#asahi-dev · 2026-10-10 09:13
chadmedi think its specific to something in the battlefront shaders. gta iv works fine with dxvk here#asahi-dev · 2026-10-10 08:34
MichaelReeves[m]weirdly, for battlefront, dxvk did work for me (with lavapipe) so I guess the wow64 issues with it are just with real GPUs#asahi-dev · 2026-10-10 08:33
MichaelReeves[m]trying Fallout: New Vegas (2010) with p3s next#asahi-dev · 2026-10-10 07:47
jannaunot really an issue but more than I expected#asahi-dev · 2026-10-10 07:47
jannauhuh, the alignment and statically allocated smp stacks increased m1n1.bin's size by factor 3. from 1.1M to 3.7M. the increase for the macho is modest: 864K to 1.1M#asahi-dev · 2026-10-10 05:42
MichaelReeves[m]oh and also just checked and with the workaround applied it all functions fine#asahi-dev · 2026-10-10 05:26
MichaelReeves[m]once we've all done it then I'll reply on-list with the results#asahi-dev · 2026-10-10 05:26
MichaelReeves[m]have you (and yuka) had a chance to try it get on your devices?#asahi-dev · 2026-10-10 05:26
MichaelReeves[m]tried on my end and it causes the host to lock up and then a SoC reset, so I think that confirms t8122 has the buggy impl#asahi-dev · 2026-10-10 05:25
MichaelReeves[m]thanks for sending it through#asahi-dev · 2026-10-10 02:31
myriaAh, fun fun#asahi-dev · 2026-10-10 02:29
MichaelReeves[m]I think he might have DM'ed my IRC nick but I don't have a bouncer so I didnt get them when I was offline#asahi-dev · 2026-10-10 02:28
MichaelReeves[m]yuka: did you get them?#asahi-dev · 2026-10-10 02:28
MichaelReeves[m]sorry#asahi-dev · 2026-10-10 02:27
MichaelReeves[m]I didn't get any so I'm not sure#asahi-dev · 2026-10-10 02:26
myriaMichaelReeves[m], yuka: can one of you send me the testing instructions and I'll try running them on my M3#asahi-dev · 2026-10-10 00:33
JamesCalligeros[m]llvm 23 introduced a new optimisation path that broke SEH unwind for all windows binaries lmao #asahi-dev · 2026-10-10 00:33
JamesCalligeros[m]yeah im still on 11.16 #asahi-dev · 2026-10-10 00:26
MichaelReeves[m]it's just a 1 line fix and probably fixes other platforms as well so ill send it upstream#asahi-dev · 2026-10-10 00:25
MichaelReeves[m]bisected it to https://github.com/wine-mirror/wine/commit/52857120a9a5c22095580232b2982f6e45043077#asahi-dev · 2026-10-10 00:24
MichaelReeves[m]chadmed: found out why you didnt need the wine patch, the buggy code was introduced in 11.19 (came out last week) so you might have been on 11.18 or older#asahi-dev · 2026-10-09 18:41
jannaupr-700 broke because the value of lp-sip0 isn't identical#asahi-dev · 2026-10-09 18:36
yukathe main use case was the MacBook Neo#asahi-dev · 2026-10-09 18:35
yukaYes#asahi-dev · 2026-10-09 18:35
jannauyuka: did you test enable_debugusb on spmi devices?#asahi-dev · 2026-10-09 17:22
jannauconfirmed as fixed#asahi-dev · 2026-10-09 17:11
jannauI think it just broke the framebuffer#asahi-dev · 2026-10-09 17:01
yukaThe absence of notch properties caused the chainload problem?#asahi-dev · 2026-10-09 16:44
jannausorry for the noise that experiment caused#asahi-dev · 2026-10-09 16:41
jannaugithub stacked PRs seem to require that branches reside in the target repo. At least when using the webui#asahi-dev · 2026-10-09 16:36
jannauimacs do not have a notch#asahi-dev · 2026-10-09 16:03
jannaubuild fix for rust 1.99.0 https://github.com/AsahiLinux/m1n1/pull/702#asahi-dev · 2026-10-09 14:28
jannauhttps://github.com/AsahiLinux/m1n1/pull/700#asahi-dev · 2026-10-09 13:56
jannaupossibly but I suspect its just defensive programming#asahi-dev · 2026-10-09 13:45
yukamaybe not on all machines / iboot versions?#asahi-dev · 2026-10-09 13:44
jannaudoes anyone remembers why i2c_init() has pmgr_adt_power_enable()? the i2c controller for the dfu tipd is enabled at boot#asahi-dev · 2026-10-09 13:31
jannauyes#asahi-dev · 2026-10-09 13:30
yukaexcept when dcp is completely disabled / not present like the M4 Pro Mac Mini?#asahi-dev · 2026-10-09 13:29
jannaudcp can probably be freely used for any output since M2*, display cross-bridge is fully switchable#asahi-dev · 2026-10-09 13:29
JamesCalligeros[m]Yeah seems that way, there are weird bandwidth constraints with dcp itself#asahi-dev · 2026-10-09 13:28
jannauchadmed: I suspect this could be memory bandwidth limited in macos or firmware#asahi-dev · 2026-10-09 13:27
yuka> Vielen Dank für Ihre Rückmeldung. / Thanks for your feedback#asahi-dev · 2026-10-09 13:27
yukaalso reported the inaccuracy#asahi-dev · 2026-10-09 13:27
yukaGood to know that these support pages are not to be trusted#asahi-dev · 2026-10-09 13:26
jannauyour support page is inconsistant and wrong. tech specs pages were so always correct#asahi-dev · 2026-10-09 13:26
chadmed????? #asahi-dev · 2026-10-09 13:26
chadmed"up to 3 displays if you use this very specific arrangement of displays and modes but otherwise 4 displays unless you also use this mode in which case its two displays"#asahi-dev · 2026-10-09 13:26
chadmedi think the way they express supported display configs is really confusing though #asahi-dev · 2026-10-09 13:25
chadmedthe specs page looks right to me, dcp + 4 dcpexts #asahi-dev · 2026-10-09 13:25
chadmedyeah