On 07/01/2025 14:48, Lorenzo Stoakes wrote:

On Tue, Jan 07, 2025 at 02:29:35PM +0000, Ryan Roberts wrote:

...

The recently introduced guard-pages mm selftest uses the process_madvise() syscall, a wrapper for which was added to glibc v2.36. For those of us stuck with older distributions this causes a compile error when compiling the mm selftests. For example Ubuntu 22.04 uses glibc 2.35, which does not have the wrapper.

Ah oops! I didn't check glibc and had erroneously assumed this would be trivially available, but perhaps using a rolling release distro has warped my perceptions on this rather...

At any rate you're entire correct, this is very much needed, cheers!

...

To workaround the issue, let's introduce our own static process_madvise() wrapper that uses glibc's syscall() helper.

While we are at it, add the guard-page test suite to run_vmtests.sh so that it can be automatically run by CI systems.

Oops part 2... I was sure I had added it there... thanks!

...

Signed-off-by: Ryan Roberts ryan.roberts@arm.com

Reviewed-by: Lorenzo Stoakes lorenzo.stoakes@oracle.com

No worries, thanks for the quick review!

...

---

Applies on top of mm-unstable (f349e79bfbf3)

Thanks, Ryan

tools/testing/selftests/mm/guard-pages.c | 10 ++++++++-- tools/testing/selftests/mm/run_vmtests.sh | 5 +++++ 2 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/tools/testing/selftests/mm/guard-pages.c b/tools/testing/selftests/mm/guard-pages.c index d8f8dee9ebbd..ece37212a8a2 100644 --- a/tools/testing/selftests/mm/guard-pages.c +++ b/tools/testing/selftests/mm/guard-pages.c @@ -55,6 +55,12 @@ static int pidfd_open(pid_t pid, unsigned int flags) return syscall(SYS_pidfd_open, pid, flags); }

+static ssize_t sys_process_madvise(int pidfd, const struct iovec *iovec,

• 		   size_t n, int advice, unsigned int flags)
  

+{

• return syscall(__NR_process_madvise, pidfd, iovec, n, advice, flags);

+}

/*

• Enable our signal catcher and try to read/write the specified buffer. The
• return value indicates whether the read/write succeeds without a fatal

@@ -419,7 +425,7 @@ TEST_F(guard_pages, process_madvise) ASSERT_EQ(munmap(&ptr_region[99 * page_size], page_size), 0);

/* Now guard in one step. */

• count = process_madvise(pidfd, vec, 6, MADV_GUARD_INSTALL, 0);

• count = sys_process_madvise(pidfd, vec, 6, MADV_GUARD_INSTALL, 0);

  /* OK we don't have permission to do this, skip. */ if (count == -1 && errno == EPERM)

@@ -440,7 +446,7 @@ TEST_F(guard_pages, process_madvise) ASSERT_FALSE(try_read_write_buf(&ptr3[19 * page_size]));

/* Now do the same with unguard... */

• count = process_madvise(pidfd, vec, 6, MADV_GUARD_REMOVE, 0);

• count = sys_process_madvise(pidfd, vec, 6, MADV_GUARD_REMOVE, 0);

  /* ...and everything should now succeed. */

diff --git a/tools/testing/selftests/mm/run_vmtests.sh b/tools/testing/selftests/mm/run_vmtests.sh index 2fc290d9430c..00c3f07ea100 100755 --- a/tools/testing/selftests/mm/run_vmtests.sh +++ b/tools/testing/selftests/mm/run_vmtests.sh @@ -45,6 +45,8 @@ separated by spaces: vmalloc smoke tests

• hmm hmm smoke tests

+- madv_guard

• test madvise(2) MADV_GUARD_INSTALL and MADV_GUARD_REMOVE options

• madv_populate test memadvise(2) MADV_POPULATE_{READ,WRITE} options
• memfd_secret

@@ -375,6 +377,9 @@ CATEGORY="mremap" run_test ./mremap_dontunmap

CATEGORY="hmm" run_test bash ./test_hmm.sh smoke

+# MADV_GUARD_INSTALL and MADV_GUARD_REMOVE tests +CATEGORY="madv_guard" run_test ./guard-pages

# MADV_POPULATE_READ and MADV_POPULATE_WRITE tests CATEGORY="madv_populate" run_test ./madv_populate

-- 2.43.0