[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <a72d3264eaa352652a4ebaf4b1baa664@trvn.ru>
Date: Thu, 27 Feb 2025 14:18:09 +0500
From: Nikita Travkin <nikita@...n.ru>
To: Rob Herring <robh@...nel.org>, Bryan O'Donoghue
<bryan.odonoghue@...aro.org>
Cc: Bjorn Andersson <andersson@...nel.org>, Konrad Dybcio
<konradybcio@...nel.org>, Krzysztof Kozlowski <krzk+dt@...nel.org>, Conor
Dooley <conor+dt@...nel.org>, Konrad Dybcio
<konrad.dybcio@....qualcomm.com>, linux-arm-msm@...r.kernel.org,
devicetree@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/2] arm64: dts: qcom: qrb5165-rb5-vision-mezzanine: Fix
broken overlay root
Rob Herring писал(а) 27.02.2025 00:22:
> On Wed, Feb 26, 2025 at 8:30 AM Nikita Travkin <nikita@...n.ru> wrote:
>>
>> When converting to the overlay format, it was missed that "/" in the
>> overlay corresponds to the overlay's own root node and not the fragment
>> targeted to update root of the base dts, which should be "&{/}" instead.
>> This results in the cma node never actually being applied by libfdt.
>>
>> Fix the overlay to use correct target node.
>>
>> Fixes: 231c03c6119d ("arm64: dts: qcom: qrb5165-rb5-vision-mezzanine: Convert mezzanine riser to dtbo")
>> Signed-off-by: Nikita Travkin <nikita@...n.ru>
>> ---
>> arch/arm64/boot/dts/qcom/qrb5165-rb5-vision-mezzanine.dtso | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/arch/arm64/boot/dts/qcom/qrb5165-rb5-vision-mezzanine.dtso b/arch/arm64/boot/dts/qcom/qrb5165-rb5-vision-mezzanine.dtso
>> index ae256c713a36078afdadc67193f381a19ea8e5d3..254df3d518d8cbfb1082511f38e132435b7fdf59 100644
>> --- a/arch/arm64/boot/dts/qcom/qrb5165-rb5-vision-mezzanine.dtso
>> +++ b/arch/arm64/boot/dts/qcom/qrb5165-rb5-vision-mezzanine.dtso
>> @@ -9,7 +9,7 @@
>> #include <dt-bindings/clock/qcom,camcc-sm8250.h>
>> #include <dt-bindings/gpio/gpio.h>
>>
>> -/ {
>> +&{/} {
>> reserved-memory {
>
> IMO, this should be applied to the /reserved-memory node rather than
Hm, considering pretty much no qcom dtb will be without reserved-memory,
you're right...
> the root node. Though I also think using overlays to set CMA size is
> questionable. It's much easier to change the kernel command line than
> apply an overlay.
...and deleting the node altogether would be effectively keeping the
existing behavior, which, I just now realized, was introduced in an
extra commit after the conversion (which I assumed was the cause of
the mistake) so my Fixes: for both commits are wrong.
I think I will leave it to Bryan to decide whether this should be
fixed or original commit to be reverted.
>
>> linux,cma {
>> compatible = "shared-dma-pool";
>>
>> --
>> 2.48.1
>>
Powered by blists - more mailing lists