On Sun, Jan 30, 2022 at 6:47 AM Guo Ren <guoren@xxxxxxxxxx> wrote:
>
> On Sun, Jan 30, 2022 at 5:56 AM Arnd Bergmann <arnd@xxxxxxxx> wrote:
> >
> > On Sat, Jan 29, 2022 at 1:17 PM <guoren@xxxxxxxxxx> wrote:
> > > +
> > > +#define COMPAT_RLIM_INFINITY 0x7fffffff
> > >
> > > +#define F_GETLK64 12
> > > +#define F_SETLK64 13
> > > +#define F_SETLKW64 14
> >
> > These now come from the generic definitions I think. The flock definitions
> > are just the normal ones,
> Yes, it could be removed after Christoph Hellwig's patch merged.
Rgiht, I keep forgetting that this is a separate series, so this is fine.
> > and AFAICT the RLIM_INIFINITY definition here
> > is actually wrong and should be the default 0xffffffffu to match the
> > native (~0UL) definition.
> Yes, native rv32 used ~0UL, although its task_size is only 2.4GB.
The rlimit range has very little to do with the virtual memory address
limits, it is used for a number of other things that are typically more
limited in practice.
> I would remove #define COMPAT_RLIM_INFINITY 0x7fffffff
Ok.
Arnd
