mm: alloc pages for pmem from peer node

euleros inclusion
category: feature
feature: etmem
bugzilla: 48246

-------------------------------------------------

Pmem is slower than dram. So alloc pages from pmem's peer
dram node to accelerate access to pmem page struct.
Signed-off-by: NShijie Luo <luoshijie1@huawei.com>
Signed-off-by: Kemeng Shi<shikemeng@huawei.com>
Reviewed-by: louhongxiang <louhongxiang@huawei.com
Signed-off-by: NZheng Zengkai <zhengzengkai@huawei.com>

include/linux/mm.h

@@ -3056,7 +3056,7 @@ extern int sysctl_memory_failure_recovery;
 extern void shake_page(struct page *p, int access);
 extern atomic_long_t num_poisoned_pages __read_mostly;
 extern int soft_offline_page(unsigned long pfn, int flags);
-
+extern unsigned int sysctl_vmemmap_block_from_dram;
 
 /*
  * Error handlers for various types of pages.

kernel/sysctl.c

@@ -3158,6 +3158,15 @@ static struct ctl_table vm_table[] = {
 		.extra2         = SYSCTL_ONE,
 	},
 #endif
+	{
+		.procname       = "vmemmap_block_from_dram",
+		.data           = &sysctl_vmemmap_block_from_dram,
+		.maxlen         = sizeof(sysctl_vmemmap_block_from_dram),
+		.mode           = 0600,
+		.proc_handler   = proc_dointvec_minmax,
+		.extra1         = SYSCTL_ZERO,
+		.extra2         = SYSCTL_ONE,
+	},
 	{ }
 };
 

mm/memory_hotplug.c

@@ -914,6 +914,7 @@ static pg_data_t __ref *hotadd_new_pgdat(int nid)
 	/* we can use NODE_DATA(nid) from here */
 	pgdat->node_id = nid;
 	pgdat->node_start_pfn = 0;
+	pgdat->peer_node = find_best_peer_node(nid);
 
 	/* init node's zones as empty zones, we don't have any present pages.*/
 	free_area_init_core_hotplug(nid);

mm/sparse-vmemmap.c

@@ -30,6 +30,8 @@
 #include <asm/dma.h>
 #include <asm/pgalloc.h>
 
+unsigned int sysctl_vmemmap_block_from_dram;
+
 /*
  * Allocate a block of memory to be used to back the virtual memory map
  * or to back the page tables that are used to create the mapping.
@@ -54,6 +56,12 @@ void * __meminit vmemmap_alloc_block(unsigned long size, int node)
 		static bool warned;
 		struct page *page;
 
+		/* if node is pmem node, alloc_pages from it's peer
+		 * dram node to accelerate access to page struct
+		 */
+		if (is_node_pmem(node) && sysctl_vmemmap_block_from_dram)
+			node = NODE_DATA(node)->peer_node;
+
 		page = alloc_pages_node(node, gfp_mask, order);
 		if (page)
 			return page_address(page);