Re: [PATCH v2 00/13] selftests/mm: fix failures and robustness improvements

[Venkat Rao Bagalkote]

On 13/03/26 7:46 pm, Sayali Patil wrote:

On 13/03/26 03:35, Andrew Morton wrote:
On Thu, 12 Mar 2026 17:55:27 +0530 Sayali Patil 
<saya...@linux.ibm.com> wrote:

Hi all,

Powerpc systems with a 64K base page size exposed several issues while
running mm selftests. Some tests assume specific hugetlb 
configurations,
use incorrect interfaces, or fail instead of skipping when the required
kernel features are not available.

This series fixes these issues and improves test robustness.


Hello Sayali,


I still see, below test failure. Apart from this, everyting else is 
working as expected.


# # ------------------------------------------------
# # running ./hugetlb_reparenting_test.sh -cgroup-v2
# # ------------------------------------------------
# # cleanup
# #
# # Test charge, rmdir, uncharge
# # mkdir
# # write
# # Writing to this path: /mnt/huge/test
# # Writing this size: 26843545600
# # Populating.
# # Not writing to memory.
# # Using method=0
# # Shared mapping.
# # RESERVE mapping.
# # Allocating using HUGETLBFS.
# #
# # rmdir
# # uncharge
# # cleanup
# # done
# #
# #
# # Test child only hugetlb usage
# # setup
# # write
# # Writing to this path: /mnt/huge/test2
# # Writing this size: 26843545600
# # Populating.
# # Not writing to memory.
# # Using method=0
# # Shared mapping.
# # RESERVE mapping.
# # Allocating using HUGETLBFS.
# #
# # Assert memory charged correctly for child only use.
# # actual = 11 MB
# # expected = 0 MB
# # FAIL
# # cleanup
# # [FAIL]
# not ok 12 hugetlb_reparenting_test.sh -cgroup-v2 # exit=1


Regards,

Venkat.

Thanks, I updated mm.git's mm-new branch to this version.

---
v1->v2
   - For "selftests/mm: ensure destination is hugetlb-backed in 
hugepage-mremap":
     update FLAGS definition to MAP_HUGETLB | MAP_SHARED | 
MAP_POPULATE and
     used it for mmap() calls as suggested during review.
That was a really small change (below).  It's better to wait a while
(half to one week) before resending an entire series.  So as to
hopefully accumulate more changes.



--- a/tools/testing/selftests/mm/hugepage-mremap.c~b
+++ a/tools/testing/selftests/mm/hugepage-mremap.c
@@ -31,7 +31,7 @@
  #define MB_TO_BYTES(x) (x * 1024 * 1024)
    #define PROTECTION (PROT_READ | PROT_WRITE | PROT_EXEC)
-#define FLAGS (MAP_SHARED | MAP_ANONYMOUS)
+#define FLAGS (MAP_HUGETLB | MAP_SHARED | MAP_POPULATE)
    static void check_bytes(char *addr)
  {
@@ -121,23 +121,20 @@ int main(int argc, char *argv[])
        /* mmap to a PUD aligned address to hopefully trigger pmd 
sharing. */
      unsigned long suggested_addr = 0x7eaa40000000;
-    void *haddr = mmap((void *)suggested_addr, length, PROTECTION,
-               MAP_HUGETLB | MAP_SHARED | MAP_POPULATE, fd, 0);
+    void *haddr = mmap((void *)suggested_addr, length, PROTECTION, 
FLAGS, fd, 0);
      ksft_print_msg("Map haddr: Returned address is %p\n", haddr);
      if (haddr == MAP_FAILED)
          ksft_exit_fail_msg("mmap1: %s\n", strerror(errno));
        /* mmap again to a dummy address to hopefully trigger pmd 
sharing. */
      suggested_addr = 0x7daa40000000;
-    void *daddr = mmap((void *)suggested_addr, length, PROTECTION,
-               MAP_HUGETLB | MAP_SHARED | MAP_POPULATE, fd, 0);
+    void *daddr = mmap((void *)suggested_addr, length, PROTECTION, 
FLAGS, fd, 0);
      ksft_print_msg("Map daddr: Returned address is %p\n", daddr);
      if (daddr == MAP_FAILED)
          ksft_exit_fail_msg("mmap3: %s\n", strerror(errno));
        suggested_addr = 0x7faa40000000;
-    void *vaddr = mmap((void *)suggested_addr, length, PROTECTION,
-            MAP_HUGETLB | MAP_SHARED | MAP_POPULATE, fd, 0);
+    void *vaddr = mmap((void *)suggested_addr, length, PROTECTION, 
FLAGS, fd, 0);
      ksft_print_msg("Map vaddr: Returned address is %p\n", vaddr);
      if (vaddr == MAP_FAILED)
          ksft_exit_fail_msg("mmap2: %s\n", strerror(errno));
_


Thanks for the guidance. I'll make sure to follow this approach from 
next time.