> -----Original Message----- > From: Wood Scott-B07421 > Sent: Tuesday, February 19, 2013 9:39 AM > To: Bhushan Bharat-R65777 > Cc: Alex Williamson; Yoder Stuart-B08248; Wood Scott-B07421; Graf Alexander- > B36701; qemu-devel@nongnu.org; qemu-...@nongnu.org > Subject: Re: VFIO: Not require to make VFIO_IOMMU_MAP_DMA for MMAPED PCI bars > > On 02/18/2013 10:05:20 PM, Bhushan Bharat-R65777 wrote: > > > > > > > -----Original Message----- > > > From: Alex Williamson [mailto:alex.william...@redhat.com] > > > Sent: Thursday, February 14, 2013 11:57 PM > > > To: Bhushan Bharat-R65777 > > > Cc: Yoder Stuart-B08248; Wood Scott-B07421; Graf Alexander-B36701; > > qemu- > > > de...@nongnu.org; qemu-...@nongnu.org > > > Subject: Re: VFIO: Not require to make VFIO_IOMMU_MAP_DMA for > > MMAPED PCI bars > > > > > > On Thu, 2013-02-14 at 08:22 +0000, Bhushan Bharat-R65777 wrote: > > > > +static bool vfio_listener_skipped_section(VFIOContainer > > *container, > > > > + MemoryRegionSection > > > > +*section) { > > > > + if (!memory_region_is_ram(section->mr)) > > > > + return 1; > > > > > > Need some kind of conditional here for platforms that support > > peer-to-peer. > > > Thanks, > > > > What about adding a iommu_type will help here (VFIO_TYPE2_IOMMU), > > which does not require BARs to be mapped in iommu ? > > What specifically does "type 2" mean again? If we're going to use it as the > thing to key off of for all sorts of PAMU quirks, might as well just call it > VFIO_IOMMU_FSL_PAMU.
Just thought of a common name so in case some other IOMMU also fall in this category and add a comment where this will be defined. I am fine with this name as well :) Thanks -Bharat
