[v6,03/12] mm/sparsemem: Add helpers track active portions of a section at boot
diff mbox series

Message ID 155552635098.2015392.5460028594173939000.stgit@dwillia2-desk3.amr.corp.intel.com
State New
Headers show
Series
  • mm: Sub-section memory hotplug support
Related show

Commit Message

Dan Williams April 17, 2019, 6:39 p.m. UTC
Prepare for hot{plug,remove} of sub-ranges of a section by tracking a
section active bitmask, each bit representing 2MB (SECTION_SIZE (128M) /
map_active bitmask length (64)). If it turns out that 2MB is too large
of an active tracking granularity it is trivial to increase the size of
the map_active bitmap.

The implications of a partially populated section is that pfn_valid()
needs to go beyond a valid_section() check and read the sub-section
active ranges from the bitmask.

Cc: Michal Hocko <mhocko@suse.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: Logan Gunthorpe <logang@deltatee.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 include/linux/mmzone.h |   29 ++++++++++++++++++++++++++++-
 mm/page_alloc.c        |    4 +++-
 mm/sparse.c            |   48 ++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 79 insertions(+), 2 deletions(-)

Comments

Oscar Salvador April 25, 2019, 2:43 p.m. UTC | #1
On Wed, 2019-04-17 at 11:39 -0700, Dan Williams wrote:
> Prepare for hot{plug,remove} of sub-ranges of a section by tracking a
> section active bitmask, each bit representing 2MB (SECTION_SIZE
> (128M) /
> map_active bitmask length (64)). If it turns out that 2MB is too
> large
> of an active tracking granularity it is trivial to increase the size
> of
> the map_active bitmap.
> 
> The implications of a partially populated section is that pfn_valid()
> needs to go beyond a valid_section() check and read the sub-section
> active ranges from the bitmask.
> 
> Cc: Michal Hocko <mhocko@suse.com>
> Cc: Vlastimil Babka <vbabka@suse.cz>
> Cc: Logan Gunthorpe <logang@deltatee.com>
> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Hi Dan,

I am still going through the patchset but:
 
> +static unsigned long section_active_mask(unsigned long pfn,
> +		unsigned long nr_pages)
> +{
> +	int idx_start, idx_size;
> +	phys_addr_t start, size;
> +
> +	if (!nr_pages)
> +		return 0;
> +
> +	start = PFN_PHYS(pfn);
> +	size = PFN_PHYS(min(nr_pages, PAGES_PER_SECTION
> +				- (pfn & ~PAGE_SECTION_MASK)));

We already picked the lowest value in section_active_init, didn't we?
This min() operations seems redundant to me here.

> +	size = ALIGN(size, SECTION_ACTIVE_SIZE);
> +
> +	idx_start = section_active_index(start);
> +	idx_size = section_active_index(size);
> +
> +	if (idx_size == 0)
> +		return -1;
> +	return ((1UL << idx_size) - 1) << idx_start;
> +}
> +
> +void section_active_init(unsigned long pfn, unsigned long nr_pages)
> +{
> +	int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
> +	int i, start_sec = pfn_to_section_nr(pfn);
> +
> +	if (!nr_pages)
> +		return;
> +
> +	for (i = start_sec; i <= end_sec; i++) {
> +		struct mem_section *ms;
> +		unsigned long mask;
> +		unsigned long pfns;

s/pfns/nr_pfns/ instead?

> +		pfns = min(nr_pages, PAGES_PER_SECTION
> +				- (pfn & ~PAGE_SECTION_MASK));
> +		mask = section_active_mask(pfn, pfns);
> +
> +		ms = __nr_to_section(i);
> +		pr_debug("%s: sec: %d mask: %#018lx\n", __func__, i,
> mask);
> +		ms->usage->map_active = mask;
> +
> +		pfn += pfns;
> +		nr_pages -= pfns;
> +	}
> +}

Although the code is not very complicated, it could use some comments
here and there I think.

> +
>  /* Record a memory area against a node. */
>  void __init memory_present(int nid, unsigned long start, unsigned
> long end)
>  {
>
Oscar Salvador April 26, 2019, 12:57 p.m. UTC | #2
On Wed, Apr 17, 2019 at 11:39:11AM -0700, Dan Williams wrote:
> Prepare for hot{plug,remove} of sub-ranges of a section by tracking a
> section active bitmask, each bit representing 2MB (SECTION_SIZE (128M) /
> map_active bitmask length (64)). If it turns out that 2MB is too large
> of an active tracking granularity it is trivial to increase the size of
> the map_active bitmap.
> 
> The implications of a partially populated section is that pfn_valid()
> needs to go beyond a valid_section() check and read the sub-section
> active ranges from the bitmask.
> 
> Cc: Michal Hocko <mhocko@suse.com>
> Cc: Vlastimil Babka <vbabka@suse.cz>
> Cc: Logan Gunthorpe <logang@deltatee.com>
> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
[...]  
> +static unsigned long section_active_mask(unsigned long pfn,
> +		unsigned long nr_pages)
> +{
> +	int idx_start, idx_size;
> +	phys_addr_t start, size;
> +
> +	if (!nr_pages)
> +		return 0;
> +
> +	start = PFN_PHYS(pfn);
> +	size = PFN_PHYS(min(nr_pages, PAGES_PER_SECTION
> +				- (pfn & ~PAGE_SECTION_MASK)));
> +	size = ALIGN(size, SECTION_ACTIVE_SIZE);

I am probably missing something, and this is more a question than anything else, but:
is there a reason for shifting pfn and pages to get the size and the address?
Could not we operate on pfn/pages, so we do not have to shift every time?
(even for pfn_section_valid() calls)

Something like:

#define SUB_SECTION_ACTIVE_PAGES        (SECTION_ACTIVE_SIZE / PAGE_SIZE)

static inline int section_active_index(unsigned long pfn)
{
	return (pfn & ~(PAGE_SECTION_MASK)) / SUB_SECTION_ACTIVE_PAGES;
}

> +
> +	idx_start = section_active_index(start);
> +	idx_size = section_active_index(size);
> +
> +	if (idx_size == 0)
> +		return -1;

What about turning that into something more intuitive?
Since -1 represents here a full section, we could define something like:

#define FULL_SECTION	(-1UL)

Or a better name, it is just that I find "-1" not really easy to interpret.

