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
crabbedhaloablut has quit [Ping timeout: 480 seconds]
yuyichao has quit [Quit: Konversation terminated!]
nsklaus has quit [Ping timeout: 480 seconds]
<marcan> mps: mesa doesn't care about machine differences and if you had kernel GPU problems I would expect faults in dmesg, so that sounds like a random userspace problem
<marcan> I haven't tested on all machines yet, but if I can't repro it anywhere I think you'll have to debug that one yourself. could be a musl thing.
gabuscus has quit []
gabuscus has joined #asahi-dev
crabbedhaloablut has joined #asahi-dev
nicolas17 has joined #asahi-dev
ncopa has quit []
roxfan2 has joined #asahi-dev
ellyq_ has joined #asahi-dev
roxfan has quit [Ping timeout: 480 seconds]
roxfan has joined #asahi-dev
roxfan2 has quit [Ping timeout: 480 seconds]
ellyq has quit [Ping timeout: 480 seconds]
ellyq_ is now known as ellyq
mkurz has quit [Ping timeout: 480 seconds]
nimprod3l has joined #asahi-dev
nimprod3l has quit [Ping timeout: 480 seconds]
<mps> marcan: binaries on both machines are same versions and same build. Even I could try to boot both from external usb ssd to check
<mps> and I forgot to mention that X crashes with sigabort when starting with asahi-mesa 20230603
<mps> while sway/wayland works fine with it
<mps> only non standard thing is local patches which enables speakers in kernel
<mps> but to repeat, this is not much important to me because I've got it working with older asahi-mesa (for which I forgot to write exact git commit from which I built it)
ncopa has joined #asahi-dev
mkurz has joined #asahi-dev
<mps> marcan: sorry for noise. it works fine on j293 when booted from usb ssd which is full copy of rootFS from j314s
<mps> and thank you for the hints
<mps> (I guessing 'apk upgrade' skipped something)
<marcan> crashing X sounds like what happens when you build regular mesa with the asahi driver enabled
<marcan> (particularly older versions, that got fixed recently)
<mps> I expected UAPI difference warning but didn't seen it
<mps> for long I use only asahi-mesa on these machines and not regular mesa from alpine repo
<mps> btw, with asahi kernel 6.3-5 battery lasts longer
<marcan> than with -4?
<marcan> compared to what?
<mps> compared to 6.2 series
<marcan> well yes, that's kinda the point :p
<marcan> < mps> I expected UAPI difference warning but didn't seen it <- if you build non-asahi mesa with asahi enabled there is no UAPI, that code is broken and never supposed to run at all
<mps> heh yes. just wanted to report
<marcan> that's what the lovely manjaro people enabled for us which broke the ALARM mesa
<mps> ah
<marcan> newer mesa properly neutered that driver entirely so now it's just dead code
<mps> I tried just once (about month ago) to build regular mesa and use with asahi kernel but (as expected) it didn't worked and I deleted it
<mps> now everything is very fine. thank you all devs
cylm has quit [Ping timeout: 480 seconds]
odak_ has joined #asahi-dev
odak_ has quit []
odak_ has joined #asahi-dev
bps3 has joined #asahi-dev
nsklaus has joined #asahi-dev
odak_ has quit [Quit: odak_]
bps3 has quit [Ping timeout: 480 seconds]
mkurz has quit [Ping timeout: 480 seconds]
mkurz has joined #asahi-dev
chadmed_ has joined #asahi-dev
chadmed has quit [Ping timeout: 480 seconds]
r0ni has quit [Ping timeout: 480 seconds]
bps3 has joined #asahi-dev
Dementor has quit [Read error: Connection reset by peer]
Dementor has joined #asahi-dev
i509vcb has joined #asahi-dev
<axboe> is it known the the latest kernels (update from a today, few days ago, and the one asahi-wip before that) don't wotk with X and accelleration?
<axboe> I just get llvm-pipe on those
<axboe> f72b509e14dcaea5aa7fcd21083e25d60a20bcc3 seems like the last known good one on my end in terms of that
<mps> axboe: you need new asahi-mesa
<axboe> ah shoot
<axboe> why didn't I think of that
<axboe> building...
maria has quit [Remote host closed the connection]
<axboe> looks like I need a newer directx-headers first
<psykose> heh
<psykose> is there a reason to enable that at all? i don't think it has uses outside of WSL integration
<psykose> (specifically gallium d3d12 thing)
<psykose> idk if anything else uses those
<axboe> was thinking that too, maybe I can just disable directx
<psykose> for that one it's just not adding d3d12 to -Dgallium-drivers
<psykose> yep, quick grep shows it's just that, experimental-vulkan (also for wsl), and then platform==windows stuff
<axboe> building, let's see...
maria has joined #asahi-dev
<mps> axboe: here is my ugly hacked alpine APKBUILD https://tpaste.us/JRLq
<mps> but it works
maria has quit [Remote host closed the connection]
maria has joined #asahi-dev
<psykose> could probably delete half of that :p
<mps> psykose: yes, but who cares, it build fast anyway
<psykose> of course :)
<mps> actually I had smaller one but in last two builds added some options to be closer to regular asahi
<psykose> it's a bit unintuitive to get such an override to work with apk but that provides= per name works, so you got it, it's what i would have done
<mps> imagine working in frenzy ;-)
zachole has joined #asahi-dev
maria has quit [Remote host closed the connection]
maria has joined #asahi-dev
maria has quit [Remote host closed the connection]
maria has joined #asahi-dev
brolin has joined #asahi-dev
<axboe> mps: yep that worked
<mps> nice
brolin has quit [Ping timeout: 480 seconds]
<mps> axboe: you are using X?
<axboe> yep, thanks for the hint. not sure why I didn't think of a mesa upgrade...
<axboe> using X, yep
<mps> does TearFree option works for you
<axboe> I think I tried it back when it was a separate patch, but haven't run it since
<mps> I still have sometimes in some apps cursor flickering
<axboe> yeah that one I have too for some web pages
Guest2187 has quit [Quit: Bridge terminating on SIGTERM]
rhysmdnz has quit [Quit: Bridge terminating on SIGTERM]
<mps> yes, also I have it just for some pages
<mps> s/just/only/
* mps is self taught in english
<axboe> does the tearfree fix it for you?
<axboe> I didn't have any noticeable tearing in videos etc, and the flickering for some pages wasn't bad enough for me to care about testing it any further
Jamie has joined #asahi-dev
brolin has joined #asahi-dev
rhysmdnz has joined #asahi-dev
<mps> axboe: no, Tearfree doesn't help for those web pages in my case
Jamie is now known as Guest2261
<axboe> ok
<axboe> I'm on xfce4, looks like they do have some experimental wayland support going. once that works, I'll likely just switch away from X, I have no sentimental attachment to it :)
brolin has quit [Ping timeout: 480 seconds]
<mps> I'm contemplating build hyprland for alpine and test it. if it works fine maybe also will switch from X
<mps> but don't have enough time for such things for now
<psykose> i could built it for you but it's mostly a pile of shit
brolin has joined #asahi-dev
<mps> psykose: hm, could be, I have very little experience with wayland and only used sway for some testing
<mps> what is your advice for heavy and long time awesome wm user
<psykose> hmm
<psykose> i don't think there's something quite like awesome
<mps> to me hyprland looks close but I"m not sure
<psykose> the most polished non-major-de wm is sway which you already tried, ofc nowhere near the same
<mps> I like lua scripting with awesome
<psykose> depending on what the scripting does (i don't remember, though i used awesome for a year before) most of it is implementable with whatever hooks/shortcuts that then interact via api client (i.e. on sway, swaymsg sends stuff, there's various 3rdparty things to make sway do stuff, ..) and most of the compositors have some ctl command
<mps> psykose: anyway, here is where I reached with hyprland, https://tpaste.us/rjNj
<psykose> that said, a bit offtopic
<psykose> sure, sure, i'll build it
<mps> it will need patch for glvnd which we don't have in alpine
<mps> and right, we are really off-topic here, sorry
brolin has quit [Ping timeout: 480 seconds]
brolin has joined #asahi-dev
r0ni has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
amarioguy has joined #asahi-dev
brolin has joined #asahi-dev
cylm has joined #asahi-dev
axboe has quit [Remote host closed the connection]
axboe has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
brolin has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
brolin has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
brolin has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
mkurz has quit [Ping timeout: 480 seconds]
aafeke has joined #asahi-dev
brolin has joined #asahi-dev
bps3 has quit [Ping timeout: 480 seconds]
aafeke has quit [Quit: aafeke]
brolin has quit [Ping timeout: 480 seconds]
aafeke_ has joined #asahi-dev
brolin has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
<jannau> PCIe looks weird on the mac pro. the HDMI ports are on Apple I/O card on a 4 lane PCIe Gen3 slot. are they using unused contacts for DP/HDMI?
