ChanServ changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Chrome OS Trogdor Devices - Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630 - Lenovo ThinkPad X13s - and various other snapdragon laptops) - https://oftc.irclog.whitequark.org/aarch64-laptops
tobhe has joined #aarch64-laptops
tobhe_ has quit [Ping timeout: 480 seconds]
hexdump01 has joined #aarch64-laptops
hexdump0815 has quit [Ping timeout: 480 seconds]
tobhe_ has joined #aarch64-laptops
tobhe has quit [Ping timeout: 480 seconds]
Segfault[m] has joined #aarch64-laptops
<Segfault[m]> huh neat i wasn't expecting the x13s webcam to "just work" so easily, i guess libcamera and pipewire help a lot
<Segfault[m]> the quality isn't great but it seems like a lot of that is stuff like bloom and bad focus from a crap lens which no amount of dsp magic will fix lol
<KieranBingham[m]> Segfault: yeah the quality can/will/is being worked on. First step was to get it to run and plumbing everywhere.
<KieranBingham[m]> Segfault: what os have you installed to use it so far? And have you also run in Firefox/video calls?
<Segfault[m]> fedora silverblue 41, and yes after i set the pipewire flag the camera does work in firefox
<Segfault[m]> aside from better processing what improvements are there still to make?
<KieranBingham[m]> Currently everything is being handled through the CPU, although the CPU usage isn't high "yet". Pipewire is getting a very small crop of the image so you get a big face in video calls.
<KieranBingham[m]> GPU acceleration is being worked on, and also when I get to it I'll take tuning instead so we can work on colour corrections.
<KieranBingham[m]> There's lots that "can be done" just about finding time and resources
<KieranBingham[m]> I'm happy to hear of other testers ahem users though :-)
<Segfault[m]> heh
<Segfault[m]> i noticed the video was kinda blurry even at the full 5mp, is that just a limit of the lens in this camera or is there something missing from the sensor driver?
<KieranBingham[m]> I don't know at the minute. I didn't think the was a vcm focus but I haven't looked!
<KieranBingham[m]> Can you share a screen shot / picture?
<Segfault[m]> tbf i was taking a fairly far off picture and i think this is a fixed focus camera so maybe it's just like that lol
<KieranBingham[m]> Quite possible. Definitely don't expect high quality yet. Think of it as a built in camera from 25 years ago but it should be upgraded with software updates :-)
<Segfault[m]> gnome's snapshot program defaults to 1280x1080 for some reason, after patching abgr8888 support into the libcamera test app i was able to get a preview of the full 5mp mode
<KieranBingham[m]> What did you have to patch?
<Segfault[m]> the sdl preview code, for some reason it only supports nv12, yuv and mjpeg
<KieranBingham[m]> Oh. Using cam ? I guess that hasn't had much use. Could you submit your patch to the list please?
<Segfault[m]> yeah i can clean it up and submit it, it's a fairly simple change, basically just copying this but s/NV12/ABGR8888 https://patchwork.libcamera.org/patch/17008/
<KieranBingham[m]> Great, yeah we only add formats as they are tested, so no one has added "every format" to that SDL interface.
patrickm has quit [Remote host closed the connection]
hogliux has joined #aarch64-laptops
<hogliux> macc24: great work on the EC driver for the yoga 7x. Suspend now works perfectly for me
<hogliux> macc24: you said that you would use the yoga 7x' camera as a lid switch: you have the webcam working on the 7x? Which module do I need to enable to get it working?
hogliux has quit [Quit: Page closed]
jhovold has joined #aarch64-laptops
macc24 has joined #aarch64-laptops
<Segfault[m]> <KieranBingham[m]> "Quite possible. Definitely don't..." <- do you know if libcamera ever uses the 1296x972 pixel binned mode or does it just use 2592x1944 and downscale?
crisma has joined #aarch64-laptops
<KieranBingham[m]> Segfault: Looks like it's selecting 1296x972 : https://paste.debian.net/1330765/
mcbridematt has joined #aarch64-laptops
<macc24> does anyone have pics of yoga slim 7x motherboard?
patrickm has joined #aarch64-laptops
<JensGlathe[m]> SoC is X1E-78-100
SpieringsAE has joined #aarch64-laptops
<JensGlathe[m]> nice cable mgt
<SpieringsAE> I didnt time the times I compiled the kernel on my asus, but I feel it was faster, I guess I was compiling a more trimmed down kernel though.
<SpieringsAE> It was the jhovold fork with some slight tweaks
<JensGlathe[m]> oh thats the full fat everything what ubuntu has but qcom only kernel. Its massive
<SpieringsAE> jeah that makes sense, I only really compile kernels at work for our imx8 platform, and that one is quite small aswell
<SpieringsAE> its like 2 min on my ryzen 5900x
<JensGlathe[m]> the 3950x is blowing nearly 200W when fully loaded, so its ~ 150+W
<macc24> HdkR: thanks, i forgot to mention that i was asking about *
<macc24> pic of a *bare* motherboard in case there would be some labeled test points
<JensGlathe[m]> like the micro-usb dbg_uart?
SpieringsAE has quit [Remote host closed the connection]
<macc24> yeah but for the i2c bus the ec sits on
<HdkR> macc24: Ah, would need to tear it down more for that
<kuruczgy[m]> macc24: FYI if you don't already know, the hall sensor is near the left-side front USB-C port, according to my tests using a magnet.
<kuruczgy[m]> I have been thinking about how to get good board pics or schematics, but I don't see an easy solution... I think everyone here is too afraid to tear down their only laptop, and searching google/youtube for teardowns did not turn up any results.
<HdkR> I'm mostly too lazy to get thermal paste out again
<kuruczgy[m]> I don't know of any non-destructive methods for obtaining schematics, AFAIK even CT is not high enough res for the PCB traces
<kuruczgy[m]> One option would be to buy a broken motherboard for (hopefully) cheap, but I think the laptop is too new to have broken ones floating around yet
<kuruczgy[m]> HdkR: Well if you ever take it apart more for whatever reason please make sure to take lots of hi-res photos :)
<JensGlathe[m]> FCC ID replacement part maybe?
<kuruczgy[m]> Hm good idea actually, I don't know the FRU number for the mb itself but maybe just searching for the laptop itself might be enough
<macc24> kuruczgy[m]: well i would tear it down if i had a good thermal paste for replacement :/
<calebccff> what commands are people running to dump ACPI tables for the laptops? like https://github.com/aarch64-laptops/build/tree/656d4d51ec9c580c435fb7f64f6cebabc62fd4c3/misc/acer-gateway-gwtn133-1/acpi
<JensGlathe[m]> acpidump.exe on Windows, with the intel acpi toolkit
<kuruczgy[m]> I could not find the FCC id for the laptop itself, only the WiFi/BT module.
<kuruczgy[m]> This alibaba listing has some photos of the mb, but it's quite low res: https://www.alibaba.com/product-detail/FOR-For-Yoga-Slim-7-14Q8X9_1601230180305.html
<kuruczgy[m]> Also here is the repair manual, again quite low res photos, but might be useful: https://download.lenovo.com/consumer/mobiles_pub/yoga_slim_7_14_5_9_hmm.pdf
<kuruczgy[m]> macc24: But even if you tear it down, and find the hall sensor, how do you find out what's it connected to? I think this board is too dense to see traces, so you have to poke around with a multimeter. And even then, you are out of luck if it's connected to something with a BGA package.
<JensGlathe[m]> wow the thing is mostly PMICs
<kuruczgy[m]> How do you see that?
<JensGlathe[m]> Chip surrounded by capacitors (lots) and one big ferrite coil as one square group. Count them.
