ChanServ changed the topic of #asahi-dev to: Asahi Linux: porting Linux to Apple Silicon macs | Non-development talk: #asahi | General development | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Logs: https://alx.sh/l/asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
chrisl has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
mini_ has quit [Quit: ZNC closing...]
mini_ has joined #asahi-dev
mini_ has quit [Quit: ZNC closing...]
mini_ has joined #asahi-dev
pb17 has quit [Remote host closed the connection]
chrisl has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
pb17 has joined #asahi-dev
mischa85 has joined #asahi-dev
tobhe_ has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
tobhe has quit [Ping timeout: 480 seconds]
chrisl has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
chrisl has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
DannyBee has joined #asahi-dev
DannyB has quit [Ping timeout: 480 seconds]
pb17 has quit [Ping timeout: 480 seconds]
nora_ has joined #asahi-dev
pb17 has joined #asahi-dev
nora has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
runxiyu has quit [Ping timeout: 480 seconds]
runxiyu_ has quit [Read error: Connection reset by peer]
runxiyu has joined #asahi-dev
chrisl has joined #asahi-dev
mattia013_ has joined #asahi-dev
eluks has quit [Remote host closed the connection]
eluks has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
<nickchan>
also
<nickchan>
does ACTLR TSO work for raw boot objects at least?
<chadmed>
since its part of the arm spec i dont see why theyd gate that
mattia013_ has quit [Ping timeout: 480 seconds]
<nickchan>
that and HACR traps
<nickchan>
we use those in Linux
JayBeeFOSS has quit [Ping timeout: 480 seconds]
JayBeeFOSS has joined #asahi-dev
umaga has joined #asahi-dev
umaga is now known as yussef
yussef is now known as umaga
<nickchan>
and the HACR one is *not* compliant at least the PMUv3 part
mattia013 has quit [Remote host closed the connection]
mattia013 has joined #asahi-dev
mattia013_ has joined #asahi-dev
umaga has quit [Quit: umaga]
umaga has joined #asahi-dev
<chadmed>
hopefully they left it as-is since those blas jokers didnt bring it up
chadmed has quit [Quit: Konversation terminated!]
chadmed has joined #asahi-dev
pb17 has quit [Ping timeout: 480 seconds]
umaga has quit [Quit: umaga]
chadmed has quit [Quit: Konversation terminated!]
pb17 has joined #asahi-dev
umaga has joined #asahi-dev
<sven>
another thing that’s going to be annoying: DART etc is inside SPTM so we won’t be able to trace that
<nickchan>
sven: cpu power management is probably inside sptm too
<nickchan>
sigh so it's either not being able to trace some things or playing cat and mouse
mattia013 has quit [Remote host closed the connection]
mattia013_ is now known as mattia013
<nickchan>
and the second option means keeping some things secret which is not very open source (though it should be possible to keep the shipped binaries fully open source as long as apple did not mess with hacr or actlr tso)
mattia013_ has joined #asahi-dev
mattia013_ has quit [Remote host closed the connection]
mischa85 has joined #asahi-dev
mattia013_ has joined #asahi-dev
<sven>
there’s also the issue that booting on >15.1 seems to fail. maybe they accidentally broke raw boot objects and didn’t notice 🙃
<sven>
but anyway, I’m not going to play cat and mouse. I’ve done that ages ago on consoles and don’t want to ever do that again. it’s just no fun
umaga has quit [Ping timeout: 480 seconds]
<nickchan>
think some other people may do it however
<nickchan>
we will see
mischa85 has quit [Ping timeout: 480 seconds]
mattia013 has quit [Ping timeout: 480 seconds]
mattia013_ is now known as mattia013
<sven>
as long as they’re gonna reverse all the new hardware and write drivers then :D
<sven>
though I guess it’s just iommus and power stuff that’s hidden in SPTM and that binary isn’t very large
<sven>
for the rest patching our tracers into xnu might just work
<nickchan>
sven: xnu patching, where should it go? (python or in m1n1)
<nickchan>
and the thing with any kind of patcher is that it will neccessary contain xnu instructions either in code or in comments
<nickchan>
which would be directly from binary RE given the situations
<nickchan>
not looking forwards to the copyright implications
<sven>
it’s going to be painful, yes
<sven>
haven’t thought about it in details but probably hijack exception vectors and maybe proxy calls to SPTM to mess with page tables
chadmed has joined #asahi-dev
<chadmed>
what actually made them do this? was it literally those guys being like "we should implement AMX in linux"
<sven>
maybe that. maybe arm lawyer being annoyed just because they were bored. maybe both. :/
mattia013 has quit [Remote host closed the connection]
mattia013 has joined #asahi-dev
<nickchan>
chadmed: who is "those guys"
<chadmed>
aaaages ago some folks were asking around about adding AMX support to arm64 linux. we always maintained that we would never _ever_ support anything even remotely like that but i think upstream caught wind of it
mattia013 has quit [Remote host closed the connection]
mattia013 has joined #asahi-dev
mattia013 has quit [Remote host closed the connection]
mattia013 has joined #asahi-dev
<nickchan>
i think there's also a thing where some performance counters have some edge case regarding amx
<nickchan>
ah, *that* got posted onto the mailing list
<chadmed>
then of course there was the whole apple-specific TSO discussion that kinda went nowhere which probably didnt help in that regard especially since FEAT_LRCPC was already a thing
<chadmed>
yeah which is why i think arm probably had some words to apple and they picked the nuclear option. which is just kind of really dickish of them
<nickchan>
and after looking into xnu sources it seems that a12 and possibly a13 have TSO too, but not ACTLR, it's HID9[16]
<nickchan>
not like that matters in practice already confirmed it does not exist on a11
<chadmed>
whos going to be doing x86 emu on an iphone anyway :p
<chadmed>
wel... i guess it's good that iboot is not encrypted now :/
<chadmed>
this doesnt affect M3, only M4 (and all new socs going forward presumably) right?
