linux-kernel - Re: [PATCH v4 09/14] of: reserved-memory: Fix using wrong number of cells to get property 'alignment'

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: <20250113232551.GB1983895-robh@kernel.org>
Date: Mon, 13 Jan 2025 17:25:51 -0600
From: Rob Herring <robh@...nel.org>
To: Zijun Hu <zijun_hu@...oud.com>
Cc: Saravana Kannan <saravanak@...gle.com>,
	Maxime Ripard <mripard@...nel.org>,
	Robin Murphy <robin.murphy@....com>,
	Grant Likely <grant.likely@...retlab.ca>,
	Marc Zyngier <maz@...nel.org>,
	Andreas Herrmann <andreas.herrmann@...xeda.com>,
	Marek Szyprowski <m.szyprowski@...sung.com>,
	Catalin Marinas <catalin.marinas@....com>,
	Mike Rapoport <rppt@...nel.org>,
	Oreoluwa Babatunde <quic_obabatun@...cinc.com>,
	devicetree@...r.kernel.org, linux-kernel@...r.kernel.org,
	Zijun Hu <quic_zijuhu@...cinc.com>, stable@...r.kernel.org
Subject: Re: [PATCH v4 09/14] of: reserved-memory: Fix using wrong number of
 cells to get property 'alignment'

On Thu, Jan 09, 2025 at 09:27:00PM +0800, Zijun Hu wrote:
> From: Zijun Hu <quic_zijuhu@...cinc.com>
> 
> According to DT spec, size of property 'alignment' is based on parent
> node’s #size-cells property.
> 
> But __reserved_mem_alloc_size() wrongly uses @dt_root_addr_cells to get
> the property obviously.
> 
> Fix by using @dt_root_size_cells instead of @dt_root_addr_cells.

I wonder if changing this might break someone. It's been this way for 
a long time. It might be better to change the spec or just read 
'alignment' as whatever size it happens to be (len / 4). It's not really 
the kernel's job to validate the DT. We should first have some 
validation in place to *know* if there are any current .dts files that 
would break. That would probably be easier to implement in dtc than 
dtschema. Cases of #address-cells != #size-cells should be pretty rare, 
but that was the default for OpenFirmware.

As the alignment is the base address alignment, it can be argued that 
"#address-cells" makes more sense to use than "#size-cells". So maybe 
the spec was a copy-n-paste error.

> 
> Fixes: 3f0c82066448 ("drivers: of: add initialization code for dynamic reserved memory")
> Cc: stable@...r.kernel.org
> Signed-off-by: Zijun Hu <quic_zijuhu@...cinc.com>
> ---
>  drivers/of/of_reserved_mem.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/of/of_reserved_mem.c b/drivers/of/of_reserved_mem.c
> index 45517b9e57b1add36bdf2109227ebbf7df631a66..d2753756d7c30adcbd52f57338e281c16d821488 100644
> --- a/drivers/of/of_reserved_mem.c
> +++ b/drivers/of/of_reserved_mem.c
> @@ -409,12 +409,12 @@ static int __init __reserved_mem_alloc_size(unsigned long node, const char *unam
>  
>  	prop = of_get_flat_dt_prop(node, "alignment", &len);
>  	if (prop) {
> -		if (len != dt_root_addr_cells * sizeof(__be32)) {
> +		if (len != dt_root_size_cells * sizeof(__be32)) {
>  			pr_err("invalid alignment property in '%s' node.\n",
>  				uname);
>  			return -EINVAL;
>  		}
> -		align = dt_mem_next_cell(dt_root_addr_cells, &prop);
> +		align = dt_mem_next_cell(dt_root_size_cells, &prop);
>  	}
>  
>  	nomap = of_get_flat_dt_prop(node, "no-map", NULL) != NULL;
> 
> -- 
> 2.34.1
>