On 15.10.25 10:33, Itamar Dalal wrote:
Add a new test `migrate.cow_after_fork` that verifies correct RMAP handling
of Copy-On-Write (COW) pages after fork().
Before a write, the parent and child share the same PFN. After a write, the
child’s PFN differs. This confirms that proper COW duplication occurred and
that RMAP correctly tracks page ownership transitions during COW events.
Signed-off-by: Itamar-Dalal <[email protected]
<mailto:[email protected]>>
---
tools/testing/selftests/mm/rmap.c | 45 ++++++++++++++++++++++++++++++-
1 file changed, 44 insertions(+), 1 deletion(-)
diff --git a/tools/testing/selftests/mm/rmap.c b/tools/testing/
selftests/mm/rmap.c
index 13f7bccfd0a9..2ba3361fecf0 100644
--- a/tools/testing/selftests/mm/rmap.c
+++ b/tools/testing/selftests/mm/rmap.c
@@ -430,4 +430,47 @@ TEST_F(migrate, ksm)
propagate_children(_metadata, data);
}
-TEST_HARNESS_MAIN
+TEST_F(migrate, cow_after_fork)
+{
+ struct global_data *data = &self->data;
+ int status;
+ pid_t pid;
+ unsigned long parent_pfn, child_pfn;
+ int pagemap_fd;
+ char *region;
+
+ /* Map private anonymous memory and fault it in */
+ region = mmap(NULL, data->mapsize, PROT_READ | PROT_WRITE,
+ MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+ ASSERT_NE(region, MAP_FAILED);
+ memset(region, 0xaa, data->mapsize);
+
+ pagemap_fd = open("/proc/self/pagemap", O_RDONLY);
+ ASSERT_NE(pagemap_fd, -1);
+ parent_pfn = pagemap_get_pfn(pagemap_fd, region);
+ close(pagemap_fd);
+
+ pid = fork();
+ ASSERT_NE(pid, -1);
+
+ if (pid == 0) {
+ /* Child: write to trigger COW */
+ region[0] = 0xbb;
+
+ pagemap_fd = open("/proc/self/pagemap", O_RDONLY);
+ ASSERT_NE(pagemap_fd, -1);
+ child_pfn = pagemap_get_pfn(pagemap_fd, region);
+ close(pagemap_fd);
+
+ /* Expect PFN to differ after write (COW happened) */
+ if (child_pfn == parent_pfn)
+ _exit(FAIL_ON_CHECK);
+ _exit(0);
+ }
+
+ waitpid(pid, &status, 0);
+ ASSERT_EQ(WEXITSTATUS(status), 0);
+ munmap(region, data->mapsize);
+}
We have excessive cow tests in cow.c, that are independent of any PFN
checks.
So I don't think we need this.
BTW, I never received a mail that shows up right-aligned in my mail client.
Seems to be because of
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
<div dir=3D"rtl"
The whole attachment should be dropped. Did you send this with
git-send-email? :)
--
Cheers
David / dhildenb
