linux-kernel - Re: [PATCH v3] cma: check for memory region overlapping

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: <8c3fe8ce-4716-73f0-fb8a-ee982f212371@arm.com>
Date:   Thu, 7 Sep 2023 13:57:42 +0100
From:   Robin Murphy <robin.murphy@....com>
To:     Binglei Wang <l3b2w1@...il.com>, hch@....de,
        m.szyprowski@...sung.com
Cc:     iommu@...ts.linux.dev, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3] cma: check for memory region overlapping

On 26/07/2023 3:28 pm, Binglei Wang wrote:
> From: Binglei Wang <l3b2w1@...il.com>
> 
> In the process of parsing the DTS, checks
> whether the memory region specified by the DTS CMA node area
> overlaps with the kernel text memory space reserved by memblock
> before calling early_init_fdt_scan_reserved_mem.

This description bears no relation to what the code actually does. 
rmem_cma_setup() happens well after parsing the DTS, as it is that 
initial parsing process in fdt_scan_reserved_mem() which *makes* the 
memblock reservations in the first place. Thus, as the revert patch 
demonstrates, by the time we get here to start *using* the reserved 
region via fdt_init_reserved_mem(), it is always guaranteed to overlap a 
reserved region because it trivially overlaps with itself.

Furthermore, even if the bootloader does stupidly load a non-relocatable 
kernel into memory that it's described as reserved, the kernel text 
pages should not be available for CMA to allocate from - and if they 
were, that would be a far more fundamental bug elsewhere - so what is 
the exact problem you're trying to solve here?

Thanks,
Robin.

> Maybe it's better to have some warning prompts printed.
> 
> Signed-off-by: Binglei Wang <l3b2w1@...il.com>
> ---
> 
> Notes:
>      v3: fix compile error.
>      v2: delete the logic code for handling return -EBUSY.
>      v1: return -EBUSY when detect overlapping and handle the return case.
> 
>   kernel/dma/contiguous.c | 5 +++++
>   1 file changed, 5 insertions(+)
> 
> diff --git a/kernel/dma/contiguous.c b/kernel/dma/contiguous.c
> index 6ea80ae42..dc6d2af1e 100644
> --- a/kernel/dma/contiguous.c
> +++ b/kernel/dma/contiguous.c
> @@ -410,6 +410,11 @@ static int __init rmem_cma_setup(struct reserved_mem *rmem)
>   		return -EBUSY;
>   	}
>   
> +	if (memblock_is_region_reserved(rmem->base, rmem->size)) {
> +		pr_info("Reserved memory: overlap with other memblock reserved region\n");
> +		return -EBUSY;
> +	}
> +
>   	if (!of_get_flat_dt_prop(node, "reusable", NULL) ||
>   	    of_get_flat_dt_prop(node, "no-map", NULL))
>   		return -EINVAL;