On Fri, Nov 09, 2018 at 02:26:30PM -0800, Andrew Morton wrote:
> On Sun, 14 Oct 2018 13:35:58 -0300 Ernesto A. Fernández <ernesto.mnd.fernandez@xxxxxxxxx> wrote:
>
> > The immutable, append-only and no-dump attributes can only be retrieved
> > with an ioctl; implement the ->getattr() method to return them on statx.
> > Do not return the inode birthtime yet, because the issue of how best to
> > handle the post-2038 timestamps is still under discussion.
> >
> > This patch is needed to pass xfstests generic/424.
>
> It's been a while since I looked into such things...
>
> > --- a/fs/hfsplus/inode.c
> > +++ b/fs/hfsplus/inode.c
> > @@ -270,6 +270,26 @@ static int hfsplus_setattr(struct dentry *dentry, struct iattr *attr)
> > return 0;
> > }
> >
> > +int hfsplus_getattr(const struct path *path, struct kstat *stat,
> > + u32 request_mask, unsigned int query_flags)
> > +{
> > + struct inode *inode = d_inode(path->dentry);
> > + struct hfsplus_inode_info *hip = HFSPLUS_I(inode);
> > +
> > + if (inode->i_flags & S_APPEND)
> > + stat->attributes |= STATX_ATTR_APPEND;
> > + if (inode->i_flags & S_IMMUTABLE)
> > + stat->attributes |= STATX_ATTR_IMMUTABLE;
>
> ext4_getattr() inspects the underlying ext4_inode's flags to determine
> the above. But here hfs is looking at the vfs-level inode. Which is
> correct, which is best, do we know why they differ, etc?
ext4 is not reading the flags from the ext4_inode directly, it reads them
from ext4_inode_info. hfsplus_inode_info doesn't have that information
anymore, since 722c55d13e72 ("hfsplus: remove superflous rootflags field in
hfsplus_inode_info").
My intention here was to follow what the FS_IOC_GETFLAGS ioctl is doing, so
I just copied the checks from hfsplus_ioctl_getflags() at ioctl.c.
>
> > + if (hip->userflags & HFSPLUS_FLG_NODUMP)
> > + stat->attributes |= STATX_ATTR_NODUMP;
> > +
> > + stat->attributes_mask |= STATX_ATTR_APPEND | STATX_ATTR_IMMUTABLE |
> > + STATX_ATTR_NODUMP;
> > +
> > + generic_fillattr(inode, stat);
> > + return 0;
> > +}
>
>
