Re: [RFC PATCH] overlayfs: add OVL_IOC_GETINFOFD ioctl that opens ovlinfofd

[Amir Goldstein]

On Mon, Oct 5, 2020 at 8:13 PM Alexander Mikhalitsyn
<alexander.mikhalitsyn@xxxxxxxxxxxxx> wrote:
>
> On Mon, 5 Oct 2020 10:08:42 -0700
> Randy Dunlap <rdunlap@xxxxxxxxxxxxx> wrote:
>
> > On 10/5/20 10:02 AM, Alexander Mikhalitsyn wrote:
> > >  #define    OVL_IOC_GETLWRFHNDLSNUM                 _IO('o', 1)
> > >  // DISCUSS: what if MAX_HANDLE_SZ will change?
> > >  #define    OVL_IOC_GETLWRFHNDL                     _IOR('o', 2, struct ovl_mnt_opt_fh)
> > >  #define    OVL_IOC_GETUPPRFHNDL                    _IOR('o', 3, struct ovl_mnt_opt_fh)
> > >  #define    OVL_IOC_GETWRKFHNDL                     _IOR('o', 4, struct ovl_mnt_opt_fh)
> > > +#define    OVL_IOC_GETINFOFD                       _IO('o', 5)
> >
> > Hi,
> >
> > Quoting (repeating) from
> > https://lore.kernel.org/lkml/9cd0e9d1-f124-3f2d-86e6-e6e96a1ccb1e@xxxxxxxxxxxxx/:
> >
> > This needs to have Documentation/userspace-api/ioctl/ioctl-number.rst
> > updated also.
> >
> > ...
> >
> > Are you waiting until it's past RFC stage?
> >
> > thanks.
> > --
> > ~Randy
> >
>
> Hi,
>
> thank you! I will prepare this change too when we
> decide which ioctls to add. ;)
>

Or... don't do ioctls and avoid the ABI headaches.
If you are going to expose a seqfile I think it would be much prefered
to do a /sys/fs/overlay/instance/<device minor>/layes file.

But that's just my opinion.

Thanks,
Amir.