[LTP] [PATCH v2] memcg/memcontrol04: Fix judgment for recursive_protection

[Li Wang]

Hi Jin,

On Thu, Dec 19, 2024 at 2:09 PM Jin Guojie <guojie.jin@gmail.com> wrote:

> V2:
> * Change the expected events in F depending on memory_recursiveprot
>
> In Patch v1[1], the role of recursive_protection in cgoup v2 was not
> considered.
>
> By carefully reading the algorithm in the kernel function
> effective_protection(). When memory_recursiveprot is enabled, a
> subgroup with usage > 0 can get unclaimed greater than 0.
>
> The purpose of doing this should be to achieve the essential purpose
> of recursive_protection: the sum of all subgroups' unprotected values
> is equal to parent's unprotected values.
>
> Even though the documentation does not give an explicit description
> for recursive_memoryprot, it looks like the kernel's processing
> algorithm is reasonable.
>
> Based on the idea of [2], Patch v2 is rewritten to determine whether
> memory_recursiveprot is enabled.
>
> On distributions with memory_recursiveprot enabled by default (from
> Ubuntu 22.04 to 24.10), running this passes:
>
> memcontrol04.c:208: TPASS: Expect: (C oom events=0) == 0
> memcontrol04.c:211: TPASS: Expect: (C low events=966) > 0
> memcontrol04.c:208: TPASS: Expect: (D oom events=0) == 0
> memcontrol04.c:211: TPASS: Expect: (D low events=966) > 0
> memcontrol04.c:208: TPASS: Expect: (E oom events=0) == 0
> memcontrol04.c:214: TPASS: Expect: (E low events=0) == 0
> memcontrol04.c:208: TPASS: Expect: (F oom events=0) == 0
> memcontrol04.c:211: TPASS: Expect: (F low events=874) > 0
>
> [1] https://lists.linux.it/pipermail/ltp/2024-November/040946.html
> [2] https://lore.kernel.org/ltp/20220222144511.GA12037@blackbody.suse.cz/
>
> Signed-off-by: Jin Guojie <guojie.jin@gmail.com>
> Suggested-by: Richard Palethorpe <rpalethorpe@suse.com>
> Suggested-by: Michal Koutný <mkoutny@suse.com>
> ---
>  include/tst_cgroup.h                              |  2 ++
>  lib/tst_cgroup.c                                  | 12 ++++++++++++
>  testcases/kernel/controllers/memcg/memcontrol04.c |  2 +-
>  3 files changed, 15 insertions(+), 1 deletion(-)
>
> diff --git a/include/tst_cgroup.h b/include/tst_cgroup.h
> index d23a8e652..068ff8306 100644
> --- a/include/tst_cgroup.h
> +++ b/include/tst_cgroup.h
> @@ -256,4 +256,6 @@ int safe_cg_occursin(const char *file, const int
> lineno,
>                          const char *const file_name,
>                          const char *const needle);
>
> +int tst_cg_memory_recursiveprot(struct tst_cg_group *cg);
> +
>  #endif /* TST_CGROUP_H */
> diff --git a/lib/tst_cgroup.c b/lib/tst_cgroup.c
> index 6055015eb..9e3b21ed0 100644
> --- a/lib/tst_cgroup.c
> +++ b/lib/tst_cgroup.c
> @@ -76,6 +76,8 @@ struct cgroup_root {
>         int we_mounted_it:1;
>         /* cpuset is in compatability mode */
>         int no_cpuset_prefix:1;
> +
> +       int memory_recursiveprot:1;
>

I prefer to move it up to put it together with the nsdelegate.



>  };
>
>  /* Controller sub-systems */
> @@ -592,6 +594,7 @@ static void cgroup_root_scan(const char *const
> mnt_type,
>         }
>         for (tok = strtok(mnt_opts, ","); tok; tok = strtok(NULL, ",")) {
>                 nsdelegate |= !strcmp("nsdelegate", tok);
> +               root->memory_recursiveprot |=
> !strcmp("memory_recursiveprot", tok);
>


Why not define a single variable memory_recursiveprot to save the value
and move the root->memory_recursiveprot assignment to the "backref" part?
(just like what we did for nsdelegate)



>         }
>
>         if (root->ver && ctrl_field == root->ctrl_field)
> @@ -718,6 +721,7 @@ mount:
>                 tst_res(TINFO, "Mounted V2 CGroups on %s", mnt_path);
>                 tst_cg_scan();
>                 roots[0].we_mounted_it = 1;
> +               roots[0].memory_recursiveprot = 1;
>

This is completely wrong, as the first mount operation may fail,
while CGroupV2 falls back to the default mount and succeeds.

Hence we should remove the line here, as tst_cg_scan() helps
to automatically detect and set "root->memory_recursiveprot"
to the correct value.



>                 return;
>         }
>
> @@ -1509,3 +1513,11 @@ int safe_cg_occursin(const char *const file,
> const int lineno,
>
>         return !!strstr(buf, needle);
>  }
> +
> +int tst_cg_memory_recursiveprot(struct tst_cg_group *cg)
> +{
> +       if (cg && cg->dirs_by_ctrl[0]->dir_root)
> +               return cg->dirs_by_ctrl[0]->dir_root->memory_recursiveprot;
> +       return 0;
> +}
> +
> diff --git a/testcases/kernel/controllers/memcg/memcontrol04.c
> b/testcases/kernel/controllers/memcg/memcontrol04.c
> index 1b8d115f8..9e9d6ab6e 100644
> --- a/testcases/kernel/controllers/memcg/memcontrol04.c
> +++ b/testcases/kernel/controllers/memcg/memcontrol04.c
> @@ -207,7 +207,7 @@ static void test_memcg_low(void)
>
>                 TST_EXP_EXPR(oom == 0, "(%c oom events=%ld) == 0", id,
> oom);
>
> -               if (i < E) {
> +               if (i < E || ((i == F) &&
> tst_cg_memory_recursiveprot(leaf_cg[F]))) {
>                         TST_EXP_EXPR(low > 0,
>                                      "(%c low events=%ld) > 0", id, low);
>                 } else {
> --
> 2.34.1
>
> --
> Mailing list info: https://lists.linux.it/listinfo/ltp
>


-- 
Regards,
Li Wang