04:34 <Molyuu[m]> <Molyuu[m]> "Anyone can help about my DSI..." <- I'm tried to digging into how qualcomm firmware generates pps cmd, it seems it is same as the one that generated by Linux kernel. As the init seq is taken from DSDT, I think it is correct.

04:41 <Molyuu[m]> The rest one I haven't check is the pixel clock. I found that firmware automatically divide the freq by compression ratio. And that's seems processed by msm driver in kernel.

06:21 <craftyguy> has anyone seen usb devices not show up at all (nothing in kernel log, nothing in lsusb) on the usb-c port closest to the hinge on the x13s? tried several usb2 and usb3 things :(

06:22 <abby> for me hotplug doesn't always work in either port

06:22 <craftyguy> hotplugging usb works in the other port for me

06:23 <craftyguy> I ran the EFi diag app that comes on the x13s, and it correctly ID'd a usb storage device was connected, so everything seems fine in EFI(?)

06:26 <JensGlathe[m]> <craftyguy> "for those with an x13s running a..." <- I have a lenovo dock, seems to work okay. Need to give it a new round. But when on external display, I use the builtin dock of the display (Iiyama QSN2792 or sth). Works, but "only" gives USB-2 side channel to the hub. Soes the same on Windows, though, it assigns 4 lanes to the DP.

06:39 <steev> craftyguy: works for me here - this is using the pinephone usb-c dock https://www.irccloud.com/pastebin/hYGsNGew/

11:25 <kuruczgy[m]> What's the status of X1E mesa support? Is 24.2.0 without any patches enough?

11:25 <kuruczgy[m]> (I also see that the jhovold kernel has the chipid override, I assume that's still necessary?)

12:24 <JosDehaes[m]> You can revert the chip id override if you use 24.2

12:24 <JosDehaes[m]> Works perfectly, 10 000 in glmark2

12:29 <konradybcio> glmark2 is very unstable

12:29 <konradybcio> inconsistent is the word i was looking for

12:37 <JosDehaes[m]> It varies a bit yes, I just mean it works quite well

12:39 <kuruczgy[m]> Thanks, I will try 24.2 then!

12:39 <kuruczgy[m]> Btw do we have some sort of community wiki where things like this could be documented? (And also updated quickly since things are still changing rapidly.)

12:45 <konradybcio> Jos Dehaes: so far my golden shot is running minecraft.. it's java so it works fine

12:45 <konradybcio> i use prism launcher (oss alternative to the official electron-based one)

12:45 <JosDehaes[m]> I never played that, too old 😅

12:46 <JosDehaes[m]> I mean me😁

12:47 <mlau> its really fun for a few days

12:47 <mlau> then the novelty wears off, at least for me

13:02 <mlau> what is the correct ath12k firmware for the yoga slim 7?

13:02 <mlau> I cannot get ath12k to work

13:02 <konradybcio> well that was a solid few years for me and peers :p

13:03 <mlau> driver says "failed to receive wmi unified ready event"

13:03 <mlau> failed to start core

13:04 <mlau> fw_version 0x100301e1 2023-12-06 WLAN.HMT.1.0.c5-00481-QCAHMTSWPL_V1.0_V2.0_SILICONZ-3

13:05 <mlau> chip_id 0x2 chip_family 0x4 board_id 0xff soc_id 0x40170200

13:12 <mlau> I took the ath12k phy and main firmware from the windows driver, it's newer (fw_version 0x110b0010) but doesn't work either

13:13 <mlau> also fails with "qmi failed to set mode request, mode: 4 err = -110"

13:30 <JosDehaes[m]> I use the firmware from the repo I sent you

13:32 <mlau> hm ath12k is really wonky on with my kernels. It generates pci aer messages

13:32 <mlau> I now had a 5x bootloop, it rebooted hard every time ath12k initialized

13:32 <mlau> JosDehaes: they are the same files, just different names

14:25 <kuruczgy[m]> Updated to mesa 24.2.0, but it's still not working for me.

14:25 <kuruczgy[m]> both kmscube and vulkaninfo only see llvmpipe (kmscube does render with it though)

14:25 <kuruczgy[m]> msm module is loaded, /dev/dri/card0 is present, zap shader should be in the right place

14:25 <kuruczgy[m]> (I still have udev disabled, but I hope it doesn't affect graphics?)

14:25 <kuruczgy[m]> Not sure what could be going wrong, are there any extra steps I am missing?

14:34 <JosDehaes[m]> Maybe you still have the env variable set to zink?