> +	return ((1UL << idx_size) - 1) << idx_start;
> +}
> +
> +void section_active_init(unsigned long pfn, unsigned long nr_pages)
> +{
> +	int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
> +	int i, start_sec = pfn_to_section_nr(pfn);
> +
> +	if (!nr_pages)
> +		return;
> +
> +	for (i = start_sec; i <= end_sec; i++) {
> +		struct mem_section *ms;
> +		unsigned long mask;
> +		unsigned long pfns;
> +
> +		pfns = min(nr_pages, PAGES_PER_SECTION
> +				- (pfn & ~PAGE_SECTION_MASK));
> +		mask = section_active_mask(pfn, pfns);
> +
> +		ms = __nr_to_section(i);
> +		pr_debug("%s: sec: %d mask: %#018lx\n", __func__, i, mask);
> +		ms->usage->map_active = mask;
> +
> +		pfn += pfns;
> +		nr_pages -= pfns;
> +	}
> +}
> +
>  /* Record a memory area against a node. */
>  void __init memory_present(int nid, unsigned long start, unsigned long end)
>  {
>
Pavel Tatashin May 2, 2019, 4:12 p.m. UTC | #3
On Wed, Apr 17, 2019 at 2:53 PM Dan Williams <dan.j.williams@intel.com> wrote:
>
> Prepare for hot{plug,remove} of sub-ranges of a section by tracking a
> section active bitmask, each bit representing 2MB (SECTION_SIZE (128M) /
> map_active bitmask length (64)). If it turns out that 2MB is too large
> of an active tracking granularity it is trivial to increase the size of
> the map_active bitmap.

Please mention that 2M on Intel, and 16M on Arm64.

>
> The implications of a partially populated section is that pfn_valid()
> needs to go beyond a valid_section() check and read the sub-section
> active ranges from the bitmask.
>
> Cc: Michal Hocko <mhocko@suse.com>
> Cc: Vlastimil Babka <vbabka@suse.cz>
> Cc: Logan Gunthorpe <logang@deltatee.com>
> Signed-off-by: Dan Williams <dan.j.williams@intel.com>
> ---
>  include/linux/mmzone.h |   29 ++++++++++++++++++++++++++++-
>  mm/page_alloc.c        |    4 +++-
>  mm/sparse.c            |   48 ++++++++++++++++++++++++++++++++++++++++++++++++
>  3 files changed, 79 insertions(+), 2 deletions(-)
>
> diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> index 6726fc175b51..cffde898e345 100644
> --- a/include/linux/mmzone.h
> +++ b/include/linux/mmzone.h
> @@ -1175,6 +1175,8 @@ struct mem_section_usage {
>         unsigned long pageblock_flags[0];
>  };
>
> +void section_active_init(unsigned long pfn, unsigned long nr_pages);
> +
>  struct page;
>  struct page_ext;
>  struct mem_section {
> @@ -1312,12 +1314,36 @@ static inline struct mem_section *__pfn_to_section(unsigned long pfn)
>
>  extern int __highest_present_section_nr;
>
> +static inline int section_active_index(phys_addr_t phys)
> +{
> +       return (phys & ~(PA_SECTION_MASK)) / SECTION_ACTIVE_SIZE;

How about also defining SECTION_ACTIVE_SHIFT like this:

/* BITS_PER_LONG = 2^6 */
#define BITS_PER_LONG_SHIFT 6
#define SECTION_ACTIVE_SHIFT (SECTION_SIZE_BITS - BITS_PER_LONG_SHIFT)
#define SECTION_ACTIVE_SIZE (1 << SECTION_ACTIVE_SHIFT)

The return above would become:
return (phys & ~(PA_SECTION_MASK)) >> SECTION_ACTIVE_SHIFT;

> +}
> +
> +#ifdef CONFIG_SPARSEMEM_VMEMMAP
> +static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
> +{
> +       int idx = section_active_index(PFN_PHYS(pfn));
> +
> +       return !!(ms->usage->map_active & (1UL << idx));
> +}
> +#else
> +static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
> +{
> +       return 1;
> +}
> +#endif
> +
>  #ifndef CONFIG_HAVE_ARCH_PFN_VALID
>  static inline int pfn_valid(unsigned long pfn)
>  {
> +       struct mem_section *ms;
> +
>         if (pfn_to_section_nr(pfn) >= NR_MEM_SECTIONS)
>                 return 0;
> -       return valid_section(__nr_to_section(pfn_to_section_nr(pfn)));
> +       ms = __nr_to_section(pfn_to_section_nr(pfn));
> +       if (!valid_section(ms))
> +               return 0;
> +       return pfn_section_valid(ms, pfn);
>  }
>  #endif
>
> @@ -1349,6 +1375,7 @@ void sparse_init(void);
>  #define sparse_init()  do {} while (0)
>  #define sparse_index_init(_sec, _nid)  do {} while (0)
>  #define pfn_present pfn_valid
> +#define section_active_init(_pfn, _nr_pages) do {} while (0)
>  #endif /* CONFIG_SPARSEMEM */
>
>  /*
> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> index f671401a7c0b..c9ad28a78018 100644
> --- a/mm/page_alloc.c
> +++ b/mm/page_alloc.c
> @@ -7273,10 +7273,12 @@ void __init free_area_init_nodes(unsigned long *max_zone_pfn)
>
>         /* Print out the early node map */
>         pr_info("Early memory node ranges\n");
> -       for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid)
> +       for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
>                 pr_info("  node %3d: [mem %#018Lx-%#018Lx]\n", nid,
>                         (u64)start_pfn << PAGE_SHIFT,
>                         ((u64)end_pfn << PAGE_SHIFT) - 1);
> +               section_active_init(start_pfn, end_pfn - start_pfn);
> +       }
>
>         /* Initialise every node */
>         mminit_verify_pageflags_layout();
> diff --git a/mm/sparse.c b/mm/sparse.c
> index f87de7ad32c8..5ef2f884c4e1 100644
> --- a/mm/sparse.c
> +++ b/mm/sparse.c
> @@ -210,6 +210,54 @@ static inline unsigned long first_present_section_nr(void)
>         return next_present_section_nr(-1);
>  }
>
> +static unsigned long section_active_mask(unsigned long pfn,
> +               unsigned long nr_pages)
> +{
> +       int idx_start, idx_size;
> +       phys_addr_t start, size;
> +
> +       if (!nr_pages)
> +               return 0;
> +
> +       start = PFN_PHYS(pfn);
> +       size = PFN_PHYS(min(nr_pages, PAGES_PER_SECTION
> +                               - (pfn & ~PAGE_SECTION_MASK)));
> +       size = ALIGN(size, SECTION_ACTIVE_SIZE);
> +
> +       idx_start = section_active_index(start);
> +       idx_size = section_active_index(size);
> +
> +       if (idx_size == 0)
> +               return -1;
> +       return ((1UL << idx_size) - 1) << idx_start;
> +}
> +
> +void section_active_init(unsigned long pfn, unsigned long nr_pages)
> +{
> +       int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
> +       int i, start_sec = pfn_to_section_nr(pfn);
> +
> +       if (!nr_pages)
> +               return;
> +
> +       for (i = start_sec; i <= end_sec; i++) {
> +               struct mem_section *ms;
> +               unsigned long mask;
> +               unsigned long pfns;
> +
> +               pfns = min(nr_pages, PAGES_PER_SECTION
> +                               - (pfn & ~PAGE_SECTION_MASK));
> +               mask = section_active_mask(pfn, pfns);
> +
> +               ms = __nr_to_section(i);
> +               pr_debug("%s: sec: %d mask: %#018lx\n", __func__, i, mask);
> +               ms->usage->map_active = mask;
> +
> +               pfn += pfns;
> +               nr_pages -= pfns;
> +       }
> +}

