#
✕ nick:marcan
#asahi · 2023-03-11 09:34
marcan(unless you use WEP)
#asahi · 2023-03-11 09:34
marcancompletely broken
#asahi · 2023-03-11 09:34
marcanyeah that's broken
#asahi · 2023-03-11 09:34
marcanthat cached_keys stuff looks for WEP only, but then when it's not set they go and wipe all the crypto info?
#asahi · 2023-03-11 09:30
marcanbrcmfmac is hilariously undermaintained, it's missing some major features
#asahi · 2023-03-11 09:30
marcan(which is not upstream yet)
#asahi · 2023-03-11 09:30
marcanlike another one
#asahi · 2023-03-11 09:30
marcanwe have a patch for WPA3 support
#asahi · 2023-03-11 09:29
marcanhas to do with offload etc, might have been SME
#asahi · 2023-03-11 09:29
marcan(wpa_supplicant doesn't or didn't when I first tested)
#asahi · 2023-03-11 09:29
marcannote that iwd uses a different codepath on brcmfmac and we *specifically* chose it because it works with wpa3 there
#asahi · 2023-03-11 09:27
marcanI wonder if that's a userspace usage problem then
#asahi · 2023-03-11 09:26
marcanjannau: oh so it's that one
#asahi · 2023-03-11 09:26
marcanok so that patch is replacing a schedule that calls cfg80211_conn_work() with only a direct call to cfg80211_conn_do_work()
#asahi · 2023-03-11 09:23
marcanhm no, wait
#asahi · 2023-03-11 09:22
marcanpretty sure that patch is broken
#asahi · 2023-03-11 09:16
marcanyes
#asahi · 2023-03-11 09:12
marcanlinux-wireless mailing list I guess
#asahi · 2023-03-11 09:11
marcanI'm not sure if the patch is broken or the fix is exposing another bug, but there you go
#asahi · 2023-03-11 09:06
marcanI bet on f4b6a138efb8 / 015b8cc5e7c
#asahi · 2023-03-11 09:00
marcanK900_: yeah I bet on a cfg80211 change doing it
#asahi · 2023-03-11 08:58
marcanoutside any drivers
#asahi · 2023-03-11 08:58
marcanchanges
#asahi · 2023-03-11 08:58
marcanah wait, I do see cfg80211
#asahi · 2023-03-11 08:57
marcanwell, all the driver commits
#asahi · 2023-03-11 08:56
marcanI went through all the commits in 6.2.3 and didn't see anything notable though
#asahi · 2023-03-11 08:45
marcanmaybe something else broke it? sdio controller, something in the 80211 core...
#asahi · 2023-03-11 08:45
marcanit kind of sounds like you're looking in the wrong place, I don't think any of the brcmfmac changes in 6.2.3 make any sense as breaking anything
#asahi · 2023-03-11 08:44
marcandoes the device probe?
#asahi · 2023-03-11 08:41
marcanK900_: I see mostly non-asahi changes from 6.1 to 6.2, or are you saying it broke in stable 6.2.3?
#asahi · 2023-03-08 03:38
marcanabby: right, what I meant was that the regular repos don't have archives
#asahi · 2023-03-08 03:37
marcanelvishjerricco: I don't think macOS supports nested virt yet (it's on the list of things we support *before* macOS does!)
#asahi · 2023-03-08 03:37
marcanabby: that's not ARM
#asahi · 2023-03-08 02:51
marcannot easily, Arch Linux generally does not do version archives and I'm not aware of any public ALARM one.
#asahi · 2023-03-06 05:50
marcanhttp://archive.raspberrypi.org/debian/pool/main/w/widevine/libwidevinecdm0_4.10.2252.0+1_arm64.deb
#asahi · 2023-03-06 05:49
marcanfound it
#asahi · 2023-03-06 05:45
marcanI looked at the rpios repos and I can't find the package at all, but I don't have an actual insall to verify
#asahi · 2023-03-06 05:44
marcanhttps://www.techworm.net/2022/02/raspberry-pi-os-64-bit.html says it's still 32bit only
#asahi · 2023-03-05 15:35
marcanpatreon has paypal AIUI
#asahi · 2023-03-04 18:27
marcanwe should probably hardcode most of that stuff in at least the UCM if not outright in the kernel. too easy to mess up...
#asahi · 2023-03-04 14:54
marcanmort_: "120Hz" mode ends up capped at 60Hz anyway for some reason, so yes
#asahi · 2023-03-04 14:54
marcancould also be an appleism though
#asahi · 2023-03-04 14:54
marcanbut possibly
#asahi · 2023-03-04 14:54
marcanunclear why exactly AP mode is broken yet
#asahi · 2023-03-04 14:53
marcanEAP works afaik (it always worked except we had a missing kernel compile option for key management...). AP mode does not.
#asahi · 2023-03-04 14:47
marcanApple seem to be pretty good at locking down actual pstate parameters into pmgr in iboot (that's how it works for CPUs), so I wouldn't be surprised if the GPU is the same and the firmware can't actually touch that, only make policy decisions
#asahi · 2023-03-04 14:46
marcanyou can definitely control the min/max pstate at least
#asahi · 2023-03-04 14:45
marcanlina knows more but that is data passed to the firmware, I'm not sure if anyone knows for sure exactly what the firmware controls (i.e. whether you can *actually* change the clocks/voltages or it's just informational data for the power controller to do its math)
#asahi · 2023-03-04 14:44
marcanbut I might just a version his way with the if(cyw) conditional and if he doesn't have time to test it he better just ack it, because he already confirmed the existing code doesn't work either so there is nothing to regress
#asahi · 2023-03-04 14:44
marcanright now I'm mostly waiting for the brcm guy to test whether my wpa3 patch actually works on his chips (because apparently the existing code was a hack cypress added for their firmware fork)

plus ›