marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | General project discussion | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Topics: #asahi-dev #asahi-re #asahi-gpu #asahi-stream #asahi-offtopic | Keep things on topic | Logs: https://alx.sh/l/asahi
<rkjnsn[m]>
I believe teared is the past tense of tear (to produce clear salty liquid from one's eyes), while torn is the past participle of tear (to rip).
<opticron>
hmm, that introduces an interesting play on words
<opticron>
"I was all torn up"
<opticron>
but yes, you are correct
malvo has quit [Ping timeout: 480 seconds]
malvo has joined #asahi
PhilippvK has joined #asahi
bgb has joined #asahi
marvin24 has joined #asahi
marvin24_ has quit [Ping timeout: 480 seconds]
bgb_ has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
bgb has joined #asahi
bgb_ has quit [Ping timeout: 480 seconds]
<bgb>
any changes for usb thing? I can not get usb device ttyACM0 ttyACM1 on my host, tried C-C C-A cables
VinDuv_ has joined #asahi
VinDuv has quit [Ping timeout: 480 seconds]
VinDuv_ is now known as VinDuv
bgb_ has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
bgb has joined #asahi
bgb_ has quit [Ping timeout: 480 seconds]
bgb has quit [Ping timeout: 480 seconds]
bgb has joined #asahi
<j_ey>
bgb: youre just trying to run m1n1?
<bgb>
j_ey: I want to try marcan's smp-hpervisor. But I just found UEFI bug msg from dmesg of host, and m1n1 works fine with another linux machine
<j_ey>
bgb: so, is it working now? (on the other machine)
<bgb>
works fine for USB, not reach the hypervisor point
<j_ey>
have you got a kernel and roots/initramrfs?
<bgb>
j_ey: a bit complicated, my new host does not have desktop, inconvenient. I'm struggling to install gentoo desktop.
<bgb>
I'll back when env is ok, thanks
bradfier has quit [Quit: Leaving...]
bradfier has joined #asahi
zopieux has quit [Ping timeout: 480 seconds]
zopieux has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
gladiac is now known as Guest235
gladiac has joined #asahi
bgb has joined #asahi
Guest235 has quit [Ping timeout: 480 seconds]
chadmed has quit [Remote host closed the connection]
chadmed has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
bgb has joined #asahi
chadmed has quit [Remote host closed the connection]
chadmed has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
<alyssa>
j_ey: now would be a godo time to plug, we're putting together an unofficial debian host for rootfs images
<alyssa>
there's a lot of folks who run with arch linux arm because of the wget and go for arm64, for devices which can't run a full debian installer due to broken bootloaders
<alyssa>
no reason the exact same workflow can't work on debian
<j_ey>
cool! im currently just using an initramfs cos im not even trying to use the mac for anything more than messing aruound
<sven>
i recommend putting that onto nvme. makes booting kernels much faster :D
<j_ey>
my m1n1-payload.macho (m1n1, linux, initramfs) is only 7mb!
<j_ey>
(and thats with kasan, i think it's like 4mb without)
<alyssa>
j_ey: smaller than the DCP firmware!
<j_ey>
^_^
<sven>
alright.. time to send out the tipd series i guess :>
X-Scale has joined #asahi
* j_ey
takes a final look
<sven>
i essentially fixed alyssa's nits
<sven>
i guess the rest can happen on the ML anyway
<j_ey>
typo: The major difference>>s<< are the changed interrupt numbers
<sven>
thanks, fixed that as well!
M320mb[m]1 has left #asahi [#asahi]
<j_ey>
sven: as an outsider, I like the series, so lets see what Heikko thinks!
<j_ey>
*Heikki
<alyssa>
sven: I guess I should cherry-pick your whole tipd+i2c series?
<alyssa>
sven: so it looks like the pinctrl driver in your branch might be newer than the one in mine?
<sven>
uh. maybe?
* sven
blames j_ey
<alyssa>
eh, I just need to be careful to not break PCIe along the way
<j_ey>
alyssa: yes, its newer, but it's just api cleanups, NFC
<alyssa>
ack
<alyssa>
will still update so you get more testing
<j_ey>
ok ty
* alyssa
is the designated test monkey
<alyssa>
as the only one crazy enough to run this in prod :-p
* kettenis
wishes I would find the OpenBSD bug that causes SMP crashes
* alyssa
does git surgery
<alyssa>
ok, have removed the old pinctrl commits and DT, now to cherry-pick the new
<sven>
we should really try to have some AsahiLinux/next branch
bgb has quit [Ping timeout: 480 seconds]
<sven>
hrm, and i should also figure out why macos doesn't seem to like the linux serial gadget driver
<kettenis>
I think we need to trick marcan into coordinating some of the DT additions
<alyssa>
sven: once I finish smashing together PCIe and type-C hotplug I can branch off a new date branch (without DCP)
<marcan>
you all keep giving me things to do instead of getting kernel trees in order :p
<alyssa>
kettenis: stop distracting marcan :p
<alyssa>
marcan: write kernel code :p
<sven>
can we vote on priorities? :D
<marcan>
lol
<sven>
i vote for clock/powermanagement then!