For some reasons the above code is confusing to me. It seems all the
code supposed to do is set all map_active to -1, and trim the first
and last sections (can be the same section of course). So, I would
replace the above two functions with one function like this:

void section_active_init(unsigned long pfn, unsigned long nr_pages)
{
        int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
        int i, idx, start_sec = pfn_to_section_nr(pfn);
        struct mem_section *ms;

        if (!nr_pages)
                return;

        for (i = start_sec; i <= end_sec; i++) {
                ms = __nr_to_section(i);
                ms->usage->map_active = ~0ul;
        }

        /* Might need to trim active pfns from the beginning and end */
        idx = section_active_index(PFN_PHYS(pfn));
        ms = __nr_to_section(start_sec);
        ms->usage->map_active &= (~0ul << idx);

        idx = section_active_index(PFN_PHYS(pfn + nr_pages -1));
        ms = __nr_to_section(end_sec);
        ms->usage->map_active &= (~0ul >> (BITS_PER_LONG - idx - 1));
}
Dan Williams May 4, 2019, 7:26 p.m. UTC | #4
On Thu, May 2, 2019 at 9:12 AM Pavel Tatashin <pasha.tatashin@soleen.com> wrote:
>
> On Wed, Apr 17, 2019 at 2:53 PM Dan Williams <dan.j.williams@intel.com> wrote:
> >
> > Prepare for hot{plug,remove} of sub-ranges of a section by tracking a
> > section active bitmask, each bit representing 2MB (SECTION_SIZE (128M) /
> > map_active bitmask length (64)). If it turns out that 2MB is too large
> > of an active tracking granularity it is trivial to increase the size of
> > the map_active bitmap.
>
> Please mention that 2M on Intel, and 16M on Arm64.
>
> >
> > The implications of a partially populated section is that pfn_valid()
> > needs to go beyond a valid_section() check and read the sub-section
> > active ranges from the bitmask.
> >
> > Cc: Michal Hocko <mhocko@suse.com>
> > Cc: Vlastimil Babka <vbabka@suse.cz>
> > Cc: Logan Gunthorpe <logang@deltatee.com>
> > Signed-off-by: Dan Williams <dan.j.williams@intel.com>
> > ---
> >  include/linux/mmzone.h |   29 ++++++++++++++++++++++++++++-
> >  mm/page_alloc.c        |    4 +++-
> >  mm/sparse.c            |   48 ++++++++++++++++++++++++++++++++++++++++++++++++
> >  3 files changed, 79 insertions(+), 2 deletions(-)
> >
> > diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> > index 6726fc175b51..cffde898e345 100644
> > --- a/include/linux/mmzone.h
> > +++ b/include/linux/mmzone.h
> > @@ -1175,6 +1175,8 @@ struct mem_section_usage {
> >         unsigned long pageblock_flags[0];
> >  };
> >
> > +void section_active_init(unsigned long pfn, unsigned long nr_pages);
> > +
> >  struct page;
> >  struct page_ext;
> >  struct mem_section {
> > @@ -1312,12 +1314,36 @@ static inline struct mem_section *__pfn_to_section(unsigned long pfn)
> >
> >  extern int __highest_present_section_nr;
> >
> > +static inline int section_active_index(phys_addr_t phys)
> > +{
> > +       return (phys & ~(PA_SECTION_MASK)) / SECTION_ACTIVE_SIZE;
>
> How about also defining SECTION_ACTIVE_SHIFT like this:
>
> /* BITS_PER_LONG = 2^6 */
> #define BITS_PER_LONG_SHIFT 6
> #define SECTION_ACTIVE_SHIFT (SECTION_SIZE_BITS - BITS_PER_LONG_SHIFT)
> #define SECTION_ACTIVE_SIZE (1 << SECTION_ACTIVE_SHIFT)
>
> The return above would become:
> return (phys & ~(PA_SECTION_MASK)) >> SECTION_ACTIVE_SHIFT;
>
> > +}
> > +
> > +#ifdef CONFIG_SPARSEMEM_VMEMMAP
> > +static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
> > +{
> > +       int idx = section_active_index(PFN_PHYS(pfn));
> > +
> > +       return !!(ms->usage->map_active & (1UL << idx));
> > +}
> > +#else
> > +static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
> > +{
> > +       return 1;
> > +}
> > +#endif
> > +
> >  #ifndef CONFIG_HAVE_ARCH_PFN_VALID
> >  static inline int pfn_valid(unsigned long pfn)
> >  {
> > +       struct mem_section *ms;
> > +
> >         if (pfn_to_section_nr(pfn) >= NR_MEM_SECTIONS)
> >                 return 0;
> > -       return valid_section(__nr_to_section(pfn_to_section_nr(pfn)));
> > +       ms = __nr_to_section(pfn_to_section_nr(pfn));
> > +       if (!valid_section(ms))
> > +               return 0;
> > +       return pfn_section_valid(ms, pfn);
> >  }
> >  #endif
> >
> > @@ -1349,6 +1375,7 @@ void sparse_init(void);
> >  #define sparse_init()  do {} while (0)
> >  #define sparse_index_init(_sec, _nid)  do {} while (0)
> >  #define pfn_present pfn_valid
> > +#define section_active_init(_pfn, _nr_pages) do {} while (0)
> >  #endif /* CONFIG_SPARSEMEM */
> >
> >  /*
> > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > index f671401a7c0b..c9ad28a78018 100644
> > --- a/mm/page_alloc.c
> > +++ b/mm/page_alloc.c
> > @@ -7273,10 +7273,12 @@ void __init free_area_init_nodes(unsigned long *max_zone_pfn)
> >
> >         /* Print out the early node map */
> >         pr_info("Early memory node ranges\n");
> > -       for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid)
> > +       for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
> >                 pr_info("  node %3d: [mem %#018Lx-%#018Lx]\n", nid,
> >                         (u64)start_pfn << PAGE_SHIFT,
> >                         ((u64)end_pfn << PAGE_SHIFT) - 1);
> > +               section_active_init(start_pfn, end_pfn - start_pfn);
> > +       }
> >
> >         /* Initialise every node */
> >         mminit_verify_pageflags_layout();
> > diff --git a/mm/sparse.c b/mm/sparse.c
> > index f87de7ad32c8..5ef2f884c4e1 100644
> > --- a/mm/sparse.c
> > +++ b/mm/sparse.c
> > @@ -210,6 +210,54 @@ static inline unsigned long first_present_section_nr(void)
> >         return next_present_section_nr(-1);
> >  }
> >
> > +static unsigned long section_active_mask(unsigned long pfn,
> > +               unsigned long nr_pages)
> > +{
> > +       int idx_start, idx_size;
> > +       phys_addr_t start, size;
> > +
> > +       if (!nr_pages)
> > +               return 0;
> > +
> > +       start = PFN_PHYS(pfn);
> > +       size = PFN_PHYS(min(nr_pages, PAGES_PER_SECTION
> > +                               - (pfn & ~PAGE_SECTION_MASK)));
> > +       size = ALIGN(size, SECTION_ACTIVE_SIZE);
> > +
> > +       idx_start = section_active_index(start);
> > +       idx_size = section_active_index(size);
> > +
> > +       if (idx_size == 0)
> > +               return -1;
> > +       return ((1UL << idx_size) - 1) << idx_start;
> > +}
> > +
> > +void section_active_init(unsigned long pfn, unsigned long nr_pages)
> > +{
> > +       int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
> > +       int i, start_sec = pfn_to_section_nr(pfn);
> > +
> > +       if (!nr_pages)
> > +               return;
> > +
> > +       for (i = start_sec; i <= end_sec; i++) {
> > +               struct mem_section *ms;
> > +               unsigned long mask;
> > +               unsigned long pfns;
> > +
> > +               pfns = min(nr_pages, PAGES_PER_SECTION
> > +                               - (pfn & ~PAGE_SECTION_MASK));
> > +               mask = section_active_mask(pfn, pfns);
> > +
> > +               ms = __nr_to_section(i);
> > +               pr_debug("%s: sec: %d mask: %#018lx\n", __func__, i, mask);
> > +               ms->usage->map_active = mask;
> > +
> > +               pfn += pfns;
> > +               nr_pages -= pfns;
> > +       }
> > +}
>
> For some reasons the above code is confusing to me. It seems all the
> code supposed to do is set all map_active to -1, and trim the first
> and last sections (can be the same section of course). So, I would
> replace the above two functions with one function like this:
>
> void section_active_init(unsigned long pfn, unsigned long nr_pages)
> {
>         int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
>         int i, idx, start_sec = pfn_to_section_nr(pfn);
>         struct mem_section *ms;
>
>         if (!nr_pages)
>                 return;
>
>         for (i = start_sec; i <= end_sec; i++) {
>                 ms = __nr_to_section(i);
>                 ms->usage->map_active = ~0ul;
>         }
>
>         /* Might need to trim active pfns from the beginning and end */
>         idx = section_active_index(PFN_PHYS(pfn));
>         ms = __nr_to_section(start_sec);
>         ms->usage->map_active &= (~0ul << idx);
>
>         idx = section_active_index(PFN_PHYS(pfn + nr_pages -1));
>         ms = __nr_to_section(end_sec);
>         ms->usage->map_active &= (~0ul >> (BITS_PER_LONG - idx - 1));
> }

