RE: [PATCH v2 7/9] dpaa_eth: fix iova handling for contiguous frames

2019-05-02 Thread Laurentiu Tudor



> -Original Message-
> From: Christoph Hellwig 
> Sent: Saturday, April 27, 2019 7:46 PM
> 
> On Sat, Apr 27, 2019 at 10:10:29AM +0300, laurentiu.tu...@nxp.com wrote:
> > From: Laurentiu Tudor 
> >
> > The driver relies on the no longer valid assumption that dma addresses
> > (iovas) are identical to physical addressees and uses phys_to_virt() to
> > make iova -> vaddr conversions. Fix this by adding a function that does
> > proper iova -> phys conversions using the iommu api and update the code
> > to use it.
> > Also, a dma_unmap_single() call had to be moved further down the code
> > because iova -> vaddr conversions were required before the unmap.
> > For now only the contiguous frame case is handled and the SG case is
> > split in a following patch.
> > While at it, clean-up a redundant dpaa_bpid2pool() and pass the bp
> > as parameter.
> 
> Err, this is broken.  A driver using the DMA API has no business
> call IOMMU APIs.  Just save the _virtual_ address used for the mapping
> away and use that again.  We should not go through crazy gymnastics
> like this.

I think that due to the particularity of this hardware we don't have a way of 
saving the VA, but I'd let my colleagues maintaining this driver to comment 
more on why we need to do this.

---
Best Regards, Laurentiu
___
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu


Re: [PATCH v2 7/9] dpaa_eth: fix iova handling for contiguous frames

2019-04-27 Thread Christoph Hellwig
On Sat, Apr 27, 2019 at 10:10:29AM +0300, laurentiu.tu...@nxp.com wrote:
> From: Laurentiu Tudor 
> 
> The driver relies on the no longer valid assumption that dma addresses
> (iovas) are identical to physical addressees and uses phys_to_virt() to
> make iova -> vaddr conversions. Fix this by adding a function that does
> proper iova -> phys conversions using the iommu api and update the code
> to use it.
> Also, a dma_unmap_single() call had to be moved further down the code
> because iova -> vaddr conversions were required before the unmap.
> For now only the contiguous frame case is handled and the SG case is
> split in a following patch.
> While at it, clean-up a redundant dpaa_bpid2pool() and pass the bp
> as parameter.

Err, this is broken.  A driver using the DMA API has no business
call IOMMU APIs.  Just save the _virtual_ address used for the mapping
away and use that again.  We should not go through crazy gymnastics
like this.
___
iommu mailing list
iommu@lists.linux-foundation.org
https://lists.linuxfoundation.org/mailman/listinfo/iommu