#asahi · 2023-09-28 13:17
leioI was hoping to get a proxy from stage1 with the boot-args, but yeah, something's up with that#asahi · 2023-09-28 13:13
leioI ran csrutil disable and chose the fedora APFS out of the choices#asahi · 2023-09-28 13:12
leiook, I'll keep that in mind after I've got a patched u-boot binary#asahi · 2023-09-28 13:12
leioI understood nvram option should give me the opportunity to connect to ttyACM and not have to do the "delete boot.bin" dances#asahi · 2023-09-28 13:10
leiogoes looking how u-boot is built for fedora#asahi · 2023-09-28 13:10
leioright, so that explains why it doesn't boot, but not why I have no boot-args=-v honoring#asahi · 2023-09-28 13:09
leiothe nvram boot-args=-v stuff didn't work for me either - if boot.bin (stage2) is found, I never get any ttyACM* anymore#asahi · 2023-09-28 13:09
leioso I need to check into that#asahi · 2023-09-28 13:09
leioand it has M2 ultra support as last commit to board.c, which I'm not sure whether is included in my binary or not#asahi · 2023-09-28 13:08
leiohmm, looking into it I clone the u-boot repo and looked at history#asahi · 2023-09-28 13:04
leioI guess I should make sure m1n1 stage2 does go through the cur_boot_args.video.base conditional that sets up a bunch of framebuffer things with fdt_setprop calls#asahi · 2023-09-28 13:00
leioI need to drop in some vars to get u-boot serial console?#asahi · 2023-09-28 13:00
leio(that output and logo are from stage2 then, as the current m1n1 patches don't set up the framebuffer base or something, so only the stage2 gets that going to display stuff on monitor)#asahi · 2023-09-28 12:58
leioI also installed a chainloading patched m1n1 as stage1 in the hope that the display is initialized (and that's why I see the logo now and some m1n1 output scroll by)#asahi · 2023-09-28 12:58
leiofor update-m1n1 I modified the sysconfig file in fedora to point to a patched m1n1 instead and just re-ran it. The dtb's it picks up already come from the new kernel package and include the new dtb's. But afterwards I just get no output beyond some m1n1 scroll-through and the logo on anything to know what's wrong#asahi · 2023-09-28 12:56
leioso my conundrum freshly summarized is: I have built a patched fedora kernel rpm (for j475d dtb and hdmi and other support on it) and installed it, but after update-m1n1 I'm not getting any output on anything and can't know what's going on. I see a asahi logo in the middle of the screen, I think it comes from m1n1#asahi · 2023-09-28 12:54
leioand doesn't have any output support over a ttyACM thing like m1n1?#asahi · 2023-09-28 12:52
leiois u-boot supposed to show any output on monitors? I don't really have a comparison to know from other hardware#asahi · 2023-09-28 12:48
leiocan I somehow grab the Image.gz equivalent (to feed into linux.py or run_guest_kernel) out of a EFI UKI kernel image?#asahi · 2023-09-27 19:28
leiochainloading the u-boot binary yield the same result of no output on anything, I guess that's my problem (on top of for some reason not getting m1n1 output unless payload is missing), but can't be sure as maybe I need to load it dtb's or something too that I don't know how to#asahi · 2023-09-27 19:01
leioif I move away boot.bin to force stage1 into proxy mode anyways, I get in there (due to no payload, etc). Chainloading with proxyclient that boot.bin makes it not able to reconnect afterwards, so I don't get any debug either. If I chainload only the m1n1 part (what I fed to update-m1n1), then it works fine#asahi · 2023-09-27 18:53
leiomy stage1 isn't even a release build#asahi · 2023-09-27 18:45
leio(because the hdmi output patches for M2 ultra Mac Studio only get picture visible on second run of m1n1 right now)#asahi · 2023-09-27 18:44
leiowhich suggests that it does load stage2#asahi · 2023-09-27 18:44
leioyeah, I don't have any, I'm just staring at asahi logo now :(#asahi · 2023-09-27 18:43
leioI did `csrutil disable`, I have nvram boot-args=-v done, but I'm not getting any ttyACM devices anymore when I have patched m1n1.bin in both stage1 and stage2. I think something is probably up with the kernel boot or something, but I can't debug it without any output#asahi · 2023-09-27 16:56
leiodo I need to patch u-boot as well for M2 ultra if it's used or something? I don't get any picture with only stage2 m1n1 supporting HDMI out, it doesn't boot and I don't get any debug serial now like this#asahi · 2023-09-27 16:43
leioso I got the fedora system booting again and see that update-m1n1 will have all the components available, so yeah, I'll just point that to my patched m1n1#asahi · 2023-09-27 16:40
leioI removed it to get it to work again, yes#asahi · 2023-09-27 16:39
leionvm, something with the higher resolution hdmi again, it looks like, I'll deal with it#asahi · 2023-09-27 16:37
leiomeh, now after second m1n1 chainload I'm not getting /dev/m1n1 back :(#asahi · 2023-09-27 16:34
leioand I don't know where I'd get the other components from it#asahi · 2023-09-27 16:34
leiotbh, I'm not sure what I'm using, I think fedora used that, but it doesn't have the patched m1n1 there#asahi · 2023-09-27 16:30
leiois there some tool to take apart a boot.bin to its components, so I could quickly just replace the m1n1 stage2 part?#asahi · 2023-09-26 17:33
leiook, if there's no underlying issue here beyond 4K just taking more time, it's good, don't let me delay what I want to have ready to test too :)#asahi · 2023-09-26 17:33
leiothat EPIC IOP stuff is visible on the monitor itself too#asahi · 2023-09-26 17:32
leioit eventually works, changed some counter from 100 to 150 in proxy.py and it went fine#asahi · 2023-09-26 17:26
leioedited proxy.py to try a bit longer for now#asahi · 2023-09-26 17:25
leiojanneg: it feels like some kind of new "EPIC: IOP returned 0xe00002c0" spam followed by "display: waited 100000 ms for display status display: failed to get display status" is taking too long time with the higher resolution and the proxyclient scripts time out#asahi · 2023-09-26 16:30
leiowhich studio? :)#asahi · 2023-09-26 16:26
leiobut got 4K software rendered GNOME, nice#asahi · 2023-09-26 16:23
leiojanneg: looks like the 4K resolution display var has some weird behaviour for the chainload (e.g. often can't reconnect after the chainload), but looks like I got the higher resolution and also reconnect sometimes. Might be unrelated wonkliness#asahi · 2023-09-25 22:18
leioI hope lack of HDMI audio is just about getting the kernel vs initramfs sorted, not something that needs more work#asahi · 2023-09-25 22:16
leiojanneg: thanks, I'll try the resolution tomorrow; let me know if I can test anything re gpu or otherwise, or be a coding slave of some sort (but probably hard to be useful that way), especially in areas scratching my itches too#asahi · 2023-09-25 21:46
leiojanneg: I'm in Fedora, thanks a lot! Of course after all this, the natural question is - can I get a higher resolution than 1920x1080 and where are things with GPU for these? :)#asahi · 2023-09-25 20:38
leiojanneg: me copying the firmware over LAN to build it into the kernel temporarily would be kosher by Apple? :)#asahi · 2023-09-25 20:36
leioCONFIG_FW_LOADER_USER_HELPER was already enabled#asahi · 2023-09-25 20:35
leioI see re u-boot#asahi · 2023-09-25 20:34
leioI can't spot anything for this deferred stuff#asahi · 2023-09-25 20:31
leiothat's builtin in fedora asahi config already too