linux-kernel - Re: [PATCH] dma-debug: Check scatterlist segments

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]

Date:   Wed, 25 Apr 2018 07:58:56 +0200
From:   Christoph Hellwig <hch@....de>
To:     Robin Murphy <robin.murphy@....com>
Cc:     hch@....de, m.szyprowski@...sung.com,
        iommu@...ts.linux-foundation.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] dma-debug: Check scatterlist segments

This looks interesting.  I suspect it is going to blow up in
quite a few places, so maybe at least for now it might make sense
to have a separate config option?

On Tue, Apr 24, 2018 at 05:12:19PM +0100, Robin Murphy wrote:
> Drivers/subsystems creating scatterlists for DMA should be taking care
> to respect the scatter-gather limitations of the appropriate device, as
> described by dma_parms. A DMA API implementation cannot feasibly split
> a scatterlist into *more* entries than originally passed, so it is not
> well defined what they should do when given a segment larger than the
> limit they are also required to respect.
> 
> Conversely, devices which are less limited than the rather conservative
> defaults, or indeed have no limitations at all (e.g. GPUs with their own
> internal MMU), should be encouraged to set appropriate dma_parms, as
> they may get more efficient DMA mapping performance out of it.
> 
> Signed-off-by: Robin Murphy <robin.murphy@....com>
> ---
>  lib/dma-debug.c | 26 ++++++++++++++++++++++++++
>  1 file changed, 26 insertions(+)
> 
> diff --git a/lib/dma-debug.c b/lib/dma-debug.c
> index 7f5cdc1e6b29..9f158941004d 100644
> --- a/lib/dma-debug.c
> +++ b/lib/dma-debug.c
> @@ -1293,6 +1293,30 @@ static void check_sync(struct device *dev,
>  	put_hash_bucket(bucket, &flags);
>  }
>  
> +static void check_sg_segment(struct device *dev, struct scatterlist *sg)
> +{
> +	unsigned int max_seg = dma_get_max_seg_size(dev);
> +	dma_addr_t start, end, boundary = dma_get_seg_boundary(dev);
> +
> +	/*
> +	 * Either the driver forgot to set dma_parms appropriately, or
> +	 * whoever generated the list forgot to check them.
> +	 */
> +	if (sg->length > max_seg)
> +		err_printk(dev, NULL, "DMA-API: mapping sg segment longer than device claims to support [len=%u] [max=%u]\n",
> +			   sg->length, max_seg);
> +	/*
> +	 * In some cases this could potentially be the DMA API
> +	 * implementation's fault, but it would usually imply that
> +	 * the scatterlist was built inappropriately to begin with.
> +	 */
> +	start = sg_dma_address(sg);
> +	end = start + sg_dma_len(sg) - 1;
> +	if ((start ^ end) & ~boundary)
> +		err_printk(dev, NULL, "DMA-API: mapping sg segment across boundary [start=0x%016llx] [end=0x%016llx] [boundary=0x%016llx]\n",
> +			   start, end, boundary);
> +}
> +
>  void debug_dma_map_page(struct device *dev, struct page *page, size_t offset,
>  			size_t size, int direction, dma_addr_t dma_addr,
>  			bool map_single)
> @@ -1423,6 +1447,8 @@ void debug_dma_map_sg(struct device *dev, struct scatterlist *sg,
>  			check_for_illegal_area(dev, sg_virt(s), sg_dma_len(s));
>  		}
>  
> +		check_sg_segment(dev, s);
> +
>  		add_dma_entry(entry);
>  	}
>  }
> -- 
> 2.17.0.dirty
---end quoted text---