#
#asahi-dev · 2025-05-25 11:04
cy8aerand it occurs around every 90s
#asahi-dev · 2025-05-25 10:59
cy8aerI get something like this: https://paste.debian.net/1376491/
#asahi-dev · 2025-05-25 10:57
cy8aersven: I try, thanks
#asahi-dev · 2025-05-25 10:56
sventry to get a log with btmon while that happens
#asahi-dev · 2025-05-25 10:48
cy8aerI have messages like "Unable to get Hands-Free Voice gateway SDP record: Host is down" for some time now. And with this bluetooth headsets are knocked out for some time (for me Momentum 4 or Linkbuds S). I wonder if this is a bluez problem or really a problem of the actual bluetooth stack on the kernel side?
#asahi-dev · 2025-05-25 06:34
jannaucomparing them in m1n1 with the ADT wouldn't a bad idea as well
#asahi-dev · 2025-05-25 06:33
jannauif we were to keep them I'd create .dtsi files with all gpu properties and include them directly in device .dts
#asahi-dev · 2025-05-25 06:31
jannauI think splitting the values over multiple .dtsi/.dts is not a good idea
#asahi-dev · 2025-05-25 06:29
jannauchaos_princess: yes. also our dts' were buggy. see 25a6e10c983dc and b8047d540dca1
#asahi-dev · 2025-05-24 23:23
allmazzi was going to capture a trace, but... this is my only machine now
#asahi-dev · 2025-05-24 23:17
allmazzmaybe there are some assumptions about the difference between aop on m2 and m2 max?
#asahi-dev · 2025-05-24 22:14
chaos_princessalso, probably fixed cs/afr init, affects t602x
#asahi-dev · 2025-05-24 21:44
chaos_princessjannau: am i reading this right that ppm-ki/kp and similar are different on the studio/pro?
#asahi-dev · 2025-05-24 12:34
alxdrlwill move to #ci-tron
#asahi-dev · 2025-05-24 12:33
alxdrlcs+pico assembled and working, flux really helps for clean solder joints lol
#asahi-dev · 2025-05-24 11:10
svennow let's see about atcphy and suspend
#asahi-dev · 2025-05-24 11:10
svensmc-hid needs minimal changes but works just fine with the smc that's on list then: https://git.kernel.org/pub/scm/linux/kernel/git/sven/linux.git/log/?h=wip/smc-hid
#asahi-dev · 2025-05-24 10:46
sventhis almost looks reasonable now :D
#asahi-dev · 2025-05-24 10:43
svenso i'll just drop it and keep it enabled to make the mmio trace match whatever xnu does
#asahi-dev · 2025-05-24 10:43
svenand i think that's what the parameter switches to
#asahi-dev · 2025-05-24 10:43
sveni can do the pipehandler switch with dwc3 in reset without the lock dance or BIST thingy
#asahi-dev · 2025-05-24 10:43
svenokay, so the pipehandler/BIST dance clearly smells like a workaround for _something_
#asahi-dev · 2025-05-24 09:32
sventhey also seem to get away without the entire dwc3/atcphy sync mess
#asahi-dev · 2025-05-24 08:37
jannauof course different because tipd has no integrated retimer
#asahi-dev · 2025-05-24 08:33
jannaumost of the x1 elite devices seems to use https://www.paradetech.com/products/ps8830/ https://lore.kernel.org/linux-arm-msm/20241112-x1e80100-ps8830-v5-0-4ad83af4d162@linaro.org/ as type-c controller
#asahi-dev · 2025-05-24 08:20
svenand those also make sense to describe the hardware I think. there have to be some lanes from the PHY to both
#asahi-dev · 2025-05-24 08:20
svenI might need one to the tbt nhi and maybe even to dwc3 for usb4 tunneling
#asahi-dev · 2025-05-24 08:17
svenI think I'll just start with the connection to the type-c controller and see if we need the others later
#asahi-dev · 2025-05-24 08:15
svenok, ports subnode. looks reasonable
#asahi-dev · 2025-05-24 08:13
sveni think that one's still missing the connection to tipd, etc.
#asahi-dev · 2025-05-24 08:13
svennice, let's see how that compares to the atcphy binding I made up
#asahi-dev · 2025-05-24 08:11
jannauthe phy is Documentation/devicetree/bindings/phy/qcom,sc8280xp-qmp-usb43dp-phy.yaml but not usb4/pcie-tunneling support there either
#asahi-dev · 2025-05-24 08:10
jannauI can't find any usb4 patches
#asahi-dev · 2025-05-24 07:55
svenpity, would be great if this was based on the same phy ip *and* if they implemented tbt for us :D
#asahi-dev · 2025-05-24 07:48
jannauin 6.15 there seems to be just usb3 and dp-alt support but it's dwc3 based
#asahi-dev · 2025-05-24 07:46
svenhuh, i somehow missed that had usb4 support these day
#asahi-dev · 2025-05-24 07:42
jannauwhat does qualcomm use? they use at least dwc3 and their eusb2 hub seems to be synopsys based as well
#asahi-dev · 2025-05-24 07:41
chadmed_they have versions for tsmc and samsung pdks apparently so who knows
#asahi-dev · 2025-05-24 07:40
chadmed_they got it certified on an fpga, i dont think there are any hw implementations at all yet lol
#asahi-dev · 2025-05-24 07:38
sventhey claim to have usb4 ip but I haven’t seen that used anywhere last time I checked
#asahi-dev · 2025-05-24 07:37
svenI still wonder if atcphy was done in collaboration with synopsys
#asahi-dev · 2025-05-24 07:25
chadmed_i figured amd would tape out their own thunderbolt block at some point but theyll probably just copy intels homework if they ever do at this point
#asahi-dev · 2025-05-24 07:24
chadmed_yeah
#asahi-dev · 2025-05-24 07:22
sventhat’s the other argument for this: everyone else is likely just going to stuff the common Intel chip in there and call it a day ;)
#asahi-dev · 2025-05-24 07:22
chadmed_let me cope
#asahi-dev · 2025-05-24 07:22
svenriiiiight
#asahi-dev · 2025-05-24 07:22
chadmed_surely, suuureeeely at some point someone else is going to come along with a similarly non-intel-reference-design tbt platform
#asahi-dev · 2025-05-24 07:21
chadmed_let someone get paid to do that
#asahi-dev · 2025-05-24 07:21
sventhe logic there is still brittle and spread across tipd,dwc3 and atcphy but it’s the best we can do without a major refactor of everything usb I think
#asahi-dev · 2025-05-24 07:19
svenbut the atcphy side of tbt is included

plus ›