On Mon, 2026-01-19 at 09:52 +0200, Leon Romanovsky wrote:
> On Sun, Jan 18, 2026 at 03:16:25PM +0100, Thomas Hellström wrote:
> > Hi, Leon,
> > 
> > On Sun, 2026-01-18 at 14:08 +0200, Leon Romanovsky wrote:
> > > Changelog:
> > > v2:
> > >  * Changed series to document the revoke semantics instead of
> > >    implementing it.
> > > v1:
> > > https://patch.msgid.link/[email protected]
> > > 
> > > -----------------------------------------------------------------
> > > ----
> > > ----
> > > This series documents a dma-buf “revoke” mechanism: to allow a
> > > dma-
> > > buf
> > > exporter to explicitly invalidate (“kill”) a shared buffer after
> > > it
> > > has
> > > been distributed to importers, so that further CPU and device
> > > access
> > > is
> > > prevented and importers reliably observe failure.
> > > 
> > > The change in this series is to properly document and use
> > > existing
> > > core
> > > “revoked” state on the dma-buf object and a corresponding
> > > exporter-
> > > triggered
> > > revoke operation. Once a dma-buf is revoked, new access paths are
> > > blocked so
> > > that attempts to DMA-map, vmap, or mmap the buffer fail in a
> > > consistent way.
> > 
> > This sounds like it does not match how many GPU-drivers use the
> > move_notify() callback.
> 
> No change for them.
> 
> > 
> > move_notify() would typically invalidate any device maps and any
> > asynchronous part of that invalidation would be complete when the
> > dma-
> > buf's reservation object becomes idle WRT DMA_RESV_USAGE_BOOKKEEP
> > fences.
> 
> This part has not changed and remains the same for the revocation
> flow as well.
> 
> > 
> > However, the importer could, after obtaining the resv lock, obtain
> > a
> > new map using dma_buf_map_attachment(), and I'd assume the CPU maps
> > work in the same way, I.E. move_notify() does not *permanently*
> > revoke
> > importer access.
> 
> This part diverges by design and is documented to match revoke
> semantics.  
> It defines what must occur after the exporter requests that the
> buffer be  
> "killed". An importer that follows revoke semantics will not attempt
> to call  
> dma_buf_map_attachment(), and the exporter will block any remapping
> attempts  
> regardless. See the priv->revoked flag in the VFIO exporter.
> 
> In addition, in this email thread, Christian explains that revoke
> semantics already exists, with the combination of dma_buf_pin and
> dma_buf_move_notify, just not documented:
> https://lore.kernel.org/all/[email protected]/


Hmm,

Considering 

https://elixir.bootlin.com/linux/v6.19-rc5/source/drivers/infiniband/core/umem_dmabuf.c#L192

this sounds like it's not just undocumented but also in some cases
unimplemented. The xe driver for one doesn't expect move_notify() to be
called on pinned buffers, so if that is indeed going to be part of the
dma-buf protocol,  wouldn't support for that need to be advertised by
the importer?

Thanks,
Thomas

> 
> Thanks
> 
> > 
> > /Thomas
> > 
> > 
> > > 
> > > Thanks
> > > 
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > Cc: [email protected]
> > > To: Sumit Semwal <[email protected]>
> > > To: Christian König <[email protected]>
> > > To: Alex Deucher <[email protected]>
> > > To: David Airlie <[email protected]>
> > > To: Simona Vetter <[email protected]>
> > > To: Gerd Hoffmann <[email protected]>
> > > To: Dmitry Osipenko <[email protected]>
> > > To: Gurchetan Singh <[email protected]>
> > > To: Chia-I Wu <[email protected]>
> > > To: Maarten Lankhorst <[email protected]>
> > > To: Maxime Ripard <[email protected]>
> > > To: Thomas Zimmermann <[email protected]>
> > > To: Lucas De Marchi <[email protected]>
> > > To: Thomas Hellström <[email protected]>
> > > To: Rodrigo Vivi <[email protected]>
> > > To: Jason Gunthorpe <[email protected]>
> > > To: Leon Romanovsky <[email protected]>
> > > To: Kevin Tian <[email protected]>
> > > To: Joerg Roedel <[email protected]>
> > > To: Will Deacon <[email protected]>
> > > To: Robin Murphy <[email protected]>
> > > To: Alex Williamson <[email protected]>
> > > 
> > > ---
> > > Leon Romanovsky (4):
> > >       dma-buf: Rename .move_notify() callback to a clearer
> > > identifier
> > >       dma-buf: Document revoke semantics
> > >       iommufd: Require DMABUF revoke semantics
> > >       vfio: Add pinned interface to perform revoke semantics
> > > 
> > >  drivers/dma-buf/dma-buf.c                   |  6 +++---
> > >  drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c |  4 ++--
> > >  drivers/gpu/drm/virtio/virtgpu_prime.c      |  2 +-
> > >  drivers/gpu/drm/xe/tests/xe_dma_buf.c       |  6 +++---
> > >  drivers/gpu/drm/xe/xe_dma_buf.c             |  2 +-
> > >  drivers/infiniband/core/umem_dmabuf.c       |  4 ++--
> > >  drivers/infiniband/hw/mlx5/mr.c             |  2 +-
> > >  drivers/iommu/iommufd/pages.c               | 11 +++++++++--
> > >  drivers/vfio/pci/vfio_pci_dmabuf.c          | 16
> > > ++++++++++++++++
> > >  include/linux/dma-buf.h                     | 25
> > > ++++++++++++++++++++++---
> > >  10 files changed, 60 insertions(+), 18 deletions(-)
> > > ---
> > > base-commit: 9ace4753a5202b02191d54e9fdf7f9e3d02b85eb
> > > change-id: 20251221-dmabuf-revoke-b90ef16e4236
> > > 
> > > Best regards,
> > > --  
> > > Leon Romanovsky <[email protected]>
> > > 
> > 

Reply via email to