On Wed, Mar 27, 2013 at 08:14:58AM +0100, Andreas Degert wrote:
> 2013/3/26 Joerg Roedel <[email protected]>:
> > Okay, thanks for trying this out. I think I know where the problem is,
> > can you try the attached patch (without my debug-patch) and report me
> > whether it fixes the issue?
>
> Yes, works. What is the value of archdata.dma_ops before the assignment
> and where is it set? (my hack to set the global variable dma_ops seems to
> change that value.)
It is a little bit different. When the per-device dma_ops is not set the
global one will be used. Your fix set the global one which hides the
bug, that the per-device one was not set in this case.
Joerg
_______________________________________________
iommu mailing list
[email protected]
https://lists.linuxfoundation.org/mailman/listinfo/iommu
