Merge: MM: revert POSIX_FADV_NOREUSE

MR: https://gitlab.com/redhat/rhel/src/kernel/rhel-9/-/merge_requests/3517

JIRA: https://issues.redhat.com/browse/RHEL-80655
Upstream Status: RHEL-only

This series reverts the introduction of POSIX_FADV_NOREUSE in 9.6.

As documented in the jira ticket, it was found that downstream commit
d908e31 seems responsible for a
10-20x slow down in OCP's etcd compaction operation which may be due
to increased OCP API latency.

We believe that d908e31 is regressing
MADV_RANDOM in a way that causes performance degradation for applications
using this hint, as after this commit the pages backing the VMAs that
are marked for random access will not receive a second chance to be
re-activated once they are in the LRU inactive list.

Signed-off-by: Luiz Capitulino <[email protected]>

Approved-by: Rafael Aquini <[email protected]>
Approved-by: Aristeu Rozanski <[email protected]>
Approved-by: Waiman Long <[email protected]>
Approved-by: CKI KWF Bot <[email protected]>

Merged-by: CKI KWF Bot <[email protected]>

Author: CKI KWF Bot

include/linux/fs.h

@@ -163,8 +163,6 @@ typedef int (dio_iodone_t)(struct kiocb *iocb, loff_t offset,
 /* File supports DIRECT IO */
 #define	FMODE_CAN_ODIRECT	((__force fmode_t)0x400000)
 
-#define	FMODE_NOREUSE		((__force fmode_t)0x800000)
-
 /* File supports non-exclusive O_DIRECT writes from multiple threads */
 #define FMODE_DIO_PARALLEL_WRITE	((__force fmode_t)0x1000000)
 

include/linux/mm_inline.h

@@ -579,15 +579,4 @@ pte_install_uffd_wp_if_needed(struct vm_area_struct *vma, unsigned long addr,
 #endif
 }
 
-static inline bool vma_has_recency(struct vm_area_struct *vma)
-{
-	if (vma->vm_flags & (VM_SEQ_READ | VM_RAND_READ))
-		return false;
-
-	if (vma->vm_file && (vma->vm_file->f_mode & FMODE_NOREUSE))
-		return false;
-
-	return true;
-}
-
 #endif

mm/fadvise.c

@@ -79,7 +79,7 @@ int generic_fadvise(struct file *file, loff_t offset, loff_t len, int advice)
 	case POSIX_FADV_NORMAL:
 		file->f_ra.ra_pages = bdi->ra_pages;
 		spin_lock(&file->f_lock);
-		file->f_mode &= ~(FMODE_RANDOM | FMODE_NOREUSE);
+		file->f_mode &= ~FMODE_RANDOM;
 		spin_unlock(&file->f_lock);
 		break;
 	case POSIX_FADV_RANDOM:
@@ -106,9 +106,6 @@ int generic_fadvise(struct file *file, loff_t offset, loff_t len, int advice)
 		force_page_cache_readahead(mapping, file, start_index, nrpages);
 		break;
 	case POSIX_FADV_NOREUSE:
-		spin_lock(&file->f_lock);
-		file->f_mode |= FMODE_NOREUSE;
-		spin_unlock(&file->f_lock);
 		break;
 	case POSIX_FADV_DONTNEED:
 		__filemap_fdatawrite_range(mapping, offset, endbyte,

mm/memory.c

@@ -1460,7 +1460,8 @@ static unsigned long zap_pte_range(struct mmu_gather *tlb,
 						force_flush = 1;
 					}
 				}
-				if (pte_young(ptent) && likely(vma_has_recency(vma)))
+				if (pte_young(ptent) &&
+				    likely(!(vma->vm_flags & VM_SEQ_READ)))
 					mark_page_accessed(page);
 			}
 			rss[mm_counter(page)]--;
@@ -5252,8 +5253,8 @@ static inline void mm_account_fault(struct mm_struct *mm, struct pt_regs *regs,
 #ifdef CONFIG_LRU_GEN
 static void lru_gen_enter_fault(struct vm_area_struct *vma)
 {
-	/* the LRU algorithm only applies to accesses with recency */
-	current->in_lru_fault = vma_has_recency(vma);
+	/* the LRU algorithm doesn't apply to sequential or random reads */
+	current->in_lru_fault = !(vma->vm_flags & (VM_SEQ_READ | VM_RAND_READ));
 }
 
