02:08 <dgilmore> robclark: the delay is blocking loading in the initramfs

02:54 <robclark> dgilmore: I kinda suspect there are 1000 paths to fail when it comes to probe-defer, given the # of various dependencies... in x86 world it is all just acpi and you can take that for granted

02:55 <robclark> (well, maybe not 1000 but a lot)

02:57 <dgilmore> robclark: it's certainly not as straight forward as it should be

02:58 <robclark> the arm approach of having a bunch of small drivers for a particular part of the SoC.. while cleaner, it does make things more complicated

03:00 <robclark> it would have been easier if MS forced the ACPI route on windows laptops instead of the PEP thing.. maybe not better in the long run, but easier in the short term

03:00 <dgilmore> It would be

06:26 <owc[m]> <push> "owc: i've been trying to get a..." <- Yes, same for me. No luck w/ catching anything in the logs.

06:27 <owc[m]> <push> "owc: Im confident is saying..." <- Interesting. I'm still on GNOME. So maybe the hw rendering then…

06:29 <owc[m]> <push> "owc: i've been trying to get a..." <- Kinda same: 1.57, though I totally wiped windows, installed concept and upgraded to mantic.

08:45 <danielt> Can't believe I've had one for so long and am only asking this now... but is there a commonly used trick to get the MAC address of the X13s WiFi stable?

11:27 <Bioxvirizm-x13s[m]> <danielt> "Can't believe I've had one for..." <- +1

13:23 <mkrawczuk> Hi!

13:25 <mkrawczuk> I'm trying to launch the mainline kernel on a Samsung Galaxy Book S SM-W767. I noticed that many of the features are marked as working on postmarketOS. I built and launched the kernel from https://gitlab.com/jenneron/linux/-/tree/galaxy-book-s-6.1.2 on Armbian, but many components like wifi are not working. Is there any kernel or distro image that has better support for my device?

13:27 <jenneron[m]> mkrawczuk: https://gitlab.com/sc8180x-mainline/linux and https://gitlab.com/postmarketOS/pmaports/-/blob/master/device/testing/linux-postmarketos-qcom-sc8180x/config-postmarketos-qcom-sc8180x.aarch64

13:27 <jenneron[m]> you can build distro image with pmbootstrap

13:28 <mkrawczuk> thx

14:17 <arnd> https://semiaccurate.com/2023/09/26/whats-going-on-with-qualcomms-oryon-soc/

14:25 <Jasper[m]> <arnd> "https://semiaccurate.com/2023/09..."; <- Oh no

14:26 <HdkR> It makes me think of an autoparts store commercial jingle. Hope it comes out some day :D

14:30 <Jasper[m]> Late 2024 I've heard

14:40 <robclark> tbf, s/a never had much love for qc

14:41 <clover[m]> <danielt> "Can't believe I've had one for..." <- https://github.com/ironrobin/archiso-x13s/wiki/Feature-Support#bluetooth

14:41 <clover[m]> systemd override is what most are doing these days

14:55 <push> owc[m]: try installing cinnamon and lightdm, then remove geoclue and all the gnome crap that depends on it, and make sure your linux-firmware is up-to-date

14:56 <push> owc[m]: thats what ive done in the last few couple days and it hasn't reset once, unfortunately I don't know what I actually did that fixed it

14:57 <clover[m]> oh oops you are talking about WiFi

14:58 <clover[m]> idk about that one

14:59 <push> probably just use nmcli to set the hwaddr right?

15:00 <danielt> clover: No worries at all... I'm not currently using such an elegant approach for the BT MAC either so I shall probably switch over.

15:01 <Jasper[m]> <robclark> "tbf, s/a never had much love for..." <- Hearing some noise that it's mostly bs anyways. I'm not going to bother taking it seriously.

15:04 <jhovold> danielt: i use a udev rule to set the mac address using ip

15:05 <jhovold> ACTION=="add", SUBSYSTEM=="net", KERNELS=="0006:01:00.0", RUN+="/usr/bin/ip link set dev $name address xx:xx:xx:xx:xx:xx

15:06 <jhovold> Bioxvirizm-x13s[m]: ^

15:07 <danielt> jhovold: Sounds good. Thanks!

15:09 <jhovold> dgilmore: i've posted a fix for the touchpad regression in 6.6-rc1 here:

15:09 <jhovold> https://lore.kernel.org/lkml/20230918125851.310-1-johan+linaro@kernel.org/

15:10 <dgilmore> jhovold: excellent thanks

15:11 <jhovold> you can always check out my wip braches if you run into issues like this, i don't push new ones if there are known regressions like that:

15:11 <jhovold> https://github.com/jhovold/linux/tree/wip/sc8280xp-v6.6-rc2

15:18 <jhovold> ...until i have a fix or workaround i mean

15:26 <ukleinek> the right place to set the macaddress is the dtb, but that has its own problems.

15:32 <travmurav[m]> cooperative firmware should be fine with patching the dtb with board specific details like wifi/bt mac upon boot

15:33 <jhovold> ukleinek: sure, but we don't control the fw

15:34 <jhovold> we'll probably be able to read out the addresses at some point and set it from the drivers

15:34 <jhovold> just need qualcomm to tell us how... or spend time figuring it out ourselves

15:35 <travmurav[m]> pretty sure lenovo ships some kind of dtb loader nowdays in their efi? given they seem cooperative and as a board vendor they are the ones assigning the macs, it sounds sad if they can't make firmware pass it...

15:36 <travmurav[m]> or did qcom invent something new to hide the macs now?

15:37 <jhovold> the fw can load a dtb, but that's it currently

15:38 <jhovold> still waiting on details so we can determine how to move forward wth this

15:39 <travmurav[m]> well I still think it should be trivial to add a couple of libfdt calls on the loaded dt to insert macs from the firmware so if lenovo wants they should be able to implement it, though I have no idea what's really behind the x13s effort and what bureaucracy/qcom firmware mess may intervene...

15:41 <jhovold> should be, but as you know this is a woa device and we're retrofitting linux support

15:41 <jhovold> so things aren't always as easy they may seem

15:41 <jhovold> as*

15:42 <steev> truth

15:44 <travmurav[m]> Well I get it, while considering something like aspire1 I'm working on, if anyone is ever to implement efi driver with sku detection/mac provisioning it would probably be me...

15:44 <travmurav[m]> but not sure if I got wrong impression from lenovo adding that "linux mode" switch

16:44 <konradybcio> travmurav the issue with that approach is that only lenovo machines will be fixed

16:44 <konradybcio> setting mac address from userspace is "normal" and that's why it's passed in android cmdline

16:45 <konradybcio> userspace.. uh i meant HLOS

16:45 <travmurav[m]> on the other hand every vendor has their own way of storing the macs I'd imagine

16:46 <travmurav[m]> so if you are going to implement it in the kernel, it will be a total mess

16:51 <konradybcio> my educated guesstimate is that it's in SMBIOS

16:51 <konradybcio> saying that without checking though

16:52 <travmurav[m]> konradybcio on aspire1 it's in EC and you need to tickle it properly to get the bytes out

16:52 <travmurav[m]> so I really expect it to be a total mess across vendors

16:53 <travmurav[m]> so IMO if it's possible for the bootloader to provision the mac via DT, it's reasonable to target that for a vendor that cooperates