mm/rmap: cleanup partially-mapped handling in __folio_remove_rmap()

Let's simplify and reduce code indentation. In the RMAP_LEVEL_PTE case, we already check for nr when computing partially_mapped.

For RMAP_LEVEL_PMD, it's a bit more confusing. Likely, we don't need the "nr" check, but we could have "nr < nr_pmdmapped" also if we stumbled into the "/* Raced ahead of another remove and an add? */" case. So let's simply move the nr check in there.

Note that partially_mapped is always false for small folios.

No functional change intended.

Cc Yosry Ahmed <[email protected]>

Signed-off-by David Hildenbrand <[email protected]>

---
 mm/rmap.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

Let's simplify and reduce code indentation. In the RMAP_LEVEL_PTE case, we already check for nr when computing partially_mapped.

For RMAP_LEVEL_PMD, it's a bit more confusing. Likely, we don't need the "nr" check, but we could have "nr < nr_pmdmapped" also if we stumbled into the "/* Raced ahead of another remove and an add? */" case. So let's simply move the nr check in there.

Note that partially_mapped is always false for small folios.

No functional change intended.

Cc Yosry Ahmed <[email protected]>

Signed-off-by David Hildenbrand <[email protected]>

---
 mm/rmap.c | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

mm/rmap.c +3 -2

index afec4b6800caf..aa900e46cdf82 100644
@@ -1551,11 +1551,12 @@ static __always_inline void __folio_remove_rmap(struct folio *folio,
                        }
                }
 
-               partially_mapped = nr < nr_pmdmapped;
+               partially_mapped = nr && nr < nr_pmdmapped;
                break;
        }
 
-       if (nr) {
+       if (partially_mapped && folio_test_anon(folio) &&
+           list_empty(&folio->_deferred_list))
                /*
                 * Queue anon large folio for deferred split if at least one
                 * page of the folio is unmapped and at least one page

mm/rmap.c +0 -1

        __folio_mod_stat(folio, -nr, -nr_pmdmapped);
 
        /*
--

2.45.2

Quoted 9 lines · David Hildenbrand

Let's simplify and reduce code indentation. In the RMAP_LEVEL_PTE case, we already check for nr when computing partially_mapped.

For RMAP_LEVEL_PMD, it's a bit more confusing. Likely, we don't need the "nr" check, but we could have "nr < nr_pmdmapped" also if we stumbled into the "/* Raced ahead of another remove and an add? */" case. So let's simply move the nr check in there.

Quoted 6 lines

Note that partially_mapped is always false for small folios.

No functional change intended.

Cc Yosry Ahmed <[email protected]>

Signed-off-by David Hildenbrand <[email protected]>

Makes sense to me. FWIW:

Reviewed-by Yosry Ahmed <[email protected]>

Quoted 24 lines

---
  mm/rmap.c | 10 ++++------
  1 file changed, 4 insertions(+), 6 deletions(-)

mm/rmap.c +3 -2

index afec4b6800caf..aa900e46cdf82 100644
@@ -1551,11 +1551,12 @@ static __always_inline void __folio_remove_rmap(struct folio *folio,
                         }
                 }
 
-               partially_mapped = nr < nr_pmdmapped;
+               partially_mapped = nr && nr < nr_pmdmapped;
                 break;
         }
 
-       if (nr) {
+       if (partially_mapped && folio_test_anon(folio) &&
+           list_empty(&folio->_deferred_list))
                 /*
                  * Queue anon large folio for deferred split if at least one
                  * page of the folio is unmapped and at least one page

mm/rmap.c static __always_inline void __folio_remove_rmap(struct folio *folio, · line 1567

Quoted 8 lines

mm/rmap.c +1 -4

@@ -1563,10 +1564,7 @@ static __always_inline void __folio_remove_rmap(struct folio *folio,
                  *
                  * Check partially_mapped first to ensure it is a large folio.
                  */
-               if (folio_test_anon(folio) && partially_mapped &&
-                   list_empty(&folio->_deferred_list))
-                       deferred_split_folio(folio);
-       }
+               deferred_split_folio(folio);

FWIW, I prefer moving the comment out of the one-line if block as you originally suggested in [1].

[1]https://lore.kernel.org/lkml/[email protected]/