#
#asahi · 2024-11-30 04:20
nickchanthis is the cyclone cache workaround except I used EHID instead of HID
#asahi · 2024-11-30 04:15
nickchanthen set EHID4_DISABLE_DC_MVA | EHID4_DISABLE_DC_SW_L2_OPS in SYS_IMP_APL_EHID4 again
#asahi · 2024-11-30 04:14
nickchandc civac
#asahi · 2024-11-30 04:14
nickchanreverser: try clear EHID4_DISABLE_DC_MVA | EHID4_DISABLE_DC_SW_L2_OPS from SYS_IMP_APL_EHID4
#asahi · 2024-11-29 21:19
reverser(Just accessing a bunch of addresses that have to share the space with my target memory address and cause the target address to be evicted from the cache)
#asahi · 2024-11-29 21:18
reverserAnd even when I am trying to flush or use the data cache management instructions (DC CIVAC), I do not see any changes in cache timings. However, I do see a change in timings when "flushing" the memory address by using an eviction set.
#asahi · 2024-11-29 21:17
reverserAnd Calandracas_: Not sure.
#asahi · 2024-11-29 21:17
reverserI am using a PMU to time how long it takes to access a specific memory address. If the access is below a threshold, the data is stored in the cache.
#asahi · 2024-11-29 21:17
Calandracas_are any apple silicon devices numa?
#asahi · 2024-11-29 21:15
jnhow do you check the data cache content?
#asahi · 2024-11-29 21:15
reverserI am able to evict data via eviction sets, but that approach is a bit slow. I am trying to speed things up by forcing a flush (as clflush does on Intel)
#asahi · 2024-11-29 21:14
reverserI am looking at L1 and L2
#asahi · 2024-11-29 21:14
jn(just curious. i probably won't have answers to your question)
#asahi · 2024-11-29 21:14
jnreverser: which cache level(s)?
#asahi · 2024-11-29 21:12
reverserHi, it's me again. How would I flush the data cache from an Apple M1 under Asahi linux? I am trying to empty my data cache as part of a hardware reverse engineering effort. However, when using the DC CIVAC instruction or the flush_cache_mm() kernel function: both do not seem to have any effect on the content of my data cache? Anything I am overseeing?
#asahi · 2024-11-29 13:54
j`eyit's dense code
#asahi · 2024-11-29 13:53
chaos_princessyes
#asahi · 2024-11-29 13:53
j`eychaos_princess: what does x11bridge actually do? is it an X11 server proxy?
#asahi · 2024-11-29 10:31
j`eythat's when doing 1k+ fps, so probably if I limit it to 500fps or something, it will run cool(er)
#asahi · 2024-11-29 10:31
j`eyso, expected, and nothing to worry about too much
#asahi · 2024-11-29 10:31
j`eyso the other day I said how m2+linux+wine made my computer hot, but m1+macos+wine didn't.. well I made the m1 use the same third-party ddraw.dll replacement, and it gets hot too
#asahi · 2024-11-29 05:11
runxiyuI can't even switch vttys
#asahi · 2024-11-29 05:11
runxiyuHow should I begin to debug a situation where my graphics are completelymlcoked up?
#asahi · 2024-11-28 10:37
chadmedmrosorensen: yeah, i also have the buzzing sound intermittently. not sure what we are doing differently at this stage, other than maybe wifi 2.4ghz coexistence?
#asahi · 2024-11-28 09:01
yukaLet me switch to the dev channe
#asahi · 2024-11-28 09:01
yukaAh wait
#asahi · 2024-11-28 09:01
yukaHow did you obtain the uart base?
#asahi · 2024-11-28 09:00
yukaFair, just joined it
#asahi · 2024-11-28 08:59
jannauyuka: #asahi-dev is the better suited channel for this topic
#asahi · 2024-11-28 08:55
jannauyuka: https://gist.github.com/jannau/3c4514584c6d155b76a25e34ae451a1e was working and producing output on debug uart
#asahi · 2024-11-28 08:53
jannauwhich chip_id and EARLY_UART_BASE do you use?
