[dpdk-dev] [PATCH v3] vfio: fix sPAPR IOMMU DMA window size

Burakov, Anatoly anatoly.burakov at intel.com
Tue Aug 8 11:43:43 CEST 2017


> From: Jonas Pfefferle1 [mailto:JPF at zurich.ibm.com]
> Sent: Tuesday, August 8, 2017 10:30 AM
> To: Burakov, Anatoly <anatoly.burakov at intel.com>
> Cc: aik at ozlabs.ru; dev at dpdk.org
> Subject: RE: [PATCH v3] vfio: fix sPAPR IOMMU DMA window size
> 
> "Burakov, Anatoly" <anatoly.burakov at intel.com> wrote on 08/08/2017
> 11:15:24 AM:
> 
> > From: "Burakov, Anatoly" <anatoly.burakov at intel.com>
> > To: Jonas Pfefferle <jpf at zurich.ibm.com>
> > Cc: "dev at dpdk.org" <dev at dpdk.org>, "aik at ozlabs.ru" <aik at ozlabs.ru>
> > Date: 08/08/2017 11:18 AM
> > Subject: RE: [PATCH v3] vfio: fix sPAPR IOMMU DMA window size
> >
> > From: Jonas Pfefferle [mailto:jpf at zurich.ibm.com]
> > > Sent: Tuesday, August 8, 2017 9:41 AM
> > > To: Burakov, Anatoly <anatoly.burakov at intel.com>
> > > Cc: dev at dpdk.org; aik at ozlabs.ru; Jonas Pfefferle <jpf at zurich.ibm.com>
> > > Subject: [PATCH v3] vfio: fix sPAPR IOMMU DMA window size
> > >
> > > DMA window size needs to be big enough to span all memory segment's
> > > physical addresses. We do not need multiple levels of IOMMU tables
> > > as we already span ~70TB of physical memory with 16MB hugepages.
> > >
> > > Signed-off-by: Jonas Pfefferle <jpf at zurich.ibm.com>
> > > ---
> > > v2:
> > > * roundup to next power 2 function without loop.
> > >
> > > v3:
> > > * Replace roundup_next_pow2 with rte_align64pow2
> > >
> > >  lib/librte_eal/linuxapp/eal/eal_vfio.c | 13 ++++++++++---
> > >  1 file changed, 10 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/lib/librte_eal/linuxapp/eal/eal_vfio.c
> > > b/lib/librte_eal/linuxapp/eal/eal_vfio.c
> > > index 946df7e..550c41c 100644
> > > --- a/lib/librte_eal/linuxapp/eal/eal_vfio.c
> > > +++ b/lib/librte_eal/linuxapp/eal/eal_vfio.c
> > > @@ -759,10 +759,12 @@ vfio_spapr_dma_map(int vfio_container_fd)
> > >        return -1;
> > >     }
> > >
> > > -   /* calculate window size based on number of hugepages configured
> > > */
> > > -   create.window_size = rte_eal_get_physmem_size();
> > > +   /* physicaly pages are sorted descending i.e. ms[0].phys_addr is max
> > > */
> >
> > Do we always expect that to be the case in the future? Maybe it
> > would be safer to walk the memsegs list.
> >
> > Thanks,
> > Anatoly
> 
> I had this loop in before but removed it in favor of simplicity.
> If we believe that the ordering is going to change in the future
> I'm happy to bring back the loop. Is there other code which is
> relying on the fact that the memsegs are sorted by their physical
> addresses?

I don't think there is. In any case, I think making assumptions about particulars of memseg organization is not a very good practice.

I seem to recall us doing similar things in other places, so maybe down the line we could introduce a new API (or internal-only) function to get a memseg with min/max address. For now I think a loop will do. 

> 
> >
> > > +   /* create DMA window from 0 to max(phys_addr + len) */
> > > +   /* sPAPR requires window size to be a power of 2 */
> > > +   create.window_size = rte_align64pow2(ms[0].phys_addr +
> > > ms[0].len);
> > >     create.page_shift = __builtin_ctzll(ms->hugepage_sz);
> > > -   create.levels = 2;
> > > +   create.levels = 1;
> > >
> > >     ret = ioctl(vfio_container_fd, VFIO_IOMMU_SPAPR_TCE_CREATE,
> > > &create);
> > >     if (ret) {
> > > @@ -771,6 +773,11 @@ vfio_spapr_dma_map(int vfio_container_fd)
> > >        return -1;
> > >     }
> > >
> > > +   if (create.start_addr != 0) {
> > > +      RTE_LOG(ERR, EAL, "  DMA window start address != 0\n");
> > > +      return -1;
> > > +   }
> > > +
> > >     /* map all DPDK segments for DMA. use 1:1 PA to IOVA mapping */
> > >     for (i = 0; i < RTE_MAX_MEMSEG; i++) {
> > >        struct vfio_iommu_type1_dma_map dma_map;
> > > --
> > > 2.7.4
> >


More information about the dev mailing list