Re: [PATCH v5 6/9] vfio/pci: Provide a user-facing name for BAR mappings

[Matt Evans <matt@ozlabs.org>]

Hi Alex,

On 29/07/2026 18:52, Alex Williamson wrote:
> On Wed, 15 Jul 2026 18:47:29 +0100
> Matt Evans <matt@ozlabs.org> wrote:
> 
>> Since converting BAR mmap()s to using DMABUFs, we lose the original
>> device path in /proc/<pid>/maps, lsof, etc.  Generate a debug-oriented
>> synthetic 'filename' based on the cdev, plus BDF, plus resource index.
>>
>> This applies only to BAR mappings via the VFIO device fd, as
>> explicitly-exported DMABUFs are named by userspace via the
>> DMA_BUF_SET_NAME ioctl.
>>
>> Signed-off-by: Matt Evans <matt@ozlabs.org>
>> Reviewed-by: Pranjal Shrivastava <praan@google.com>
>> ---
>>  drivers/vfio/pci/vfio_pci_dmabuf.c | 22 ++++++++++++++++++++--
>>  1 file changed, 20 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/vfio/pci/vfio_pci_dmabuf.c b/drivers/vfio/pci/vfio_pci_dmabuf.c
>> index 46b3cf5dc94b..51606ff3cdc6 100644
>> --- a/drivers/vfio/pci/vfio_pci_dmabuf.c
>> +++ b/drivers/vfio/pci/vfio_pci_dmabuf.c
>> @@ -4,6 +4,7 @@
>>  #include <linux/dma-buf-mapping.h>
>>  #include <linux/pci-p2pdma.h>
>>  #include <linux/dma-resv.h>
>> +#include <uapi/linux/dma-buf.h>
> 
> Is this required?  It seems unused.

No, good catch.  That originally pulled in DMA_BUF_NAME_LEN in the
previous version snprintf'ing here.  Removed.

>>  
>>  #include "vfio_pci_priv.h"
>>  
>> @@ -486,6 +487,7 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
>>  {
>>  	struct vfio_pci_dma_buf *priv;
>>  	unsigned long vma_pgoff = vma->vm_pgoff & (VFIO_PCI_OFFSET_MASK >> PAGE_SHIFT);
>> +	char *bufname;
>>  	int ret;
>>  
>>  	priv = kzalloc_obj(*priv);
>> @@ -498,6 +500,15 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
>>  		goto err_free_priv;
>>  	}
>>  
>> +	bufname = kasprintf(GFP_KERNEL, "%s:%s/%x",
>> +			    dev_name(&vdev->vdev.device), pci_name(vdev->pdev),
>> +			    res_index);
>> +
>> +	if (!bufname) {
>> +		ret = -ENOMEM;
>> +		goto err_free_phys;
>> +	}
>> +
>>  	/*
>>  	 * The DMABUF begins from the mmap()'s BAR offset, i.e. the
>>  	 * start of the VMA corresponds to byte 0 of the DMABUF and
>> @@ -516,7 +527,7 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
>>  	priv->provider = pcim_p2pdma_provider(vdev->pdev, res_index);
>>  	if (!priv->provider) {
>>  		ret = -EINVAL;
>> -		goto err_free_phys;
>> +		goto err_free_name;
>>  	}
>>  
>>  	priv->phys_vec[0].paddr = phys_start + ((u64)vma_pgoff << PAGE_SHIFT);
>> @@ -524,7 +535,7 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
>>  
>>  	ret = vfio_pci_dmabuf_export(vdev, priv, O_CLOEXEC | O_RDWR);
>>  	if (ret)
>> -		goto err_free_phys;
>> +		goto err_free_name;
>>  
>>  	/*
>>  	 * Ownership of the DMABUF file transfers to the VMA so that
>> @@ -539,8 +550,15 @@ int vfio_pci_core_mmap_prep_dmabuf(struct vfio_pci_core_device *vdev,
>>  	vma->vm_file = priv->dmabuf->file;
>>  	vma->vm_private_data = priv;
>>  
>> +	spin_lock(&priv->dmabuf->name_lock);
>> +	kfree(priv->dmabuf->name);
>> +	priv->dmabuf->name = bufname;
>> +	spin_unlock(&priv->dmabuf->name_lock);
> 
> This is simple, but shouldn't it really have a sanctioned, exported
> dmabuf helper rather than open coding this handling?  Thanks,

Hm, okay that's reasonable.  I've refactored dma_buf_set_name() and used
it from here (separating out a userspace-string-copying version for the
original ioctl usage).

Thanks for looking at the rest of v5 also.


Matt


> 
> Alex
> 
>> +
>>  	return 0;
>>  
>> +err_free_name:
>> +	kfree(bufname);
>>  err_free_phys:
>>  	kfree(priv->phys_vec);
>>  err_free_priv:
>