Re: [PATCH v5 03/18] gfs2: add compat_ioctl support

["Andreas Grünbacher" <andreas.gruenbacher@gmail.com>]

Am So., 18. Aug. 2019 um 21:32 Uhr schrieb Arnd Bergmann <arnd@arndb.de>:
> On Fri, Aug 16, 2019 at 7:32 PM Andreas Gruenbacher <agruenba@redhat.com> wrote:
> > On Wed, Aug 14, 2019 at 10:45 PM Arnd Bergmann <arnd@arndb.de> wrote:
> > > +       /* These are just misnamed, they actually get/put from/to user an int */
> > > +       switch(cmd) {
> > > +       case FS_IOC32_GETFLAGS:
> > > +               cmd = FS_IOC_GETFLAGS;
> > > +               break;
> > > +       case FS_IOC32_SETFLAGS:
> > > +               cmd = FS_IOC_SETFLAGS;
> > > +               break;
> >
> > I'd like the code to be more explicit here:
> >
> >         case FITRIM:
> >         case FS_IOC_GETFSLABEL:
> >               break;
> >         default:
> >               return -ENOIOCTLCMD;
>
> I've looked at it again: if we do this, the function actually becomes
> longer than the native gfs2_ioctl(). Should we just make a full copy then?

I don't think the length of gfs2_compat_ioctl is really an issue as
long as the function is that simple.

> static long gfs2_compat_ioctl(struct file *filp, unsigned int cmd,
> unsigned long arg)
> {
>         switch(cmd) {
>         case FS_IOC32_GETFLAGS:
>                 return gfs2_get_flags(filp, (u32 __user *)arg);
>         case FS_IOC32_SETFLAGS:
>                 return gfs2_set_flags(filp, (u32 __user *)arg);
>         case FITRIM:
>                 return gfs2_fitrim(filp, (void __user *)arg);
>         case FS_IOC_GETFSLABEL:
>                 return gfs2_getlabel(filp, (char __user *)arg);
>         }
>
>         return -ENOTTY;
> }

Don't we still need the compat_ptr conversion? That seems to be the
main point of having a compat_ioctl operation.

> > Should we feed this through the gfs2 tree?
>
> A later patch that removes the FITRIM handling from fs/compat_ioctl.c
> depends on it, so I'd like to keep everything together.

Ok, fine for me.

Thanks,
Andreas