On Tue, May 17, 2022 at 09:47:48AM +0200, Oscar Salvador wrote:
> On Mon, May 16, 2022 at 06:22:07PM +0800, Muchun Song wrote:
> > We are almost running out of free slots, only one bit is available in the
>
> I would be more precise about what are we running out of. Free slots of
> what?
>
> > worst case (powerpc with 256k pages). However, there are still some free
> > slots on other architectures (e.g. x86_64 has 10 bits available, arm64
> > has 8 bits available with worst case of 64K pages). We have hard coded
> > those numbers in code, it is inconvenient to use those bits on other
> > architectures except powerpc. So transfer those section flags to
> > enumeration to make it easy to add new section flags in the future. Also,
> > move SECTION_TAINT_ZONE_DEVICE into the scope of CONFIG_ZONE_DEVICE
> > to save a bit on non-zone-device case.
> >
> > Signed-off-by: Muchun Song <songmuchun@xxxxxxxxxxxxx>
> > ---
> ...
> > --- a/include/linux/mmzone.h
> > +++ b/include/linux/mmzone.h
> > @@ -1418,16 +1418,37 @@ extern size_t mem_section_usage_size(void);
> > * (equal SECTION_SIZE_BITS - PAGE_SHIFT), and the
> > * worst combination is powerpc with 256k pages,
> > * which results in PFN_SECTION_SHIFT equal 6.
> > - * To sum it up, at least 6 bits are available.
> > + * To sum it up, at least 6 bits are available on all architectures.
> > + * However, we can exceed 6 bits on some other architectures except
> > + * powerpc (e.g. 15 bits are available on x86_64, 13 bits are available
> > + * with the worst case of 64K pages on arm64) if we make sure the
> > + * exceeded bit is not applicable to powerpc.
> > */
> > -#define SECTION_MARKED_PRESENT (1UL<<0)
> > -#define SECTION_HAS_MEM_MAP (1UL<<1)
> > -#define SECTION_IS_ONLINE (1UL<<2)
> > -#define SECTION_IS_EARLY (1UL<<3)
> > -#define SECTION_TAINT_ZONE_DEVICE (1UL<<4)
> > -#define SECTION_MAP_LAST_BIT (1UL<<5)
> > +#define ENUM_SECTION_FLAG(MAPPER) \
> > + MAPPER(MARKED_PRESENT) \
> > + MAPPER(HAS_MEM_MAP) \
> > + MAPPER(IS_ONLINE) \
> > + MAPPER(IS_EARLY) \
> > + MAPPER(TAINT_ZONE_DEVICE, CONFIG_ZONE_DEVICE) \
> > + MAPPER(MAP_LAST_BIT)
> > +
> > +#define __SECTION_SHIFT_FLAG_MAPPER_0(x)
> > +#define __SECTION_SHIFT_FLAG_MAPPER_1(x) SECTION_##x##_SHIFT,
> > +#define __SECTION_SHIFT_FLAG_MAPPER(x, ...) \
> > + __PASTE(__SECTION_SHIFT_FLAG_MAPPER_, IS_ENABLED(__VA_ARGS__))(x)
> > +
> > +#define __SECTION_FLAG_MAPPER_0(x)
> > +#define __SECTION_FLAG_MAPPER_1(x) SECTION_##x = BIT(SECTION_##x##_SHIFT),
> > +#define __SECTION_FLAG_MAPPER(x, ...) \
> > + __PASTE(__SECTION_FLAG_MAPPER_, IS_ENABLED(__VA_ARGS__))(x)
> > +
> > +enum {
> > + ENUM_SECTION_FLAG(__SECTION_SHIFT_FLAG_MAPPER)
> > + ENUM_SECTION_FLAG(__SECTION_FLAG_MAPPER)
> > +};
> > +
> > #define SECTION_MAP_MASK (~(SECTION_MAP_LAST_BIT-1))
> > -#define SECTION_NID_SHIFT 6
> > +#define SECTION_NID_SHIFT SECTION_MAP_LAST_BIT_SHIFT
>
> Is this really worth the extra code? And it might be me that I am not
> familiar with all this magic, but it looks overcomplicated.
> Maybe some comments here and there help clarifying what it is going on
> here.
>
Yeah, it's a little complicated. All the magic aims to generate
two enumeration from one MAPPER(xxx, config), one is SECTION_xxx_SHIFT,
another is SECTION_xxx = BIT(SECTION_xxx_SHIFT) if the 'config' is
configured. If we want to add a new flag, like the follow patch, just
one line could do that.
MAPPER(CANNOT_OPTIMIZE_VMEMMAP, CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP)
Without those magic, we have to add 4 lines like follows to do the
similar thing.
#ifdef CONFIG_HUGETLB_PAGE_OPTIMIZE_VMEMMAP
SECTION_CANNOT_OPTIMIZE_VMEMMAP_SHIFT,
#define SECTION_CANNOT_OPTIMIZE_VMEMMAP BIT(SECTION_CANNOT_OPTIMIZE_VMEMMAP_SHIFT)
#endif
I admit it is more clear but not simplified as above approach.
Both two approaches are fine to me. If we choose the simplified
one, I agree with you I should add more comments to explain
what happens here.
Thanks.
