[PATCH v13 mm-new 14/16] khugepaged: avoid unnecessary mTHP collapse attempts
From: Nico Pache <hidden>
Date: 2025-12-01 17:49:54
Also in:
linux-doc, linux-mm, lkml
Subsystem:
memory management, memory management - thp (transparent huge page), the rest · Maintainers:
Andrew Morton, David Hildenbrand, Linus Torvalds
There are cases where, if an attempted collapse fails, all subsequent orders are guaranteed to also fail. Avoid these collapse attempts by bailing out early. Signed-off-by: Nico Pache <redacted> --- mm/khugepaged.c | 35 ++++++++++++++++++++++++++++++++++- 1 file changed, 34 insertions(+), 1 deletion(-)
diff --git a/mm/khugepaged.c b/mm/khugepaged.c
index fcb927cd5456..daf39aab7e86 100644
--- a/mm/khugepaged.c
+++ b/mm/khugepaged.c@@ -1503,9 +1503,42 @@ static int mthp_collapse(struct mm_struct *mm, unsigned long address, ret = collapse_huge_page(mm, collapse_address, referenced, unmapped, cc, mmap_locked, order); - if (ret == SCAN_SUCCEED) { + + switch (ret) { + /* Cases were we continue to next collapse candidate */ + case SCAN_SUCCEED: collapsed += nr_pte_entries; + fallthrough; + case SCAN_PTE_MAPPED_HUGEPAGE: continue; + /* Cases were lower orders might still succeed */ + case SCAN_LACK_REFERENCED_PAGE: + case SCAN_EXCEED_NONE_PTE: + case SCAN_EXCEED_SWAP_PTE: + case SCAN_EXCEED_SHARED_PTE: + case SCAN_PAGE_LOCK: + case SCAN_PAGE_COUNT: + case SCAN_PAGE_LRU: + case SCAN_PAGE_NULL: + case SCAN_DEL_PAGE_LRU: + case SCAN_PTE_NON_PRESENT: + case SCAN_PTE_UFFD_WP: + case SCAN_ALLOC_HUGE_PAGE_FAIL: + goto next_order; + /* Cases were no further collapse is possible */ + case SCAN_CGROUP_CHARGE_FAIL: + case SCAN_COPY_MC: + case SCAN_ADDRESS_RANGE: + case SCAN_NO_PTE_TABLE: + case SCAN_ANY_PROCESS: + case SCAN_VMA_NULL: + case SCAN_VMA_CHECK: + case SCAN_SCAN_ABORT: + case SCAN_PAGE_ANON: + case SCAN_PMD_MAPPED: + case SCAN_FAIL: + default: + return collapsed; } }
--
2.51.1