diff --git a/mm/percpu-km.c b/mm/percpu-km.c
index c84a9f781a6c51c09f0a8065aba0383600ec8168..01e31bd55860254ec9821e93eadadb8e7c014109 100644
--- a/mm/percpu-km.c
+++ b/mm/percpu-km.c
@@ -32,6 +32,12 @@
 
 #include <linux/log2.h>
 
+static void pcpu_post_unmap_tlb_flush(struct pcpu_chunk *chunk,
+				      int page_start, int page_end)
+{
+	/* nothing */
+}
+
 static int pcpu_populate_chunk(struct pcpu_chunk *chunk,
 			       int page_start, int page_end, gfp_t gfp)
 {
diff --git a/mm/percpu-vm.c b/mm/percpu-vm.c
index c75f6f24f2d5a00469ca08b069a671cf7f435202..ba07eb5e014b80fa7d80d107ba502b853f2a4cdc 100644
--- a/mm/percpu-vm.c
+++ b/mm/percpu-vm.c
@@ -302,6 +302,9 @@ static int pcpu_populate_chunk(struct pcpu_chunk *chunk,
  * For each cpu, depopulate and unmap pages [@page_start,@page_end)
  * from @chunk.
  *
+ * Caller is required to call pcpu_post_unmap_tlb_flush() if not returning the
+ * region back to vmalloc() which will lazily flush the tlb.
+ *
  * CONTEXT:
  * pcpu_alloc_mutex.
  */
@@ -323,8 +326,6 @@ static void pcpu_depopulate_chunk(struct pcpu_chunk *chunk,
 
 	pcpu_unmap_pages(chunk, pages, page_start, page_end);
 
-	/* no need to flush tlb, vmalloc will handle it lazily */
-
 	pcpu_free_pages(chunk, pages, page_start, page_end);
 }
 
diff --git a/mm/percpu.c b/mm/percpu.c
index ceae1ed8b5cfe0a6b29a7c32fb4941887f38a08a..42d95c0f5c4feca98a8db58c187211abb57e1894 100644
--- a/mm/percpu.c
+++ b/mm/percpu.c
@@ -1579,6 +1579,7 @@ static void pcpu_chunk_depopulated(struct pcpu_chunk *chunk,
  *
  * pcpu_populate_chunk		- populate the specified range of a chunk
  * pcpu_depopulate_chunk	- depopulate the specified range of a chunk
+ * pcpu_post_unmap_tlb_flush	- flush tlb for the specified range of a chunk
  * pcpu_create_chunk		- create a new chunk
  * pcpu_destroy_chunk		- destroy a chunk, always preceded by full depop
  * pcpu_addr_to_page		- translate address to physical address
@@ -1590,6 +1591,8 @@ static void pcpu_depopulate_chunk(struct pcpu_chunk *chunk,
 				  int page_start, int page_end);
 static struct pcpu_chunk *pcpu_create_chunk(enum pcpu_chunk_type type,
 					    gfp_t gfp);
+static void pcpu_post_unmap_tlb_flush(struct pcpu_chunk *chunk,
+				      int page_start, int page_end);
 static void pcpu_destroy_chunk(struct pcpu_chunk *chunk);
 static struct page *pcpu_addr_to_page(void *addr);
 static int __init pcpu_verify_alloc_info(const struct pcpu_alloc_info *ai);
@@ -2137,11 +2140,12 @@ static void pcpu_reclaim_populated(enum pcpu_chunk_type type)
 	struct list_head *pcpu_slot = pcpu_chunk_list(type);
 	struct pcpu_chunk *chunk;
 	struct pcpu_block_md *block;
+	int freed_page_start, freed_page_end;
 	int i, end;
+	bool reintegrate;
 
 	spin_lock_irq(&pcpu_lock);
 
-restart:
 	/*
 	 * Once a chunk is isolated to the to_depopulate list, the chunk is no
 	 * longer discoverable to allocations whom may populate pages.  The only
@@ -2157,6 +2161,9 @@ static void pcpu_reclaim_populated(enum pcpu_chunk_type type)
 		 * Scan chunk's pages in the reverse order to keep populated
 		 * pages close to the beginning of the chunk.
 		 */
+		freed_page_start = chunk->nr_pages;
+		freed_page_end = 0;
+		reintegrate = false;
 		for (i = chunk->nr_pages - 1, end = -1; i >= 0; i--) {
 			/* no more work to do */
 			if (chunk->nr_empty_pop_pages == 0)
@@ -2165,8 +2172,8 @@ static void pcpu_reclaim_populated(enum pcpu_chunk_type type)
 			/* reintegrate chunk to prevent atomic alloc failures */
 			if (pcpu_nr_empty_pop_pages[type] <
 			    PCPU_EMPTY_POP_PAGES_HIGH) {
-				pcpu_reintegrate_chunk(chunk);
-				goto restart;
+				reintegrate = true;
+				goto end_chunk;
 			}
 
 			/*
@@ -2195,16 +2202,29 @@ static void pcpu_reclaim_populated(enum pcpu_chunk_type type)
 			spin_lock_irq(&pcpu_lock);
 
 			pcpu_chunk_depopulated(chunk, i + 1, end + 1);
+			freed_page_start = min(freed_page_start, i + 1);
+			freed_page_end = max(freed_page_end, end + 1);
 
 			/* reset the range and continue */
 			end = -1;
 		}
 
-		if (chunk->free_bytes == pcpu_unit_size)
+end_chunk:
+		/* batch tlb flush per chunk to amortize cost */
+		if (freed_page_start < freed_page_end) {
+			spin_unlock_irq(&pcpu_lock);
+			pcpu_post_unmap_tlb_flush(chunk,
+						  freed_page_start,
+						  freed_page_end);
+			cond_resched();
+			spin_lock_irq(&pcpu_lock);
+		}
+
+		if (reintegrate || chunk->free_bytes == pcpu_unit_size)
 			pcpu_reintegrate_chunk(chunk);
 		else
-			list_move(&chunk->list,
-				  &pcpu_slot[pcpu_sidelined_slot]);
+			list_move_tail(&chunk->list,
+				       &pcpu_slot[pcpu_sidelined_slot]);
 	}
 
 	spin_unlock_irq(&pcpu_lock);