#asahi · 2025-12-17 10:27
pitustyeah#asahi · 2025-12-17 10:27
chaos_princessis this after the kasan complaint?#asahi · 2025-12-17 10:27
pitustthe caller of __kvmalloc_node_noprof is <asahi::driver::AsahiDriver as kernel::drm::driver::Driver>::IOCTLS::ASAHI_VM_BIND#asahi · 2025-12-17 10:26
pitustwell the oops is unrecoverable#asahi · 2025-12-17 10:25
JamesB192probably, armchair idiocy suggests building, restarting, then stepping aside for a drink, a snack, and reading a pulped dead tree slice#asahi · 2025-12-17 10:22
pitust"Unable to handle kernel paging request at virtual address 8b9da2eca289d532"#asahi · 2025-12-17 10:22
pitustI also hit an oops in __kvmalloc_node_noprof (which is much more worrying, to be honest)#asahi · 2025-12-17 10:21
pitustHmm, I'm wondering if there is another issue too#asahi · 2025-12-17 10:16
pitustand a non-broken package for clang-tools#asahi · 2025-12-17 10:16
pitustlike, uh, linux with kasan#asahi · 2025-12-17 10:16
pitustnixpkgs and the binary cache is actually shockingly complete, and i had to build only a few things myself#asahi · 2025-12-17 10:16
pitustheh#asahi · 2025-12-17 10:15
JamesB192jokes "probably more complete on amd64 and (4k arm64) with default {c,ld}flags."#asahi · 2025-12-17 10:11
pitustwhich i guess gentoo also has, but idk how complete it is#asahi · 2025-12-17 10:11
chaos_princesswe have bindist too#asahi · 2025-12-17 10:11
pitustthey have a binary cache#asahi · 2025-12-17 10:11
chaos_princessi thought nix is also a "compile everything locally", no?#asahi · 2025-12-17 10:10
pitustyeah no#asahi · 2025-12-17 10:10
pitusti mean im not reinstalling, thats too much pain even for me#asahi · 2025-12-17 10:10
chaos_princessgentoo (on xfs)#asahi · 2025-12-17 10:10
pitustwhich distro then?#asahi · 2025-12-17 10:10
pitustah okay#asahi · 2025-12-17 10:10
chaos_princessthe point is that i am not on fedora either#asahi · 2025-12-17 10:10
pitustidk my experience with btrfs on fedora was... not good#asahi · 2025-12-17 10:10
chaos_princessso do i :P#asahi · 2025-12-17 10:09
pitustwhich happens shockingly often (quite possibly because of this bug, to be honest)#asahi · 2025-12-17 10:09
pitustwell, at least i have the option to not use a filesystem which loses data whenever i have to hard reset#asahi · 2025-12-17 10:07
chaos_princessright, nixos :P#asahi · 2025-12-17 10:07
pitustchaos_princess: sure, once i figure out how to get nixos to accept a kernel i built by hand, and the bug triggers again (in probably another like 12 hours..)#asahi · 2025-12-17 10:06
chaos_princessi wonder if it is just some packet type we aren't handling and assuming it is a touch packet#asahi · 2025-12-17 10:06
chaos_princessmind also adding a printk of z2->rx_buf[0] ?#asahi · 2025-12-17 10:03
pitustit doesn't help that the bug is rare and nondeterministic, either#asahi · 2025-12-17 10:03
pitustokay so now i get to recompile the kernel from scratch, because nixos doesn't really do the whole incremental compilation thing :/#asahi · 2025-12-17 10:02
chaos_princessjn: the call stack is more than enough w/o lines, as there aren't other choices for most places#asahi · 2025-12-17 10:02
pitustand it's in the data copyin loop#asahi · 2025-12-17 10:01
jnah, fair#asahi · 2025-12-17 10:01
pitustjn: apple_spi_rx only contains one 1-byte write anyway#asahi · 2025-12-17 10:01
jnyou can track down the call sites of functions that were normally called, but the interesting part would be the line number that apple_spi_rx.part.0+0x26c/0x3c0 resolves to. addr2line might still provide it of vmlinux is unstripped, i'm not sure; but the kernel lacks that information for its own use#asahi · 2025-12-17 09:58
pitustjn: the log is missing line numbers but you can track down the call chain pretty easily#asahi · 2025-12-17 09:58
pitustit has to be writing *something* into the buffer#asahi · 2025-12-17 09:58
pitustyeah, idk then#asahi · 2025-12-17 09:56
jnah hm, missing line number info#asahi · 2025-12-17 09:54
chaos_princessit doesn't#asahi · 2025-12-17 09:54
pitustmy guess is that during firmware boot it writes logs or something into the buffer#asahi · 2025-12-17 09:54
chaos_princessyea, i guess it could. it shouldn't, but who knows what happened there.#asahi · 2025-12-17 09:53
pitusthttps://paste.sr.ht/~pitust/fdc7bea24b10bd60c568cf308d9c326b1584034c#asahi · 2025-12-17 09:52
jnoh. out of curiosity, how does the kasan output look?#asahi · 2025-12-17 09:49
pitustso i guess depending on *something*, the device may decide to make an oversized packet in some cases?#asahi · 2025-12-17 09:47
pitust(it takes a length from a received packet from apple_z2, and then passes it with no size checks into spi_read as a length...)#asahi · 2025-12-17 09:46
pitustuuuuuuuuuh. that feels wrong: https://elixir.bootlin.com/linux/v6.17.7/source/drivers/input/touchscreen/apple_z2.c#L145