Hi Alex,

On 29/07/2026 18:52, Alex Williamson wrote:
> On Wed, 15 Jul 2026 18:47:26 +0100
> Matt Evans <[email protected]> wrote:
> 
>> Add vfio_pci_dma_buf_find_pfn(), which a VMA fault handler can use to
>> find a PFN.
>>
>> This supports multi-range DMABUFs, which typically would be used to
>> represent scattered spans but might even represent overlapping or
>> aliasing spans of PFNs.
>>
>> Because this is intended to be used in vfio_pci_core.c, we also need
>> to expose the struct vfio_pci_dma_buf in the vfio_pci_priv.h header.
> 
> Nit, it would be useful to mention the addition of vma_pgoff_adjust
> here and justification for it added here as a static zero versus later
> in the series where its value is updated.

Sure, good point.

>>
>> Signed-off-by: Matt Evans <[email protected]>
>> ---
>>  drivers/vfio/pci/vfio_pci_dmabuf.c | 153 ++++++++++++++++++++++++++---
>>  drivers/vfio/pci/vfio_pci_priv.h   |  20 ++++
>>  2 files changed, 160 insertions(+), 13 deletions(-)
>>
>> diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c 
>> b/drivers/vfio/pci/vfio_pci_dmabuf.c
>> index c16f460c01d6..7c047400dfd1 100644
>> --- a/drivers/vfio/pci/vfio_pci_dmabuf.c
>> +++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
>> @@ -9,19 +9,6 @@
>>  
>>  MODULE_IMPORT_NS("DMA_BUF");
>>  
>> -struct vfio_pci_dma_buf {
>> -    struct dma_buf *dmabuf;
>> -    struct vfio_pci_core_device *vdev;
>> -    struct list_head dmabufs_elm;
>> -    size_t size;
>> -    struct phys_vec *phys_vec;
>> -    struct p2pdma_provider *provider;
>> -    u32 nr_ranges;
>> -    struct kref kref;
>> -    struct completion comp;
>> -    u8 revoked : 1;
>> -};
>> -
>>  static int vfio_pci_dma_buf_attach(struct dma_buf *dmabuf,
>>                                 struct dma_buf_attachment *attachment)
>>  {
>> @@ -106,6 +93,146 @@ static const struct dma_buf_ops vfio_pci_dmabuf_ops = {
>>      .release = vfio_pci_dma_buf_release,
>>  };
>>  
>> +int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *priv,
>> +                          struct vm_area_struct *vma,
>> +                          unsigned long fault_addr,
>> +                          unsigned int order,
>> +                          unsigned long *out_pfn)
>> +{
>> +    /*
>> +     * Given a VMA (start, end, pgoffs) and a fault address,
>> +     * search the corresponding DMABUF's phys_vec[] to find the
>> +     * range representing the address's offset into the VMA, and
>> +     * its PFN.
>> +     *
>> +     * The phys_vec[] ranges represent contiguous spans of VAs
>> +     * upwards from the buffer offset 0; the actual PFNs might be
>> +     * in any order, overlap/alias, etc.  Calculate an offset of
>> +     * the desired page given VMA start/pgoff and address, then
>> +     * search upwards from 0 to find which span contains it.
>> +     *
>> +     * On success, a valid PFN for a page sized by 'order' is
>> +     * returned into out_pfn.
>> +     *
>> +     * Failure occurs if:
>> +     * - A hugepage would cross the edge of the VMA,
>> +     * - A hugepage isn't entirely contained within a range
>> +     *   (including where it straddles the boundary between
>> +     *   ranges),
>> +     * - We find a range, but the final PFN isn't aligned to the
>> +     *   requested order.
>> +     *
>> +     * Upon failure, -EAGAIN is returned and the caller is
>> +     * expected to try again with a smaller order, which will
>> +     * eventually succeed (order=0 will always work).
> 
> We typically use -EAGAIN to indicate a transient error where the caller
> should try again with the same args.  We're documenting here that
> -EAGAIN means to try again with different args.  Why wouldn't we use
> something like -ERANGE to indicate that?

-EAGAIN didn't seem all that inappropriate, but you're correct that it
isn't being used as a transient error here.  I'll change it to -ERANGE
as suggested.

