selftests/mm: uffd-missing-wp: add GC related functionality test
Subsystems (4)
PATCH 1/1 selftests/mm: uffd-missing-wp: add GC related functionality test
Add a test that asserts that functionality needed for generational garbage collection works. Namely to implement a write barrier using userfaultfd instead of using segmentation faults. The planned target is the GC in SBCL.
The real GC will initially fill heap pages by using on-demand loading of pages from an image file, either via mmap(2) or using userfaultfd to decompress a page read from an image file made from individually compressed blocks. This test doesn't do that so that it needs no disk I/O and no decompression library. Instead, a fault will be filled by copying page content from prepared pages in the heap.
--- tools/testing/selftests/mm/Makefile | 1 + tools/testing/selftests/mm/uffd-missing-wp.c | 413 +++++++++++++++++++ 2 files changed, 414 insertions(+) create mode 100644 tools/testing/selftests/mm/uffd-missing-wp.c
tools/testing/selftests/mm/Makefile +1 -0
@@ -92,6 +92,7 @@ TEST_GEN_FILES += transhuge-stress TEST_GEN_FILES += uffd-stress TEST_GEN_FILES += uffd-unit-tests TEST_GEN_FILES += uffd-wp-mremap +TEST_GEN_FILES += uffd-missing-wp TEST_GEN_FILES += split_huge_page_test TEST_GEN_FILES += ksm_tests TEST_GEN_FILES += ksm_functional_tests
tools/testing/selftests/mm/uffd-missing-wp.c new file +413 -0
413 lines not on this page · raw message
Signature · 1 line
2.47.3
Andrew Morton
maintainerQuoted 11 lines · On Sun, 27 Sep 2026 23:26:31 -0400 Martin Cracauer <[email protected]> wrote:
Add a test that asserts that functionality needed for generational garbage collection works. Namely to implement a write barrier using userfaultfd instead of using segmentation faults. The planned target is the GC in SBCL.
The real GC will initially fill heap pages by using on-demand loading of pages from an image file, either via mmap(2) or using userfaultfd to decompress a page read from an image file made from individually compressed blocks. This test doesn't do that so that it needs no disk I/O and no decompression library. Instead, a fault will be filled by copying page content from prepared pages in the heap.
Thanks. Sashiko asked a couple of things:
https://sashiko.dev/#/patchset/[email protected]
David Hildenbrand (Arm)
maintainerQuoted 12 lines · Martin Cracauer
Add a test that asserts that functionality needed for generational garbage collection works. Namely to implement a write barrier using userfaultfd instead of using segmentation faults. The planned target is the GC in SBCL.
The real GC will initially fill heap pages by using on-demand loading of pages from an image file, either via mmap(2) or using userfaultfd to decompress a page read from an image file made from individually compressed blocks. This test doesn't do that so that it needs no disk I/O and no decompression library. Instead, a fault will be filled by copying page content from prepared pages in the heap.
Quoted 7 lines
Assisted-by Anthropic:Opus-5.5Signed-off-by Martin Cracauer <[email protected]>--- tools/testing/selftests/mm/Makefile | 1 + tools/testing/selftests/mm/uffd-missing-wp.c | 413 +++++++++++++++++++ 2 files changed, 414 insertions(+) create mode 100644 tools/testing/selftests/mm/uffd-missing-wp.c
Isn't something like that a better fit for LTP?
Signature · 3 lines
Cheers, David