| Message ID | 20210622181056.27632-1-petr.vorel@gmail.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v2,1/1] arm64: dts: qcom: msm8994-angler: Fix cont_splash_mem | expand |
Ah yes, I think I know the issue.. Angler's firmware (probably Huawei's job, but unsure) seems to make IOMMUs (another guesstimate) not report where the memory is allocated. If it's an issue only on these boards, I reckon it would be better to just add "/delete-node/ &cont_splash_mem;" in angler/bullhead DTSI, as other 8992/4 devices seem to not have this issue. But as I've said, bullhead could probably work anyway, but if it doesn't, it's probably Google's job after all.. P.S I did not suggest redefining the node, as the memory@0 is reserved for SoC registers and your firmware is just playing tricks on you. Your RAM starts @ 0x80000000. Konrad
Hi Konrad, > Ah yes, I think I know the issue.. > Angler's firmware (probably Huawei's job, but unsure) seems to make IOMMUs (another guesstimate) > not report where the memory is allocated. If it's an issue only on these boards, I reckon > it would be better to just add > "/delete-node/ &cont_splash_mem;" Makes sense, I'll send v3. I didn't know proper syntax, that's why used memory@3800000 instead of alias. > in angler/bullhead DTSI, as other 8992/4 devices seem to not have this issue. But as I've said, bullhead could > probably work anyway, but if it doesn't, it's probably Google's job after all.. Although bullhead is from LG, it might be affected as well. Hope somebody with this device test it and send patch if needed. > P.S > I did not suggest redefining the node, as the memory@0 is reserved for SoC registers > and your firmware is just playing tricks on you. Your RAM starts @ 0x80000000. Thanks for an explanation. > Konrad Kind regards, Petr
diff --git a/arch/arm64/boot/dts/qcom/msm8994-angler-rev-101.dts b/arch/arm64/boot/dts/qcom/msm8994-angler-rev-101.dts index ffe1a9bd8f70..4f8a5f5cbbaa 100644 --- a/arch/arm64/boot/dts/qcom/msm8994-angler-rev-101.dts +++ b/arch/arm64/boot/dts/qcom/msm8994-angler-rev-101.dts @@ -1,6 +1,7 @@ // SPDX-License-Identifier: GPL-2.0-only /* Copyright (c) 2015, Huawei Inc. All rights reserved. * Copyright (c) 2016, The Linux Foundation. All rights reserved. + * Copyright (c) 2021, Petr Vorel <petr.vorel@gmail.com> */ /dts-v1/; @@ -31,6 +32,15 @@ serial@f991e000 { pinctrl-1 = <&blsp1_uart2_sleep>; }; }; + + reserved-memory { + /delete-node/ memory@3800000; + + cont_splash_mem: memory@0 { + reg = <0 0 0 0x1600000>; + no-map; + }; + }; }; &tlmm {
As the default definition breaks booting angler: [ 1.862561] printk: console [ttyMSM0] enabled [ 1.872260] msm_serial: driver initialized D - 15524 - pm_driver_init, Delta cont_splash_mem was introduced in 74d6d0a145835, but the problem manifested after 86588296acbf ("fdt: Properly handle "no-map" field in the memory region"). Using new definition from downstream kernel: [ 0.000000] cma: Found cont_splash_mem@0, memory base 0x0000000000000000, size 16 MiB, limit 0x0000000000000000 [ 0.000000] cma: CMA: reserved 16 MiB at 0x0000000000000000 for cont_splash_mem Fixes: 74d6d0a145835 ("arm64: dts: qcom: msm8994/8994-kitakami: Fix up the memory map") Signed-off-by: Petr Vorel <petr.vorel@gmail.com> --- Changes v1->v2: * redefine address instead of deleting it (as suggested by Konrad) Kind regards, Petr [1] https://wiki.postmarketos.org/wiki/Google_Nexus_5X_(lg-bullhead) arch/arm64/boot/dts/qcom/msm8994-angler-rev-101.dts | 10 ++++++++++ 1 file changed, 10 insertions(+)