ChanServ changed the topic of #asahi-dev to: Asahi Linux: porting Linux to Apple Silicon macs | General development | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Logs: https://alx.sh/l/asahi-dev
chadmed has joined #asahi-dev
yuyichao_ has joined #asahi-dev
nirgo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
chadmed has quit [Quit: Konversation terminated!]
chadmed has joined #asahi-dev
pFalken has quit [Ping timeout: 480 seconds]
pFalken has joined #asahi-dev
PhilippvK has joined #asahi-dev
nyx_o has joined #asahi-dev
nirgo has joined #asahi-dev
nirgo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
BenPetterborg[m] has joined #asahi-dev
nirgo has joined #asahi-dev
nirgo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
the_lanetly_052 has joined #asahi-dev
<jannau> still does not work with all dart-disp0 as reserved regions/mappings
<jannau> sven: I install the reserved mappings in apple_dart_setup_translation() before apple_dart_hw_enable_translation()
<jannau> using apple_dart_map_pages() which does not complains, addresses and sinze is what I expect
<sven> hrm, do you get dart pagefaults?
<sven> and doesn’t the dart driver disable all translations during probe time?
<jannau> if the page faults are logged I don't see anything. kernel seems to hang after dcp-piodma attaches to the iommu
<jannau> yes, it resets during probe so the simple solution to do this during device attache will not work
* jannau should have looked closer at the other patches in the thread
bpye has quit [Quit: The Lounge - https://thelounge.chat]
bpye has joined #asahi-dev
<sven> if the framebuffer dart needs more than just the framebuffer maybe we need to introduce something like apple,keep-entries in the DT instead that just recreate whatever mappings it finds during probe time whenever a domain is attached
<sven> and leaves that SID alone otherwise during probe time and also only allows DMA and UNMANAGED domains
<sven> that feels like a bit of hack though but oh well :-(
<jannau> it's not too bad, there are two regions for for stream 0 and one region for stream 4
ChaosPrincess has quit [Quit: WeeChat 3.4]
ChaosPrincess has joined #asahi-dev
<jannau> all are described in "/chosen/carveout-memory-map" and I have m1n1 code which verifies that the region is continously mapped in the dart stream we expect
MajorBiscuit has joined #asahi-dev
<jannau> currently just by checking if the last io address translates to the last physical address
<jannau> I've have not yet thouroughly checked if the region ids are stable over devices/iboot versions
<sven> hm.. i wonder what those non-framebuffer regions are
<sven> even for the DCP DART (the locked one) most (all?) of the mapped regions are just the DCP firmware sections described in the ADT fwiw
<jannau> both additional buffers are all 0 after iboot
<jannau> still zero after a full linux boot, so those might not be used/needed (for normal operation)
littledust3403[m] is now known as MarxLandRover[m]
MarxLandRover[m] is now known as NotHere[m]
the_lanetly_052 has quit [Ping timeout: 480 seconds]
the_lanetly_052 has joined #asahi-dev
the_lanetly_052 has quit [Ping timeout: 480 seconds]
ViniciusSantos[m] has joined #asahi-dev
<sven> jannau: https://f.svpe.de/0be7ceb398bd41885a0bb017f6ff7b01439a39d209ac783c5d62e57253c426a6_locked-darts-braindump.txt just wrote that during my lunch break, hope it makes at least some sense
<j`ey> sven: "you can generally attach a single to domain" a single device?
<sven> ah, no
<j`ey> or just remove 'to' in that sentence actually
<sven> yup
<sven> my lunch break is over now though ;)
os0 has quit [Quit: The Lounge - https://thelounge.chat]
os has joined #asahi-dev
<alyssa> sven: "apple,keep-entries in the DT instead that just recreate whatever mappings it finds"
<alyssa> so... apple,force-locked then? :-p
kov has joined #asahi-dev
nirgo has joined #asahi-dev
<sven> alyssa: one half of that, yes
nirgo has quit [Ping timeout: 480 seconds]
jbowen has quit [Quit: Lost terminal]
the_lanetly_052 has joined #asahi-dev
the_lanetly_052__ has joined #asahi-dev
the_lanetly_052 has quit [Ping timeout: 480 seconds]
<maz> is there a branch with an SMC driver somewhere? I may try and have a go at implementing PCIe hotplug during the weekend.
