On Fri, Sep 4, 2026 at 5:38 AM Cédric Le Goater <[email protected]> wrote:
>
> Hi Naman,
>
> Thanks for the series. However, the multiple FDs are a vfio-user
> transport detail. The core VFIO only needs to know "which fd to
> mmap for this subregion" and it doesn't need to manage arrays of
> FDs per region. So, I'd prefer :
Thanks for the review Cedric. That makes sense to me, I'll refactor accordingly.
>
> 1. No change.
>
> 2. Add 'fd' and 'fd_offset' to VFIOMmap :
>
> typedef struct VFIOMmap {
> ...
> + int fd; /* per-mmap fd, or -1 to use the region fd */
> + uint64_t fd_offset; /* offset within fd for the mmap call */
> } VFIOMmap;
>
> This is the only core VFIO change.
>
> 3. Add a 'setup_sparse_mmaps' hook to VFIODeviceIOOps
>
> 4. vfio-user: support multiple FDs for sparse mmap regions
> All multi-fd logic stays in hw/vfio-user/device.c
>
> Thanks,
>
> C.
>
>
> On 9/3/26 19:46, Naman Gulati wrote:
> > Gentle ping on this series incase it got lost in the noise :)
> >
> > On Tue, Aug 25, 2026 at 10:00 PM Naman Gulati <[email protected]>
> > wrote:
> >>
> >> This series extends the vfio-user protocol specification to allow
> >> servers to pass multiple file descriptors for different sparse memory
> >> areas to be mmap'ed within a single device region.
> >>
> >> This extension is intended for exposing paravirtualized PCI devices via
> >> a vfio-user server that may have multiple unique sub-regions in a BAR
> >> backed by some mmap'able buffers sourced from distinct file descriptors.
> >>
> >> This change introduces a new capability:
> >> VFIO_REGION_INFO_CAP_SPARSE_MMAP_FDS
> >>
> >> Unlike VFIO_REGION_INFO_CAP_SPARSE_MMAP (which maps all sparse areas
> >> from a single open file descriptor), this new capability defines an
> >> array of sparse mmap areas where each area contains an index into the
> >> array of file descriptors.
> >>
> >> When all sparse areas share a single file descriptor, servers may
> >> continue to emit VFIO_REGION_INFO_CAP_SPARSE_MMAP and send exactly one
> >> file descriptor in SCM_RIGHTS for backwards compatibility with older
> >> clients.
> >>
> >> The extensions in this series also lays the groundwork to make use of
> >> the in-development series [1] extending vfio-pci to return distinct
> >> dmabuf FDs for discrete regions of a device's BAR. A vfio-user server
> >> may choose to return dmabuf FDs for various regions of a device's BAR to
> >> map a vfio passthrough device into the guest, in order to retain control
> >> of the device and be able to revoke mappings passed to the VM.
> >>
> >> libvfio-user server side changes WIP on github [2].
> >>
> >>
> >> [1] https://lore.kernel.org/all/[email protected]
> >> [2] https://github.com/NamanGulati/libvfio-user/tree/multi-region-fd
> >>
> >> ---
> >> Changes from RFC:
> >> - Refactored the get_region_info callback to keep vbasedev constant
> >> and take in a VFIORegionFDs* out parameter to store the received fds.
> >> - Link to RFC:
> >> https://lore.kernel.org/qemu-devel/[email protected]
> >> ---
> >>
> >> Naman Gulati (3):
> >> vfio-user: add multi-fd region mmap capability
> >> vfio: mmap sparse regions backed by multiple FDs
> >> vfio-user: implement client support for multi-fd region mmap
> >>
> >> docs/interop/vfio-user.rst | 71 +++++++++++++++++++++++++++++++++--
> >> hw/vfio-user/device.c | 66 ++++++++++++++++++++++++++++++--
> >> hw/vfio-user/protocol.h | 20 ++++++++++
> >> hw/vfio/device.c | 49 ++++++++++++++----------
> >> hw/vfio/region.c | 41 ++++++++++++++++----
> >> hw/vfio/vfio-region.h | 8 ++++
> >> include/hw/vfio/vfio-device.h | 32 ++++++++++++++--
> >> 7 files changed, 249 insertions(+), 38 deletions(-)
> >>
> >> --
> >> 2.55.0.860.g4b6b3295ed-goog
> >>
> >
>