>> +     *
>> +     * It's suboptimal if DMABUFs are created with neighbouring
>> +     * ranges that are physically contiguous, since hugepages
>> +     * can't straddle range boundaries.  (The construction of the
>> +     * ranges should merge them in this case.)
>> +     *
>> +     * Finally, vma_pgoff_adjust is used with a DMABUF created for
>> +     * a VFIO BAR mmap: a BAR mapped with vm_pgoff > 0 creates a
>> +     * DMABUF such that byte 0 of the VMA corresponds to byte 0 of
>> +     * the DMABUF and byte 'vm_pgoff << PAGE_SHIFT' into the BAR.
>> +     * To avoid double-offsetting in this scenario, subtracting
>> +     * vma_pgoff_adjust from this (non-zero) vm_pgoff generates
>> +     * the effective offset.
>> +     */
>> +
>> +    const unsigned long pagesize = PAGE_SIZE << order;
>> +    unsigned long vma_off = ((vma->vm_pgoff - priv->vma_pgoff_adjust) <<
>> +                             PAGE_SHIFT) & VFIO_PCI_OFFSET_MASK;
>> +    unsigned long rounded_page_addr = ALIGN_DOWN(fault_addr, pagesize);
>> +    unsigned long rounded_page_end = rounded_page_addr + pagesize;
>> +    unsigned long fault_offset;
>> +    unsigned long fault_offset_end;
>> +    unsigned long range_start_offset = 0;
>> +    unsigned int i;
>> +    int ret;
>> +
> 
> AIUI, find_pfn should only act on a live dmabuf, but we're relying on
> the caller to guarantee that it's live and not racing a revoke.  The
> dmabuf is live if !revoked under memory_lock.  Should this annotate and
> validate that requirement by adding a vdev arg so we can do:
> 
>       lockdep_assert_held(&vdev->memory_lock);
>       if (priv->revoked) // later updated to status
>               return -ENODEV;

Yes, it makes no sense to call this when the buffer's revoked, and it's
a good idea to both explicitly document/assert this.

I think your point regarding the extra vdev parameter -- as opposed to a
local 'vdev = priv->vdev' dereference -- is that it should also assert
that vdev is non-NULL, but one can't just read priv->vdev.  Dealing with
the races having only priv takes some careful steps (see
vfio_pci_mmap_huge_fault), so make it the caller's responsibility to
provide a vdev.  This function can both check it's non-NULL, and verify
it relates to priv, and then annotate vdev->memory_lock etc.

> This also removes one level of dereferencing in the below dev_warns,
> which is otherwise a bit suspect without the validation.  I realize the
> mmap path does the validation, but this is an exposed internal API, not
> a static function there.  Thanks,

[s/mmap path/the fault path/]  Yep, checks here make sense, thanks for
the suggestions.


