thp: use mm_file_counter to determine update which rss counter
authorYang Shi <yang.shi@linux.alibaba.com>
Fri, 17 Aug 2018 22:44:55 +0000 (15:44 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 17 Aug 2018 23:20:28 +0000 (16:20 -0700)
Since commit eca56ff906bd ("mm, shmem: add internal shmem resident
memory accounting"), MM_SHMEMPAGES is added to separate the shmem
accounting from regular files.  So, all shmem pages should be accounted
to MM_SHMEMPAGES instead of MM_FILEPAGES.

And, normal 4K shmem pages have been accounted to MM_SHMEMPAGES, so
shmem thp pages should be not treated differently.  Account them to
MM_SHMEMPAGES via mm_counter_file() since shmem pages are swap backed to
keep consistent with normal 4K shmem pages.

This will not change the rss counter of processes since shmem pages are
still a part of it.

The /proc/pid/status and /proc/pid/statm counters will however be more
accurate wrt shmem usage, as originally intended.  And as eca56ff906bd
("mm, shmem: add internal shmem resident memory accounting") mentioned,
oom also could report more accurate "shmem-rss".

Link: http://lkml.kernel.org/r/1529442518-17398-1-git-send-email-yang.shi@linux.alibaba.com
Signed-off-by: Yang Shi <yang.shi@linux.alibaba.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Hugh Dickins <hughd@google.com>
Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/huge_memory.c
mm/memory.c

index 1ce44e87f4949c3cf4a0dc9f338b2e583957d815..064a9d78879d1a99bcd44cb8cd037cb17558191f 100644 (file)
@@ -1740,7 +1740,7 @@ int zap_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
                } else {
                        if (arch_needs_pgtable_deposit())
                                zap_deposited_table(tlb->mm, pmd);
-                       add_mm_counter(tlb->mm, MM_FILEPAGES, -HPAGE_PMD_NR);
+                       add_mm_counter(tlb->mm, mm_counter_file(page), -HPAGE_PMD_NR);
                }
 
                spin_unlock(ptl);
@@ -2090,7 +2090,7 @@ static void __split_huge_pmd_locked(struct vm_area_struct *vma, pmd_t *pmd,
                        SetPageReferenced(page);
                page_remove_rmap(page, true);
                put_page(page);
-               add_mm_counter(mm, MM_FILEPAGES, -HPAGE_PMD_NR);
+               add_mm_counter(mm, mm_counter_file(page), -HPAGE_PMD_NR);
                return;
        } else if (is_huge_zero_pmd(*pmd)) {
                /*
index 7c3bd119fccaa3755fb99636caf69b1308f11123..d7b5b22a1a0a35fbab402099ffe29fa53b74d928 100644 (file)
@@ -3400,7 +3400,7 @@ static int do_set_pmd(struct vm_fault *vmf, struct page *page)
        if (write)
                entry = maybe_pmd_mkwrite(pmd_mkdirty(entry), vma);
 
-       add_mm_counter(vma->vm_mm, MM_FILEPAGES, HPAGE_PMD_NR);
+       add_mm_counter(vma->vm_mm, mm_counter_file(page), HPAGE_PMD_NR);
        page_add_file_rmap(page, true);
        /*
         * deposit and withdraw with pmd lock held