I like the cleanup, but one of the fixes in v7 resulted in the
realization that a given section may be populated twice at init time.
For example, enabling that pr_debug() yields:

    section_active_init: sec: 12 mask: 0x00000003ffffffff
    section_active_init: sec: 12 mask: 0xe000000000000000

So, the implementation can't blindly clear bits based on the current
parameters. However, I'm switching this code over to use bitmap_*()
helpers which should help with the readability.
Pavel Tatashin May 4, 2019, 7:40 p.m. UTC | #5
On Sat, May 4, 2019, 3:26 PM Dan Williams <dan.j.williams@intel.com> wrote:

> On Thu, May 2, 2019 at 9:12 AM Pavel Tatashin <pasha.tatashin@soleen.com>
> wrote:
> >
> > On Wed, Apr 17, 2019 at 2:53 PM Dan Williams <dan.j.williams@intel.com>
> wrote:
> > >
> > > Prepare for hot{plug,remove} of sub-ranges of a section by tracking a
> > > section active bitmask, each bit representing 2MB (SECTION_SIZE (128M)
> /
> > > map_active bitmask length (64)). If it turns out that 2MB is too large
> > > of an active tracking granularity it is trivial to increase the size of
> > > the map_active bitmap.
> >
> > Please mention that 2M on Intel, and 16M on Arm64.
> >
> > >
> > > The implications of a partially populated section is that pfn_valid()
> > > needs to go beyond a valid_section() check and read the sub-section
> > > active ranges from the bitmask.
> > >
> > > Cc: Michal Hocko <mhocko@suse.com>
> > > Cc: Vlastimil Babka <vbabka@suse.cz>
> > > Cc: Logan Gunthorpe <logang@deltatee.com>
> > > Signed-off-by: Dan Williams <dan.j.williams@intel.com>
> > > ---
> > >  include/linux/mmzone.h |   29 ++++++++++++++++++++++++++++-
> > >  mm/page_alloc.c        |    4 +++-
> > >  mm/sparse.c            |   48
> ++++++++++++++++++++++++++++++++++++++++++++++++
> > >  3 files changed, 79 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
> > > index 6726fc175b51..cffde898e345 100644
> > > --- a/include/linux/mmzone.h
> > > +++ b/include/linux/mmzone.h
> > > @@ -1175,6 +1175,8 @@ struct mem_section_usage {
> > >         unsigned long pageblock_flags[0];
> > >  };
> > >
> > > +void section_active_init(unsigned long pfn, unsigned long nr_pages);
> > > +
> > >  struct page;
> > >  struct page_ext;
> > >  struct mem_section {
> > > @@ -1312,12 +1314,36 @@ static inline struct mem_section
> *__pfn_to_section(unsigned long pfn)
> > >
> > >  extern int __highest_present_section_nr;
> > >
> > > +static inline int section_active_index(phys_addr_t phys)
> > > +{
> > > +       return (phys & ~(PA_SECTION_MASK)) / SECTION_ACTIVE_SIZE;
> >
> > How about also defining SECTION_ACTIVE_SHIFT like this:
> >
> > /* BITS_PER_LONG = 2^6 */
> > #define BITS_PER_LONG_SHIFT 6
> > #define SECTION_ACTIVE_SHIFT (SECTION_SIZE_BITS - BITS_PER_LONG_SHIFT)
> > #define SECTION_ACTIVE_SIZE (1 << SECTION_ACTIVE_SHIFT)
> >
> > The return above would become:
> > return (phys & ~(PA_SECTION_MASK)) >> SECTION_ACTIVE_SHIFT;
> >
> > > +}
> > > +
> > > +#ifdef CONFIG_SPARSEMEM_VMEMMAP
> > > +static inline int pfn_section_valid(struct mem_section *ms, unsigned
> long pfn)
> > > +{
> > > +       int idx = section_active_index(PFN_PHYS(pfn));
> > > +
> > > +       return !!(ms->usage->map_active & (1UL << idx));
> > > +}
> > > +#else
> > > +static inline int pfn_section_valid(struct mem_section *ms, unsigned
> long pfn)
> > > +{
> > > +       return 1;
> > > +}
> > > +#endif
> > > +
> > >  #ifndef CONFIG_HAVE_ARCH_PFN_VALID
> > >  static inline int pfn_valid(unsigned long pfn)
> > >  {
> > > +       struct mem_section *ms;
> > > +
> > >         if (pfn_to_section_nr(pfn) >= NR_MEM_SECTIONS)
> > >                 return 0;
> > > -       return valid_section(__nr_to_section(pfn_to_section_nr(pfn)));
> > > +       ms = __nr_to_section(pfn_to_section_nr(pfn));
> > > +       if (!valid_section(ms))
> > > +               return 0;
> > > +       return pfn_section_valid(ms, pfn);
> > >  }
> > >  #endif
> > >
> > > @@ -1349,6 +1375,7 @@ void sparse_init(void);
> > >  #define sparse_init()  do {} while (0)
> > >  #define sparse_index_init(_sec, _nid)  do {} while (0)
> > >  #define pfn_present pfn_valid
> > > +#define section_active_init(_pfn, _nr_pages) do {} while (0)
> > >  #endif /* CONFIG_SPARSEMEM */
> > >
> > >  /*
> > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > index f671401a7c0b..c9ad28a78018 100644
> > > --- a/mm/page_alloc.c
> > > +++ b/mm/page_alloc.c
> > > @@ -7273,10 +7273,12 @@ void __init free_area_init_nodes(unsigned long