<nickchan>
chadmed: another thing
<nickchan>
from some of the m1n1 m4 support code i think wfi is also broken somehow
<nickchan>
deep wfi
<nickchan>
and we use that and not having that would make for a really bad experience
mattia013 has quit [Remote host closed the connection]
<chadmed>
yeah id rather not go back to space heater laptops
<chadmed>
theres no way thats being trapped to sptm or anything right
<nickchan>
chadmed: i think power management things and cpu start stop things like that are handled by sptm because of state loss
mattia013 has joined #asahi-dev
<chadmed>
so if sptm is torn down due to a raw boot object we just lose deep wfi?
<nickchan>
it does look like it assuming the m4 forks is not making a horribly incorrect assumption
<chadmed>
thats fucking stupid and surely not intended
<chaos_princess>
Is there any m4 fork that does chicken bits and wfi configuration correctly
<nickchan>
chaos_princess: CYC_OVRD is supposedly completely inaccessible so there is no wfi configuration to speak of
<nickchan>
and yuka's fork forced the a7-a11 code path on wfi which is a shallow wfi
<nickchan>
and that means on m4 by default wfi is already the shallow wfi
<chaos_princess>
Would be interesting to see if cyc_ovrd is inaccessible just in el2
<chadmed>
if apple hid it in gl2 that would kinda defeat the purpose of doing this to make raw boot objects only see the in-spec architecture
<chadmed>
dunno if arm would be happy about that
pb17 has quit [Ping timeout: 480 seconds]
<nickchan>
chadmed: why would hiding cyc_ovrd defeat the purpose?
<nickchan>
though I can see that will make people much more willing to play cat and mouse
<nickchan>
space heater is unfun
<nickchan>
there are some alternatives that have not been tried however like turning off cpu power domains in pmgr
<chadmed>
nickchan: oh thats an apple register
ece314378925355451680698427415 has quit [Ping timeout: 480 seconds]
chrisl has joined #asahi-dev
mischa85 has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
jannau_ is now known as jannau
pb17 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
<sven>
any of the deeper wfi states being broken would be very bad :(
alessaba has joined #asahi-dev
Mary has quit [Quit: .]
Mary has joined #asahi-dev
mischa85 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
<maz>
jannau: what was preventing this from working before my changes? I didn't really expect any functional difference with them.
<jannau>
maz: it has 2 root ports which was incompatible with the bus_notifier approach
mischa85 has joined #asahi-dev
<maz>
jannau: huh, right. oh well, less code anyway, and additional functionality. too bad I can't afford that box! :)
mischa85 has quit [Ping timeout: 480 seconds]
PaulFertser has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
<jannau>
I got it refurbished for a little bit over half price. still expensive though
<jannau>
pcie setup is not fully working. the switch management endpoint is not working and the devices behind the switch only come up with PCIe gen1
<maz>
yeah. it would take my M1 studio to die on me to be able to justify it. and even then, a 128 core Altra could make more sense, given what I do with it.
mischa85 has quit [Ping timeout: 480 seconds]
<maz>
I usually have to play all sort of command-line tricks to deal with PCIe switches...
<maz>
do we have a native driver for the switch in this box? or relying on the generic Linux code?
<jannau>
yes, switchtec. it's a microchip PM40100
<chaos_princess>
jannau: oh, wow, i salute you and your wallet for that sacrifice.
<chaos_princess>
did they fix device mappings on pcie in that box?