<j`ey> no, marcan never pushed it
<alyssa> i had a big ol' pile of hacks somewhere but um
espo has joined #asahi-dev
jbowen has joined #asahi-dev
espo_ has joined #asahi-dev
espo has quit [Ping timeout: 480 seconds]
rafaelmartins has quit [Quit: https://rgm.io/]
rafaelmartins has joined #asahi-dev
MajorBiscuit has quit [Quit: WeeChat 3.3]
the_lanetly_052__ has quit [Ping timeout: 480 seconds]
espo_ has quit [Quit: Leaving]
<alyssa> so ummm
<alyssa> i accidentally hit backspace and gnome made the annoying chimg..
<alyssa> *chime
<alyssa> ...on the internal speaker
<alyssa> why did it probe this time when it didn't in the past? i don't know!
<alyssa> povik: I smell a race
<alyssa> cs42l42 probed fine this time around
<alyssa> tas2770 still unhappy, tapple-mca still spams about dx DMA channels, um
<alyssa> apple-admac added to iommu before cs42l, maybe that's different
<tpw_rules> alyssa: what system?
<alyssa> tpw_rules: M1 Mini
<alyssa> only thing I can think of is I just did a warm reboot from macOS
<tpw_rules> i've noticed some sort of race condition too. i only get sound maybe one out of every 5 reboots
<alyssa> so maybe there's early hardware init that macOS did
<tpw_rules> i didn't try to diagnose, but the problem isn't just you
<alyssa> and hopefully 12.x iBoot does this
<tpw_rules> i was on 12.0.1 so it must not
<povik> uh oh
<alyssa> Hfm
<povik> i am running 11.x fwiw
<alyssa> k
<alyssa> definitely some nondeterminism happening
<alyssa> that causes cs42l42 not to probe, sometimes
<povik> can't say i am happy about that, to be honest
<alyssa> (specifcally causes all communication with cs42l42 to fail, starting with the ID/revision reads.)
<alyssa> how does the AP talk to cs42l? is that i2c?
<tpw_rules> it seems to probe to me, just no sound comes out
<povik> alyssa: yes
<alyssa> so maybe an i2c issue?
<povik> yours is a failure mode i haven't seen yet
<povik> maybe i should try running out of hypervisor a couple times
<povik> tpw_rules: post dmesg whenever you get to it
<alyssa> latest issue blocking valhall support in mesa ... my uh type-c ethernet is pretty dead
<povik> hmm, ran 5 times out of hypervisor, didn't see the probe fails
<alyssa> hmm
<alyssa> povik: at any rate I am currently, err, testing the driver with the Rent soundtrack
<povik> let me look that up
<alyssa> ;-D
<alyssa> they call me, they call me, Mimi!
<povik> it wasn't for me :-p
<povik> anyway
<povik> look who showed up
<povik> > [ 1.379196] tas2770 1-0031: ASoC: error at soc_component_write_no_lock on tas2770.1-0031 for register: [0x00000001] -6
<alyssa> ooh
<povik> i wanted to chase that the other day but didn't get any
<alyssa> (it was not a recommendation. unless you like bisexual 90s Queen Elsa)
<povik> figured
<povik> who wants to bet it's this write
<povik> "Once all supplies are stable the SDZ pin can be set high to initialize the part. After a hardware or software reset additional commands to the device should be delayed for 1 mS to allow the OTP to load."
<povik> from tas2770 datasheet
<povik> yeah, that's not happening here
<povik> doesn't help us with the cs42l42 probe unfortunatelly
<kettenis> yeah, you defenitely need a 1ms after powering the tas5770 chip up
<kettenis> otherwise it doesn't respond
<tpw_rules> maybe 2ms just to be safe? how accurate is msleep
<povik> i assume it's guaranteed to be at least 1 ms
<Glanzmann> tpw_rules: I don't have all day.
<jannau> it will not be shorter but can much longer, iirc another function is recommended for low single digit msleep
<jannau> usleep_range()
<povik> i wonder what that unused TAS2770_MDELAY was there for in the driver
<povik> anyway will switch to usleep_range then
<povik> alyssa: when you have the chance please try p.read32(0x23d1f002c) after fresh boot and send me the value
<povik> that's the cs42l reset gpio
<povik> improbable there will be a surprise but worth checking
commandoline_ has quit [Quit: Bye!]
commandoline has joined #asahi-dev