Re: [PATCH 5/7] swiotlb: (re)Create swiotlb_unmap_single

[Kumar Gala <galak@kernel.crashing.org>]

On Apr 8, 2009, at 7:43 AM, Ingo Molnar wrote:

>
> * FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp> wrote:
>
>>> Same here.  It makes it easier for us to work on the powerpc arch
>>> specific changes for .31 if we can get these into .30.  What are you
>>> looking at for .31?
>>
>> I need to finish the dma_mapping_ops cleanups and cross-arch
>> unification stuff:
>>
>> http://marc.info/?l=linux-kernel&m=123827903216314&w=2
>>
>> But the changes to swiotlb is minor. BTW, I have the patches for
>> powerpc too.
>>
>>> Ingo, any comments on that?
>
> if it's genuine fixes we can do it post-rc1, but it looks a tad wide
> for that:
>
> arch/x86/kernel/pci-swiotlb.c |    2 +-
> include/linux/swiotlb.h       |    3 +-
> lib/swiotlb.c                 |  115 +++++++++++++++++++++ 
> +-------------------
>
> the patches came right in the merge window - that's too late for
> IOMMU bits, the patch cutoff is generally the beginning of the merge
> window. (Which was on March 23 in this window)

Fair.

> But i can queue them up in the .31 queue if it has all the Acks from
> you folks. Becky, mind resending the latest version, with all the
> acks in place?

I can do this in Becky's abscense this week.  The only "acks" I've  
seen are from FUJITA Tomonori.

Can we ask there be a unique swiotlb branch in your tree to make it  
easier for us to continue additional swiotlb work as well as arch  
specific work.

- k