Re: Cache Invalidation Solution for Nested IOMMU

[Jason Gunthorpe <jgg@nvidia.com>]

On Tue, Apr 11, 2023 at 10:07:40AM +0100, Jean-Philippe Brucker wrote:
> On Sun, Apr 09, 2023 at 06:08:25PM -0700, Nicolin Chen wrote:
> > On Thu, Apr 06, 2023 at 08:40:04AM -0300, Jason Gunthorpe wrote:
> > > On Thu, Apr 06, 2023 at 02:23:17PM +0800, Zhangfei Gao wrote:
> > > 
> > > > We are using ioctl method now.
> > > > From the testing, the TLB miss impacts performance a lot, so we use
> > > > huge page method.
> > > > After using huge page method, guest can achieve comparable performance
> > > > with host.
> > > 
> > > Looks like these tests are not stressing the MM, just measuring pure
> > > BW of the DMA, so they don't get into the invalidation regime..
> > > 
> > > You need to measure a more real application that is actually using the
> > > MM (eg alloc/free memory, fork, etc) while it operates and turn on SVA.
> > 
> > Would an iommu map/unmap benchmark test be useful here?
> 
> You can reuse and improve dma_map_benchmark, already upstream. It supports
> multi-threads and reports stddev. For some comparisons the report
> resolution (.1 us) is insufficient and needs to be increased, but for
> comparing guest performance it might be alright.
> 
> https://lore.kernel.org/linux-iommu/20201102080646.2180-2-song.bao.hua@hisilicon.com/

I'm not talking about kernel dma_map/unmap, I'm talking about MM
activities like mmap and munmap that become slower once SVA is enabled

Jason