#aarch64-laptops on 2023-11-01 — irc logs at oftc.irclog.whitequark.org

2023-07-24 16:55 robclark changed the topic of #aarch64-laptops to: Linux support for AArch64 Laptops (Chrome OS Trogdor Devices - Asus NovaGo TP370QL - HP Envy x2 - Lenovo Mixx 630 - Lenovo Yoga C630 - Lenovo ThinkPad X13s - and various other snapdragon laptops) - https://oftc.irclog.whitequark.org/aarch64-laptops

00:09 <jenneron[m]> bamse: i found one more bug, in 6.6.0 UFS sometimes dies on high disk load https://dpaste.com/F3JPKD3PN, can be reproduced with gnome-disks benchmark with sample size 1000 and big number of samples e.g. 50+

00:09 <jenneron[m]> it doesn't happen on 6.2.8

00:09 <jenneron[m]> also on 6.6.0 the UFS speed is higher but less stable

00:21 <robclark> drive by.. mmu strict the same in both cases? I could see that as something that also effects ufs..

00:26 <jenneron[m]> i use the same cmdline on both 6.6.0 and 6.2.8 https://dpaste.com/8ZB7YY55N

00:31 <robclark> try w/out iommu.strict=0

00:35 <jenneron[m]> robclark: is it supposed to fix it?

00:35 <jenneron[m]> because it can just workaround it by making UFS 10 times slower?

00:35 <jenneron[m]> i will try in a bit

00:46 <jenneron[m]> robclark: without iommu.strict=0 it still happens, even faster

00:47 <robclark> ok, good to rule that out.. with dpu safe threshold patch it shouldn't be 10x slower... but there can potentially be memory corruption issues exposed by not using iommu in strict mode so good to rule that out as a possible cause

00:51 <jenneron[m]> yeah it's not the cause

00:52 <jenneron[m]> meanwhile, i should probably try disabling strict iommu for eMMC on trogdor chromebooks in postmarketOS, like chrome os does

01:00 <jenneron[m]> it's probably not going to work though

01:16 <robclark> I think we might still be setting non-strict via udev rules.. but sc7180/sc7280 both already have the dpu bits to make strict not hurt perf as much

01:24 <jenneron[m]> robclark: on which stage do these udev rules trigger? probing? because we use initramfs without udev, so all these devices (sdhci, usb) get probed before loading udev

01:25 <jenneron[m]> also, do you mean they have dpu bits in upstream or chrome os kernel

01:26 <robclark> dpu bits are upstream... but we discovered recently that sc8180* and sc8280* missed the dpu bits.. doug posted this link to the udev rules https://chromium.googlesource.com/chromiumos/overlays/board-overlays/+/refs/heads/main/baseboard-trogdor/chromeos-base/chromeos-bsp-baseboard-trogdor/files/98-qcom-nonstrict-iommu.rules

01:27 <jenneron[m]> i see, so i don't have to do anything for sc7180 devices

01:27 <robclark> https://patchwork.freedesktop.org/series/125778/ is sc8280xp dpu part.. sc8180xp needs something similar still.. otherwise all upstream

01:28 <robclark> no kernel patches needed.. you might want udev rules to eek out that last bit of perf (but the already upstream dpu bits should get you 95% of the way there)

01:30 <jenneron[m]> well, it's not a big deal to add udev rules to the package, but i'm afraid it won't work because of udev loading after initramfs when these peripherals had already been probed

01:30 Ablu has quit [Remote host closed the connection]

01:30 <jenneron[m]> also, i'm not sure i can include that file in a package under MIT license...

01:31 Ablu has joined #aarch64-laptops

01:31 <robclark> I won't claim to be udev expert.. maybe there is another way to accomplish the same thing?

01:32 <robclark> I'm mostly the 2ndhand messenger here ;-)

01:45 <jenneron[m]> it increases eMMC read rate from 203 MB/s to 231 MB/s

01:46 <jenneron[m]> so +15%

02:01 <robclark> hmm, ok so 85% of the way there (but IIRC it was 0% to 15 or maybe 20% depending on benchmark)

02:05 <jenneron[m]> <robclark> "I won't claim to be udev expert...." <- yep, doing it another way https://gitlab.com/postmarketOS/pmaports/-/merge_requests/4526, without udev

02:06 <jenneron[m]> basically running 2 commands in initramfs script

02:06 <jenneron[m]> do you know if sc7180, sc8180* sc8280* are all SoCs affected or are there more?

02:06 <bamse> jenneron[m]: is that with the iommu.strict=0?

02:07 <bamse> jenneron[m]: the ufs issue above that is

02:07 <jenneron[m]> bamse: with and without

02:08 <bamse> wonder if it's the undervolt issue i spotted on some other platform...

02:11 <bamse> jenneron[m]: please copy the power-domains and required-opps from the ufs_mem_hc node in sc8280xp.dtsi...looks like they should be the same

02:12 <bamse> i mean, that is missing in sc8180x and the information should be the same across the two platforms

02:12 <jenneron[m]> will do

02:19 <jenneron[m]> bamse: it still happens

02:20 alpernebbi has quit [Ping timeout: 480 seconds]

02:21 alpernebbi has joined #aarch64-laptops

02:34 <bamse> jenneron[m]: :( thanks for checking though

05:01 hexdump01 has joined #aarch64-laptops

05:03 hexdump0815 has quit [Ping timeout: 480 seconds]

05:21 Ablu has quit [Remote host closed the connection]

05:22 Ablu has joined #aarch64-laptops

06:17 Ablu has quit [Remote host closed the connection]

06:36 Ablu has joined #aarch64-laptops

06:38 Ablu has quit [Remote host closed the connection]

06:52 Ablu has joined #aarch64-laptops

06:54 Ablu has quit [Remote host closed the connection]

06:58 Ablu has joined #aarch64-laptops

07:00 Ablu has quit [Remote host closed the connection]

07:12 Ablu has joined #aarch64-laptops

07:12 srinik has quit [Killed (NickServ (Too many failed password attempts.))]

