[PATCH v2 0/8] io-pgtable lock removal

On 26/06/17 12:35, John Garry wrote:

On 23/06/2017 10:58, Robin Murphy wrote:

quoted

On 23/06/17 09:47, John Garry wrote:

quoted

On 22/06/2017 16:53, Robin Murphy wrote:

quoted

The feedback has been promising, so v2 is just a final update to cover
a handful of memory ordering and cosmetic tweaks that came up when Will
and I went through this offline.

Thanks,
Robin.

Hi Robin,

Is it worth us retesting this patchset?

If yes, as for a branch (for convenience), does this one now match v2:
git://linux-arm.org/linux-rm  iommu/pgtable

I saw it was rewritten recently.

Indeed, I pushed the branch out before posting. Functionally it ought to
be identical to v1, so this was more just a heads-up - feel free to
double-check I've not broken anything, but I wouldn't say you need to
throw the full test lab at it again.

I saw Will has already sent the pull request. But, FWIW, we are seeing
roughly the same performance as v1 patchset. For PCI NIC, Zhou again
found performance drop goes from ~15->8% with SMMU enabled, and for
integrated storage controller [platform device], we still see a drop of
about 50%, depending on datarates (Leizhen has been working on fixing
this).

Thanks for confirming. Following Joerg's suggestion that the storage
workloads may still depend on rbtree performance - it had slipped my
mind that even with small block sizes those could well be grouped into
scatterlists large enough to trigger a >64-page IOVA allocation - I've
taken the liberty of cooking up a simplified version of Leizhen's rbtree
optimisation series in the iommu/iova branch of my tree. I'll follow up
on that after the merge window, but if anyone wants to play with it in
the meantime feel free.

Robin.