannauI'm still getting 302 but that's probably still be cached
#asahi-dev · 2026-03-19 12:13
jannaucan't we just use fetch_uncached for the installer-data as well to avoid the CD pipeline?
#asahi-dev · 2026-03-19 12:09
jannausee `curl -v https://alx.sh/`
#asahi-dev · 2026-03-19 12:08
jannauI get the prod installer directly from 2400:52e0:1e00:2::1332:1 which belongs to bunny
#asahi-dev · 2026-03-19 12:04
jannauare you sure? https://alx.sh/ comes from bunny while https://alx.sh/dev redirects to github
#asahi-dev · 2026-03-19 11:59
jannauthe dev installer only uses INSTALLER_DATA=...github..., let's keep it that way until someone complains
#asahi-dev · 2026-03-19 11:55
jannauI just remembered to check if we have to update manually or if it automatically mirrors the source
#asahi-dev · 2026-03-19 11:49
jannauchadmed_: was https://alx.sh/installer_data.json (INSTALLER_DATA_ALT) intentionally removed? It was added for places that block github
#asahi-dev · 2026-03-18 10:45
jannauthere are still cases like EARLY_UART_BASE where the ability to send VDMs from the host is helpful
#asahi-dev · 2026-03-18 10:38
jannaum1n1 needs support for the USB port controller to enable it. so there is more code required for early bringup without the ability to send VDMs
#asahi-dev · 2026-03-18 10:23
jannauand $soon (with debug usb / dockchannel-uart) the only advantage of a second mac/cs board will be triggering reboots via usb pd
#asahi-dev · 2026-03-17 16:10
jannaudcp and isp have it iirc. smc is probably fine
#asahi-dev · 2026-03-17 16:07
jannauit's unfortunate that it might require adapting the bindings for each firmware abi but it's a price I'm willing to pay / argue for
#asahi-dev · 2026-03-17 16:06
jannauwe're already hiding the complexity behind the current bindings. I don't think it makes sense to add to the complexity behind the scene
#asahi-dev · 2026-03-17 16:04
jannauI think I'd prefer to being explicit and not trying to retrofit changes into the existing bindings
#asahi-dev · 2026-03-17 08:40
jannau.get_reg() does that
#asahi-dev · 2026-03-17 08:39
jannauxarkes: addresses in the ADT needs to be translated based on /arm-io/ranges
#asahi-dev · 2026-03-15 17:38
jannaufmzakari: Hej
#asahi-dev · 2026-03-15 16:36
jannauwith a mini as target you could try if it recovers after being off power. I was using a laptop so hard to be sure that it is without power
#asahi-dev · 2026-03-15 16:35
jannaunoopwafel: I think I hit the same or similar thing. Not sure which of the two port controllers were confused. I suspect it was the the one in the target device as reboot worked as soon as booted into recovery
#asahi-dev · 2026-03-15 14:10
jannausome of the proxyclient commands already reset the panic counter
#asahi-dev · 2026-03-15 12:33
jannauisn't it mostly independent and we just need to ensure that the fifo still works with mtp
#asahi-dev · 2026-03-15 12:28
jannauhave you started dockchannel-uart linux (and u-boot) drivers? if not I guess we could add those to the yak list
#asahi-dev · 2026-03-15 12:08
jannau`tuxvdmtool -c "back left" reboot debugusb` will be a little nicer than manually looking up i2c bus and address. https://github.com/AsahiLinux/tuxvdmtool/pull/8
#asahi-dev · 2026-03-14 20:38
jannau+w
#asahi-dev · 2026-03-14 20:37
jannauwonders why "dmem_regioning.c" wasn't hashed no that I can look at the wonderful iboot serial output
#asahi-dev · 2026-03-14 20:29
jannauanother advantage of debugusb, it works with Apple's USB 2.0 charging cables, less relevant now as everything except the macbook neo has magsafe
#asahi-dev · 2026-03-14 19:31
jannauwe probably want to add `{mac,tux}vdmtool reboot debugusb`
#asahi-dev · 2026-03-14 19:11
jannaunicolas17: unclear. at least 2 and probably not more than 4 but I suppose exclave dcp will always make it work
#asahi-dev · 2026-03-14 19:07
jannauthe two dcpext on M4 wasn't widely reported before m4 macbooks launched although it was already in the m4 ipad pro ADT
#asahi-dev · 2026-03-14 19:05
jannauI still wonder why apparently none of apple leaks/rumor sources look at the ADT
#asahi-dev · 2026-03-14 19:02
jannaunickchan: and tons of :sotra_h17d_* conditionals
#asahi-dev · 2026-03-14 18:59
jannausven is instead focusing on thunderbolt rdma support in m1n1 :D
#asahi-dev · 2026-03-14 18:57
jannaucdc acm is more than 22 times faster on m1, 112 Mbps
#asahi-dev · 2026-03-14 18:52
jannaum1 has faster debug usb, ~5Mbps. possibly CPU frequency related
#asahi-dev · 2026-03-14 17:44
jannauand hopefully we can use it on all devices which have not this extra cursed usb setup
#asahi-dev · 2026-03-14 17:43
jannaucdc acm device mode is much faster
#asahi-dev · 2026-03-14 17:42
jannaudoes it matter? I'd expect that the
#asahi-dev · 2026-03-14 16:36
jannaudebug USB on A18 Pro is faster than uart. I measured 2.7Mbps for reading the framebuffer. so clearly faster than the max 1.5Mbps via uart, especially since reconfiguring the baudrate doesn't work
#asahi-dev · 2026-03-14 16:31
jannauxarkes: how are you experimenting? via proxy shell? does running m1n1 as guest still work? at least the latter case SErrors are recoverable. Not sure about the shell
#asahi-dev · 2026-03-14 08:33
jannauyou can probably guess the cpufreq registers with a little bit of experimentation

plus ›