On Sun, Jun 02, 2024 at 09:00:40PM -0600, Alex Williamson wrote: > On Sat, 1 Jun 2024 07:41:15 +0800 > Yan Zhao <yan.y.zhao@xxxxxxxxx> wrote: > > > On Wed, May 29, 2024 at 10:52:30PM -0600, Alex Williamson wrote: > > ... > > > diff --git a/drivers/vfio/group.c b/drivers/vfio/group.c > > > index 610a429c6191..ded364588d29 100644 > > > --- a/drivers/vfio/group.c > > > +++ b/drivers/vfio/group.c > > > @@ -286,6 +286,13 @@ static struct file *vfio_device_open_file(struct vfio_device *device) > > > */ > > > filep->f_mode |= (FMODE_PREAD | FMODE_PWRITE); > > > > > Instead of using anon_inode_getfile(), is it possible to get the filep like > > filep = alloc_file_pseudo_noaccount(device->inode, > > get_vfs_mount(),"[vfio-device]", O_RDWR, &vfio_device_fops); > > > > > + /* > > > + * Use the pseudo fs inode on the device to link all mmaps > > > + * to the same address space, allowing us to unmap all vmas > > > + * associated to this device using unmap_mapping_range(). > > > + */ > > > + filep->f_mapping = device->inode->i_mapping; > > Then this is not necessary here. > > Maybe? The code paths are not obviously equivalent to me, so I think > this adds risk to a series being proposed for the rc cycle. Would you > like to propose this as a change on top of this series for v6.11? > Thanks, Ok, let me have a try :)