PaulFertser has joined #asahi-dev
<jannau>
I haven't checked beyond quickly testing that old radeon hd4350
<jannau>
maz: mem assigment breaks down with hotplug but if all devices are powered up before booting it mostly works after putting "bus-range = <0 8>;" in the device tree
mischa85 has joined #asahi-dev
<jannau>
not sure how to model the reset and power enable gpios of devices behind the switch in the devicetree
<maz>
huh, what a mess.
<maz>
I guess this should be part of describing the slots past the switch, but how to express that numbering is a mystery. wonder if there's prior art around that.
<maz>
robher: any clue?
mischa85 has quit [Ping timeout: 480 seconds]
<jannau>
arch/arm64/boot/dts/qcom/sa8540p-ride.dts has a PM40028 switch but doesn't model devices behind the switch
pb17 has quit [Ping timeout: 480 seconds]
calebccff has quit [Quit: ZNC 1.8.2+deb3.1+deb12u1 - https://znc.in]
calebccff has joined #asahi-dev
<fl0_id>
why do you think they even still allow raw / custom boot objects, if they apparentyl locked down M4 so tightly?
mischa85 has joined #asahi-dev
pb17 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
<chadmed>
fl0_id: they locked down apple's extensions to the ARM ISA
<chadmed>
the platform is designed explicitly to allow custom boot objects to exist. arm got shitty with them for letting those custom objects "see" all of apples not-ARM things
mischa85 has joined #asahi-dev
<fl0_id>
ah ok.
<fl0_id>
thx
<fl0_id>
and these just happen to be important bits I guess
<sven>
it’s mostly annoying because our Hv for reverse engineering won’t work without these extensions
<sven>
plus deep sleep states might be affected which would actually be important for running Linux
<LarstiQ>
I don't suppose the blessed Arm ISA has working alternatives for those?
<fl0_id>
sven yeah deep sleep states would be rly bad :/
mischa85 has quit [Ping timeout: 480 seconds]
<sven>
yeah, the rest we can (painfully) work around probably
<sven>
but sleep might just be a bug, ie they just disabled everything and didn’t realize that the sleep msr is very important
<sven>
or maybe it just works differently on m4. We don’t really know yet
<fl0_id>
sure. thx for explaining
<sven>
still all pretty far away since we need to get m1/2 upstream and then ideally bring up m3 directly upstream as well
mischa85 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
<fl0_id>
yeah for sure
<fl0_id>
main issue for me is time and that my m3 is also my work machine, soo I cannot boot it into m1n1 that often
chadmed has quit [Quit: Konversation terminated!]
chadmed has joined #asahi-dev
mischa85 has joined #asahi-dev
<sven>
time is probably the issue for all of us :D
<fl0_id>
yeah, not saying I'm unique haha
mischa85 has quit [Ping timeout: 480 seconds]
johey has quit [Ping timeout: 480 seconds]
johey has joined #asahi-dev
mischa85 has joined #asahi-dev
pb17 has quit [Ping timeout: 480 seconds]
mischa85 has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
pb17 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
yuyichao_ has quit [Ping timeout: 480 seconds]
chrisl has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
chrisl has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
<robher>
maz: Certainly possible. PPC boots with all the PCI hierarchy described. The kernel can in fact generate the DT PCI nodes for you. Boot with CONFIG_PCI_DYNAMIC_OF_NODES enabled. Add what you need from the DT to your dts. Then add the GPIOs you need to handle.
<robher>
maz: There's also a switch example on hikey970 (or 960 maybe?).
john-cabaj has joined #asahi-dev
<robher>
maz: there's also the linux,pci-probe-only flag if keeping the firmware configuration is needed.
