Message ID | 20240319-setlease-v1-1-5997d67e04b3@kernel.org (mailing list archive) |
---|---|
State | New |
Headers | show |
Series | vboxsf: explicitly deny setlease attempts | expand |
Hi, On 3/19/24 5:32 PM, Jeff Layton wrote: > vboxsf does not break leases on its own, so it can't properly handle the > case where the hypervisor changes the data. Don't allow file leases on > vboxsf. > > Signed-off-by: Jeff Layton <jlayton@kernel.org> Thanks, patch looks good to me: Reviewed-by: Hans de Goede <hdegoede@redhat.com> I have added this to my local vboxsf branch now and I'll send out a pull-request with this and a couple of other vboxsf fixes soon. Regards, Hans > --- > Looking over the comments in the code around cache coherency, it seems > like it ought to deny file locks as well? We could add a stub ->lock > routine that just returns -ENOLCK or something. > --- > fs/vboxsf/file.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/fs/vboxsf/file.c b/fs/vboxsf/file.c > index 2307f8037efc..118dedef8ebe 100644 > --- a/fs/vboxsf/file.c > +++ b/fs/vboxsf/file.c > @@ -218,6 +218,7 @@ const struct file_operations vboxsf_reg_fops = { > .release = vboxsf_file_release, > .fsync = noop_fsync, > .splice_read = filemap_splice_read, > + .setlease = simple_nosetlease, > }; > > const struct inode_operations vboxsf_reg_iops = { > > --- > base-commit: 0a7b0acecea273c8816f4f5b0e189989470404cf > change-id: 20240319-setlease-ce31fb8777b0 > > Best regards,
diff --git a/fs/vboxsf/file.c b/fs/vboxsf/file.c index 2307f8037efc..118dedef8ebe 100644 --- a/fs/vboxsf/file.c +++ b/fs/vboxsf/file.c @@ -218,6 +218,7 @@ const struct file_operations vboxsf_reg_fops = { .release = vboxsf_file_release, .fsync = noop_fsync, .splice_read = filemap_splice_read, + .setlease = simple_nosetlease, }; const struct inode_operations vboxsf_reg_iops = {
vboxsf does not break leases on its own, so it can't properly handle the case where the hypervisor changes the data. Don't allow file leases on vboxsf. Signed-off-by: Jeff Layton <jlayton@kernel.org> --- Looking over the comments in the code around cache coherency, it seems like it ought to deny file locks as well? We could add a stub ->lock routine that just returns -ENOLCK or something. --- fs/vboxsf/file.c | 1 + 1 file changed, 1 insertion(+) --- base-commit: 0a7b0acecea273c8816f4f5b0e189989470404cf change-id: 20240319-setlease-ce31fb8777b0 Best regards,