14:36 <kuruczgy[m]> I tried both with and without MESA_LOADER_DRIVER_OVERRIDE=zink, not sure what it should be set to

14:38 <JosDehaes[m]> Should not set it at all

14:39 <JosDehaes[m]> How did you build mesa? Did you include freedreno gallium driver?

14:44 <kuruczgy[m]> Cherry picked this PR: https://github.com/NixOS/nixpkgs/pull/332413

14:44 <kuruczgy[m]> And `/run/opengl-driver/lib/libvulkan_freedreno.so` does exist. But this is the only freedreno file I can find, is there supposed to also be a separate opengl driver?

14:45 <JosDehaes[m]> You don't need any other patch on top of 24.2

14:45 <JosDehaes[m]> You just need to configure to build freedreno

14:46 <kuruczgy[m]> JosDehaes[m]: The patch is for nixpkgs to build mesa 24.2, not for mesa itself. (I am using NixOS just to clarify.)

14:47 <JosDehaes[m]> Don't know anything about nix, but you need to make sure it builds freedreno

14:47 <kuruczgy[m]> Ok, thanks, I will double check

14:48 <JosDehaes[m]> Add the freedreno to the list of gallium drivers

14:48 <JosDehaes[m]> You can remove any Intel or Radeon drivers

15:30 <kuruczgy[m]> Is a freedreno_dri.so supposed to exist? Because I can confirm that -Dgallium-drivers is getting passed freedreno, but I only see the vulkan driver being built.

15:31 <kuruczgy[m]> Hm wait I think just the name is different, I do have msm_dri.so which I suspect is the freedreno driver

15:32 <HdkR> Indeed

15:37 <kuruczgy[m]> With VK_LOADER_DEBUG=error,warn,info I see a bunch of "Removing driver [...]/libvulkan_*.so due to not having any physical devices" messages, this seems like it could be the issue...

20:36 <steev> the vulkan driver is libvulkan_freedreno

20:37 <steev> freedreno's vulkan driver is called turnip (idk why)

20:37 <HdkR> root vegetables are very popular these days

21:01 <kuruczgy[m]> My suspicion right now is that the msm driver is not actually getting loaded, and /dev/dri/card0 belongs to a thing called simpledrm. (Which is strange because the msm module is nevertheless loaded.)

21:01 <kuruczgy[m]> Also it seems that /dev/dri/renderD* is missing, which AFAICT is a hard requirement for Vulkan.

21:02 <kuruczgy[m]> The only thing that comes up when grepping the kernel log for msm is msm-mdss failing to probe. If this is a dependency of the msm drm driver I guess it could explain things though...

21:03 <HdkR> Double check if you have msm drm enabled in the kernel? :D

21:04 <kuruczgy[m]> What's the name of the kernel module, and of the CONFIG_ option? It's a bit hard to keep track of the at least three (msm, freedreno, turnip) names this driver stack has.

21:04 <kuruczgy[m]> The msm module is loaded, are there also other ones that should be loaded?

21:04 <HdkR> DRM_MSM is the kernel option

21:05 <HdkR> Maybe DRM_MSM_DPU as well is necessary?

21:08 <kuruczgy[m]> I have CONFIG_DRM_MSM=m and CONFIG_DRM_MSM_DPU=y too

21:10 <HdkR> Smells like something still wrong in the kernel if you don't have msm_dpu talking about firmwares getting loaded or something. msm-mdss message sounds related

21:13 <kuruczgy[m]> Ok, then I will try to look into the msm-mdss error next. It says "deferred probe timeout, ignoring dependency", then it fails to probe with error code -110

21:14 <HdkR> I don't get that message on my Yoga, only msm-mdss message is about adding to an iommu group

21:38 <akaWolf> hi. I've just tried latest kernel in ironrobin repo (steev's 6.2.10 branch) and it doesnt see my drive: hanging up at initrd with unable to mount root

21:38 <akaWolf> what can be the problem?

21:40 <kuruczgy[m]> Is the drive you are trying to mount usb or nvme? You need a few select kernel modules in the initrd for both I think. (For usb I have phy_qcom_qmp_combo, phy_qcom_snps_eusb2, phy_qcom_eusb2_repeater, tcsrcc_x1e80100 in my initrd on my yoga.)

21:41 <akaWolf> nvme

21:47 <akaWolf> with other kernel the same config of the system and initrd is working

23:46 <kuruczgy[m]> The problematic driver seems to be dispcc-x1e80100. It not being loaded seems to be the blocker in the driver dependency graph. But if I do modprobe dispcc-x1e80100 manually, the screen goes black.

23:47 <kuruczgy[m]> So probably this is the same driver I am avoiding loading by disabling udev, so I guess my two problems have a common root cause, I guess I should be happy about that...