mischa85 has quit [Ping timeout: 480 seconds]
pjakobsson has quit []
pb17 has quit [Ping timeout: 480 seconds]
pb17 has joined #asahi-dev
Halian has quit [Ping timeout: 480 seconds]
alessaba has quit [Quit: Connection closed for inactivity]
yuyichao_ has joined #asahi-dev
chrisl has joined #asahi-dev
swiftraccoon has quit [Ping timeout: 480 seconds]
chrisl has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
chrisl has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
chrisl has quit [Ping timeout: 480 seconds]
pb17 has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
chrisl has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
chrisl has quit [Ping timeout: 480 seconds]
pb17 has joined #asahi-dev
mischa85 has joined #asahi-dev
tanty has quit [Ping timeout: 480 seconds]
mischa85 has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
tanty has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
mischa85 has joined #asahi-dev
chrisl has joined #asahi-dev
<mischa85>
well if there's a will there's a way
tanty has quit [Ping timeout: 480 seconds]
<mischa85>
nickchan: you mentioned trying disabling cpu power domains to gain write access to cyc_ovrd. how to? considering there's no ps-regs and the power registers are still unknown
mischa85_ has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
mischa85_ has quit [Ping timeout: 480 seconds]
___nick___ has joined #asahi-dev
___nick___ has quit []
___nick___ has joined #asahi-dev
tanty has joined #asahi-dev
chrisl has joined #asahi-dev
swiftraccoon has joined #asahi-dev
alessaba has joined #asahi-dev
<kettenis>
maz, robher: there also is CONFIG_PCI_PWRCTL now
chrisl has quit [Ping timeout: 480 seconds]
<kettenis>
I believe that is the preferred way to power on PCI devices that are soldered onto a system board
pb17 has quit [Ping timeout: 480 seconds]
CME has quit [Read error: Connection reset by peer]
CME has joined #asahi-dev
pb17 has joined #asahi-dev
chrisl has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
mischa85_ has joined #asahi-dev
mischa85_ has quit [Ping timeout: 480 seconds]
mischa85 has quit [Ping timeout: 480 seconds]
checkfoc_us9 has quit [Remote host closed the connection]
checkfoc_us9 has joined #asahi-dev
cyrinux9490 has quit []
cyrinux9490 has joined #asahi-dev
mischa85 has joined #asahi-dev
pb17 has quit [Ping timeout: 480 seconds]
mischa85_ has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
chrisl has joined #asahi-dev
WindowPain has quit [Read error: Connection reset by peer]
WindowPain has joined #asahi-dev
mischa85 has joined #asahi-dev
mischa85_ has quit [Ping timeout: 480 seconds]
chrisl has quit [Ping timeout: 480 seconds]
___nick___ has quit [Remote host closed the connection]
pb17 has joined #asahi-dev
mischa85 has quit [Ping timeout: 480 seconds]
alessaba has quit [Quit: Connection closed for inactivity]
mischa85 has joined #asahi-dev
mischa85_ has joined #asahi-dev
tanty has quit [Quit: Ciao!]
mischa85 has quit [Ping timeout: 480 seconds]
tanty has joined #asahi-dev
chrisl has joined #asahi-dev
tanty has quit []
tanty has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
tanty has quit [Quit: Ciao!]
tanty has joined #asahi-dev
tanty has quit [Quit: Ciao!]
tanty has joined #asahi-dev
mischa85_ has quit [Ping timeout: 480 seconds]
pb17 has quit [Ping timeout: 480 seconds]
chrisl has joined #asahi-dev
pb17 has joined #asahi-dev
chrisl has quit [Ping timeout: 480 seconds]
tanty has quit [Quit: Ciao!]
tanty has joined #asahi-dev
swiftraccoon has quit [Read error: Connection reset by peer]
<chadmed>
for context: the black box is kinda useless now that the visense stuff is stable. id rather rip that out before merging this to eliminate some churn
tubtub_ has joined #asahi-dev
tubtub_ has quit [Remote host closed the connection]
<chaos_princess>
i am not clear as to how it fixes things? even with the boundary directory bit, are you still not leaving files owned by unassigned uid?
tanty has quit [Remote host closed the connection]
<chadmed>
seems that way yeah
<chadmed>
the black box is also not really "state"
<chaos_princess>
this feels like a feature that is for files that are owned by the unit and not accessed by anything else
<chadmed>
you wouldnt call any other log file "state"
tanty has joined #asahi-dev
tanty has quit []
mischa85 has joined #asahi-dev
red_rider has quit [Ping timeout: 480 seconds]
tanty has joined #asahi-dev
<flokli>
chadmed: what's in the black box
<flokli>
Can we improve logging sufficiently, pipe everything needed to the journal during the crash handler?
<flokli>
Ah, now reading "visense stuff is stable, we can rip it out"
<chaos_princess>
blackbox is kinda like a core dump file
<chaos_princess>
(it is a binary)
<flokli>
Mmmh
<chadmed>
no we cant pipe everything to the journal on crash because its PCM samples :p
<flokli>
Heh
<chadmed>
it doesnt matter, i will be ripping it out at some point sooni(ish)
<chadmed>
then we can look at dynamic users
<chaos_princess>
chadmed: json array of floats :P
mischa85 has quit [Ping timeout: 480 seconds]
<chadmed>
chaos_princess: you only need one column type in your database and it's JSONB
<flokli>
chadmed: I'm happy to update the PR to drop the blackbox functionality, or rebase on top of such a change if you want to do it.
<chadmed>
flokli: yeah hold off until i rip it out. ive got a bunch of other stuff to take care of first
<chaos_princess>
Just let it sit for now, we can rebase it later
<chaos_princess>
Not like there is a ton of changes happening to those files
<chadmed>
i need to submit v4 of the codec patches upstream today without the tdm idle mask stuff. since that was the last controversial bit hopefully those get merged without a fuss
tanty has quit [Quit: Ciao!]
<chadmed>
i am deeply concerned about the fact that the upstream-approved tdm idle mask stuff ended up not working but i'll worry about that when it's worth worrying about