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
bcrumb has joined #asahi-dev
bcrumb has quit []
Hibyehello has joined #asahi-dev
DragoonAethis has quit [Quit: hej-hej!]
DragoonAethis has joined #asahi-dev
chadmed has quit [Remote host closed the connection]
chadmed has joined #asahi-dev
<chadmed>
is anyone game/able/has the time to sanity check some DSP improvements for j314? you will need to build a downstream branch of wireplumber
<chadmed>
pros: waaaay less cpu use (~2% of an ecore at 600mhz), sounds better, can theoretically hide the raw output from users
<chadmed>
cons: hiding the raw output breaks clients which expect to be able to access the entire node graph (e.g. carla)
nsklaus has quit [Ping timeout: 480 seconds]
gabuscus has quit []
chadmed has quit [Remote host closed the connection]
ourdumbfuture has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
ourdumbfuture has joined #asahi-dev
chadmed has joined #asahi-dev
gabuscus has joined #asahi-dev
<marcan>
chadmed: is that because it sees one end only? how does it break?
<chadmed>
carla can still see the physical node, but when it is open all sound stops playing. the "hiding" implementation is to set the permissions on the target node such that only wireplumber can access it, so i assume something carla wants to do with routing is being blocked by whatever permissions are set
<chadmed>
interestingly though, after commenting out the policy which does that, everything works again but the node is still hidden from plasma
<chadmed>
either way there are other issues with that functionality upstream so its disabled in dmitry's branch, i just re-enabled it to see what it would look like to end users when its eventually production ready
ourdumbfuture has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<marcan>
cool
<chadmed>
ive pushed the latest version for anyone whos interested in testing. no help installing it though unless i know i can trust you to not blow up your computer as per the rules in the readme ;)
<jannau>
nice: systemd has "ConditionFirmware=|device-tree-compatible(apple,j293)"
<jannau>
service is autostarted from udev via the touch bar digitizer input device anyway but the conditions should make it safe to install it unconditionally on all devices
WildCatFish has joined #asahi-dev
WildCatFish has left #asahi-dev [#asahi-dev]
WildCatFish has joined #asahi-dev
millenialhacker_ has joined #asahi-dev
millenialhacker has quit [Ping timeout: 480 seconds]
millenialhacker_ is now known as millenialhacker
dsharshakov has joined #asahi-dev
<dsharshakov>
Hello chadmed! WP filter-chain management is on my list now
<dsharshakov>
we can now develop that in online mode. what's next? should we polish up what's present without hiding the nodes (that's certainly not upstreamable because of being racy)?
millenialhacker has quit [Quit: Konversation terminated!]
millenialhacker has joined #asahi-dev
<chadmed>
dsharshakov: i havent really found anything wrong with the non-hiding stuff as it currently is tbh, works basically perfectly in my testing
<chadmed>
if i were being picky id say that its annoying that i cant set a rule to put the card into Pro Audio on startup and the filter chain having to be a one-liner is annoying for development but neither is _really_ a problem all things considered
<chadmed>
if you need a Tested-by you can use me :)
<chadmed>
one interesting thing i did notice (as above) is that the node i set to hidden is still hidden even after i comment out the rule
<chadmed>
but that might just be plasma caching something
<chadmed>
or .local/state/wireplumber
<chadmed>
i havent really investigated
<millenialhacker>
Hi, has anyone experienced dead/not working keyboard after updating asahi kernel ?
