01:55 <shawnguo> ardb: the efi fix on -next doesn't help flex5g poweroff. so indeed the resetsystem is broken on flex5g (and x13s) in different way from spx.

01:57 <shawnguo> actually the flex5g behavior seems different from spx. reboot seems fine, but poweroff is broken and ended up with a machine reboot without crash dump.

03:02 <bamse> shawnguo: but isn't crash dump ju bypassed, as it's a production unit? so a crash will look just like a normal reboot

03:41 <shawnguo> bamse: ah, possibly, but qzed has reported crash dump on spx with resetsystem, so there is a difference between Lenovo and MSFT devices.

03:46 <bamse> shawnguo: interesting

04:05 <jenneron[m]> steev: uhh, yeah, that was wlanmdsp.mbn

04:05 <jenneron[m]> thank you

04:05 <jenneron[m]> next is modem and trying to reproduce everything on galaxy book s with alpine

05:00 <jenneron[m]> bamse: flex 5g doesn't have anything like this? https://github.com/andersson/kernel/commit/da829268ad3857e0175722e2bf67cb992c26b787#diff-3d46e4f0129a7c33647235580664bebb5aeaf859fda43844e1c8f494bf582a5dR78

05:10 <bamse> jenneron[m]: there should be some pcie bus, for the 5g...but iirc not for wifi

05:36 <jenneron[m]> bamse: I mean regulator for modem

05:36 <jenneron[m]> afaiu I need pcie and mhi, pcie works, but mhi doesn't get probed

09:07 <shawnguo> qzed: I appreciate your great work on efivars/qseecom support! I tested it on c630/sdm850, flex5g/sc8180x and devkit2023/sc8280xp using acpi (with a couple of changes). it just worked! With the play with acpi, I would discuss a couple of things with you.

09:09 <shawnguo> 1. QCOM_QSEECOM driver is calling devm_mfd_add_devices(), MFD_CORE might need to be selected in Kconfig, otherwise we may have build issue.

09:11 <shawnguo> 2. Is platform specific qseecom_data really needed? as far as I can see (and tested), the driver is generic for SDM850, SC8180X and SC8280XP.

09:21 <shawnguo> 3. As there is no such qseecom device in acpi dsdt table, I'm wondering if we can register a platform device for qseecom from scm driver, and save such "virtual" device from device tree. In that case, we will have qseecom device created/registered in the same way for both acpi and dt. Also as scm will be the parent device of qseecom, the scm

09:21 <shawnguo> lookup in qseecom_setup_scm_link() will be simplified?

09:47 <qzed> Sure

09:47 <qzed> 1) right, forgot about that...

09:48 <qzed> 2) people don't wanted a generic compatible because firmware might change / be different, so I guess that's the best way to do that

09:49 <qzed> 3) what about QCOM0476 (Qualcomm System Manager TrEE Device)? I think that would be the ACPI equivalent

09:49 <qzed> shawnguo: ^

09:58 <shawnguo> qzed: hmm, honestly, I'm not sure what that device is for. how does scm device need a GpioInt resource?

09:58 <qzed> not sure, but there's a lot that I don't know about it

10:00 <shawnguo> s/scm/qseecom

10:01 <qzed> based on .inf files: ACPI\QCOM0476 is using QcTrEE8180.sys

10:02 <qzed> and this in turn is (based on ghidra and inf files) the driver managing all secure apps

10:02 <qzed> so essentially the one that provides/manages the qseecom interface

10:05 <shawnguo> Ok, good to know, thanks! I will test QCOM0476 probe then :)

10:07 <qzed> the naming is just qseecom instead of qctree (or anything TrEE) because Srinivas and ndec (I think) mentioned that there is a new TrEE interface

10:07 <qzed> and qseecom is the downstream name for that older one (vs smcinvoke for the newer I think)

17:29 <steev> new x13s bios - http://paste.debian.net/1268443/

17:31 <steev> also some new "boot critical drivers" package that enables D3Hot support

17:32 <steev> also new trackpoint firmware

17:38 <steev> whatever that package is... it takes centuries to install

19:26 <rfs613> steev: I had windows running on mine (finally figured out the strange sleep issue) and it decided to download and update that BIOS all by itself.

19:27 <rfs613> i haven't really noticed any differences afterwards...

19:37 <steev> that's... probably good?

19:53 <rfs613> well other than the forced update without any notice ;-)

20:46 <clover[m]> consent is important!