 static void lru_gen_exit_fault(void)

mm/rmap.c

@@ -821,15 +821,25 @@ static bool folio_referenced_one(struct folio *folio,
 		}
 
 		if (pvmw.pte) {
-			if (lru_gen_enabled() &&
-			    pte_young(ptep_get(pvmw.pte))) {
+			if (lru_gen_enabled() && pte_young(ptep_get(pvmw.pte)) &&
+			    !(vma->vm_flags & (VM_SEQ_READ | VM_RAND_READ))) {
 				lru_gen_look_around(&pvmw);
 				referenced++;
 			}
 
 			if (ptep_clear_flush_young_notify(vma, address,
-						pvmw.pte))
-				referenced++;
+						pvmw.pte)) {
+				/*
+				 * Don't treat a reference through
+				 * a sequentially read mapping as such.
+				 * If the folio has been used in another mapping,
+				 * we will catch it; if this other mapping is
+				 * already gone, the unmap path will have set
+				 * the referenced flag or activated the folio.
+				 */
+				if (likely(!(vma->vm_flags & VM_SEQ_READ)))
+					referenced++;
+			}
 		} else if (IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE)) {
 			if (pmdp_clear_flush_young_notify(vma, address,
 						pvmw.pmd))
@@ -863,20 +873,7 @@ static bool invalid_folio_referenced_vma(struct vm_area_struct *vma, void *arg)
 	struct folio_referenced_arg *pra = arg;
 	struct mem_cgroup *memcg = pra->memcg;
 
-	/*
-	 * Ignore references from this mapping if it has no recency. If the
-	 * folio has been used in another mapping, we will catch it; if this
-	 * other mapping is already gone, the unmap path will have set the
-	 * referenced flag or activated the folio in zap_pte_range().
-	 */
-	if (!vma_has_recency(vma))
-		return true;
-
-	/*
-	 * If we are reclaiming on behalf of a cgroup, skip counting on behalf
-	 * of references from different cgroups.
-	 */
-	if (memcg && !mm_match_cgroup(vma->vm_mm, memcg))
+	if (!mm_match_cgroup(vma->vm_mm, memcg))
 		return true;
 
 	return false;
@@ -907,7 +904,6 @@ int folio_referenced(struct folio *folio, int is_locked,
 		.arg = (void *)&pra,
 		.anon_lock = folio_lock_anon_vma_read,
 		.try_lock = true,
-		.invalid_vma = invalid_folio_referenced_vma,
 	};
 
 	*vm_flags = 0;
@@ -923,6 +919,15 @@ int folio_referenced(struct folio *folio, int is_locked,
 			return 1;
 	}
 
+	/*
+	 * If we are reclaiming on behalf of a cgroup, skip
+	 * counting on behalf of references from different
+	 * cgroups
+	 */
+	if (memcg) {
+		rwc.invalid_vma = invalid_folio_referenced_vma;
+	}
+
 	rmap_walk(folio, &rwc);
 	*vm_flags = pra.vm_flags;
 

mm/vmscan.c

@@ -3161,10 +3161,7 @@ static int should_skip_vma(unsigned long start, unsigned long end, struct mm_wal
 	if (is_vm_hugetlb_page(vma))
 		return true;
 
-	if (!vma_has_recency(vma))
-		return true;
-
-	if (vma->vm_flags & (VM_LOCKED | VM_SPECIAL))
+	if (vma->vm_flags & (VM_LOCKED | VM_SPECIAL | VM_SEQ_READ | VM_RAND_READ))
 		return true;
 
 	if (vma == get_gate_vma(vma->vm_mm))