Matt

> 
> Alex
> 
>> +    if (rounded_page_addr < vma->vm_start || rounded_page_end > 
>> vma->vm_end) {
>> +            if (order > 0)
>> +                    return -EAGAIN;
>> +
>> +            /* A fault address outside of the VMA is absurd. */
>> +            dev_warn_ratelimited(
>> +                    &priv->vdev->pdev->dev,
>> +                    "Fault addr 0x%lx outside VMA 0x%lx-0x%lx\n",
>> +                    fault_addr, vma->vm_start, vma->vm_end);
>> +            return -EFAULT;
>> +    }
>> +
>> +    /*
>> +     * fault_offset[_end] is the span within the DMABUF
>> +     * corresponding to the faulting page:
>> +     */
>> +    if (unlikely(check_add_overflow(rounded_page_addr - vma->vm_start,
>> +                                    vma_off, &fault_offset) ||
>> +                 check_add_overflow(fault_offset, pagesize,
>> +                                    &fault_offset_end)))
>> +            return -EFAULT;
>> +
>> +    /*
>> +     * Iterate over ranges in the buffer, summing their lengths:
>> +     * range_start_offset represents the current range's starting
>> +     * offset in the buffer (from 0 upwards).
>> +     *
>> +     * A failure for order == 0 is unexpected, and triggers a
>> +     * fault/warn.
>> +     */
>> +    ret = (order == 0) ? -EFAULT : -EAGAIN;
>> +
>> +    for (i = 0; i < priv->nr_ranges; i++) {
>> +            size_t range_len = priv->phys_vec[i].len;
>> +
>> +            /* Early exit if range starts after the page end */
>> +            if (fault_offset_end <= range_start_offset)
>> +                    break;
>> +
>> +            if (fault_offset >= range_start_offset &&
>> +                fault_offset_end <= range_start_offset + range_len) {
>> +                    /*
>> +                     * The faulting page is wholly contained
>> +                     * within the span represented by this range,
>> +                     * so validate PFN alignment for the order.
>> +                     * The if() condition ensures the pfn
>> +                     * arithmetic won't overflow.
>> +                     */
>> +                    unsigned long pfn =
>> +                            ((fault_offset - range_start_offset) +
>> +                             priv->phys_vec[i].paddr) >> PAGE_SHIFT;
>> +
>> +                    if (IS_ALIGNED(pfn, 1 << order)) {
>> +                            *out_pfn = pfn;
>> +                            ret = 0;
>> +                    }
>> +                    /*
>> +                     * Else order > 0; EAGAIN retries with smaller
>> +                     * order
>> +                     */
>> +                    break;
>> +            }
>> +            range_start_offset += range_len;
>> +    }
>> +
>> +    if (order == 0 && ret != 0)
>> +            /*
>> +             * The address fell outside of the span represented by
>> +             * the (concatenated) ranges.  As setup of a mapping
>> +             * ensures that the VMA is <= the total size of the
>> +             * ranges this should never happen.  If it does, warn
>> +             * and SIGBUS.
>> +             */
>> +            dev_warn_ratelimited(
>> +                    &priv->vdev->pdev->dev,
>> +                    "No range for addr 0x%lx, order %d: VMA 0x%lx-0x%lx 
>> pgoff 0x%lx, %u ranges, size 0x%zx\n",
>> +                    fault_addr, order, vma->vm_start, vma->vm_end,
>> +                    vma->vm_pgoff, priv->nr_ranges, priv->size);
>> +
>> +    return ret;
>> +}
>> +
>>  /*
>>   * This is a temporary "private interconnect" between VFIO DMABUF and 
>> iommufd.
>>   * It allows the two co-operating drivers to exchange the physical address 
>> of
>> diff --git a/drivers/vfio/pci/vfio_pci_priv.h 
>> b/drivers/vfio/pci/vfio_pci_priv.h
>> index fca9d0dfac90..c8f6f959056a 100644
>> --- a/drivers/vfio/pci/vfio_pci_priv.h
>> +++ b/drivers/vfio/pci/vfio_pci_priv.h
>> @@ -23,6 +23,20 @@ struct vfio_pci_ioeventfd {
>>      bool                    test_mem;
>>  };
>>  
>> +struct vfio_pci_dma_buf {
>> +    struct dma_buf *dmabuf;
>> +    struct vfio_pci_core_device *vdev;
>> +    struct list_head dmabufs_elm;
>> +    size_t size;
>> +    struct phys_vec *phys_vec;
>> +    struct p2pdma_provider *provider;
>> +    u32 nr_ranges;
>> +    struct kref kref;
>> +    struct completion comp;
>> +    unsigned long vma_pgoff_adjust;
>> +    u8 revoked : 1;
>> +};
>> +
>>  bool vfio_pci_intx_mask(struct vfio_pci_core_device *vdev);
>>  void vfio_pci_intx_unmask(struct vfio_pci_core_device *vdev);
>>  
>> @@ -114,6 +128,12 @@ static inline bool vfio_pci_is_vga(struct pci_dev *pdev)
>>      return (pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA;
>>  }
>>  
>> +int vfio_pci_dma_buf_find_pfn(struct vfio_pci_dma_buf *vpdmabuf,
>> +                          struct vm_area_struct *vma,
>> +                          unsigned long address,
>> +                          unsigned int order,
>> +                          unsigned long *out_pfn);
>> +
>>  #ifdef CONFIG_VFIO_PCI_DMABUF
>>  int vfio_pci_core_feature_dma_buf(struct vfio_pci_core_device *vdev, u32 
>> flags,
>>                                struct vfio_device_feature_dma_buf __user 
>> *arg,
> 

Reply via email to