07:16 srinik has joined #aarch64-laptops

07:23 suihkulo1ki is now known as suihkulokki

07:29 hightower2 has quit [Remote host closed the connection]

07:29 hightower2 has joined #aarch64-laptops

08:08 Lucanis has quit [Ping timeout: 480 seconds]

08:52 jlinton_ has joined #aarch64-laptops

08:53 broonie_ has joined #aarch64-laptops

08:53 lvrp16_ has joined #aarch64-laptops

08:53 ardb_ has joined #aarch64-laptops

08:53 pundir_ has joined #aarch64-laptops

08:53 ldts___ has joined #aarch64-laptops

08:53 steev_ has joined #aarch64-laptops

08:53 jhugo_ has joined #aarch64-laptops

08:53 sibis0 has joined #aarch64-laptops

08:53 echanude_ has joined #aarch64-laptops

08:53 _alice_ has joined #aarch64-laptops

08:53 exit70_ has joined #aarch64-laptops

08:53 apalos_ has joined #aarch64-laptops

08:54 vkoul| has joined #aarch64-laptops

08:57 DanaG has quit [resistance.oftc.net weber.oftc.net]

08:57 broonie has quit [resistance.oftc.net weber.oftc.net]

08:57 vkoul has quit [resistance.oftc.net weber.oftc.net]

08:57 exit70 has quit [resistance.oftc.net weber.oftc.net]

08:57 apalos has quit [resistance.oftc.net weber.oftc.net]

08:57 ndec has quit [resistance.oftc.net weber.oftc.net]

08:57 echanude has quit [resistance.oftc.net weber.oftc.net]

08:57 pneuhardt has quit [resistance.oftc.net weber.oftc.net]

08:57 steev has quit [resistance.oftc.net weber.oftc.net]

08:57 ardb has quit [resistance.oftc.net weber.oftc.net]

08:57 dgilmore has quit [resistance.oftc.net weber.oftc.net]

08:57 pundir has quit [resistance.oftc.net weber.oftc.net]

08:57 zspec has quit [resistance.oftc.net weber.oftc.net]

08:57 jhugo has quit [resistance.oftc.net weber.oftc.net]

08:57 ldts__ has quit [resistance.oftc.net weber.oftc.net]

08:57 jlinton has quit [resistance.oftc.net weber.oftc.net]

08:57 _alice has quit [resistance.oftc.net weber.oftc.net]

08:57 lvrp16 has quit [resistance.oftc.net weber.oftc.net]

08:57 sibis has quit [resistance.oftc.net weber.oftc.net]

08:57 jlinton_ is now known as jlinton

08:57 ardb_ is now known as ardb

08:57 steev_ is now known as steev

08:58 ndec has joined #aarch64-laptops

08:58 zspec has joined #aarch64-laptops

08:58 pneuhardt has joined #aarch64-laptops

08:58 dgilmore has joined #aarch64-laptops

08:58 DanaG has joined #aarch64-laptops

10:06 pundir_ is now known as pundir

10:28 DanaG has quit [Server closed connection]

10:28 DanaG has joined #aarch64-laptops

10:54 dgilmore has quit [Server closed connection]

10:54 dgilmore has joined #aarch64-laptops

11:19 martiert has quit [Remote host closed the connection]

11:19 martiert has joined #aarch64-laptops

11:37 zspec has quit [Server closed connection]

11:37 zspec has joined #aarch64-laptops

11:39 mkrawczuk_ has quit [Ping timeout: 480 seconds]

11:42 ndec has quit [Server closed connection]

11:42 ndec has joined #aarch64-laptops

11:46 sibis0 has quit []

11:46 sibis0 has joined #aarch64-laptops

11:48 sibis0 is now known as sibis

11:48 sibis has quit []

11:48 sibis has joined #aarch64-laptops

11:49 mkrawczuk has joined #aarch64-laptops

12:01 pneuhardt has quit [Server closed connection]

12:01 pneuhardt has joined #aarch64-laptops

12:54 xroumegue has quit [Ping timeout: 480 seconds]

12:58 broonie_ is now known as broonie

13:01 iivanov has joined #aarch64-laptops

13:03 xroumegue has joined #aarch64-laptops

13:05 iivanov__ has joined #aarch64-laptops

13:05 iivanov__ has quit [Remote host closed the connection]

13:10 hightower2 has quit [Ping timeout: 480 seconds]

14:10 jhugo_ is now known as jhugo

15:24 <jenneron[m]> we have this piece of ACPI tables for lenovo flex 5g touchscreen https://dpaste.com/9PX98KG2A, i understand the regulators part and i can see this operating with gpios 53 and 54, but i don't understand where is setting gpio to active state and where is to low. can someone give me insight on this?

20:35 hexdump01 has quit []

20:36 hexdump0815 has joined #aarch64-laptops

22:22 agl7 is now known as moria

22:52 moria is now known as agl7

22:53 agl7 is now known as moria

22:57 moria is now known as agl7

22:58 agl7 is now known as moria

23:04 <moria> I have changed my Nick from "agl7" to "moria"

23:32 moria has left #aarch64-laptops [Leaving]

23:49 agl7 has joined #aarch64-laptops

23:54 agl7 has left #aarch64-laptops [Leaving]

23:56 agl7 has joined #aarch64-laptops