linux-kernel - Re: [PATCH 0/2] {vision/navigation}-mezzanine: Fix overlay root node

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <d12ba10b-6eb3-4344-af50-dba80c79791e@linaro.org>
Date: Thu, 27 Feb 2025 11:12:33 +0000
From: Bryan O'Donoghue <bryan.odonoghue@...aro.org>
To: Nikita Travkin <nikita@...n.ru>, Bjorn Andersson <andersson@...nel.org>,
 Konrad Dybcio <konradybcio@...nel.org>, Rob Herring <robh@...nel.org>,
 Krzysztof Kozlowski <krzk+dt@...nel.org>, Conor Dooley <conor+dt@...nel.org>
Cc: Konrad Dybcio <konrad.dybcio@....qualcomm.com>,
 linux-arm-msm@...r.kernel.org, devicetree@...r.kernel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH 0/2] {vision/navigation}-mezzanine: Fix overlay root node

On 26/02/2025 14:29, Nikita Travkin wrote:
> While considering to propose WoA EL2 dt overlays upstream I was looking
> at existing overlays and noticed that some of them are broken: they put
> seemingly meaningful fixups into the overlay's "/" node, which places
> them into the overlay "metadata" itself, not into a fixup fragment to be
> applied to the actual dtb. This series fixes those two by changing to
> full path "&{/}" which should work as it was initially intended.
> 
> See demonstration of the problem below:
> 
> $ cat base.dts
> /dts-v1/;
> / {
> 	compatible = "fake,board";
> 	fake,value = <42>;
> };
> 
> $ cat extra.dtso
> /dts-v1/;
> /plugin/;
> 
> / {
> 	foo;
> 	bar { baz; };
> };
> &{/} { whatever-comes-next-after-baz; };
> 
> $ dtc base.dts -o base.dtb
> $ dtc extra.dtso -o extra.dtbo
> $ fdtoverlay -i base.dtb -o combine.dtb extra.dtbo
> $ dtc base.dtb
> /dts-v1/;
> 
> / {
> 	compatible = "fake,board";
> 	fake,value = <0x2a>;
> };
> 
> $ dtc extra.dtbo
> /dts-v1/;
> 
> / {
> 	foo;
> 
> 	bar {
> 		baz;
> 	};
> 
> 	fragment@0 {
> 		target-path = "/";
> 
> 		__overlay__ {
> 			whatever-comes-next-after-baz;
> 		};
> 	};
> };
> 
> $ dtc combine.dtb
> /dts-v1/;
> 
> / {
> 	whatever-comes-next-after-baz;
> 	compatible = "fake,board";
> 	fake,value = <0x2a>;
> };
> 
> In the resulting dtb foo bar and baz are missing.
> 
> Signed-off-by: Nikita Travkin <nikita@...n.ru>
> ---
> Nikita Travkin (2):
>        arm64: dts: qcom: qrb5165-rb5-vision-mezzanine: Fix broken overlay root
>        arm64: dts: qcom: sdm845-db845c-navigation-mezzanine: Fix the overlay root
> 
>   arch/arm64/boot/dts/qcom/qrb5165-rb5-vision-mezzanine.dtso       | 2 +-
>   arch/arm64/boot/dts/qcom/sdm845-db845c-navigation-mezzanine.dtso | 2 +-
>   2 files changed, 2 insertions(+), 2 deletions(-)
> ---
> base-commit: 8433c776e1eb1371f5cd40b5fd3a61f9c7b7f3ad
> change-id: 20250226-qcom-nonroot-overlays-bfe21d33be8c
> 
> Best regards,

This is a good find however, we should just drop these additional CMA 
heap nodes as the reason for their original addition has been superseded.

libcamera can now use UDMA buf for the same purpose as the CMA heap, 
without the security concerns of CMA to distros.

You only need the CMA heap when you require uncached physically 
contiguous buffers - for example Hantro on iMX but, this is not the case 
on qcom hardware.

We can safely drop these CMA nodes for CAMSS.

---
bod