linux-kernel - Re: [PATCH] dma-direct: Fix potential NULL pointer dereference

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: <da9ec51d-aab5-695d-e388-5ae7c0bb30ea@oracle.com>
Date:   Tue, 15 Sep 2020 10:40:39 -0400
From:   Thomas Tai <thomas.tai@...cle.com>
To:     Christoph Hellwig <hch@....de>
Cc:     konrad.wilk@...cle.com, m.szyprowski@...sung.com,
        robin.murphy@....com, iommu@...ts.linux-foundation.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH] dma-direct: Fix potential NULL pointer dereference


On 2020-09-15 10:26 a.m., Christoph Hellwig wrote:
> On Tue, Sep 15, 2020 at 10:11:51AM -0400, Thomas Tai wrote:
>>
>>
>> On 2020-09-15 10:07 a.m., Christoph Hellwig wrote:
>>> On Tue, Sep 15, 2020 at 08:03:14AM -0600, Thomas Tai wrote:
>>>> When booting the kernel v5.9-rc4 on a VM, the kernel would panic when
>>>> printing a warning message in swiotlb_map(). It is because dev->dma_mask
>>>> can potentially be a null pointer. Using the dma_get_mask() macro can
>>>> avoid the NULL pointer dereference.
>>>
>>> dma_mask must not be zero.  This means drm is calling DMA API functions
>>> on something weird.  This needs to be fixed in the caller.
>>>
>>
>> Thanks, Christoph for your comment. The caller already fixed the null
>> pointer in the latest v5.9-rc5. I am thinking that if we had used the
>> dma_get_mask(), the kernel couldn't panic and could properly print out the
>> warning message.
> 
> If we want to solve this something like this patch is probably the
> right way:
> 
> 
> 
> diff --git a/include/linux/dma-direct.h b/include/linux/dma-direct.h
> index 6e87225600ae35..064870844f06c1 100644
> --- a/include/linux/dma-direct.h
> +++ b/include/linux/dma-direct.h
> @@ -62,9 +62,6 @@ static inline bool dma_capable(struct device *dev, dma_addr_t addr, size_t size,
>   {
>   	dma_addr_t end = addr + size - 1;
>   
> -	if (!dev->dma_mask)
> -		return false;
> -

I am concerned that some drivers may rely on this NULL checking. Would 
you think we can keep this checking and use the following WARN_ON_ONCE()?

>   	if (is_ram && !IS_ENABLED(CONFIG_ARCH_DMA_ADDR_T_64BIT) &&
>   	    min(addr, end) < phys_to_dma(dev, PFN_PHYS(min_low_pfn)))
>   		return false;
> diff --git a/kernel/dma/mapping.c b/kernel/dma/mapping.c
> index 0d129421e75fc8..2b01d8f7baf160 100644
> --- a/kernel/dma/mapping.c
> +++ b/kernel/dma/mapping.c
> @@ -144,6 +144,10 @@ dma_addr_t dma_map_page_attrs(struct device *dev, struct page *page,
>   	dma_addr_t addr;
>   
>   	BUG_ON(!valid_dma_direction(dir));
> +
> +	if (WARN_ON_ONCE(!dev->dma_mask))
> +		return DMA_MAPPING_ERROR;
> +
>   	if (dma_map_direct(dev, ops))
>   		addr = dma_direct_map_page(dev, page, offset, size, dir, attrs);
>   	else
> @@ -179,6 +183,10 @@ int dma_map_sg_attrs(struct device *dev, struct scatterlist *sg, int nents,
>   	int ents;
>   
>   	BUG_ON(!valid_dma_direction(dir));
> +
> +	if (WARN_ON_ONCE(!dev->dma_mask))
> +		return 0;
> +
>   	if (dma_map_direct(dev, ops))
>   		ents = dma_direct_map_sg(dev, sg, nents, dir, attrs);
>   	else
> @@ -217,6 +225,9 @@ dma_addr_t dma_map_resource(struct device *dev, phys_addr_t phys_addr,
>   	if (WARN_ON_ONCE(pfn_valid(PHYS_PFN(phys_addr))))
>   		return DMA_MAPPING_ERROR;
>   
> +	if (WARN_ON_ONCE(!dev->dma_mask))
> +		return DMA_MAPPING_ERROR;
> +
>   	if (dma_map_direct(dev, ops))
>   		addr = dma_direct_map_resource(dev, phys_addr, size, dir, attrs);
>   	else if (ops->map_resource)
>