> *max_zone_pfn)
> > >
> > >         /* Print out the early node map */
> > >         pr_info("Early memory node ranges\n");
> > > -       for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn,
> &nid)
> > > +       for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn,
> &nid) {
> > >                 pr_info("  node %3d: [mem %#018Lx-%#018Lx]\n", nid,
> > >                         (u64)start_pfn << PAGE_SHIFT,
> > >                         ((u64)end_pfn << PAGE_SHIFT) - 1);
> > > +               section_active_init(start_pfn, end_pfn - start_pfn);
> > > +       }
> > >
> > >         /* Initialise every node */
> > >         mminit_verify_pageflags_layout();
> > > diff --git a/mm/sparse.c b/mm/sparse.c
> > > index f87de7ad32c8..5ef2f884c4e1 100644
> > > --- a/mm/sparse.c
> > > +++ b/mm/sparse.c
> > > @@ -210,6 +210,54 @@ static inline unsigned long
> first_present_section_nr(void)
> > >         return next_present_section_nr(-1);
> > >  }
> > >
> > > +static unsigned long section_active_mask(unsigned long pfn,
> > > +               unsigned long nr_pages)
> > > +{
> > > +       int idx_start, idx_size;
> > > +       phys_addr_t start, size;
> > > +
> > > +       if (!nr_pages)
> > > +               return 0;
> > > +
> > > +       start = PFN_PHYS(pfn);
> > > +       size = PFN_PHYS(min(nr_pages, PAGES_PER_SECTION
> > > +                               - (pfn & ~PAGE_SECTION_MASK)));
> > > +       size = ALIGN(size, SECTION_ACTIVE_SIZE);
> > > +
> > > +       idx_start = section_active_index(start);
> > > +       idx_size = section_active_index(size);
> > > +
> > > +       if (idx_size == 0)
> > > +               return -1;
> > > +       return ((1UL << idx_size) - 1) << idx_start;
> > > +}
> > > +
> > > +void section_active_init(unsigned long pfn, unsigned long nr_pages)
> > > +{
> > > +       int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
> > > +       int i, start_sec = pfn_to_section_nr(pfn);
> > > +
> > > +       if (!nr_pages)
> > > +               return;
> > > +
> > > +       for (i = start_sec; i <= end_sec; i++) {
> > > +               struct mem_section *ms;
> > > +               unsigned long mask;
> > > +               unsigned long pfns;
> > > +
> > > +               pfns = min(nr_pages, PAGES_PER_SECTION
> > > +                               - (pfn & ~PAGE_SECTION_MASK));
> > > +               mask = section_active_mask(pfn, pfns);
> > > +
> > > +               ms = __nr_to_section(i);
> > > +               pr_debug("%s: sec: %d mask: %#018lx\n", __func__, i,
> mask);
> > > +               ms->usage->map_active = mask;
> > > +
> > > +               pfn += pfns;
> > > +               nr_pages -= pfns;
> > > +       }
> > > +}
> >
> > For some reasons the above code is confusing to me. It seems all the
> > code supposed to do is set all map_active to -1, and trim the first
> > and last sections (can be the same section of course). So, I would
> > replace the above two functions with one function like this:
> >
> > void section_active_init(unsigned long pfn, unsigned long nr_pages)
> > {
> >         int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
> >         int i, idx, start_sec = pfn_to_section_nr(pfn);
> >         struct mem_section *ms;
> >
> >         if (!nr_pages)
> >                 return;
> >
> >         for (i = start_sec; i <= end_sec; i++) {
> >                 ms = __nr_to_section(i);
> >                 ms->usage->map_active = ~0ul;
> >         }
> >
> >         /* Might need to trim active pfns from the beginning and end */
> >         idx = section_active_index(PFN_PHYS(pfn));
> >         ms = __nr_to_section(start_sec);
> >         ms->usage->map_active &= (~0ul << idx);
> >
> >         idx = section_active_index(PFN_PHYS(pfn + nr_pages -1));
> >         ms = __nr_to_section(end_sec);
> >         ms->usage->map_active &= (~0ul >> (BITS_PER_LONG - idx - 1));
> > }
>
> I like the cleanup, but one of the fixes in v7 resulted in the
> realization that a given section may be populated twice at init time.
> For example, enabling that pr_debug() yields:
>
>     section_active_init: sec: 12 mask: 0x00000003ffffffff
>     section_active_init: sec: 12 mask: 0xe000000000000000
>
> So, the implementation can't blindly clear bits based on the current
> parameters. However, I'm switching this code over to use bitmap_*()
> helpers which should help with the readability.
>

Yes, bitmap_* will help, and I assume you will also make active_map size
scalable?

