Anshuman Khandual <khandual@xxxxxxxxxxxxxxxxxx> writes: > On 04/16/2018 07:28 PM, David Miller wrote: >> From: Anshuman Khandual <khandual@xxxxxxxxxxxxxxxxxx> >> Date: Mon, 16 Apr 2018 14:26:07 +0530 >> >>> On 04/15/2018 08:29 PM, Christoph Hellwig wrote: >>>> This code is only used by sparc, and all new iommu drivers should use the >>>> drivers/iommu/ framework. Also remove the unused exports. >>>> >>>> Signed-off-by: Christoph Hellwig <hch@xxxxxx> >>> >>> Right, these functions are used only from SPARC architecture. Simple >>> git grep confirms it as well. Hence it makes sense to move them into >>> arch code instead. >> >> Well, we put these into a common location and used type friendly for >> powerpc because we hoped powerpc would convert over to using this >> common piece of code as well. >> >> But nobody did the powerpc work. Sorry. >> If you look at the powerpc iommu support, it's the same code basically >> for entry allocation. > > I understand. But there are some differences in iommu_table structure, > how both regular and large IOMMU pools are being initialized etc. So > if the movement of code into SPARC help cleaning up these generic config > options in general, I guess we should do that. But I will leave it upto > others who have more experience in this area. > > +mpe This is the first I've heard of it, I guess it's probably somewhere on Ben's append-only TODO list. Some of the code does look very similar, but not 100%. So someone would need to do some work to reconcile the two and test the result. TBH I doubt we're going to get around to it any time soon. Unless we have a volunteer? cheers