<alyssa>
sven: naming conflict on the DT stuff -- gpio vs pinctrl_ap
<marcan>
well I want to figure out... yeah that
<marcan>
because it's a dep of *everything*
<sven>
yup
<sven>
especially i2c
<alyssa>
sven: pinctrl_ap seems like the better approach because of pinctrl_nub and whatever but
<j_ey>
sven: I vote whatever makes marcan the most motivated!
<marcan>
lol :D
<marcan>
j_ey: you're always too nice :>
<sven>
i2c needs a real clock (which is easy, i know about that one) but also that power/clock gate
<kettenis>
alyssa: yes that is why I picked those names
<j_ey>
marcan: i have to balance out sven, who is always too mean :P
<alyssa>
kettenis:
<marcan>
I'm going to shave the cpufreq yak first because it's easy, but also because looking into how that memory controller thing is represented will probably give me ideas for the more general power management stuff
<marcan>
I'm still not sure if the soc voltage stuff is handled by macos or by magic
<marcan>
if it's handled by macos then we need to figure out how
<marcan>
but given it seems to be magic for the cpu, maybe it's magic for everything
<sven>
alyssa: see the commit title "HACK/DONOTMERGETHIS: usb dt"
<sven>
alyssa: it's literally just copied/pasted together to make it work.
<sven>
that single commit should also be at least three or so separate ones
<marcan>
and yeah, if anyone missed it... cpufreq is one register per cluster. Plus 8*2 (repeated) if you want to bump the memory controller perf mode in lockstep with the pcore state.
<marcan>
this thing really is high level like all heck, it also does the boost modes automatically
<kettenis>
frankly, that makes sense
<marcan>
I still don't know *what* does this too. there has to be firmware involved.
<marcan>
it's not PMP though
<alyssa>
sven: yes, I know. I'm just asking -- are you attached to the gpio name? I've changed it all to pinctrl_ap in my current branch
<sven>
i'm attached to nothing in that commit :D
<alyssa>
:)
<kettenis>
I believe pinctrl_xxx labels are preferred since these are pinctrl nodes
<kettenis>
(masquerading as gpio)
<alyssa>
👍
<j_ey>
alyssa: those names came from me came from corellium
<kettenis>
not sure if the dt schema stuff checks for the names
* alyssa
is making a mess of the local DT but that was a given.. sigh
<j_ey>
marcan: maybe just nothing exposed in the dt
<kettenis>
I butchered the u-boot device trees a bit as well when adding the j293 device tree
<marcan>
I guess it is possible it's all state machines though
<marcan>
like maybe iBoot just programs the operating point tables including whatever commands need to be sent to change voltages into registers, and then after that it's all state machines?
<alyssa>
sven: ok, pcie and type-c hotplug and dcp work for me on the same tree now =)
<j_ey>
kettenis: oh yeah, i noticed using the u-boot dts caused some issues with the pinctrl driver
<sven>
alyssa: now add nvme!
<kettenis>
j_ey: in what sense?
<alyssa>
sven: one thing at a time
<alyssa>
sven: when I unplug the mini cable (which became a gadget mode Ethernet) dwc3 has a WARN
<alyssa>
j_ey: that needs a sign off from a real name author
<sven>
alyssa: does hotplug work as long as you stick to host mode for the other port though?
<alyssa>
sven: yes!
<j_ey>
alyssa: yeah, someone will have to take it over
<sven>
alright. some good news at least :D
<kettenis>
j_ey: pushed my diff that adds more pinctrl stuff to the U-Boot DT
<alyssa>
sven: but the primary benefit of hotplug for me would being able to free up the port used to transfer the kernel at boot time :-p
<alyssa>
so uh
<alyssa>
can't give the series a t-b :(
<j_ey>
kettenis: cool
<alyssa>
....yet ;)
<sven>
feel free to add a commit that fixes that part :-p
bgb has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
zigmars has joined #asahi
bgb has joined #asahi
<alyssa>
sven: i have to fix DCP first
bgb has quit [Ping timeout: 480 seconds]
X-Scale` has joined #asahi
bgb has joined #asahi
X-Scale has quit [Ping timeout: 480 seconds]
bgb has quit [Ping timeout: 480 seconds]
bgb has joined #asahi
<alyssa>
sven: ...I never said I didn't have IRC on my school desktop 😇
bgb has quit [Ping timeout: 480 seconds]
<zigmars>
j_ey: thanks for helping me out the other day with hv os x kernel. Shall I update the wiki sidebar with smth like this https://pastebin.com/raw/fX1fjkhA (added 'Unsorted' section) so that it's harder to miss hypervisor wiki link now?
<j_ey>
zigmars: maybe just add the hypervisor link under "Places to start:"
<zigmars>
j_ey: ok
<zigmars>
is it expected that under hv, the mb-air built-in keyboard & trackpad don't work when os x is booted?
<alyssa>
You'd think the novelty of using M1 as my honest-to-goodness desktop would have worn off my now, but it hasn't :3
<j_ey>
that isnt expected, I think they worked for me
<j_ey>
-think, they deinitely worked since I logged in!
<zigmars>
well, i plugged in external mouse & kb, and that worked
<sven>
alyssa: go do your homework :P
<alyssa>
sven: im staring at it
<j_ey>
zigmars: weird, no idea why that would be
<zigmars>
hmmm, i'll then probably try putting this kernel back in place of m1n1 and double check that it works then
<j_ey>
(I havent tried the hv with smp + macs, but i doubt thats the issue)
<zigmars>
j_ey: I grabbed m1n1's commits from today & yesterday and internal kb & mouse work now through hv. :)
bgb has joined #asahi
<j_ey>
zigmars: weird, nothing in m1n1 should affect it
<zigmars>
j_ey: I did other things as well: I re-did `kmutil configure-boot ...` with HEAD of m1n1 (and so didn't need to do chainload)
<zigmars>
also don't have OS X 12 beta, but rather 11.5.x or smth like that
<j_ey>
same
<zigmars>
might have grabbed the wrong mach-o kernel - does that matter?
<j_ey>
I think theres only one
<zigmars>
anyway kb & trackpad problem solved. mystery stays though. :D probably did do smth wrong.
bgb has quit [Ping timeout: 480 seconds]
bgb has joined #asahi
* tophevich[m]
is looking forward to tomorrows marcan teardown stream he "promised"
bgb has quit [Ping timeout: 480 seconds]
<alyssa>
zigmars: failing to match m1n1 version is an easy way to break stuff
<steev>
alyssa: to be fair, the m1 performance as an arm64 chip is second to none
<steev>
re: novelty not wearing off
<steev>
my daily driver is a lenovo yoga c630 which is probably the second best (maybe a chromebook lazor would be better, but then you have the chromebook keyboard, so you really lose anyway)
sg94 has joined #asahi
<alyssa>
steev: hey i don't mind chromebook keyboards
<alyssa>
well, samsung chromebook plus's anyway. definitely mind some of the veyrons
<steev>
nothing wrong with liking bad things. we all have tastes like that :P
<steev>
oddly, i like typing on mini, and nyan-big, but less so on snow/daisy
<steev>
minnie*
<steev>
peach also feels nice to type on
<steev>
but overall, i dislike all the missing keys
sg94 has quit [Ping timeout: 480 seconds]
mlq has joined #asahi
boardwalk has quit [Quit: Ping timeout (120 seconds)]
<j_ey>
there's some weirdness, my bottom right hand of cluster1 is.. very different to yours
<j_ey>
hm, my max freq was 2184MHz
<marcan>
pretty sure you're throttling
<alyssa>
macbook air power management differences?
<marcan>
either that or we actually do need more init on the air
<alyssa>
j_ey: are you connected to mains?
<marcan>
alyssa: fwiw it will do 3.2 on battery on macos
<marcan>
(briefly)
<j_ey>
im not on the mains no
bgb has joined #asahi
<marcan>
there might be some voodoo needed to get things working on the air, if this isn't just a throttling thing
<marcan>
j_ey: use cpu_pstates.py to just test the highest states. there's a pile of commented out stuff there. see if uncommenting any/all of it fixes it.
<j_ey>
stilla lot better than 600mhz!
<marcan>
also, this could be a firmware difference
<j_ey>
marcan: "== Final CPU frequencies ==", for cpu in range(7):, that should be 8
<marcan>
I assume you're not on 12.0 like I am?
<marcan>
j_ey: no
<j_ey>
correct, I'm on 11.5
<marcan>
actually, that was a hack
<marcan>
j_ey: fixed
<marcan>
I was using cpu7 in parallel to test the boost thing, forgot to remove it
<alyssa>
if it's firmware, i'll know when i try on my 11.x mini
bgb has quit [Ping timeout: 480 seconds]
<marcan>
alyssa: so you're doing the DCP driver on the 11.x ABI?
<marcan>
I thought you were on 12.x, with the 4 slots for surfaces and such
<marcan>
(that was kind of the point of going to 12.x, getting ahead on those ABI changes, which is why I don't want to support 11.x at all for the initial merge of this)
<marcan>
dcp.py is all tested on 12.0 beta5 (I should update to beta6...)
