On Tue, Jan 30, 2018 at 01:16:11PM +0100, Michal Hocko wrote:
> On Thu 18-01-18 17:50:48, Ram Pai wrote:
> [...]
> > @@ -851,9 +848,13 @@ static int show_smap(struct seq_file *m, void *v, int is_pid)
> > (unsigned long)(mss->pss >> (10 + PSS_SHIFT)));
> >
> > if (!rollup_mode) {
> > - arch_show_smap(m, vma);
> > +#ifdef CONFIG_ARCH_HAS_PKEYS
> > + if (arch_pkeys_enabled())
> > + seq_printf(m, "ProtectionKey: %8u\n", vma_pkey(vma));
> > +#endif
> > show_smap_vma_flags(m, vma);
> > }
> > +
>
> Why do you need to add ifdef here? The previous patch should make
> arch_pkeys_enabled == F when CONFIG_ARCH_HAS_PKEYS=n.
You are right. it need not be wrapped in CONFIG_ARCH_HAS_PKEYS. I had to do it
because vma_pkey(vma) is not defined in some architectures.
I will provide a generic vma_pkey() definition for architectures that do
not support PKEYS.
> Btw. could you
> merge those two patches into one. It is usually much easier to review a
> new helper function if it is added along with a user.
ok.
Thanks,
RP
