Re: [Xen-devel] [PATCH 2/2] swiotlb-xen: xen_swiotlb_map_page: do not error out if dma_capable fails

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On Tue, Nov 12, 2013 at 02:12:00PM +0000, Stefano Stabellini wrote:
> Many ARM devices do not set the dma_mask correctly today.
> As a consequence dma_capable fails for them regardless of the address
> passed to it.

Wouldn't the DMA API debug warn of bad usage.

> In xen_swiotlb_map_page we currently use dma_capable to check if the
> address returned by the swiotlb is good for dma for the device.

Right..
> However the check would often fail even if the address is correct.

 .. and they will fail b/c the device hasn't set its dma mask so
we use the platform default right? What is the platform default?

> Considering that we know that the swiotlb buffer has a low address,
> skip the check.

I am not following that sentence. Could you please explain to me
how the SWIOTLB buffer low address guarantees that we don't need
the check?
> 
> Signed-off-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>

What are the drivers that are busted. Does DMA API debug flag them?

Thanks!
> ---
>  drivers/xen/swiotlb-xen.c |    7 -------
>  1 files changed, 0 insertions(+), 7 deletions(-)
> 
> diff --git a/drivers/xen/swiotlb-xen.c b/drivers/xen/swiotlb-xen.c
> index 1eac073..543e30b 100644
> --- a/drivers/xen/swiotlb-xen.c
> +++ b/drivers/xen/swiotlb-xen.c
> @@ -402,13 +402,6 @@ dma_addr_t xen_swiotlb_map_page(struct device *dev, struct page *page,
>  					map & ~PAGE_MASK, size, dir, attrs);
>  	dev_addr = xen_phys_to_bus(map);
>  
> -	/*
> -	 * Ensure that the address returned is DMA'ble
> -	 */
> -	if (!dma_capable(dev, dev_addr, size)) {
> -		swiotlb_tbl_unmap_single(dev, map, size, dir);
> -		dev_addr = 0;
> -	}
>  	return dev_addr;
>  }
>  EXPORT_SYMBOL_GPL(xen_swiotlb_map_page);
> -- 
> 1.7.2.5
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel