00:33 <robclark> mmediouni[m], kuruczgy[m]: there was an mdt_loader issue.. https://lore.kernel.org/all/20250730-mdt-loader-shentsize-zero-v1-1-04f43186229c@oss.qualcomm.com/

00:34 <robclark> I guess if that hits zap fw, that would cause EL1 booting issues

00:35 <kuruczgy[m]> I am just looking at the linaro kernel branch, seems this is already included there. I guess I should also switch to that branch.

00:36 <robclark> IIRC the fix should be in -rc5? But yeah, linaro branch probably has some fixes that haven't landed in linus's tree yet

05:55 <travmurav[m]> deathmist: tbh I'm now getting more and more convinced that ALL ec firmware is fragile mess, just that on x86 we didn't mess with it as much so it was not going off from a beaten path to crash xD

05:56 <travmurav[m]> at least vendors that put a reset hole aren't blind to it xD

05:56 <travmurav[m]> definitely had to disassemble and yank battery from an x86 laptop with ec that went mad recently

05:59 <mmediouni[m]> travmurav[m]: A lot of the Snapdragon laptops use the exact same EC as a corresponding x86 machine

08:38 <deathmist> I don't remember having to deal with stuff like this on any prior x86 laptop though, did the EC just not do as much and things just worked because of that?

08:57 <travmurav[m]> deathmist: I'd guess on x86 linux and windows work similarly enough, i.e. with all the funny stuff inside of acpi dsdt that talks to ec, and on linux on arm the environment is perhaps different enough for it to diverge from whatever OEM tested on windows

09:43 <anthony25> I had a dell laptop stuck at 800MHz because of the Intel EC hitting a bug, and I had to disconnect the battery to reset it, so I can confirm

09:44 <anthony25> x86 dell laptop, I meant

09:50 <anthony25> But yeah on x1e I've hit multiple bugs, and one was particularly annoying, the laptop couldn't boot any distro at all, disk access seemed really slow and everything was timing out

09:51 <anthony25> Though the Lenovo recovery mode and just running the tests there seems to fix every problem I have

11:03 <Pengyu[m]> krzk: Do you mind sharing kernel config for sm8750 platform with me? I noticed you are working on this platform.

11:05 <krzk> Pengyu[m]: it's a defconfig

11:07 <krzk> Pengyu[m]: I am usually customizing it more, because of running sometimes without modules in initramfs, but it should not matter for normal case. If you need that customizations, then on my github/krzk/tools take it from linux/build.sh

11:07 <Pengyu[m]> Got it. Thanks

13:37 <janrinze> JensGlathe[m]: I managed to 'revive' the x13s with another SSD. That one was built using the manual on https://gitlab.com/TheOneWithTheBraid/x13s-firmware-update and I had the internal SSD moved to a USB enclosure. It actually tried to boot windows. (which tried to mount the internal SSD and failed)

13:47 <janrinze> JensGlathe[m]: But this is where it ends for now with windows. I need to setup the new internal SSD to boot Linux. So I will try to see if I can partition the new SSD to have UEFI , boot and rootfs partitions. However the Debian iso does not boot..

14:36 <JensGlathe[m]> sounds like a re-swap

14:53 <JensGlathe[m]> I still have the recovery USB for X13s on GoogleDrive: https://drive.google.com/drive/folders/1Rjdo1gisOxvQXqBrOmba2aK4oBATeJFQ?usp=sharing

14:59 <Treibholz[m]> Uhh, found another (minor) disadvantage of a 16k kernel: every file in a tmpfs now also needs at least 16k, so a Maildir might need more space in memory, then on disk. This might not hurt on a powerful notebook, but I was irritated, that I couldn't copy a 2.8GB Maildir on a 4GB tmpfs on a Raspberry Pi 5.

15:00 <Treibholz[m]> Still no reason, to switch back to 4k :-)

15:09 <janrinze> Treibholz[m]: I think that 16k pages is like a server thing and big desktops.. Those run apps with huge memory requirements and reducing MMU tables by 4x should give it some performance boost. Most apps won't see much benefit. Perhaps Chrome is faster since that is very memory hungry :-D

15:10 <Treibholz[m]> janrinze: all my measures on my X13s have shown 5-10% more performance (even glmark!). And everything "feels" faster.

15:11 <Treibholz[m]> I have 32GiB RAM and I want them used! :-)

15:13 <janrinze> Treibholz[m]: 5-10% is nice. I have 32GB too. When I have linux back working on it I will run a few tests :-D

15:40 <robclark> fwiw, I think the servers are going for 64kb pages (although android seems interested in 16kb)

15:41 <robclark> (16kb page size is optional for aarch64, 4kb and 64kb are mandatory.. x1 doesn't support 16kb, idk about newer oryon cores)

15:55 <janrinze> JensGlathe[m]: the original SSD is not working properly. In the USB enclosure the bandwidth is dropping to 1MB/s or less.. Either a coincidence or the actual cause of my troubles in the first place..

16:50 <JensGlathe[m]> What a coincidence

16:53 <mmediouni[m]> <robclark> "fwiw, I think the servers are..." <- not a rule :) 64KB is also not mandatory in the AArch6r spec and notably not implemented on modern Apple cores

16:54 <mmediouni[m]> Apple also shipped quite some gens without 4KB page support on mobile but anybody on PC isn't gonna want to do that

17:13 <janrinze> JensGlathe[m]: after install of Debian 13 (using net install iso from USB) we still require a work-around for loading the dtb by adding a line in grub.

17:14 <janrinze> JensGlathe[m]: I finally have a linux desktop again!

18:00 <robclark> mmediouni[m]: apple is probably not the thing you should point to when talking about spec compliance.. they kinda do their own thing ;-)

18:00 <robclark> 4kb and 64kb are required.. others are implementation defined.. at least for "A" cores.. https://developer.arm.com/documentation/101811/0104/Translation-granule

18:01 <mmediouni[m]> robclark: The newer arm arms don't mandate any granule size since quite a while now

18:01 <Treibholz[m]> janrinze: Just copy /boot/dtb to /boot/efi/sc8280xp-lenovo-thinkpad-x13s.dtb, then you don't need the dtb-line

18:01 <mmediouni[m]> See that doc page: says > The granule sizes that a processor supports are IMPLEMENTATION DEFINED and are reported by ID_AA64MMFR0_EL1.

18:01 <mmediouni[m]> nowadays

18:03 <mmediouni[m]> The SBBR mandates support for the page sizes you'd expect though - but a vendor doesn't care about SBSA they can just ignore that

18:04 <robclark> I'm going by "All Arm Cortex-A processors support 4KB and 64KB".. maybe that is superseded and/or requirement moved into a different spec.. maz could possibly correct me

18:15 <mmediouni[m]> <robclark> "I'm going by "All Arm Cortex-A..." <- Cortex-A just means Arm's own cores :) as a reference point

18:20 <robclark> I could be wrong about which combination of specs... but pretty sure the net result is that 4kb and 64kb are required, others are optional

18:33 <mmediouni[m]> <robclark> "I could be wrong about which..." <- CBSA says: https://developer.arm.com/documentation/den0151/a/?lang=en... (full message at <https://matrix.org/oftc/media/v1/media/download/Abu2vh52zmKbkI6Gq2jkrm_0hVorMKucnIyy0MdbRCvuzcbqJlA0cqIIE4lAg2j0Ke1tI_gBRxzzKmkkBjfh84NCeZbY0eZQAG1hdHJpeC5vcmcvTHRSZWZSSmRzSW9kSHlJWkpzRWNzc0pE>)

18:44 <robclark> ok, well that still leaves 4kb required, 64kb essentially required, 16kb/etc optional ;-)

20:05 <janrinze> Treibholz[m]: unfortunately when i add /boot/efi/sc8280xp-lenovo-thinkpad-x13s.dtb and remove the dtb setting from grub it won't boot anymore..

20:06 <JensGlathe[m]> Hmm do you have the Linux option (experimental) in the BIOS enabled?

20:06 <janrinze> JensGlathe[m]: indeed, that option is set.

20:06 <JensGlathe[m]> This used to work, you put the dtb file into the EFI root

20:08 <janrinze> JensGlathe[m]: I'm not going to speculate on things, it just does not work on this x13s.. never has.. unfortunately.

20:08 <JensGlathe[m]> I tested it on X13s, I have one

20:09 <janrinze> JensGlathe[m]: maybe my X13s is having an attitude and thinks differently.. :-D

20:10 <JensGlathe[m]> These things happen

20:10 <janrinze> JensGlathe[m]: This is a new debian 13 install using the installer iso on a clean SSD..

20:11 <janrinze> JensGlathe[m]: The nvme0n1 is formatted and installed using the debian 13 iso net installer.

20:34 <janrinze> JensGlathe[m]: apparently I installed 'forky' .. Hmm.. silly me.

21:14 <deathmist> SpieringsAE: I dunno if you read scrollback without being connected but with https://github.com/SpieringsAE/linux/tree/wip/x1e80100-6.17-rc1 (including rebase to rc4) on X1E Vivobook I finally got HDMI to work, but only properly under X11 (KDE Plasma), the Wayland session freaks out and crashes within 20 seconds or so, just wondering if you've seen anything like that

22:19 <steev> The T14s has been infuriating today. It feels like it can never properly exit efi services when it’s on battery power

22:37 <programmerin-wonderland> Sorry for being super inactive last 2 days have midterms coming up. Plan to resume like after wednesday, didn't mean to ghost anyone who's pinged me

22:41 <Jasper[m]> @[programmerin-wonderland] how could you :'(

22:41 <Jasper[m]> No worries though, midterms are more important than wacky laptops

22:41 <Jasper[m]> Well I lie, for some it's a job

22:46 <programmerin-wonderland> honestly I would love to do this for work but I barely (well, I'm starting to now) understand devicetrees

22:47 <programmerin-wonderland> but it would be awesome to one day work at qualcomm or something like that

22:48 <programmerin-wonderland> but thanks! I hope to log back in thursday with maybe good news about bluetooth depending on how dedicated I am about getting it working LOL