[LTP] [PATCH] memcg/memcontrol04: Fix the judgment error in test_memcg_low()
Jin Guojie
guojie.jin@gmail.com
Mon Nov 25 14:19:02 CET 2024
Hi Andera,
Thank you for the reply.
Here are some additional comments,
1. In the function test_memcg_low(), due to the judgment of if (i ==
E), it is obvious that no memory is allocated for E. For other groups,
50MB memory is allocated.
---- test_memcg_low() ----
for (i = 0; i < ARRAY_SIZE(leaf_cg); i++) { // leaf_cg[] contains C,D,E,F
leaf_cg[i] = tst_cg_group_mk(trunk_cg[B], "leaf_%c", 'C' + i);
if (i == E) // no memory allocated in E
continue;
alloc_pagecache_in_child(leaf_cg[i], MB(50)); // 50MB allocated for C, D, F
}
Therefore, it's obvious that the change from if (i < E) to if (i != E)
in my patch is consistent with the memory allocation .
2. Since this problem is relatively clear and reproducible in various
distributions, it should be easy to check if someone has discussed any
topics related to my patch.
Could you copy the text in
https://bugzilla.suse.com/show_bug.cgi?id=1196298 and send it to me
separately?
Kind regards,
Jin Guojie
On Mon, Nov 25, 2024 at 5:45 PM Andrea Cervesato
<andrea.cervesato@suse.com> wrote:
>
> Hi Jin,
>
> I can't share the bugzilla content, but the test has been analyzed quite
> a lot and people didn't figure out a proper fix for it.
> Linux is highly asynchronous and the memory management is not always
> predictable. Most of the times, testing memory requires a good knowledge
> of the underlying algorithms and, even so, it's not so obvious to write
> tests for it. Sometimes we end up having a test result based on
> **average reply** from the test (i.e. if we obtain TPASS 70 times out of
> 100, we can consider test pass).
>
> As far as I see, devs think this test should be rewritten or simplified,
> in order to reduce the noise in LTP, testing only features which can be
> actually tested.
>
> Feel free to take a look if you think it's worth the energy, but I also
> suggest to take a look at the kernel, because it could be there's an
> underlying bug somewhere.
>
> Kind regards,
> Andrea Cervesato
>
> On 11/21/24 14:18, Jin Guojie wrote:
> > When I visit https://bugzilla.suse.com/show_bug.cgi?id=1196298, an
> > error occurs showing that
> >
> > "You are not authorized to access bug #1196298."
> >
> > I tried to register an account by myself, but the error did not change.
> >
> > So far, I still can't view the link.
> >
> > Please see if there is a way to give me permission to view this bug.
> >
> > Or, could you please provide me the explanation of this bug in the email？
> >
> > On Thu, Nov 21, 2024 at 5:44 PM Andrea Cervesato
> > <andrea.cervesato@suse.com> wrote:
> >> Hi!
> >>
> >> Thanks for checking this test, but did you take a look at the
> >> explanation given in the test "tags" ? ->
> >> https://bugzilla.suse.com/show_bug.cgi?id=1196298
> >>
> >> Regards,
> >> Andrea
> >>
> >> On 11/21/24 04:05, Jin Guojie wrote:
> >>> When running memcontrol04, TFAIL results will appear on various Linux
> >>> distributions, kernel versions, and CPUs:
> >>>
> >>> (1) Test platform
> >>>
> >>> * Linux distribution: Ubuntu 24.10
> >>> * CPU: X86_64, Arm64
> >>> * Kernel: 6.6 longterm
> >>> * glibc: 2.40
> >>> * LTP version: commit ec4161186e5, Oct 24 12:18:17 2024
> >>>
> >>> (2) Error logs
> >>>
> >>> During the operation of memcontrol04, file systems such as ext2, ext3,
> >>> ext4, xfs, ntfs, and vfat will be tested.
> >>> For any of the file system, the same TFAIL result will appear:
> >>>
> >>> root@vm:~/ltp/testcases/kernel/controllers/memcg# ./memcontrol04
> >>>
> >>> tst_test.c:1823: TINFO: === Testing on ext2 ===
> >>> memcontrol04.c:208: TPASS: Expect: (C oom events=0) == 0
> >>> memcontrol04.c:211: TPASS: Expect: (C low events=437) > 0
> >>> memcontrol04.c:208: TPASS: Expect: (D oom events=0) == 0
> >>> memcontrol04.c:211: TPASS: Expect: (D low events=437) > 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:214: TFAIL: Expect: (F low events=412) == 0
> >>>
> >>> tst_test.c:1823: TINFO: === Testing on ext3 ===
> >>> memcontrol04.c:208: TPASS: Expect: (C oom events=0) == 0
> >>> memcontrol04.c:211: TPASS: Expect: (C low events=437) > 0
> >>> memcontrol04.c:208: TPASS: Expect: (D oom events=0) == 0
> >>> memcontrol04.c:211: TPASS: Expect: (D low events=437) > 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:214: TFAIL: Expect: (F low events=411) == 0
> >>>
> >>> ......
> >>>
> >>> Summary:
> >>> passed 55
> >>> failed 5
> >>> broken 0
> >>> skipped 0
> >>> warnings 0
> >>>
> >>> It looks like there is an error in the processing logic of cgroup F.
> >>>
> >>> (3) Cause analysis
> >>>
> >>> In the test_memcg_low() function, 4 subgroups (C, D, E, F) are created under B,
> >>> and 50MB pagecache is allocated in C, D, and F. Therefore, when checking whether
> >>> it is successful at the end, only E should be judged to have low_events==0,
> >>> and the judgment conditions for all other subgroups should be low_events > 0.
> >>>
> >>> (4) Fix issure
> >>>
> >>> #1209
> >>> https://github.com/linux-test-project/ltp/issues/1209
> >>>
> >>> Signed-off-by: Jin Guojie <guojie.jin@gmail.com>
> >>>
> >>> ---
> >>> testcases/kernel/controllers/memcg/memcontrol04.c | 2 +-
> >>> 1 file changed, 1 insertion(+), 1 deletion(-)
> >>>
> >>> diff --git a/testcases/kernel/controllers/memcg/memcontrol04.c
> >>> b/testcases/kernel/controllers/memcg/memcontrol04.c
> >>> index 1b8d115f8..0dddb7449 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) {
> >>> TST_EXP_EXPR(low > 0,
> >>> "(%c low events=%ld) > 0", id, low);
> >>> } else {
> >>> --
> >>> 2.45.2
> >>>
More information about the ltp
mailing list