Re: [PATCH v1 4/4] iommu/tegra: gart: Optimize map/unmap

From: Dmitry Osipenko
Date: Mon May 07 2018 - 11:47:03 EST


On 07.05.2018 10:59, Joerg Roedel wrote:
> On Fri, Apr 27, 2018 at 12:02:02PM +0200, Thierry Reding wrote:
>> Joerg, I've gone over the git log and header files and I see no mention
>> of why the TLB flush interface isn't used for mapping. Do you recall any
>> special reasons why the same shouldn't be applied for mapping? Would you
>> accept any patches doing this?
>
> Yeah, the reason was that we assumed that only emulated IOMMUs ever need
> flushing in the mapping path, and there is no reason to optimize for
> that. Especially when a call to iotlb_sync() in the mapping path would
> hurt most other users.
>
> Does the tegra-gart actually need it too?

Tegra-GART is exactly an emulated IOMMU, it doesn't have anything like TLB and
it's simply a remapping table. What is actually needed for the GART - is to make
sure that the remapping table modifications reach GART HW before anything tries
to touch the modified page entries. Presumable there is some kind of a HW buffer
that accumulates multiple registers writes and issues them in bursts to improve
performance, read-after-write is the way to flush that buffer.