#asahi · 2024-11-28 08:53
yukawonderful, thanks
#asahi · 2024-11-28 08:52
jannauyuka: if this is for m4* crash causing the bootloop is in init_cpu() just return early from that function
#asahi · 2024-11-28 08:42
jannauyuka: macos logs to the uart with nvram "boot-args=-v" set but I haven't seen output with macos 15 production kernels
#asahi · 2024-11-28 08:38
yukaIs there any way to test the macvdmtool debug uart thing with a mac os installation instead of m1n1, just to be sure that that is working correctly?
#asahi · 2024-11-28 08:34
yuka... the "_start_c" function to see if that would make it just stay on instead of rebooting, but it's still bootlooping. Do you people have any ideas for what I could try?
#asahi · 2024-11-28 08:34
yukaHi there, i'm fiddling around with a not-yet-supported apple silicon chip trying to get some sign of life out of m1n1. So far, I have set up a secondary APFS container, and done the kmutil thingy with a BRINGUP build of m1n1. I have also (theoretically) connected to the debug uart with a second apple silicon device and macvdmtool. Practically, I'm not getting any output on the uart. The target device bootloops. I tried adding a busyloop in the beginning of...
#asahi · 2024-11-28 08:33
fl0_idcopticcurse see also: https://mastodon.social/@marcan@treehouse.systems/113424157737265838
#asahi · 2024-11-28 08:08
jannaurinOfTheStars: just removing the old kernels is not a solution. there are intentionally the last 3 kernels. This is a config issue on old (before the release) installations resulting in huge initramfs eating all the storage on /boot. see https://discussion.fedoraproject.org/t/128788/2 for instructions to fix this
#asahi · 2024-11-28 07:49
fl0_idcopticcurse Afaik it doesn't make it harder, but there are enough differences that it isn't super quick, and the main developers decided to focus on improving existing support first, like for improved graphics, getting microphones working etc which will later benefit m3 and m4 too. Instead of each year of device updates being busy with only getting support for that in. F.e. the gpu on m3 (and m4) is quite different from m1 and m2
#asahi · 2024-11-28 03:43
rinOfTheStarsfor posterity's sake, if anyone has this issue in the future and searches through the irc logs to find an answer for some reason: https://www.reddit.com/r/Fedora/comments/1anh07j/boot_filesystem_needs_more_space_when_updating/ has the instructions on how to fix this, but do remember that asahi linux uses kernel-16k-core instead of kernel-core
#asahi · 2024-11-28 03:41
rinOfTheStarsthat was indeed the issue, there was just barely not enough space due to the two old verisons of the kernel mucking around, lmao
#asahi · 2024-11-28 03:37
rinOfTheStarsthat's probably the issue, huh
#asahi · 2024-11-28 03:36
rinOfTheStarslooks like there's some old versions of the kernel lingering for some reason
#asahi · 2024-11-28 03:30
rinOfTheStarsLooking at the /boot fs it is indeed very full; does it keep old versions of packages around or something, or do I need to allocate it more space from the main asahi linux partition?
#asahi · 2024-11-28 03:30
rinOfTheStarsHey all. Getting an error I've never seen before when trying to dnf upgrade my Asahi Linux install: Error: Transaction test error: installing package kernel-16k-core-6.11.8-400.asahi.fc40.aarch64 needs 64MB more space on the /boot filesystem
#asahi · 2024-11-28 03:14
copticcurseI wanted to ask a noob question, what is the difference between m3 and m2 that makes it hard to port linux to m3?
#asahi · 2024-11-27 09:44
mrosorensen* sound=buzzing sound
#asahi · 2024-11-27 09:42
mrosorensenchadmed Confirming that there is no sound from MacOS on super low volume.
#asahi · 2024-11-27 09:33
mrosorensenaha Give me one sec. Are you on Fedora 41 as well ? I can't remember but I first noticed this a couple of weeks ago.

plus ›