<j_ey>
marcan: PMGR is missing in the code, I think it's PMGR = 0x23B700000, right?
bgb has joined #asahi
<marcan>
ah yes
<marcan>
this is the actual waste-of-time-on-stream poke all the things version, if you want to try it too: https://mrcn.st/p/L92lzFKJ
<marcan>
anyway, I should get some sleep :)
<j_ey>
ooh,
<j_ey>
CPU 7: 3203.84 MHz
<marcan>
ha, so either it's an Air thing or a firmware thing
<marcan>
j_ey: can you bisect what did it?
<j_ey>
I will try
<marcan>
I'm going to *guess* either the LIMIT things or the CLUSTER_DVMR thing
<j_ey>
marcan: whats the lowest value for set_pstate(1, ..) 0?
<j_ey>
I want to turn them down at the end of the script
<marcan>
0, but that's 24MHz which is slow enough it breaks the setup.py process
<marcan>
so better use 1
<j_ey>
(or maybe its just easier to reboot)
bgb has quit [Ping timeout: 480 seconds]
<marcan>
if we end up needing init, I'm tempted to just do it in m1n1 (and also default the pcores to at least 2GHz)
<marcan>
then linux doesn't have to care
* marcan
off to sleep :)
<j_ey>
marcan: you guessed right
<j_ey>
marcan: it was DVMR
zigmars has quit [Remote host closed the connection]
<alyssa>
j_ey: I can reproduce the same problem. Your patch fixes it.
<sven>
alyssa: hrm, okay. so i can confirm that the port that's used to boot the linux kernel seems to break :/
<alyssa>
sven: i thought i already confirmed that ;-p
<sven>
could've been just a weird combination of that ethernet gadget driver and your linux host :D
<sven>
but uhh.. i looks like i can go back to gadget mode
<alyssa>
marcan: ...okay, yep, GNOME is absurdly fast now :-p
<alyssa>
(running cpu_pstates.py with the CLUSTER_DVMR thing uncommented)
bgb has joined #asahi
<alyssa>
and my mesa builds are also absurdly fast now. good.
<sven>
ugh. this smells like it needs a PHY reset for some reason :/
<alyssa>
wouldn't eusprirse me
<sven>
i wonder what triggers the WARN_ON in the dwc gadget driver though
bgb has quit [Ping timeout: 480 seconds]
boardwalk has quit [Ping timeout: 480 seconds]
bgb has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
boardwalk has joined #asahi
bgb has joined #asahi
boardwalk has quit [Quit: Ping timeout (120 seconds)]
boardwalk has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
boardwalk has quit [Quit: Ping timeout (120 seconds)]
boardwalk has joined #asahi
bgb has joined #asahi
<alyssa>
lots of yak shaving later, I have a kernel with both dcp and nvme
<alyssa>
which required updating dcp to use the new rtkit api
bgb has quit [Ping timeout: 480 seconds]
<alyssa>
....how is my linux ext4 partition only 32GB
boardwalk has quit [Quit: Ping timeout (120 seconds)]
boardwalk has joined #asahi
<kettenis>
250G model?
bgb has joined #asahi
boardwalk has quit [Ping timeout: 480 seconds]
boardwalk has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
boardwalk has quit [Quit: Ping timeout (120 seconds)]
boardwalk has joined #asahi
bgb has joined #asahi
<alyssa>
kettenis: Yes, even so :)
<kettenis>
well, if you set aside 100G for Mac OS and follow the instructions on the wiki to set up the partitions to install Linux, you end up with something like that ;)
bgb has quit [Ping timeout: 480 seconds]
boardwalk has quit [Ping timeout: 480 seconds]
<j_ey>
alyssa: you added an e to my last name D: no worries, its a common typo :p
<j_ey>
and I would probably typo yours soo
boardwalk has joined #asahi
boardwalk has quit [Quit: Ping timeout (120 seconds)]
boardwalk has joined #asahi
boardwalk has quit [Quit: Ping timeout (120 seconds)]
boardwalk has joined #asahi
boardwalk has quit [Ping timeout: 480 seconds]
<alyssa>
j_ey: ELissa Roznswig?
<steev>
this is why i just always put people's nicknames
<alyssa>
perks of being a woman in tech: get to use my first name as my handle
* alyssa
apologizes to Alyssa Ross
* eta
has always known the latter as qyliss :p
<steev>
i use mine because i'm not creative enough to come up with something.... creative
<eta>
galaxy brain: change your legal name to your IRC nick
<alyssa>
eta: one name change per lifetime is enough
<eta>
sure, I used my one to change my legal name to my IRC nick >.>
<eta>
also this is very easy to do in the UK anyways
<alyssa>
cis privilege right there ;p
<eta>
>assuming I'm cis
<alyssa>
touche
<alyssa>
i'm uh
<eta>
"eta" is an okay name for a trans girl okay >.>