I will take another look at version 8.


Thank you,
Pasha
<div dir="auto"><div><br><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Sat, May 4, 2019, 3:26 PM Dan Williams &lt;<a href="mailto:dan.j.williams@intel.com">dan.j.williams@intel.com</a>&gt; wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On Thu, May 2, 2019 at 9:12 AM Pavel Tatashin &lt;<a href="mailto:pasha.tatashin@soleen.com" target="_blank" rel="noreferrer">pasha.tatashin@soleen.com</a>&gt; wrote:<br>
&gt;<br>
&gt; On Wed, Apr 17, 2019 at 2:53 PM Dan Williams &lt;<a href="mailto:dan.j.williams@intel.com" target="_blank" rel="noreferrer">dan.j.williams@intel.com</a>&gt; wrote:<br>
&gt; &gt;<br>
&gt; &gt; Prepare for hot{plug,remove} of sub-ranges of a section by tracking a<br>
&gt; &gt; section active bitmask, each bit representing 2MB (SECTION_SIZE (128M) /<br>
&gt; &gt; map_active bitmask length (64)). If it turns out that 2MB is too large<br>
&gt; &gt; of an active tracking granularity it is trivial to increase the size of<br>
&gt; &gt; the map_active bitmap.<br>
&gt;<br>
&gt; Please mention that 2M on Intel, and 16M on Arm64.<br>
&gt;<br>
&gt; &gt;<br>
&gt; &gt; The implications of a partially populated section is that pfn_valid()<br>
&gt; &gt; needs to go beyond a valid_section() check and read the sub-section<br>
&gt; &gt; active ranges from the bitmask.<br>
&gt; &gt;<br>
&gt; &gt; Cc: Michal Hocko &lt;<a href="mailto:mhocko@suse.com" target="_blank" rel="noreferrer">mhocko@suse.com</a>&gt;<br>
&gt; &gt; Cc: Vlastimil Babka &lt;<a href="mailto:vbabka@suse.cz" target="_blank" rel="noreferrer">vbabka@suse.cz</a>&gt;<br>
&gt; &gt; Cc: Logan Gunthorpe &lt;<a href="mailto:logang@deltatee.com" target="_blank" rel="noreferrer">logang@deltatee.com</a>&gt;<br>
&gt; &gt; Signed-off-by: Dan Williams &lt;<a href="mailto:dan.j.williams@intel.com" target="_blank" rel="noreferrer">dan.j.williams@intel.com</a>&gt;<br>
&gt; &gt; ---<br>
&gt; &gt;  include/linux/mmzone.h |   29 ++++++++++++++++++++++++++++-<br>
&gt; &gt;  mm/page_alloc.c        |    4 +++-<br>
&gt; &gt;  mm/sparse.c            |   48 ++++++++++++++++++++++++++++++++++++++++++++++++<br>
&gt; &gt;  3 files changed, 79 insertions(+), 2 deletions(-)<br>
&gt; &gt;<br>
&gt; &gt; diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h<br>
&gt; &gt; index 6726fc175b51..cffde898e345 100644<br>
&gt; &gt; --- a/include/linux/mmzone.h<br>
&gt; &gt; +++ b/include/linux/mmzone.h<br>
&gt; &gt; @@ -1175,6 +1175,8 @@ struct mem_section_usage {<br>
&gt; &gt;         unsigned long pageblock_flags[0];<br>
&gt; &gt;  };<br>
&gt; &gt;<br>
&gt; &gt; +void section_active_init(unsigned long pfn, unsigned long nr_pages);<br>
&gt; &gt; +<br>
&gt; &gt;  struct page;<br>
&gt; &gt;  struct page_ext;<br>
&gt; &gt;  struct mem_section {<br>
&gt; &gt; @@ -1312,12 +1314,36 @@ static inline struct mem_section *__pfn_to_section(unsigned long pfn)<br>
&gt; &gt;<br>
&gt; &gt;  extern int __highest_present_section_nr;<br>
&gt; &gt;<br>
&gt; &gt; +static inline int section_active_index(phys_addr_t phys)<br>
&gt; &gt; +{<br>
&gt; &gt; +       return (phys &amp; ~(PA_SECTION_MASK)) / SECTION_ACTIVE_SIZE;<br>
&gt;<br>
&gt; How about also defining SECTION_ACTIVE_SHIFT like this:<br>
&gt;<br>
&gt; /* BITS_PER_LONG = 2^6 */<br>
&gt; #define BITS_PER_LONG_SHIFT 6<br>
&gt; #define SECTION_ACTIVE_SHIFT (SECTION_SIZE_BITS - BITS_PER_LONG_SHIFT)<br>
&gt; #define SECTION_ACTIVE_SIZE (1 &lt;&lt; SECTION_ACTIVE_SHIFT)<br>
&gt;<br>
&gt; The return above would become:<br>
&gt; return (phys &amp; ~(PA_SECTION_MASK)) &gt;&gt; SECTION_ACTIVE_SHIFT;<br>
&gt;<br>
&gt; &gt; +}<br>
&gt; &gt; +<br>
&gt; &gt; +#ifdef CONFIG_SPARSEMEM_VMEMMAP<br>
&gt; &gt; +static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)<br>
&gt; &gt; +{<br>
&gt; &gt; +       int idx = section_active_index(PFN_PHYS(pfn));<br>
&gt; &gt; +<br>
&gt; &gt; +       return !!(ms-&gt;usage-&gt;map_active &amp; (1UL &lt;&lt; idx));<br>
&gt; &gt; +}<br>
&gt; &gt; +#else<br>
&gt; &gt; +static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)<br>
&gt; &gt; +{<br>
&gt; &gt; +       return 1;<br>
&gt; &gt; +}<br>
&gt; &gt; +#endif<br>
&gt; &gt; +<br>
&gt; &gt;  #ifndef CONFIG_HAVE_ARCH_PFN_VALID<br>
&gt; &gt;  static inline int pfn_valid(unsigned long pfn)<br>
&gt; &gt;  {<br>
&gt; &gt; +       struct mem_section *ms;<br>
&gt; &gt; +<br>
&gt; &gt;         if (pfn_to_section_nr(pfn) &gt;= NR_MEM_SECTIONS)<br>
&gt; &gt;                 return 0;<br>
&gt; &gt; -       return valid_section(__nr_to_section(pfn_to_section_nr(pfn)));<br>
&gt; &gt; +       ms = __nr_to_section(pfn_to_section_nr(pfn));<br>
&gt; &gt; +       if (!valid_section(ms))<br>
&gt; &gt; +               return 0;<br>
&gt; &gt; +       return pfn_section_valid(ms, pfn);<br>
&gt; &gt;  }<br>
&gt; &gt;  #endif<br>
&gt; &gt;<br>
&gt; &gt; @@ -1349,6 +1375,7 @@ void sparse_init(void);<br>
&gt; &gt;  #define sparse_init()  do {} while (0)<br>
&gt; &gt;  #define sparse_index_init(_sec, _nid)  do {} while (0)<br>
&gt; &gt;  #define pfn_present pfn_valid<br>
&gt; &gt; +#define section_active_init(_pfn, _nr_pages) do {} while (0)<br>
&gt; &gt;  #endif /* CONFIG_SPARSEMEM */<br>
&gt; &gt;<br>
&gt; &gt;  /*<br>
&gt; &gt; diff --git a/mm/page_alloc.c b/mm/page_alloc.c<br>
&gt; &gt; index f671401a7c0b..c9ad28a78018 100644<br>
&gt; &gt; --- a/mm/page_alloc.c<br>
&gt; &gt; +++ b/mm/page_alloc.c<br>
&gt; &gt; @@ -7273,10 +7273,12 @@ void __init free_area_init_nodes(unsigned long *max_zone_pfn)<br>
&gt; &gt;<br>
&gt; &gt;         /* Print out the early node map */<br>
&gt; &gt;         pr_info(&quot;Early memory node ranges\n&quot;);<br>
&gt; &gt; -       for_each_mem_pfn_range(i, MAX_NUMNODES, &amp;start_pfn, &amp;end_pfn, &amp;nid)<br>
&gt; &gt; +       for_each_mem_pfn_range(i, MAX_NUMNODES, &amp;start_pfn, &amp;end_pfn, &amp;nid) {<br>
&gt; &gt;                 pr_info(&quot;  node %3d: [mem %#018Lx-%#018Lx]\n&quot;, nid,<br>
&gt; &gt;                         (u64)start_pfn &lt;&lt; PAGE_SHIFT,<br>
&gt; &gt;                         ((u64)end_pfn &lt;&lt; PAGE_SHIFT) - 1);<br>
&gt; &gt; +               section_active_init(start_pfn, end_pfn - start_pfn);<br>
&gt; &gt; +       }<br>
&gt; &gt;<br>
&gt; &gt;         /* Initialise every node */<br>
&gt; &gt;         mminit_verify_pageflags_layout();<br>
&gt; &gt; diff --git a/mm/sparse.c b/mm/sparse.c<br>
&gt; &gt; index f87de7ad32c8..5ef2f884c4e1 100644<br>
&gt; &gt; --- a/mm/sparse.c<br>
&gt; &gt; +++ b/mm/sparse.c<br>
&gt; &gt; @@ -210,6 +210,54 @@ static inline unsigned long first_present_section_nr(void)<br>
&gt; &gt;         return next_present_section_nr(-1);<br>
&gt; &gt;  }<br>
&gt; &gt;<br>
&gt; &gt; +static unsigned long section_active_mask(unsigned long pfn,<br>
&gt; &gt; +               unsigned long nr_pages)<br>
&gt; &gt; +{<br>
&gt; &gt; +       int idx_start, idx_size;<br>
&gt; &gt; +       phys_addr_t start, size;<br>
&gt; &gt; +<br>
&gt; &gt; +       if (!nr_pages)<br>
&gt; &gt; +               return 0;<br>
&gt; &gt; +<br>
&gt; &gt; +       start = PFN_PHYS(pfn);<br>
&gt; &gt; +       size = PFN_PHYS(min(nr_pages, PAGES_PER_SECTION<br>
&gt; &gt; +                               - (pfn &amp; ~PAGE_SECTION_MASK)));<br>
&gt; &gt; +       size = ALIGN(size, SECTION_ACTIVE_SIZE);<br>
&gt; &gt; +<br>
&gt; &gt; +       idx_start = section_active_index(start);<br>
&gt; &gt; +       idx_size = section_active_index(size);<br>
&gt; &gt; +<br>
&gt; &gt; +       if (idx_size == 0)<br>
&gt; &gt; +               return -1;<br>
&gt; &gt; +       return ((1UL &lt;&lt; idx_size) - 1) &lt;&lt; idx_start;<br>
&gt; &gt; +}<br>
&gt; &gt; +<br>
&gt; &gt; +void section_active_init(unsigned long pfn, unsigned long nr_pages)<br>
&gt; &gt; +{<br>
&gt; &gt; +       int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);<br>
&gt; &gt; +       int i, start_sec = pfn_to_section_nr(pfn);<br>
&gt; &gt; +<br>
&gt; &gt; +       if (!nr_pages)<br>
&gt; &gt; +               return;<br>
&gt; &gt; +<br>
&gt; &gt; +       for (i = start_sec; i &lt;= end_sec; i++) {<br>
&gt; &gt; +               struct mem_section *ms;<br>
&gt; &gt; +               unsigned long mask;<br>
&gt; &gt; +               unsigned long pfns;<br>
&gt; &gt; +<br>
&gt; &gt; +               pfns = min(nr_pages, PAGES_PER_SECTION<br>
&gt; &gt; +                               - (pfn &amp; ~PAGE_SECTION_MASK));<br>
&gt; &gt; +               mask = section_active_mask(pfn, pfns);<br>
&gt; &gt; +<br>
&gt; &gt; +               ms = __nr_to_section(i);<br>
&gt; &gt; +               pr_debug(&quot;%s: sec: %d mask: %#018lx\n&quot;, __func__, i, mask);<br>
&gt; &gt; +               ms-&gt;usage-&gt;map_active = mask;<br>
&gt; &gt; +<br>
&gt; &gt; +               pfn += pfns;<br>
&gt; &gt; +               nr_pages -= pfns;<br>
&gt; &gt; +       }<br>
&gt; &gt; +}<br>
&gt;<br>
&gt; For some reasons the above code is confusing to me. It seems all the<br>
&gt; code supposed to do is set all map_active to -1, and trim the first<br>
&gt; and last sections (can be the same section of course). So, I would<br>
&gt; replace the above two functions with one function like this:<br>
&gt;<br>
&gt; void section_active_init(unsigned long pfn, unsigned long nr_pages)<br>
&gt; {<br>
&gt;         int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);<br>
&gt;         int i, idx, start_sec = pfn_to_section_nr(pfn);<br>
&gt;         struct mem_section *ms;<br>
&gt;<br>
&gt;         if (!nr_pages)<br>
&gt;                 return;<br>
&gt;<br>
&gt;         for (i = start_sec; i &lt;= end_sec; i++) {<br>
&gt;                 ms = __nr_to_section(i);<br>
&gt;                 ms-&gt;usage-&gt;map_active = ~0ul;<br>
&gt;         }<br>
&gt;<br>
&gt;         /* Might need to trim active pfns from the beginning and end */<br>
&gt;         idx = section_active_index(PFN_PHYS(pfn));<br>
&gt;         ms = __nr_to_section(start_sec);<br>
&gt;         ms-&gt;usage-&gt;map_active &amp;= (~0ul &lt;&lt; idx);<br>
&gt;<br>
&gt;         idx = section_active_index(PFN_PHYS(pfn + nr_pages -1));<br>
&gt;         ms = __nr_to_section(end_sec);<br>
&gt;         ms-&gt;usage-&gt;map_active &amp;= (~0ul &gt;&gt; (BITS_PER_LONG - idx - 1));<br>
&gt; }<br>
<br>
I like the cleanup, but one of the fixes in v7 resulted in the<br>
realization that a given section may be populated twice at init time.<br>
For example, enabling that pr_debug() yields:<br>
<br>
    section_active_init: sec: 12 mask: 0x00000003ffffffff<br>
    section_active_init: sec: 12 mask: 0xe000000000000000<br>
<br>
So, the implementation can&#39;t blindly clear bits based on the current<br>
parameters. However, I&#39;m switching this code over to use bitmap_*()<br>
helpers which should help with the readability.<br></blockquote></div></div><div dir="auto"><br></div><div dir="auto">Yes, bitmap_* will help, and I assume you will also make active_map size scalable?</div><div dir="auto"><br></div><div dir="auto">I will take another look at version 8.</div><div dir="auto"><br></div><div dir="auto"><br></div><div dir="auto">Thank you,</div><div dir="auto">Pasha</div><div dir="auto"><br></div><div dir="auto"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
</blockquote></div></div></div>

