[LTP] [PATCH 0/2] ksm06 fixes

[LTP] [PATCH 0/2] ksm06 fixes

[Cyril Hrubis]

Two minor fixes for the ksm06 test.

Also it looks like the test_ksm_merge_across_nodes() is only called from
the ksm06 test so it does not make that much sense for the code to be in
the mem.c library. I guess that we should move it to the test source so
that it's not hidden in the library as well.

Cyril Hrubis (2):
  ksm06: Free memory at the end of the test
  ksm06: lock memory about to be merged

 testcases/kernel/mem/lib/mem.c | 8 ++++++++
 1 file changed, 8 insertions(+)

-- 
2.34.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH 1/2] ksm06: Free memory at the end of the test

[Cyril Hrubis]

Fixes the kms06 run with -i 2.

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
---
 testcases/kernel/mem/lib/mem.c | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/testcases/kernel/mem/lib/mem.c b/testcases/kernel/mem/lib/mem.c
index 99d5f7f53..39e3f1791 100644
--- a/testcases/kernel/mem/lib/mem.c
+++ b/testcases/kernel/mem/lib/mem.c
@@ -594,6 +594,11 @@ void test_ksm_merge_across_nodes(unsigned long nr_pages)
 		    0, 0, 0, nr_pages * num_nodes);
 
 	SAFE_FILE_PRINTF(PATH_KSM "run", "2");
+
+	for (i = 0; i < num_nodes; i++)
+		SAFE_MUNMAP(memory[i], length);
+
+	free(memory);
 }
 
 /* THP */
-- 
2.34.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH 2/2] ksm06: lock memory about to be merged

[Cyril Hrubis]

We have observed sporadic failures of the test in OpenQA in the case
that the memory allocated by the test is large enough (ksm06_2 runs the
test with -n 8000 which on ppc64le with 64k page size amounts to 1GB in
allocated buffers in total) and in the case of bacground load being run
while the test is executed (LTP compilation with suitable -j make
parameter suffices to trigger the conditions).

The culprit is that some of the pages may end up being swapped before
they happen to be merged due to a memory pressure caused by the
background load.

After this patch applied the test seems to be passing regardless the
background load.

Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
---
 testcases/kernel/mem/lib/mem.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/testcases/kernel/mem/lib/mem.c b/testcases/kernel/mem/lib/mem.c
index 39e3f1791..f1756b34a 100644
--- a/testcases/kernel/mem/lib/mem.c
+++ b/testcases/kernel/mem/lib/mem.c
@@ -565,6 +565,9 @@ void test_ksm_merge_across_nodes(unsigned long nr_pages)
 #endif
 
 		memset(memory[i], 10, length);
+
+		if (mlock(memory[i], length))
+			tst_res(TINFO | TERRNO, "mlock() failed");
 	}
 
 	SAFE_FILE_PRINTF(PATH_KSM "sleep_millisecs", "0");
-- 
2.34.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH 1/2] ksm06: Free memory at the end of the test

[Petr Vorel]

Hi Cyril,

obviously correct.
Reviewed-by: Petr Vorel <pvorel@suse.cz>

> Fixes the kms06 run with -i 2.
typo: s/kms06/ksm06/g

Kind regards,
Petr

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH 2/2] ksm06: lock memory about to be merged

[Petr Vorel]

Hi Cyril,

Reviewed-by: Petr Vorel <pvorel@suse.cz>
Good catch!

> Signed-off-by: Cyril Hrubis <chrubis@suse.cz>
> ---
>  testcases/kernel/mem/lib/mem.c | 3 +++
>  1 file changed, 3 insertions(+)

> diff --git a/testcases/kernel/mem/lib/mem.c b/testcases/kernel/mem/lib/mem.c
> index 39e3f1791..f1756b34a 100644
> --- a/testcases/kernel/mem/lib/mem.c
> +++ b/testcases/kernel/mem/lib/mem.c
> @@ -565,6 +565,9 @@ void test_ksm_merge_across_nodes(unsigned long nr_pages)
>  #endif

>  		memset(memory[i], 10, length);
> +
> +		if (mlock(memory[i], length))
> +			tst_res(TINFO | TERRNO, "mlock() failed");
I'd consider using TWARN.

Kind regards,
Petr

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH 2/2] ksm06: lock memory about to be merged

[Li Wang]

[-- Attachment #1.1: Type: text/plain, Size: 128 bytes --]

This patchset looks pretty good if go with Petr's suggestions.

Reviewed-by: Li Wang <liwang@redhat.com>


-- 
Regards,
Li Wang

[-- Attachment #1.2: Type: text/html, Size: 643 bytes --]

[-- Attachment #2: Type: text/plain, Size: 60 bytes --]


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH 2/2] ksm06: lock memory about to be merged

[Cyril Hrubis]

Hi!
Fixed the typo and changed the TINFO to TWARN and pushed.

Thanks for the review.

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH 0/2] ksm06 fixes

[Li Wang]

[-- Attachment #1.1: Type: text/plain, Size: 523 bytes --]

Cyril Hrubis <chrubis@suse.cz> wrote:

Two minor fixes for the ksm06 test.
>
> Also it looks like the test_ksm_merge_across_nodes() is only called from
> the ksm06 test so it does not make that much sense for the code to be in
> the mem.c library. I guess that we should move it to the test source so
> that it's not hidden in the library as well.
>

Absolutely yes, that will make mem library a bit shorten to read, we can do
that in a separate patch.
(sorry for overlooking this patch cover letter)

-- 
Regards,
Li Wang

[-- Attachment #1.2: Type: text/html, Size: 1156 bytes --]

[-- Attachment #2: Type: text/plain, Size: 60 bytes --]


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH 0/2] ksm06 fixes

[Cyril Hrubis]

Hi!
> Absolutely yes, that will make mem library a bit shorten to read, we can do
> that in a separate patch.
> (sorry for overlooking this patch cover letter)

I will send another cleanup patchset later on, I wanted the test fixed
ASAP which is why I limited the first patchset strictly to obvious
fixes.

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp