01:06 <steev> and rc3 is tagged

01:07 <HdkR> Nice

01:09 <HdkR> Is there a location where the wip arm64 vdso_getrandom implementation is at? Wanting to poke at it, but I can't find where the branch lives

01:09 <steev> i have no idea what that is

01:10 <steev> https://www.phoronix.com/news/Linus-Torvalds-No-Random-vDSO did it actually make it in to 6.11?

01:11 <HdkR> It's in rc2 at least

01:12 <HdkR> Looks like it is in rc3, just checked

01:13 <HdkR> https://www.phoronix.com/news/Linux-6.11-Lands-getrandom-vDSO for the newer article

01:13 <steev> Also, though it's not ready for this pull, somebody has begun an arm64 implementation already

01:14 <steev> guess you have to wait

01:15 <HdkR> Yea, I'm prepping the code for FEX to support it, so hoping to get it in place as early as possible

02:27 <powpingdone> Hey, I'm struggling to get a gentoo install up and running on a Lenovo Yoga Slim 7x. I have tried the image at jglathe/linux_ms_dev_kit and it does work (after disabling the kmod loading in 80-drivers.rules) but I can't get a custom gentoo mincd (basically just a normal image with 6.11-rc2) to boot. Am I missing something? It seems to get stuck at udev activating devices.

05:51 <steev> bryanodonoghue: here, cheese still likes to not open the camera sometimes (after I've opened it a few times, i THINK i saw a patch to address that, not sure if i've pulled it in) - but the other thing is that it will sometimes stutter quite a bit and claim that there may be a timestamping problem, or this computer is too slow.

05:52 <steev> this is on 6.11 rc3 which is my rc2 branch just rebased on top of rc3

05:53 <steev> debian testing(ish) so it's cheese 44.1-1+b4, and gnomeshell 46.3.1-2

09:17 <Molyuu[m]> After I apply this patch to vanilla kernel (6.10) to load GPU firmware (https://gitlab.com/sc8180x-mainline/linux/-/commit/b4e971f0c6a4b27c21e7359f92a77cb84b331141), when it boots to run init process, screen comes white, and my device (Xiaomi Book S 12.4, SC8180XP) won't boot without mdss enabled anymore (That means won't boot with simple-framebuffer. Pressing Capslock on Keyboard, the lock led will light after a long time (about

09:17 <Molyuu[m]> 5s-10s)), and if I boot with mdss enabled, screen still comes while after run init process, but I can connect to device via SSH through UNDIS Network, dmesg shows that mdss initialize timeout. (If i directly boots with sc8180x-mainline kernel, it will panic with NULL pointer error when initialize clock here instead of timeout error)

09:18 <Molyuu[m]> Do I need some additional kernel patch to fix this?

10:20 <Molyuu[m]> * After I apply this patch to vanilla kernel (6.10) to enable multiports (https://lore.kernel.org/lkml/20240730-sc8180x-usb-mp-v2-2-a7dc4265b553@quicinc.com/T/), when it boots to run init process, screen comes white, and my device (Xiaomi Book S 12.4, SC8180XP) won't boot without mdss enabled anymore (That means won't boot with simple-framebuffer. Pressing Capslock on Keyboard, the lock led will light after a long time (about

10:20 <Molyuu[m]> 5s-10s)), and if I boot with mdss enabled, screen still comes while after run init process, but I can connect to device via SSH through UNDIS Network, dmesg shows that mdss initialize timeout. (If i directly boots with sc8180x-mainline kernel, it will panic with NULL pointer error when initialize clock here instead of timeout error)

12:44 <JensGlathe[m]> ouch. Yesterday was a streak of bisect bad, now its good - I guess the bug is /was in my stack of patches

14:19 <steev> sounds like you were right :D

15:01 <JensGlathe[m]> huh

15:02 <JensGlathe[m]> now bisect is in my stack of patches

17:15 <HdkR> I just noticed that the X1E-78-100 is only running the GPU at 1.1Ghz when it is rated for 1.25Ghz. What's up with that?

17:18 <robclark> HdkR: I think missing speedbin info in dts

17:18 <HdkR> bah

17:18 <robclark> presumably that will get posted at some point

17:18 <HdkR> Are teh higher end Socs not running at 1.5Ghz then?

17:19 <HdkR> Ah yep, I see the speed bns are entirely defined in the base table and cap out at 1.1Ghz

17:19 <HdkR> So every device will have limited perf

17:21 <robclark> yeah, I think no one has added the higher opp's and whatever plumbing to figure out supported hw

17:22 <robclark> travmurav[m]: btw, side note... IIRC there is an option for OLED panel on t14s.. which probably means different dtb for OLED vs non-OLED (since so far all the OLED panels seem to want samsung panel driver instead of panel-edp)

17:22 <robclark> hopefully different DMI strings for those?

17:24 <travmurav[m]> robclark: yeah, my dtbloader was kinda designed around the idea that one may need to run code after loading the dts so as long as there is a way to detect, we can pick dtb/set compatible in a node/apply overlay

17:24 <travmurav[m]> i.e. there is already a precedent upstream with the mezannine boards for dragons that one can generate overlay-enabled dtb + overlays + combined dtb+overlay

17:25 <travmurav[m]> many options xD

17:26 <travmurav[m]> but well, someone "in the field" (with the devices) would have to dump DMI/acpi/whatever to figure out how it detects the panel (if at all) and then we can implement the detection

17:32 <FarchordSteveCossette[m]> Alright guys, I'm back from vacation.... anything improved in the span of a week? XD

17:33 <travmurav[m]> I wonder if they end up reserving ~400MiB just because

17:33 <FarchordSteveCossette[m]> aka "Which kernel are we using now?" XD

17:33 <travmurav[m]> ... even on modern x elite things

17:53 <JosDehaes[m]> <FarchordSteveCossette[m]> "aka "Which kernel are we using..." <- Not much happened, same 6.11 rc1

18:12 <nirik> I'm back from flock now too... I did see a link to https://lkml.org/lkml/2024/8/9/33 go by, that might be related to the reset on boot issue I (and several others) were having? but I don't know if anyone confirmed that. :) If anyone wants to make me a new rc3 kernel rpm happy to try later today...

18:13 <JosDehaes[m]> I don't think Jens Glathe already posted updated ubuntu image to test. My system never rebooted when booted in Arch linux, but his ubuntu image did.

18:14 <JensGlathe[m]> no I didn't

18:27 <FarchordSteveCossette[m]> <nirik> "I'm back from flock now too... I..." <- I left my laptop in a non-working state and now I don't remember what to do XD meeeh.

18:30 <nirik> I'm not sure how much is still not been merged for rc3... I can easily test a rawhide rc3 kernel and see where it is...

18:34 <robclark> nirik: the pinctl patch didn't help me, I had to go back to the commented udev rule. Maybe konradybcio could confirm but it looks like that patch only matters for devices with UFS instead of nvme? (Are there any such devices?)

18:48 <nirik> ok, fair. I was getting the reset with the udev rule commented tho (on the rootfs, but present in the initramfs). Dunno. I guess I can try a rc3 and see where it's at later today.

18:50 <bamse> travmurav[m]: looking at the x13s acpi, it would seen that peeking at some registers in the EC as part of the DT selection (or update) mechanism would be appropriate as well...

19:11 <anonymix007[m]> Was X1P42100 even announced? Lenovo has (not in stock though) system boards with it listed here: https://pcsupport.lenovo.com/us/en/products/laptops-and-netbooks/thinkpad-t-series-laptops/thinkpad-t14s-gen-6-type-21n1-21n2/21n1/parts/display/compatible

19:11 <FarchordSteveCossette[m]> <nirik> "ok, fair. I was getting the..." <- I wonder if Fedora enabled the snapdragon flags in the aarch64 kernel yet.

19:14 <anonymix007[m]> BTW, what are the chances that another panel will work in T14s? I've found a few interesting options which should be much better than the stock IPS panel and not OLED

19:14 <nirik> Dunno. I am not sure what needs to be enabled... I think most of it already is.

19:14 <FarchordSteveCossette[m]> nirik: https://docs.google.com/spreadsheets/d/1cmOt9ZH3FLFoCc_dTUbFUuzrX4zyXFgK3lBArgk-I-k/edit?gid=773821052#gid=773821052

19:15 <FarchordSteveCossette[m]> I can really only spot 2 that seem to be relevant, but then again I'm not a kernel engineer XD

19:15 <FarchordSteveCossette[m]> I'm gonna reflash my drive with the latest rc3 kernel and see

19:19 <kettenis_> travmurav[m]: I wonder if for X1P you need to look at the ACPI MADT table and knock out the disable processors in the dtb

19:46 <agl_> E

19:46 <JensGlathe[m]> yeah

19:57 <JensGlathe[m]> They say the most entertaining outcome is the most likely

19:58 <JensGlathe[m]> well after bisecting to the very last commit bisect stated that the initially stated bad commit is the bad commit

19:58 <JensGlathe[m]> but its just a config change

19:58 <steev> what config change is it

19:59 <JensGlathe[m]> argh

19:59 <JensGlathe[m]> in a sec

20:04 <FarchordSteveCossette[m]> nirik: yeah, so I downloaded the latest rawhide compose (I know that's not RC3 yet) and it boots. Well, it crashes because of the udev rule, but I do get past initramfs

20:08 <nirik> Farchord (Steve Cossette): cool. Will try here hopefully later today.

20:10 <FarchordSteveCossette[m]> yeah I get to sddm, but I got no keyboard or mouse XD

20:10 <FarchordSteveCossette[m]> ah crap I forgot to backup my list of modprobes

20:12 <robclark> evdev is probably one you'd want.. I was getting no kb/trackpad without it

20:13 <FarchordSteveCossette[m]> yeah right now ill try to simply blacklist qcom_q6v5_pas

20:13 <FarchordSteveCossette[m]> or at least that's what was causing me issues 2 weeks ago

20:14 <robclark> usb boot? oddly I hadn't been having problems with usb boot, but I also don't have dsp fw in initrd

20:15 <JensGlathe[m]> -CONFIG_CMA=y

20:21 <FarchordSteveCossette[m]> <robclark> "usb boot? oddly I hadn't been..." <- Yup usb boot

20:31 <steev> cma is causing the 4k display issues?

20:35 <JensGlathe[m]> not sure yet, was the most striking diff. And no, it's not in the current (6.11rc) devkit_defconfig either

20:41 <steev> it is (afaik) needed for the webcam? not sure entirely, johan added it to his defconfig while also increasing the cma size to 128MB (the default is 32)

20:41 <steev> the defconfig should have CMA_SIZE_MBYTES=32 which would automatically select it

20:43 <steev> bryanodonoghue: do we *need* 128, or could we get by with like, 64 (arm multiconfig default and qcom arm32 default)

20:44 <JensGlathe[m]> well it def works with the HEAD^ variant of devkit_defconfig. Will isolate further, but tomorrow. Looks like CONFIG_CMA to me, at least it has lots of dependencies to DRM

21:33 <nirik> So, rawhide image from today... I managed to get to sddm (no keyboard/touchpad, but it did get there). :) Will poke at it some more and perhaps write up a quick list of what I changed...

23:24 <steev> bamse: i've applied your usb mp patches (finally found them... you and your email addresses :P ) but for some reason, the Apple, Inc. Digital AV Multiport Adapter doesn't seem to show up (the pinebook pro usb-c hub does show up though)

23:25 <steev> i tried both ports and both orientations (just in case)