linux-kernel - Re: [PATCH v3 2/2] [media] videobuf2-dc: Support cacheable MMAP

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:   Mon, 17 Sep 2018 16:41:26 +0200
From:   Hans Verkuil <hverkuil@...all.nl>
To:     Thierry Escande <thierry.escande@...labora.com>,
        Mauro Carvalho Chehab <mchehab@...nel.org>,
        Sakari Ailus <sakari.ailus@....fi>
Cc:     linux-media@...r.kernel.org, linux-kernel@...r.kernel.org,
        Pawel Osciak <pawel@...iak.com>,
        Marek Szyprowski <m.szyprowski@...sung.com>,
        Kyungmin Park <kyungmin.park@...sung.com>
Subject: Re: [PATCH v3 2/2] [media] videobuf2-dc: Support cacheable MMAP

I'm going through old patches in patchwork that fell through the
cracks, and this is one of them.

If this is still desired, please rebase and repost.

I'm marking this series as Obsoleted in patchwork, since it no longer
applies anyway.

Regards,

	Hans


On 10/26/2016 10:52 AM, Thierry Escande wrote:
> From: Heng-Ruey Hsu <henryhsu@...omium.org>
> 
> DMA allocations for MMAP type are uncached by default. But for
> some cases, CPU has to access the buffers. ie: memcpy for format
> converter. Supporting cacheable MMAP improves huge performance.
> 
> This patch enables cacheable memory for DMA coherent allocator in mmap
> buffer allocation if non-consistent DMA attribute is set and kernel
> mapping is present. Even if userspace doesn't mmap the buffer, sync
> still should be happening if kernel mapping is present.
> If not done in allocation, it is enabled when memory is mapped from
> userspace (if non-consistent DMA attribute is set).
> 
> Signed-off-by: Heng-Ruey Hsu <henryhsu@...omium.org>
> Tested-by: Heng-ruey Hsu <henryhsu@...omium.org>
> Reviewed-by: Tomasz Figa <tfiga@...omium.org>
> Signed-off-by: Thierry Escande <thierry.escande@...labora.com>
> ---
>  drivers/media/v4l2-core/videobuf2-dma-contig.c | 16 ++++++++++++++++
>  1 file changed, 16 insertions(+)
> 
> diff --git a/drivers/media/v4l2-core/videobuf2-dma-contig.c b/drivers/media/v4l2-core/videobuf2-dma-contig.c
> index 0d9665d..89b534a 100644
> --- a/drivers/media/v4l2-core/videobuf2-dma-contig.c
> +++ b/drivers/media/v4l2-core/videobuf2-dma-contig.c
> @@ -151,6 +151,10 @@ static void vb2_dc_put(void *buf_priv)
>  		sg_free_table(buf->sgt_base);
>  		kfree(buf->sgt_base);
>  	}
> +	if (buf->dma_sgt) {
> +		sg_free_table(buf->dma_sgt);
> +		kfree(buf->dma_sgt);
> +	}
>  	dma_free_attrs(buf->dev, buf->size, buf->cookie, buf->dma_addr,
>  		       buf->attrs);
>  	put_device(buf->dev);
> @@ -192,6 +196,14 @@ static void *vb2_dc_alloc(struct device *dev, unsigned long attrs,
>  	buf->handler.put = vb2_dc_put;
>  	buf->handler.arg = buf;
>  
> +	/*
> +	 * Enable cache maintenance. Even if userspace doesn't mmap the buffer,
> +	 * sync still should be happening if kernel mapping is present.
> +	 */
> +	if (!(buf->attrs & DMA_ATTR_NO_KERNEL_MAPPING) &&
> +	    buf->attrs & DMA_ATTR_NON_CONSISTENT)
> +		buf->dma_sgt = vb2_dc_get_base_sgt(buf);
> +
>  	atomic_inc(&buf->refcount);
>  
>  	return buf;
> @@ -227,6 +239,10 @@ static int vb2_dc_mmap(void *buf_priv, struct vm_area_struct *vma)
>  
>  	vma->vm_ops->open(vma);
>  
> +	/* Enable cache maintenance if not enabled in allocation. */
> +	if (!buf->dma_sgt && buf->attrs & DMA_ATTR_NON_CONSISTENT)
> +		buf->dma_sgt = vb2_dc_get_base_sgt(buf);
> +
>  	pr_debug("%s: mapped dma addr 0x%08lx at 0x%08lx, size %ld\n",
>  		__func__, (unsigned long)buf->dma_addr, vma->vm_start,
>  		buf->size);
>