The patch titled
Subject: secretmem: optimize page_is_secretmem()
has been removed from the -mm tree. Its filename was
mm-introduce-memfd_secret-system-call-to-create-secret-memory-areas-fix-3.patch
This patch was dropped because an updated version will be merged
------------------------------------------------------
From: Mike Rapoport <rppt@xxxxxxxxxxxxx>
Subject: secretmem: optimize page_is_secretmem()
Kernel test robot reported -4.2% regression of
will-it-scale.per_thread_ops due to commit "mm: introduce memfd_secret
system call to create "secret" memory areas".
The perf profile of the test indicated that the regression is caused by
page_is_secretmem() called from gup_pte_range() (inlined by
gup_pgd_range):
27.76 +2.5 30.23 perf-profile.children.cycles-pp.gup_pgd_range
0.00 +3.2 3.19 2% perf-profile.children.cycles-pp.page_mapping
0.00 +3.7 3.66 2% perf-profile.children.cycles-pp.page_is_secretmem
Further analysis showed that the slow down happens because neither
page_is_secretmem() nor page_mapping() are not inline and moreover,
multiple page flags checks in page_mapping() involve calling
compound_head() several times for the same page.
Make page_is_secretmem() inline and replace page_mapping() with page flag
checks that do not imply page-to-head conversion.
Link: https://lkml.kernel.org/r/20210420150049.14031-3-rppt@xxxxxxxxxx
Signed-off-by: Mike Rapoport <rppt@xxxxxxxxxxxxx>
Reported-by: kernel test robot <oliver.sang@xxxxxxxxx>
Cc: Hagen Paul Pfeifer <hagen@xxxxxxxx>
Cc: Alexander Viro <viro@xxxxxxxxxxxxxxxxxx>
Cc: Andy Lutomirski <luto@xxxxxxxxxx>
Cc: Arnd Bergmann <arnd@xxxxxxxx>
Cc: Borislav Petkov <bp@xxxxxxxxx>
Cc: Catalin Marinas <catalin.marinas@xxxxxxx>
Cc: Christopher Lameter <cl@xxxxxxxxx>
Cc: Dan Williams <dan.j.williams@xxxxxxxxx>
Cc: Dave Hansen <dave.hansen@xxxxxxxxxxxxxxx>
Cc: Elena Reshetova <elena.reshetova@xxxxxxxxx>
Cc: "H. Peter Anvin" <hpa@xxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: James Bottomley <jejb@xxxxxxxxxxxxx>
Cc: "Kirill A. Shutemov" <kirill@xxxxxxxxxxxxx>
Cc: Matthew Wilcox <willy@xxxxxxxxxxxxx>
Cc: Mark Rutland <mark.rutland@xxxxxxx>
Cc: Michael Kerrisk <mtk.manpages@xxxxxxxxx>
Cc: Palmer Dabbelt <palmer@xxxxxxxxxxx>
Cc: Palmer Dabbelt <palmerdabbelt@xxxxxxxxxx>
Cc: Paul Walmsley <paul.walmsley@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Rick Edgecombe <rick.p.edgecombe@xxxxxxxxx>
Cc: Roman Gushchin <guro@xxxxxx>
Cc: Shakeel Butt <shakeelb@xxxxxxxxxx>
Cc: Shuah Khan <shuah@xxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Cc: Tycho Andersen <tycho@xxxxxxxx>
Cc: Will Deacon <will@xxxxxxxxxx>
Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
---
include/linux/secretmem.h | 26 +++++++++++++++++++++++++-
mm/secretmem.c | 12 +-----------
2 files changed, 26 insertions(+), 12 deletions(-)
--- a/include/linux/secretmem.h~mm-introduce-memfd_secret-system-call-to-create-secret-memory-areas-fix-3
+++ a/include/linux/secretmem.h
@@ -4,8 +4,32 @@
#ifdef CONFIG_SECRETMEM
+extern const struct address_space_operations secretmem_aops;
+
+static inline bool page_is_secretmem(struct page *page)
+{
+ struct address_space *mapping;
+
+ /*
+ * Using page_mapping() is quite slow because of the actual call
+ * instruction and repeated compound_head(page) inside the
+ * page_mapping() function.
+ * We know that secretmem pages are not compound and LRU so we can
+ * save a couple of cycles here.
+ */
+ if (PageCompound(page) || !PageLRU(page))
+ return false;
+
+ mapping = (struct address_space *)
+ ((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS);
+
+ if (mapping != page->mapping)
+ return false;
+
+ return page->mapping->a_ops == &secretmem_aops;
+}
+
bool vma_is_secretmem(struct vm_area_struct *vma);
-bool page_is_secretmem(struct page *page);
#else
--- a/mm/secretmem.c~mm-introduce-memfd_secret-system-call-to-create-secret-memory-areas-fix-3
+++ a/mm/secretmem.c
@@ -137,22 +137,12 @@ static void secretmem_freepage(struct pa
clear_highpage(page);
}
-static const struct address_space_operations secretmem_aops = {
+const struct address_space_operations secretmem_aops = {
.freepage = secretmem_freepage,
.migratepage = secretmem_migratepage,
.isolate_page = secretmem_isolate_page,
};
-bool page_is_secretmem(struct page *page)
-{
- struct address_space *mapping = page_mapping(page);
-
- if (!mapping)
- return false;
-
- return mapping->a_ops == &secretmem_aops;
-}
-
static struct vfsmount *secretmem_mnt;
static struct file *secretmem_file_create(unsigned long flags)
_
Patches currently in -mm which might be from rppt@xxxxxxxxxxxxx are
mm-mmzoneh-simplify-is_highmem_idx.patch
docs-procrst-meminfo-briefly-describe-gaps-in-memory-accounting.patch
include-linux-mmzoneh-add-documentation-for-pfn_valid.patch
memblock-update-initialization-of-reserved-pages.patch
arm64-decouple-check-whether-pfn-is-in-linear-map-from-pfn_valid.patch
arm64-drop-pfn_valid_within-and-simplify-pfn_valid.patch
mmap-make-mlock_future_check-global.patch
riscv-kconfig-make-direct-map-manipulation-options-depend-on-mmu.patch
set_memory-allow-set_direct_map__noflush-for-multiple-pages.patch
set_memory-allow-querying-whether-set_direct_map_-is-actually-enabled.patch
mm-introduce-memfd_secret-system-call-to-create-secret-memory-areas.patch
pm-hibernate-disable-when-there-are-active-secretmem-users.patch
arch-mm-wire-up-memfd_secret-system-call-where-relevant.patch
secretmem-test-add-basic-selftest-for-memfd_secret2.patch
arch-mm-wire-up-memfd_secret-system-call-where-relevant-fix.patch
