#
#asahi-dev · 2025-06-03 12:00
jannauahdeb[mds]: the `mv -f "$TARGET" "${TARGET}.old"` is unchanged in that commit. using `cp --preserve=timestamps -f` should make the switch atomic. feel free to create a PR
#asahi-dev · 2025-06-03 11:02
nickchanwdt and serial would help bringup too
#asahi-dev · 2025-06-03 11:01
nickchan /chosen/framebuffer should be enough for display
#asahi-dev · 2025-06-03 11:00
chadmedm1n1 should fill that with the iboot framebuffer which will work with simpledrm
#asahi-dev · 2025-06-03 11:00
fl0_idsure. I think I looked at that one too. and very little if anything should be extra. I only though about adding the display, as it's a laptop, but hopfeully framebuffer works without as this needs DART or sth I think
#asahi-dev · 2025-06-03 10:59
chadmeddont we need the 1ghz clock too for m3+?
#asahi-dev · 2025-06-03 10:58
nickchanWhich can be ignored and removed in bringup
#asahi-dev · 2025-06-03 10:58
nickchanthat one's style is too old and the one i showed should be fine aside from pinctrl
#asahi-dev · 2025-06-03 10:57
fl0_idand clusters
#asahi-dev · 2025-06-03 10:57
fl0_idthough I adapted a bit to fit newer style, f.e. using cpu-map
#asahi-dev · 2025-06-03 10:57
fl0_idI was in the past pointed to this as a minimal device tree https://github.com/AsahiLinux/linux/commit/7d2d16ccf15d8eb84accfaf44a0b324f36e39588#diff-845c07f30f0f94856b22d7b50c0ff3d27a961abeffefcdfab01f2e8cd64569cc
#asahi-dev · 2025-06-03 10:56
fl0_idthx I will take a look again later. I'm very aware that there are still issues, likely I did not conisder the ranges property
#asahi-dev · 2025-06-03 10:56
nickchani guess i can think about auto generate adt cause doing it for 69 (this is the actual number of pre-M1 skus supported in linux) skus is not fun
#asahi-dev · 2025-06-03 10:55
nickchanwdt is probably fine to be in first device tree
#asahi-dev · 2025-06-03 10:54
nickchan<https://github.com/AsahiLinux/linux/commit/d23457abac1e2f309404e63c7a390d9a32bce9e6> this is what a minimal device tree would look like and id get rid of pinctrl for initial bringup even
#asahi-dev · 2025-06-03 10:54
chadmedstuff like aic and the serial port can probably happen
#asahi-dev · 2025-06-03 10:53
chadmedeven if theyre perfect
#asahi-dev · 2025-06-03 10:53
chadmedtheres very little point to expending the engineering effort until the bindings are upstream because they _will_ change
#asahi-dev · 2025-06-03 10:52
chadmedyeah
#asahi-dev · 2025-06-03 10:52
nickchaneverything else is still manual
#asahi-dev · 2025-06-03 10:52
nickchanBut over the years for the most part the only thing that materalized is pmgr_adt2dt
#asahi-dev · 2025-06-03 10:52
nickchanthe plan was in fact to generate parts of fdt from adt offline and then add the generated results to repo
#asahi-dev · 2025-06-03 10:51
chadmedonce a bunch more bindings find their way upstream it would be nice for m1n1 to generate them from the ADT where possible
#asahi-dev · 2025-06-03 10:50
nickchanranges in ADT I mean
#asahi-dev · 2025-06-03 10:50
nickchanchadmed: that's what the ranges property do
#asahi-dev · 2025-06-03 10:49
chadmedthe addresses probably all need offsetting by 0x200000000 or something like that
#asahi-dev · 2025-06-03 10:49
nickchanand t6031 aic is most definitely not compatible with "apple,aic" which refers to s5l8960x's aic
#asahi-dev · 2025-06-03 10:48
nickchansome of the address also look suspicious - make sure that you considered ADT's ranges property when computing addresses
#asahi-dev · 2025-06-03 10:48
chadmedits also just important in general because the dt maintainers are (rightfully) fastidious about that stuff
#asahi-dev · 2025-06-03 10:48
chadmedits important in compatible strings that they are consistent
#asahi-dev · 2025-06-03 10:48
nickchanfl0_id: also the 24 mhz clock should be named differently check t8103.dtsi
#asahi-dev · 2025-06-03 10:46
fl0_idnot sure how important those small differences are. I have the c, but supposedly there is also an m variant
#asahi-dev · 2025-06-03 10:46
fl0_idI at least have some inconsistency still where sometimes I have j514 but in other places like the bindings I use j514c
#asahi-dev · 2025-06-03 10:45
fl0_idI didn't give any errors after adding the bindings
#asahi-dev · 2025-06-03 10:45
fl0_idnickchan I did that
#asahi-dev · 2025-06-03 10:44
nickchannormally i would recommend running dtbs_check but the downstream tree doesn't adhere to that so i can't recommend that here
#asahi-dev · 2025-06-03 10:44
fl0_idnickchan thx.
#asahi-dev · 2025-06-03 10:43
nickchanfl0_id: #size-cells should be 2 under chosen
#asahi-dev · 2025-06-03 10:39
fl0_idstill have to test it though ...
#asahi-dev · 2025-06-03 10:39
fl0_idmade a super basic M3 device tree https://github.com/step21/linux-asahi/commit/403b936c9e97362dbffee3a01f0f3b21fbea6bf2
#asahi-dev · 2025-06-02 23:59
amarioguyjust post it whenever you're done, i read IRC backlogs often here so dw about me missing it or anything
#asahi-dev · 2025-06-02 23:58
amarioguyfl0_id: any update on the MSR?
#asahi-dev · 2025-06-02 22:38
fl0_idah ok so compatible etc is in Documentaton/devicetree/bindings I think...
#asahi-dev · 2025-06-02 22:37
fl0_idfor the second, I though the answer should be easy to find but well...
#asahi-dev · 2025-06-02 22:22
fl0_idand is there a good way to tell make dtbs_check to only redo one folder / files? make dtbs_check DTB_SCHEMA_FILES=/folder but AFAIK that is only for schema files, not dts
#asahi-dev · 2025-06-02 22:20
fl0_idok dumb question potentially. in the DTs, when you have "compatible = "apple,icestorm";" or similar, where are these defined? I naively updated them, but then dtc complains that it's not defined...
#asahi-dev · 2025-06-02 22:03
fl0_idmixi thx.
#asahi-dev · 2025-06-02 19:58
kettenisit is mostly generated by m1n1 from the ADT
#asahi-dev · 2025-06-02 19:51
amarioguykettenis: actually, i saw a commit recently adding some SMBIOS handling to m1n1 - i assume the FDTs are supposed to have necessary SMBIOS information in the future?
#asahi-dev · 2025-06-02 19:23
kettenisso is that produced as part of running rustup to install the cross-compile toolchain?

plus ›