Re: [PATCH 07/26] iommu/dma: move the arm64 wrappers to common code

Re: [PATCH 07/26] iommu/dma: move the arm64 wrappers to common code

[Jon Hunter]

On 29/04/2019 13:56, Robin Murphy wrote:
> On 22/04/2019 18:59, Christoph Hellwig wrote:
>> There is nothing really arm64 specific in the iommu_dma_ops
>> implementation, so move it to dma-iommu.c and keep a lot of symbols
>> self-contained.  Note the implementation does depend on the
>> DMA_DIRECT_REMAP infrastructure for now, so we'll have to make the
>> DMA_IOMMU support depend on it, but this will be relaxed soon.
> 
> Nothing looks objectionable, and boot testing with this much of the
> series merged has my coherent and non-coherent IOMMU-backed devices
> appearing to still work OK, so:
> 
> Acked-by: Robin Murphy <robin.murphy@arm.com>

Since next-20190529 one of our tests for MMC has started failing, where
the symptom is that the data written to the MMC does not match the
source. Bisecting this is pointing to this commit. Unfortunately, I am
not able to cleanly revert this on top of -next, but wanted to report
this if case you have any ideas.

Cheers
Jon

-- 
nvpublic

Re: [PATCH 07/26] iommu/dma: move the arm64 wrappers to common code

[Christoph Hellwig]

On Mon, Jun 03, 2019 at 08:47:57PM +0100, Jon Hunter wrote:
> Since next-20190529 one of our tests for MMC has started failing, where
> the symptom is that the data written to the MMC does not match the
> source. Bisecting this is pointing to this commit. Unfortunately, I am
> not able to cleanly revert this on top of -next, but wanted to report
> this if case you have any ideas.

Does this fix your problem?

https://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu.git/commit/?h=generic-dma-ops&id=1b961423158caaae49d3900b7c9c37477bbfa9b3

Re: [PATCH 07/26] iommu/dma: move the arm64 wrappers to common code

[Jon Hunter]

On 04/06/2019 07:05, Christoph Hellwig wrote:
> On Mon, Jun 03, 2019 at 08:47:57PM +0100, Jon Hunter wrote:
>> Since next-20190529 one of our tests for MMC has started failing, where
>> the symptom is that the data written to the MMC does not match the
>> source. Bisecting this is pointing to this commit. Unfortunately, I am
>> not able to cleanly revert this on top of -next, but wanted to report
>> this if case you have any ideas.
> 
> Does this fix your problem?
> 
> https://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu.git/commit/?h=generic-dma-ops&id=1b961423158caaae49d3900b7c9c37477bbfa9b3

Yes I can confirm with this patch on today's -next the issue is no
longer seen, and reverting this patch on top of today's -next causes the
problem to occur again. So yes this fixes my problem.

Thanks!
Jon

-- 
nvpublic