[External] [RFC PATCH v1 5/6] mm: get zone spanned pages separately for DRAM and NVDIMM

[Huaisheng HS1 Ye]

DRAM and NVDIMM are divided into separate zones, thus NVM
zone is dedicated for NVDIMMs.

During zone_spanned_pages_in_node, spanned pages of zones
are calculated separately for DRAM and NVDIMM by flags
MEMBLOCK_NONE and MEMBLOCK_NVDIMM.

Signed-off-by: Huaisheng Ye <yehs1@xxxxxxxxxx>
Signed-off-by: Ocean He <hehy1@xxxxxxxxxx>
---
 mm/nobootmem.c  |  5 +++--
 mm/page_alloc.c | 40 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 43 insertions(+), 2 deletions(-)

diff --git a/mm/nobootmem.c b/mm/nobootmem.c
index 9b02fda..19b5291 100644
--- a/mm/nobootmem.c
+++ b/mm/nobootmem.c
@@ -143,8 +143,9 @@ static unsigned long __init free_low_memory_core_early(void)
 	 *  because in some case like Node0 doesn't have RAM installed
 	 *  low ram will be on Node1
 	 */
-	for_each_free_mem_range(i, NUMA_NO_NODE, MEMBLOCK_NONE, &start, &end,
-				NULL)
+	for_each_free_mem_range(i, NUMA_NO_NODE,
+				MEMBLOCK_NONE | MEMBLOCK_NVDIMM,
+				&start, &end, NULL)
 		count += __free_memory_core(start, end);
 
 	return count;
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index d8bd20d..3fd0d95 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -4221,6 +4221,11 @@ static inline void finalise_ac(gfp_t gfp_mask,
 	 * also used as the starting point for the zonelist iterator. It
 	 * may get reset for allocations that ignore memory policies.
 	 */
+#ifdef CONFIG_ZONE_NVM
+	/* Bypass ZONE_NVM for Normal alloctions */
+	if (ac->high_zoneidx > ZONE_NVM)
+		ac->high_zoneidx = ZONE_NORMAL;
+#endif
 	ac->preferred_zoneref = first_zones_zonelist(ac->zonelist,
 					ac->high_zoneidx, ac->nodemask);
 }
@@ -5808,6 +5813,10 @@ static unsigned long __meminit zone_spanned_pages_in_node(int nid,
 					unsigned long *zone_end_pfn,
 					unsigned long *ignored)
 {
+#ifdef CONFIG_ZONE_NVM
+	unsigned long start_pfn, end_pfn;
+#endif
+
 	/* When hotadd a new node from cpu_up(), the node should be empty */
 	if (!node_start_pfn && !node_end_pfn)
 		return 0;
@@ -5815,6 +5824,26 @@ static unsigned long __meminit zone_spanned_pages_in_node(int nid,
 	/* Get the start and end of the zone */
 	*zone_start_pfn = arch_zone_lowest_possible_pfn[zone_type];
 	*zone_end_pfn = arch_zone_highest_possible_pfn[zone_type];
+
+#ifdef CONFIG_ZONE_NVM
+	/*
+	 * Use zone_type to adjust zone size again.
+	 */
+	if (zone_type == ZONE_NVM) {
+		get_pfn_range_for_nid_with_flags(nid, &start_pfn, &end_pfn,
+							MEMBLOCK_NVDIMM);
+	} else {
+		get_pfn_range_for_nid_with_flags(nid, &start_pfn, &end_pfn,
+							MEMBLOCK_NONE);
+	}
+
+	if (*zone_end_pfn < start_pfn || *zone_start_pfn > end_pfn)
+		return 0;
+	/* Move the zone boundaries inside the possile_pfn if necessary */
+	*zone_end_pfn = min(*zone_end_pfn, end_pfn);
+	*zone_start_pfn = max(*zone_start_pfn, start_pfn);
+#endif
+
 	adjust_zone_range_for_zone_movable(nid, zone_type,
 				node_start_pfn, node_end_pfn,
 				zone_start_pfn, zone_end_pfn);
@@ -6680,6 +6709,17 @@ void __init free_area_init_nodes(unsigned long *max_zone_pfn)
 		start_pfn = end_pfn;
 	}
 
+#ifdef CONFIG_ZONE_NVM
+	/*
+	 * Adjust nvm zone included in normal zone
+	 */
+	get_pfn_range_for_nid_with_flags(MAX_NUMNODES, &start_pfn, &end_pfn,
+							    MEMBLOCK_NVDIMM);
+
+	arch_zone_lowest_possible_pfn[ZONE_NVM] = start_pfn;
+	arch_zone_highest_possible_pfn[ZONE_NVM] = end_pfn;
+#endif
+
 	/* Find the PFNs that ZONE_MOVABLE begins at in each node */
 	memset(zone_movable_pfn, 0, sizeof(zone_movable_pfn));
 	find_zone_movable_pfns_for_nodes();
-- 
1.8.3.1