aafeke_ has quit [Quit: aafeke_]
mkurz has joined #asahi-dev
aafeke_ has joined #asahi-dev
<jannau> 4 8x and 2 16x PCIe Gen4 slots but no information how many lanes the m2 utra has
aafeke_ has quit []
aafeke_ has joined #asahi-dev
brolin has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
brolin has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
qyliss has quit [Quit: bye]
brolin has joined #asahi-dev
qyliss has joined #asahi-dev
brolin has quit [Ping timeout: 480 seconds]
chadmed_ is now known as chadmed
aafeke_ has quit [Quit: aafeke_]
nickchan has left #asahi-dev [#asahi-dev]
brolin has joined #asahi-dev
<marcan> the Mac Pro PCIe lane count baffles me
<marcan> I'm starting to think it's a big cheat with PCIe switches and it's all funneled down to 16 lanes on the SoC side
<marcan> like they planned this for a real, separate Mac Pro chip with actually more PCIe, canceled that, and "made it work" with M2 Ultra
<ChaosPrincess> there is something that is called a pcie-switch on the i2c bus in the dt
<marcan> lol.
<ChaosPrincess> also, it appears that they are in fact connecting most slots to 2nd die
<ChaosPrincess> in dts, it has all the slots, while the first die's apcie-ge has the io card connected
<marcan> yeah, that tracks
<ChaosPrincess> switch-mgmt-ep
<ChaosPrincess> oh, would you look at that
<ChaosPrincess> marcan: wrt nand chips, are they using pcie phys to connect those? cause i didnt think pcie and nv-ddr use the same phys
<marcan> nv-ddr?
<marcan> apple doesn't use that
<marcan> apple NANDs are PCIe
<marcan> it's custom
<marcan> they have their own controllers on the NANDs
<marcan> https://social.treehouse.systems/@marcan/110493886026958843 <- best guess as to the topology
<ChaosPrincess> oh, i thought they used off the shelf nand and just put the controller inside the m-chip
<marcan> the SSD controller is in the SoC
<marcan> but each NAND chip has its own sub-controller
<marcan> a whole ass 64-bit ARM too
<marcan> (because that's how Apple rolls)
<ChaosPrincess> lol
<ChaosPrincess> inb4 the new controller also still has broken arm pcie as usual
<marcan> I should test that on the machines I have...
<marcan> this at least explains why the changes to the PCIe controller I guess, if they were planning to use the same old APCIe for slots
brolin has quit [Ping timeout: 480 seconds]
<marcan> ohh but this is interesting
<marcan> the MMIO addresses for apcie-ge are totally different
<marcan> so maybe the digital side is a completely separate controller from "traditional" apcie
<marcan> and they just share the same PHYs
<marcan> that said, the device tree regs look exactly like normal apcie except a single port (which makes sense since it goes to the bridge chip)
<marcan> so probably easy to support
pzc is now known as zspec
zspec has quit []
zspec has joined #asahi-dev
alyssa has joined #asahi-dev
<alyssa> marcan: what's the status on releasing the new kernel+mesa?
<alyssa> are we bottlenecked on me writing my dang blog post? T_T
<marcan> alyssa: ... am I supposed to answer that question? :p
<alyssa> shit I'll write it ok ok ok
<alyssa> :p
kesslerdupont has joined #asahi-dev
roxfan has quit [Ping timeout: 480 seconds]
roxfan has joined #asahi-dev