Speed up unmapping of anonymous large folios by clearing the ptes, and
setting swap ptes, in one go.
The following benchmark (stolen from Barry at [1]) is used to measure the
time taken to swapout 256M worth of memory backed by 64K large folios:
#define _GNU_SOURCE
#include <stdio.h>
#include <stdlib.h>
#include <sys/mman.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
#include <errno.h>
#define SIZE_MB 256
#define SIZE_BYTES (SIZE_MB * 1024 * 1024)
int main() {
void *addr = mmap(NULL, SIZE_BYTES, PROT_READ | PROT_WRITE,
MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
if (addr == MAP_FAILED) {
perror("mmap failed");
return 1;
}
memset(addr, 0, SIZE_BYTES);
struct timespec start, end;
clock_gettime(CLOCK_MONOTONIC, &start);
if (madvise(addr, SIZE_BYTES, MADV_PAGEOUT) != 0) {
perror("madvise(MADV_PAGEOUT) failed");
munmap(addr, SIZE_BYTES);
return 1;
}
clock_gettime(CLOCK_MONOTONIC, &end);
long duration_ns = (end.tv_sec - start.tv_sec) * 1e9 +
(end.tv_nsec - start.tv_nsec);
printf("madvise(MADV_PAGEOUT) took %ld ns (%.3f ms)\n",
duration_ns, duration_ns / 1e6);
munmap(addr, SIZE_BYTES);
return 0;
}
Performance as measured on a Linux VM on Apple M3 (arm64):
Vanilla - Mean: 37401913 ns, std dev: 12%
Patched - Mean: 17420282 ns, std dev: 11%
No regression observed on 4K folios.
Performance as measured on bare metal x86:
Vanilla - mean: 54986286 ns, std dev: 1.5%
Patched - mean: 51930795 ns, std dev: 3%
Interestingly, no obvious improvement is observed on x86, hinting that the
benefit lies mainly in the reduction of ptep_get() calls and the reduction
of TLB flushes during contpte-unfolding, on arm64.
No regression is observed on 4K folios on x86 too.
---
Applies on mm-unstable (444fc9435e57).
v3->v4:
- Add patch 1 (page->folio conversion), patch 2 (add try_to_unmap_hugetlb_one),
patch 6 and 11 (subpage -> page renaming), and patch 10 (refactoring anon
unmap first)
- Drop patch 1 of v3, which is now in mm-hotfixes
- Patch 3: do set_ptes() restoration at call site
- Patch 4: rename the function, move to memory.c, rename pte->ptep, pteval->pte
- Patch 7,8: Remove redundant stub, add kerneldoc
- Patch 9: remove helper, do a loop in-line
v2->v3:
Mostly a resend after merge window. Some minor changes:
- Match kerneldoc parameter with function parameter (pte -> ptep)
- Mention change BUG->WARN in patch description
- Rename walk_done -> exit_walk in patch 2
-
v1->v2:
- Keep nr_pages as unsigned long
- Add patch 2
- Rename some functions, make return type bool for functions returning 0/1
- Drop page_vma_mapped_walk_jump - this is implicitly handled
- Drop likely()
- Add folio_dup/put_swap_pages, do subpage -> page
- Shorten the kerneldoc to remove unnecessary information - keep it
aligned with analogous functions
- Put clear_pages_anon_exclusive to mm.h
- Some more refactoring in last patch with finish_folio_unmap
Dev Jain (12):
mm/rmap: convert page -> folio for hwpoison checks
mm/rmap: Add try_to_unmap_hugetlb_one
mm/rmap: refactor some code around lazyfree folio unmapping
mm/memory: Batch set uffd-wp markers during zapping
mm/rmap: batch unmap folios belonging to uffd-wp VMAs
mm/swap: rename subpage->page in folio_dup_swap/folio_put_swap
mm/swapfile: Add batched version of folio_dup_swap
mm/swapfile: Add batched version of folio_put_swap
mm/rmap: Add batched version of folio_try_share_anon_rmap_pte
mm/rmap: refactor anon folio unmap in try_to_unmap_one
mm/mprotect: drop 'sub' from page_anon_exclusive_sub_batch
mm/rmap: enable batch unmapping of anonymous folios
include/linux/mm.h | 4 +
include/linux/mm_inline.h | 53 -----
include/linux/rmap.h | 52 +++--
mm/internal.h | 17 ++
mm/memory.c | 87 +++++--
mm/mprotect.c | 19 +-
mm/rmap.c | 475 ++++++++++++++++++++++++--------------
mm/shmem.c | 8 +-
mm/swap.h | 36 ++-
mm/swapfile.c | 31 ++-
10 files changed, 470 insertions(+), 312 deletions(-)
--
2.34.1