mm: filemap: batch mm counter updating in filemap_map_pages()

Like copy_pte_range()/zap_pte_range(), make mm counter batch updating in
filemap_map_pages(), since folios type are same(MM_SHMEMPAGES or
MM_FILEPAGES) in filemap_map_pages(), only check the first folio type is
enough, the 'lat_pagefault -P 1 file' test from lmbench shows 12%
improvement, and the percpu_counter_add_batch() is gone from perf flame
graph.

Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Kefeng Wang <[email protected]>
Cc: Matthew Wilcox (Oracle) <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

mm/filemap.c

@@ -3505,7 +3505,7 @@ static struct folio *next_uptodate_folio(struct xa_state *xas,
 static vm_fault_t filemap_map_folio_range(struct vm_fault *vmf,
 			struct folio *folio, unsigned long start,
 			unsigned long addr, unsigned int nr_pages,
-			unsigned int *mmap_miss)
+			unsigned long *rss, unsigned int *mmap_miss)
 {
 	vm_fault_t ret = 0;
 	struct page *page = folio_page(folio, start);
@@ -3539,8 +3539,7 @@ static vm_fault_t filemap_map_folio_range(struct vm_fault *vmf,
 skip:
 		if (count) {
 			set_pte_range(vmf, folio, page, count, addr);
-			add_mm_counter(vmf->vma->vm_mm, mm_counter_file(folio),
-				       count);
+			*rss += count;
 			folio_ref_add(folio, count);
 			if (in_range(vmf->address, addr, count * PAGE_SIZE))
 				ret = VM_FAULT_NOPAGE;
@@ -3555,7 +3554,7 @@ static vm_fault_t filemap_map_folio_range(struct vm_fault *vmf,
 
 	if (count) {
 		set_pte_range(vmf, folio, page, count, addr);
-		add_mm_counter(vmf->vma->vm_mm, mm_counter_file(folio), count);
+		*rss += count;
 		folio_ref_add(folio, count);
 		if (in_range(vmf->address, addr, count * PAGE_SIZE))
 			ret = VM_FAULT_NOPAGE;
@@ -3568,7 +3567,7 @@ static vm_fault_t filemap_map_folio_range(struct vm_fault *vmf,
 
 static vm_fault_t filemap_map_order0_folio(struct vm_fault *vmf,
 		struct folio *folio, unsigned long addr,
-		unsigned int *mmap_miss)
+		unsigned long *rss, unsigned int *mmap_miss)
 {
 	vm_fault_t ret = 0;
 	struct page *page = &folio->page;
@@ -3592,7 +3591,7 @@ static vm_fault_t filemap_map_order0_folio(struct vm_fault *vmf,
 		ret = VM_FAULT_NOPAGE;
 
 	set_pte_range(vmf, folio, page, 1, addr);
-	add_mm_counter(vmf->vma->vm_mm, mm_counter_file(folio), 1);
+	(*rss)++;
 	folio_ref_inc(folio);
 
 	return ret;
@@ -3609,7 +3608,8 @@ vm_fault_t filemap_map_pages(struct vm_fault *vmf,
 	XA_STATE(xas, &mapping->i_pages, start_pgoff);
 	struct folio *folio;
 	vm_fault_t ret = 0;
-	unsigned int nr_pages = 0, mmap_miss = 0, mmap_miss_saved;
+	unsigned long rss = 0;
+	unsigned int nr_pages = 0, mmap_miss = 0, mmap_miss_saved, folio_type;
 
 	rcu_read_lock();
 	folio = next_uptodate_folio(&xas, mapping, end_pgoff);
@@ -3628,6 +3628,8 @@ vm_fault_t filemap_map_pages(struct vm_fault *vmf,
 		folio_put(folio);
 		goto out;
 	}
+
+	folio_type = mm_counter_file(folio);
 	do {
 		unsigned long end;
 
@@ -3639,15 +3641,16 @@ vm_fault_t filemap_map_pages(struct vm_fault *vmf,
 
 		if (!folio_test_large(folio))
 			ret |= filemap_map_order0_folio(vmf,
-					folio, addr, &mmap_miss);
+					folio, addr, &rss, &mmap_miss);
 		else
 			ret |= filemap_map_folio_range(vmf, folio,
 					xas.xa_index - folio->index, addr,
-					nr_pages, &mmap_miss);
+					nr_pages, &rss, &mmap_miss);
 
 		folio_unlock(folio);
 		folio_put(folio);
 	} while ((folio = next_uptodate_folio(&xas, mapping, end_pgoff)) != NULL);
+	add_mm_counter(vma->vm_mm, folio_type, rss);
 	pte_unmap_unlock(vmf->pte, vmf->ptl);
 out:
 	rcu_read_unlock();