lkml.org 
[lkml]   [2018]   [Nov]   [30]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v3 1/1] drm: msm: Replace dma_map_sg with dma_sync_sg*
On Thu, Nov 29, 2018 at 01:57:38PM -0500, Rob Clark wrote:
> On Thu, Nov 29, 2018 at 12:24 PM Tomasz Figa <tfiga@chromium.org> wrote:
> >
> > [CC Marek]
> >
> > On Thu, Nov 29, 2018 at 9:09 AM Daniel Vetter <daniel@ffwll.ch> wrote:
> > >
> > > On Thu, Nov 29, 2018 at 5:57 PM Christoph Hellwig <hch@lst.de> wrote:
> > > >
> > > > Note that one thing I'd like to avoid is exposing these funtions directly
> > > > to drivers, as that will get us into all kinds of abuses.
> > >
> > > What kind of abuse do you expect? It could very well be that gpu folks
> > > call that "standard use case" ... At least on x86 with the i915 driver
> > > we pretty much rely on architectural guarantees for how cache flushes
> > > work very much. Down to userspace doing the cache flushing for
> > > mappings the kernel has set up.
> >
> > i915 is a very specific case of a fully contained,
> > architecture-specific hardware subsystem, where you can just hardcode
> > all integration details inside the driver, because nobody else would
> > care.
> >
> > In ARM world, you can have the same IP blocks licensed by multiple SoC
> > vendors with different integration details and that often includes the
> > option of coherency.
>
> fwiw, I believe all the GPU IP blocks that are used across multiple
> SoCs have their own GPU MMU (potentially in addition to an iommu?).
> So the dma-api is a much better fit for them.. drm/msm is a lot
> closer to drm/i915 scenario, so I don't so much care if the solution
> to our unique problem isn't something that would work for other
> drivers ;-)

Right now maybe, but I fully except the entire coherent vs. non-coherent
transactions hilarity that we have on the bigger intel socs since a few
years already to trickle down into smaller (arm based socs) eventually. I
think Apple is already there since a few generations.

So maybe we won't have to fight the iommu side of the dma-api anymore
on these, but we'll still have to fight the cache maintenance side of
dma-api. You can tell the dma-api to not flush, but then you don't have
any other way to actually flush that's acceptable for arch/arm (on x86 we
just run clflush in userspace and call it a day).
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch

\
 
 \ /
  Last update: 2018-11-30 10:40    [W:0.092 / U:0.616 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site