<millenialhacker>
(also not working touch pad)
<dsharshakov>
`put the card into Pro Audio on startup` <-- will do soon, need to integrate into policy-device-profile
<mps>
millenialhacker: did you built it or use from asahi distro
<dsharshakov>
`node i set to hidden is still hidden` <-- it's still the same node ID. You need to restart all three: pw, pw-pulse, wp
<millenialhacker>
mps I'm using my Gentoo installation (pre-chadme era)
<millenialhacker>
but I use his overlay now
<mps>
millenialhacker: then ask on #asahi-alt channel
<millenialhacker>
I just upgrade from 6.1 rc6 to 6.3
<chadmed>
dsharshakov: yeah i have killed and restarted all three multiple times, it persists across reboots even
<dsharshakov>
I want to add support for loading JSON from file into WP, so (ideally) you'll use ordinary device and node creation rules to set `wireplumber.hidden` and `wireplumber.filter-chain` or so
<dsharshakov>
Also I want to set target node when loading, so that it certainly and immutably attaches to the appropriate node
<dsharshakov>
then clean .local/state/wireplumber
<dsharshakov>
`loading JSON from file into WP` <-- essential to have a ton of props like with AutoEq headphone database. This way we'll only load four props into rulesets (device profile, hide device, hide node, attach chain)
<chadmed>
ah yeah that would be more ergonomic
<dsharshakov>
Thanks for review, will refactor some stuff for upstreaming. Better do a branch backup on your own storage as I'll be force-pushing that branch while refactoring for upstreaming
<chadmed>
ack, ill keep my local repo stale for now
millenialhacker has quit [Remote host closed the connection]
<dsharshakov>
Targeting 0.4 branch (master) without hiding nodes (impossible currently without pending changes)
<dsharshakov>
fyi PW multi-rate rate mismatch bug has been smashed. You probably didn't encounter that, but that was why I warned you about multiple allowed rates
millenialhacker has joined #asahi-dev
<dsharshakov>
(not released yet) but better keep a single rate enabled and use released version. let's keep git version mess to minimum :)
dsharshakov has quit [Quit: Page closed]
WildCatFish has quit [Remote host closed the connection]
lina has quit [Remote host closed the connection]
millenialhacker has quit [Quit: Konversation terminated!]
nsklaus has joined #asahi-dev
aafeke has joined #asahi-dev
chadmed has quit [Remote host closed the connection]
aafeke has quit [Quit: aafeke]
chadmed has joined #asahi-dev
ourdumbfuture has joined #asahi-dev
cylm has joined #asahi-dev
<sid-linux>
newbie question. Looking at the dockchannel codebase
<sid-linux>
Question: why is there not a `readl_relaxed` to get the current value of dcc->irq_base + IRQ_FLAG
<sid-linux>
before writing out BIT(hwirq) to prevent overwriting other entries ?
<sid-linux>
are the reads and writes "orthogonal" here because this is iomem?
<sven>
just guessing here because I don’t know the hardware but it looks like HW_FLAG is just „write but N to clear it“
<sven>
*IRQ_FLAG
<sven>
and *bit
<sid-linux>
(thanks for your reply -- can you please write your message with the typos fixed. I don't know where the *IRQ_FLAG and *bit needs to be subsituted)
<sid-linux>
s/HW_FLAG/IRQ_FLAG/ ?
<sven>
yes
<sven>
and s/but/bit/
<j`ey_>
(assuming what sven said is correct.) the IRQ_FLAG doesn't 'store' anything, the hw just responds to a write of that
<j`ey_>
whereas the MASK actually stores what's currently masked
<sven>
it’s just this dumb iPhone‘s „you pressed something slightly to the left of i and u is more likely than i in our ML model so I’m just gonna pretend you pressed u instead“ feature *sigh*
<j`ey_>
(and it's probably right, since it seems to be true for a lot of hw)
<ChaosPrincess>
jannau: sth weird is happening when loading tiny-dfr as a systemd daemon, libinput fails to detect the digitizer device.
<ChaosPrincess>
its probably due to load ordering, but not sure what exactly the problem is
ourdumbfuture has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<sid-linux>
j`ey and similarly since the IRQ_FLAG doesnt 'store' anything the read can ask the hardware to present it for a bit map of currently asserted IRQs? See `static void dockchannel_irq(struct irq_desc *desc)`
<ChaosPrincess>
also need to mask systemd backlight service or it messes things up
<j`ey_>
sid-linux: yeah seems to be
<sid-linux>
so here read and write achieve different outcomes in hardware...
aafeke has joined #asahi-dev
<sid-linux>
Also, I'm wondering how this can be done in a top half interrupt handler? Wont this communication with the device be subject to sleep issues?
aafeke has quit [Read error: No route to host]
ourdumbfuture has joined #asahi-dev
<sven>
what do you mean with sleep issues?
<sven>
and isn’t too half the thing that runs in irq context while bottom half is the irq thread anyway? (not that that changes anything afaict)
<sid-linux>
My understanding is that we can't do IO in the top half. We can only read from memory. IO could take a really long time
<sven>
that’s not true
<sid-linux>
Here `dockchannel_irq` is reading from the device
<sven>
you can’t sleep in irq context
<sven>
you also can‘t really access devices behind a slow bus like SPI
<sven>
but that readl is still plenty fast
<sid-linux>
but that readl is is reading from the dockchannel device
<sid-linux>
which is going to be slow??
<sid-linux>
(and therefore not permitted in the top half -- which is what `dockchannel_irq()` is in )
<sven>
still orders of magnitude faster than reading from a SPI bus which is probably what you’re thinking of
<sven>
doing sone I/O is totally fine in irq context. you usually *have* to do some I/O to e.g. read the irq and ack it in every handler
<sven>
the issue happens when you wait for I/O over a slow bus like SPI where you usually have to sleep (or wait for an irq which effectively is sleeping anyway)
<sven>
and the reason that’s a bad idea is because when you’re in irq context („top half“) many (or even all) other interrupts are disabled and will be delayed
chadmed has quit [Remote host closed the connection]
<sid-linux>
sven: thanks -- this is useful
chadmed has joined #asahi-dev
aafeke has joined #asahi-dev
chadmed has quit [Remote host closed the connection]
aafeke has quit [Quit: aafeke]
chadmed has joined #asahi-dev
ourdumbfuture has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
ourdumbfuture has joined #asahi-dev
WildCatFish has joined #asahi-dev
WildCatFish has quit [Remote host closed the connection]
ourdumbfuture has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
ourdumbfuture has joined #asahi-dev
minecrell has quit [Quit: :( ]
minecrell has joined #asahi-dev
WildCatFish has joined #asahi-dev
WildCatFish has quit [Remote host closed the connection]
WildCatFish has joined #asahi-dev
Bai-Chiang has joined #asahi-dev
Bai-Chiang has quit []
Bai-Chiang has joined #asahi-dev
Bai-Chiang has quit [Quit: bai-chiang]
gladiac has quit [Quit: k thx bye]
millenialhacker has joined #asahi-dev
Bai-Chiang has joined #asahi-dev
millenialhacker has quit [Remote host closed the connection]
millenialhacker has joined #asahi-dev
WildCatFish has quit [Remote host closed the connection]
Tomdownsouth has joined #asahi-dev
bai-chiang_ has joined #asahi-dev
WildCatFish has joined #asahi-dev
bai-chiang_ has quit []
WildCatFish has quit [Remote host closed the connection]
Tomdownsouth has quit [Quit: Konversation terminated!]
chadmed has quit [Ping timeout: 480 seconds]
WildCatFish has joined #asahi-dev
millenialhacker has quit [Remote host closed the connection]
WildCatFish has quit [Remote host closed the connection]
mlp has quit [Read error: Connection reset by peer]
mlp has joined #asahi-dev
WildCatFish has joined #asahi-dev
millenialhacker has joined #asahi-dev
millenialhacker has quit [Read error: Connection reset by peer]
ourdumbfuture has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
WildCatFish has quit [Remote host closed the connection]
irth has joined #asahi-dev
bps3 has joined #asahi-dev
ourdumbfuture has joined #asahi-dev
WildCatFish has joined #asahi-dev
WildCatFish has quit [Ping timeout: 480 seconds]
ourdumbfuture has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<amarioguy>
quick question is there a way from the ADT to check if a core is fused off or not? (long story need to be able to check if a core is fused off) or would i need to infer it being fused off based on board id or smth?
WildCatFish has joined #asahi-dev
WildCatF_ has joined #asahi-dev
WildCatFish has quit [Read error: Connection reset by peer]
Dementor has quit [Read error: Connection reset by peer]
Dementor has joined #asahi-dev
salimterryli has joined #asahi-dev
<jannau>
amarioguy: the cpus node will have only the active cores when looking at the adt from the device
SalimTer- has quit [Ping timeout: 480 seconds]
sid-linux_ has joined #asahi-dev
ourdumbfuture has joined #asahi-dev
gmarx has joined #asahi-dev
sid-linux has quit [Ping timeout: 480 seconds]
WildCatF_ has quit [Remote host closed the connection]
WildCatFish has joined #asahi-dev
gmarx has quit []
WildCatFish has quit [Ping timeout: 480 seconds]
sid-linux_ has quit [Quit: User exited]
WindowPain has joined #asahi-dev
WindowPa- has quit [Ping timeout: 480 seconds]
maximbaz has quit [Quit: bye]
maximbaz has joined #asahi-dev
<amarioguy>
jannau: ah fair, got it
c10l4844 has joined #asahi-dev
c10l4844 has quit []
c10l has joined #asahi-dev
chadmed has joined #asahi-dev
chadmed has quit [Ping timeout: 480 seconds]
zumi has quit [Remote host closed the connection]
zumi has joined #asahi-dev
zumi has quit [Remote host closed the connection]
psykose has quit [Remote host closed the connection]
psykose has joined #asahi-dev
lena6 has quit [Ping timeout: 480 seconds]
Bai-Chiang has quit [Remote host closed the connection]