<JensGlathe[m]> 15 or so
<travmurav[m]> 3-4 of those may be chargers but yeah the x platform is fun
<travmurav[m]> iirc last time I counted them for fun on the surface mobo photos, there were 3 or 4 smb* charger chips
<travmurav[m]> the rest were pm* iirc
<kuruczgy[m]> All of these PMICs are listed in x1e80100-pmics.dtsi, right? And in Qualcomm fashion a google search does not turn up anything useful about the chips, basically the only result is the dtsi file itself...
<travmurav[m]> did anyone get one of the 8 core based laptops yet?
<JensGlathe[m]> I remember somebody writing about that qcom is selling the whole set only as you need them regardless, and customers not happy about so many of these parts
<travmurav[m]> I think there is not all listed and also they're all qcom chips
<travmurav[m]> i.e. pmc8380 and technically the x1 silicon is sc8380
<travmurav[m]> but I'd say it's just "the qcom way"
<travmurav[m]> the whole platform is heavily integrated
<travmurav[m]> even 10 years ago pmic control was tied to literal cpu instructions
<JensGlathe[m]> they only play together (in qcoms application world)
<kuruczgy[m]> I guess one good thing about that is that it should be mostly uniform across the laptops then
<travmurav[m]> yeah it will be, don't think any oem will be able to customize that part of the platform
<travmurav[m]> dont' think it ever really happened for all qcom things
<JensGlathe[m]> like hall sensors 🤨
<travmurav[m]> beyond few edge cases where qcom has i.e. two allowed designs
<travmurav[m]> for pmic setup that is
<macc24> kuruczgy[m]: my plan is to tap into i2c bus where the ec sits at and trace the communication between it and windows
<kuruczgy[m]> macc24: That would be cool. Though I wonder if a software solution for that would be somehow possible, e.g. hook into the windows kernel somewhere...
<macc24> probably?
<macc24> there's stuff in the dsdt that traces i2c communication
* travmurav[m] is annoyed there is no good official name for the 8 core purwa silicon and "x plus" is ambiguous
<kuruczgy[m]> I mean I would guess that all the silicon is the same and it's just binned, no?
<travmurav[m]> to the best of my knowledge 10/12 is hamoa silicon and 8 is different "purwa" silicon
<travmurav[m]> so 10 core x plus is partial-goods but 8 core is a different chip
<kuruczgy[m]> macc24: What do you mean by "traces"? Could those traces just be found in a windows event log or something?
<travmurav[m]> so the funny question with purwa is whether they kept it software compatible or nt
<JensGlathe[m]> one would sort of assume so
<travmurav[m]> I dont think thye /have to/ sicne they can just ship different acpi/pep anyway
<JensGlathe[m]> I see only one dell and one Lenovo IdeaPad
<macc24> kuruczgy[m]: 'Debug = "something something"' stuff
<travmurav[m]> or i.e. if the difference is errata xD
<travmurav[m]> I guess the oems would be annoyed if thye need different bsp package for different skus xD
<macc24> travmurav[m]: wait isn't it even more binned down x1 elite chip?
<travmurav[m]> to the best of my knowledge, no
<macc24> o_o
<travmurav[m]> this makes it easier to see ig
* travmurav[m] is somehow surprised it was this easy to find two photos of the same mobo with both chips on
<robclark> JensGlathe[m]: re: kernel compile time.. if you are using distro config or defconfig, that is even more impressive because arm64 kernel has _way_ more drivers than x86 (because things that are in ACPI on x86 are kernel drivers on arm.. and way more arm vendors)
tobhe_ is now known as tobhe
<macc24> ^ you can disable non-needed(amdgpu, radeon, intel stuff and nvidia) drivers and non-qcom platform support for a quick way to speed up compiles
alfredo has joined #aarch64-laptops
<robclark> right.. just pointing out that kernel compile is hard to get apples-to-apples comparison unless you are cross compiling for the same arch with same config (phoronix kernel compile benchmark makes this mistake since it does native compile with the arch's defconfig.. but other compile benchmarks show the x1e78100 / yoga 7x somewhat faster than zen5)
<macc24> it definitly *feels* like my yoga 7x is faster at compiling stuff than a ryzen 6900hs
<JensGlathe[m]> I have it limited to qcom-only platform, basically. This disables quite a bunch already. But all the odd features they have enabled are usually on if there's an equivalent on the arm64 side. Sure there's a lot of drivers you could prune, but saw no need yet. it's the same config on all machines, a good reference. Try this on a raspi4 and come back 2 hours or so later.
<macc24> the only thing i'd trust a raspberry pi to do is to run my 3d printer
<JensGlathe[m]> I bootstrapped my Volterra with it
<travmurav[m]> at least you don't have to put actual work hours in to use it with linux
<travmurav[m]> I guess externally powered, compute only workloads maybe don't require too much extra work on qcom nowdays, I'm being a bit unfair
<macc24> travmurav[m]: but where's the fun in having something just work with linux?
<travmurav[m]> I mean, depends on if you want to have fun or want to... like... have device actually suspend? xD
<macc24> oh my yoga slim 7x suspends just fine
<travmurav[m]> no
<macc24> the "resume" part is sometimes a bit flaky with the display
<travmurav[m]> I mean suspend 4 days ago and resume today with 30% battery left :P
<JensGlathe[m]> getting better, but too much power loss over a day of suspend
<travmurav[m]> :^)
alfredo has quit [Quit: alfredo]
SpieringsAE has joined #aarch64-laptops
<SpieringsAE> My pinebook pro has about that graph and I turn it on/off everytime I use it hahaha
<travmurav[m]> lol
<travmurav[m]> but no, this is suspend
<SpieringsAE> But I can pretty much only use it on net power, on battery it seems to crash sometimes. I think it is my nvme ssd sucking too much power
<travmurav[m]> 7c has actually working suspend thanks to cros
* macc24 is now tempted to get a pinebook pro to fix all the stuff
<SpieringsAE> yeah thats rough
<travmurav[m]> even though I needed to fix a few things for woa, after I did, I literally never power the thing off, just close the lid and let it sleep
<SpieringsAE> the pinebook pro needs hardware to be fixed
<SpieringsAE> it just really needs better power handling, the battery or the regulators are too weak
<travmurav[m]> and aspire1 platform is not the best hw design I feel, could go lower probably
<SpieringsAE> then again it is too sad of a cpu to care much about, fun fact, I'm typing this from my pinebook lol
<travmurav[m]> so I'd expect the x elite things be at least that or better if they actually drop the votes properly before suspending
<SpieringsAE> nice!
<macc24> eh, rk3399 could be usable for some things
<SpieringsAE> I mean running firefox is a struggle for it
jhovold has quit [Ping timeout: 480 seconds]
<macc24> i used to have a gaming handheld with rk3399 and it worked surprisingly well
<SpieringsAE> even some terminals like kitty or alacritty run like doodoo
<SpieringsAE> foot seems to run well though
<macc24> huh
<SpieringsAE> It seems to me like the gpu is just very sad
<macc24> yeah
<JensGlathe[m]> x1e-78-100 seems quite decent
<SpieringsAE> would love an rk3588 in this thing
<SpieringsAE> I think that would be perfect for it
<JensGlathe[m]> hmm the dts for the rk3588 needs some love as it seems
<SpieringsAE> jeah, I believe it doesn't have 100% upstream support yet
<JensGlathe[m]> performance wise it's ~75% sc8280xp (cpu only)
<macc24> SpieringsAE: maybe you could replace your pinebook pro with a pinetab2, that thing has an rk3566/rk3568
<SpieringsAE> eh Ill have a qcom device again soon enough to do my ARM shenanigans
<SpieringsAE> kinda want that risc V frame work thing though when it comes out, sadly it has a doodoo chip aswell
<macc24> i would love to have a risc-v laptop/tablet to poke at
<SpieringsAE> Its sad how crap the application risc V chips seem to be so far
<SpieringsAE> very poor performance/watt aswell
<SpieringsAE> MCUs seem great though
<SpieringsAE> Guess I just have to be patient
<macc24> my daily driver laptop stays at a power cord(with battery charge limit at 65%) 95% of the time with its fans nearly always ramped up
<macc24> it can't be /that much/ worse than an x86 chip at power usage :D
<robclark> I something about a framework riscv laptop.. although googling for the SoC mentioned only found a couple posts about benchmarks which didn't compare all that favorably to rpi5, so maybe more of a toy than daily driver
<travmurav[m]> iirc it's something that should be compared to like 4x a53 xD
<SpieringsAE> jeah like pi3b level
<travmurav[m]> but also afaiu it's just a project/experiment of some other company and they just happen to use framework's mobo layout
<SpieringsAE> jeah, just wish they would use maybe that new spacemit M1 or something
<macc24> well there's that dc-roma ii laptop that could be faster?
<SpieringsAE> since it also actually has the version 1.0 of the vector extension
<SpieringsAE> yeah that one has the spacemit I believe
<SpieringsAE> It should definetly be faster
<SpieringsAE> 8 newer cores
<macc24> i've asked for a dev sample of that and they just refused :( that laptop sounds like a fun thing
<SpieringsAE> :(
<SpieringsAE> I guess all of them are dev samples lol
<SpieringsAE> didn't ubuntu actually support that dc roma laptop?
<SpieringsAE> I believe I read something like that
<macc24> i guess? not everyone can drop 450€ on a laptop that isn't strictly necessary to own
<SpieringsAE> yeah, the pinebook was a bit cheaper at like 220 euros or something, I expect that dc roma ii to be at its level
<SpieringsAE> maybe a bit better
<macc24> that was dc roma ii's price
<SpieringsAE> yeah I assumed so
hexdump01 has quit []
hexdump0815 has joined #aarch64-laptops
<hexdump0815> travmurav[m]: the two cpu/board screenshots above are from which socs? 8 core vs 10 core or 12 core? strange the the bigger chip then has one of the pmics not populated ...
<hexdump0815> travmurav[m]: also it looks like its the same board otherwise, so maybe the chips are pin compatible then?
<travmurav[m]> hexdump0815: afaik there are two silicones for x series: hamoa (12 core) and purwa (8 core)
<travmurav[m]> hexdump0815: the board is asus thing afaik
<travmurav[m]> hexdump0815: and hamoa can have two cores disabled to get "x plus 10 core"
<travmurav[m]> but purwa is "x plus 8 core"
<macc24> hexdump0815: there's no reason to *not* make them pin compatible
<travmurav[m]> so the big one is hamoa (10/12) and the small one is purwa (8)
<travmurav[m]> and yeah they must be pin compatible and I'd hope also software compatible
<travmurav[m]> hexdump0815: the pmic is missing on the purwa board, hamoa has all of them
<SpieringsAE> looks to me like the smaller x1p is the one with the unplaced pmic
<SpieringsAE> yeah
<hexdump0815> to me it looked the other way around: the smaller has 7 pmics and the bigger only 6 - but maybe i'm interpreting them wrong
<hexdump0815> the one with the 7 pmics has written x1p42 on it, which would be the 9 core version
<travmurav[m]> I used those two photos
<hexdump0815> 8 core
<SpieringsAE> jeah exactly, the smaller one
<SpieringsAE> I take that back
<SpieringsAE> its the one with the white looking chip that has 6
<SpieringsAE> if you switch between the two pictures you can see that is the x1p one
<macc24> travmurav[m]: huh what's that SW2100 visible on 2nd picture at wccftech website
<hexdump0815> ok - maybe better to not guess not too much from such pictures :)
<travmurav[m]> macc24: could you point me to the side of the photo a bit?
<travmurav[m]> ah 2nd
<travmurav[m]> macc24: my guess would be that it's a switch for the EDL mode
<travmurav[m]> ie to factory flash the software to the spi via usb
<travmurav[m]> it's impossible to know unlabeled things liek that if you don't have schematics sadly
<travmurav[m]> I mean without just trying to switch and see what happens xD
<SpieringsAE> *bricks firmware*
<travmurav[m]> not like you can flash it without a signed programmer blob
<SpieringsAE> jeah thats probably true
<SpieringsAE> might also have some write protection or something
<macc24> hey fun fact: did you know that poking at i2c buses can confuse the EC on slim 7x so much that you need to open it up and manually remove the battery to force a reset?
<travmurav[m]> yes
<SpieringsAE> lol
<travmurav[m]> some laptops have a nice reset hole for that ;P
<JensGlathe[m]> after you drilled it
<travmurav[m]> lol
<SpieringsAE> I have poked at a lot of busses on embedded linux devices, never had that happen hahaha
<travmurav[m]> but no, aspire1 has, x13s does, I think t14s too
<SpieringsAE> how is the t14s so expensive? Is it just thinkpad branding?
<SpieringsAE> I kinda want it, but jeez
<SpieringsAE> starts at 2k
<HdkR> SpieringsAE: Careful when Lenovo products aren't on "sale". They have anti-consumer behaviour that the products go on "sale" every other week or so and the prices become more realistic
<SpieringsAE> huh interesting
<SpieringsAE> also +490 dollars for a 1tb ssd
<SpieringsAE> when you configure it
<HdkR> Just buy a 2TB 2242 M.2 drive instead. Those are Apple prices for storage
<SpieringsAE> jeah thats what I thought, I mean I dont think even apple dares to ask almost 500 for 1 tb of storage
<SpieringsAE> wish you could order it without an m.2 drive in it at all, no windows license
<JensGlathe[m]> second hand? I got some luck with my hp x14
<SpieringsAE> I dont think there is going to be many second hand t14s for the foreseable future, spending that amount of money on something is likely to be done by someone who is sure of what they bought
<SpieringsAE> It also isn't at any of the generic retailers around here
<JensGlathe[m]> so only Linux fans, tough competition then
<SpieringsAE> got the price from the lenovo websit
<SpieringsAE> e
kettenis_ has joined #aarch64-laptops
kettenis has quit [Remote host closed the connection]
macc24 has quit [Quit: WeeChat 4.4.2]
juergh has joined #aarch64-laptops
juergh has quit [Quit: ZNC 1.8.2+deb2ubuntu0.1 - https://znc.in]
juergh has joined #aarch64-laptops
SpieringsAE has quit [Quit: Leaving]
crisma has quit [Ping timeout: 480 seconds]
<kuruczgy[m]> macc24: which EC register can mess it up so much?
<JensGlathe[m]> Is the only way to find out the panel model no. (if unknown) to open up the case?
<JensGlathe[m]> at least no more edp-panel splat in dmesg
<JensGlathe[m]> odd observation: my X14 has WCN6855, I'm loading the .41 firmware, I have yet to see the ring buffer errors here
<Segfault[m]> huh x13s suspend power draw is down to ~1.3W now, compared to about 2W when i tested a few months ago, neat
<Segfault[m]> still not as good as windows but that's a pretty big improvement
chrisl has quit [Ping timeout: 480 seconds]
<steev> re: pinebook pro, probably wouldn't help with battery life, but i have https://gitlab.com/kalilinux/build-scripts/kali-arm/-/commit/08ca486b73022a9915a35718b4734baa347fcab3 which adds a 2GHz for it (to enable you have to echo 1 to some file in sysfs that i can't recall at the moment)
<steev> well, 2GHz for the big cores, 1.5GHz for the little
<steev> 6.12 rc1 is out \o/ (and it looks like he fat fingered it and initially called it 5.12-rc1 :D )