| Message ID | 20221126010531.466046-1-pevik@seznam.cz |
|---|---|
| State | New |
| Headers | show |
| Series | [1/1] arm64: dts: qcom: msm8992-lg-bullhead: Disable cont_splash_mem | expand |
Hi all, I got reply from Dominik that he noticed dfps_data_mem (reserved in msm8994.dtsi) overlaps with the framebuffer region. I'll retest it next week, maybe removing it instead of cont_splash_mem fixes the reboot. He also noticed, that Linux doesn't reboot when loaded from uefi (his project [1]). Not sure if there is any difference. Kind regards, Petr [1] https://github.com/Dominduchami/Nexus5XPkg/
On 04/12/2022 14:16, Petr Vorel wrote: > Hi all, > > > I got reply from Dominik that he noticed dfps_data_mem (reserved in > msm8994.dtsi) overlaps with the framebuffer region. They are usually next to each other.. > I'll retest it next week, maybe removing it instead of cont_splash_mem > fixes the reboot. Maybe your bootloader reads it? And needs it to be at a certain address, maybe with a certain node name or a label? > > He also noticed, that Linux doesn't reboot when loaded from uefi (his > project [1]). > Not sure if there is any difference. Is it loaded with this DTB or does it rely on ACPI? Konrad > > Kind regards, > Petr > > [1] https://github.com/Dominduchami/Nexus5XPkg/
Hi all,
> Is it loaded with this DTB or does it rely on ACPI?
It is loaded with this DT (booted from grub with efifb enabled in kernel config)
The uefi is just Lumia950XLPKg (link : [1]) with some modifications.
It also contains code from SurfaceDuoPkg (link : [2]) that resets the platform,
and uefi then reinitializes MMU (with a more windows-suited memory map).
I'm guessing it could be the reason for the difference in behaviour.
Kind regards,
Dominik
[1] https://github.com/WOA-Project/Lumia950XLPkg
[2] https://github.com/WOA-Project/SurfaceDuoPkg
Hi all, Looks like the reboot issue was caused by a memory hole not being reserved. It's defined in the downstream kernel source (link : [1]). Adding it makes the kernel boot fine (tested when booted from lk2nd) I'll try to send a patch with it tomorrow. Kind regards, Dominik [1] https://github.com/android-linux-stable/bullhead/blob/android-msm-bullhead-3.10/arch/arm64/boot/dts/lge/msm8992-bullhead.dtsi#L144
On 10.12.2022 06:24, Dominik Kobinski wrote: > Hi all, > > Looks like the reboot issue was caused by a memory hole not being reserved. > It's defined in the downstream kernel source (link : [1]). > > Adding it makes the kernel boot fine (tested when booted from lk2nd) > > I'll try to send a patch with it tomorrow. > > Kind regards, > Dominik That looks more sane than a piece of memory being reserved causing issues :) Happy you found the fix! Konrad > > [1] https://github.com/android-linux-stable/bullhead/blob/android-msm-bullhead-3.10/arch/arm64/boot/dts/lge/msm8992-bullhead.dtsi#L144
Hi Dominik, Also congratulations for your findings, looking forward testing your patch! Setting my patch as "rejected" as it's obviously invalid. Kind regards, Petr
diff --git a/arch/arm64/boot/dts/qcom/msm8992-lg-bullhead.dtsi b/arch/arm64/boot/dts/qcom/msm8992-lg-bullhead.dtsi index 87c90e93667f..a2f69d965ae5 100644 --- a/arch/arm64/boot/dts/qcom/msm8992-lg-bullhead.dtsi +++ b/arch/arm64/boot/dts/qcom/msm8992-lg-bullhead.dtsi @@ -2,7 +2,7 @@ /* * Copyright (c) 2015, LGE Inc. All rights reserved. * Copyright (c) 2016, The Linux Foundation. All rights reserved. - * Copyright (c) 2021, Petr Vorel <petr.vorel@gmail.com> + * Copyright (c) 2021-2022, Petr Vorel <petr.vorel@gmail.com> */ /dts-v1/; @@ -11,7 +11,10 @@ #include "pm8994.dtsi" #include "pmi8994.dtsi" -/* cont_splash_mem has different memory mapping */ +/* + * cont_splash_mem has different memory mapping (0x03400000), + * but that since v6.0-rc1 it causes phone reboot somewhere in mmc module. + */ /delete-node/ &cont_splash_mem; / { @@ -46,11 +49,6 @@ ramoops@1ff00000 { ftrace-size = <0x10000>; pmsg-size = <0x20000>; }; - - cont_splash_mem: memory@3400000 { - reg = <0 0x03400000 0 0x1200000>; - no-map; - }; }; };