On Fri, Oct 11, 2024 at 4:25 PM Miklos Szeredi <miklos@xxxxxxxxxx> wrote:
>
> On Fri, 11 Oct 2024 at 15:53, Amir Goldstein <amir73il@xxxxxxxxx> wrote:
>
> > @@ -20,9 +20,18 @@ static void fuse_file_accessed(struct file *file)
> >
> > static void fuse_file_modified(struct file *file)
> > {
> > + struct fuse_file *ff = file->private_data;
> > + struct file *backing_file = fuse_file_passthrough(ff);
> > struct inode *inode = file_inode(file);
> > -
> > - fuse_invalidate_attr_mask(inode, FUSE_STATX_MODSIZE);
> > + loff_t size = i_size_read(file_inode(backing_file));
>
> What about passing iocb and res to ->end_write() instead of just the
> file, so that we don't need to touch the underlying inode?
>
I considered that. It was like this in one of my older versions.
But why do we want to avoid copying attributes from the underlying inode?
If anything, I thought that we would want to get closer to ovl_file_modified()
for backing inodes in some situations like this one.
I understand that brute copy of attributes is a problem, but I don't see the
problem with i_size = max(i_size, i_backing_size)
Can you explain the problem?
Thanks,
Amir.