Patch
diff mbox series

diff --git a/include/linux/mmzone.h b/include/linux/mmzone.h
index 6726fc175b51..cffde898e345 100644
--- a/include/linux/mmzone.h
+++ b/include/linux/mmzone.h
@@ -1175,6 +1175,8 @@  struct mem_section_usage {
 	unsigned long pageblock_flags[0];
 };
 
+void section_active_init(unsigned long pfn, unsigned long nr_pages);
+
 struct page;
 struct page_ext;
 struct mem_section {
@@ -1312,12 +1314,36 @@  static inline struct mem_section *__pfn_to_section(unsigned long pfn)
 
 extern int __highest_present_section_nr;
 
+static inline int section_active_index(phys_addr_t phys)
+{
+	return (phys & ~(PA_SECTION_MASK)) / SECTION_ACTIVE_SIZE;
+}
+
+#ifdef CONFIG_SPARSEMEM_VMEMMAP
+static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
+{
+	int idx = section_active_index(PFN_PHYS(pfn));
+
+	return !!(ms->usage->map_active & (1UL << idx));
+}
+#else
+static inline int pfn_section_valid(struct mem_section *ms, unsigned long pfn)
+{
+	return 1;
+}
+#endif
+
 #ifndef CONFIG_HAVE_ARCH_PFN_VALID
 static inline int pfn_valid(unsigned long pfn)
 {
+	struct mem_section *ms;
+
 	if (pfn_to_section_nr(pfn) >= NR_MEM_SECTIONS)
 		return 0;
-	return valid_section(__nr_to_section(pfn_to_section_nr(pfn)));
+	ms = __nr_to_section(pfn_to_section_nr(pfn));
+	if (!valid_section(ms))
+		return 0;
+	return pfn_section_valid(ms, pfn);
 }
 #endif
 
@@ -1349,6 +1375,7 @@  void sparse_init(void);
 #define sparse_init()	do {} while (0)
 #define sparse_index_init(_sec, _nid)  do {} while (0)
 #define pfn_present pfn_valid
+#define section_active_init(_pfn, _nr_pages) do {} while (0)
 #endif /* CONFIG_SPARSEMEM */
 
 /*
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index f671401a7c0b..c9ad28a78018 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -7273,10 +7273,12 @@  void __init free_area_init_nodes(unsigned long *max_zone_pfn)
 
 	/* Print out the early node map */
 	pr_info("Early memory node ranges\n");
-	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid)
+	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
 		pr_info("  node %3d: [mem %#018Lx-%#018Lx]\n", nid,
 			(u64)start_pfn << PAGE_SHIFT,
 			((u64)end_pfn << PAGE_SHIFT) - 1);
+		section_active_init(start_pfn, end_pfn - start_pfn);
+	}
 
 	/* Initialise every node */
 	mminit_verify_pageflags_layout();
diff --git a/mm/sparse.c b/mm/sparse.c
index f87de7ad32c8..5ef2f884c4e1 100644
--- a/mm/sparse.c
+++ b/mm/sparse.c
@@ -210,6 +210,54 @@  static inline unsigned long first_present_section_nr(void)
 	return next_present_section_nr(-1);
 }
 
+static unsigned long section_active_mask(unsigned long pfn,
+		unsigned long nr_pages)
+{
+	int idx_start, idx_size;
+	phys_addr_t start, size;
+
+	if (!nr_pages)
+		return 0;
+
+	start = PFN_PHYS(pfn);
+	size = PFN_PHYS(min(nr_pages, PAGES_PER_SECTION
+				- (pfn & ~PAGE_SECTION_MASK)));
+	size = ALIGN(size, SECTION_ACTIVE_SIZE);
+
+	idx_start = section_active_index(start);
+	idx_size = section_active_index(size);
+
+	if (idx_size == 0)
+		return -1;
+	return ((1UL << idx_size) - 1) << idx_start;
+}
+
+void section_active_init(unsigned long pfn, unsigned long nr_pages)
+{
+	int end_sec = pfn_to_section_nr(pfn + nr_pages - 1);
+	int i, start_sec = pfn_to_section_nr(pfn);
+
+	if (!nr_pages)
+		return;
+
+	for (i = start_sec; i <= end_sec; i++) {
+		struct mem_section *ms;
+		unsigned long mask;
+		unsigned long pfns;
+
+		pfns = min(nr_pages, PAGES_PER_SECTION
+				- (pfn & ~PAGE_SECTION_MASK));
+		mask = section_active_mask(pfn, pfns);
+
+		ms = __nr_to_section(i);
+		pr_debug("%s: sec: %d mask: %#018lx\n", __func__, i, mask);
+		ms->usage->map_active = mask;
+
+		pfn += pfns;
+		nr_pages -= pfns;
+	}
+}
+
 /* Record a memory area against a node. */
 void __init memory_present(int nid, unsigned long start, unsigned long end)
 {