From: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
To: Usama Arif <usamaarif642@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>,
david@redhat.com, linux-mm@kvack.org,
linux-fsdevel@vger.kernel.org, corbet@lwn.net, rppt@kernel.org,
surenb@google.com, mhocko@suse.com, hannes@cmpxchg.org,
baohua@kernel.org, shakeel.butt@linux.dev, riel@surriel.com,
ziy@nvidia.com, laoar.shao@gmail.com, dev.jain@arm.com,
baolin.wang@linux.alibaba.com, npache@redhat.com,
Liam.Howlett@oracle.com, ryan.roberts@arm.com, vbabka@suse.cz,
jannh@google.com, Arnd Bergmann <arnd@arndb.de>,
sj@kernel.org, linux-kernel@vger.kernel.org,
linux-doc@vger.kernel.org, kernel-team@meta.com
Subject: Re: [PATCH v4 7/7] selftests: prctl: introduce tests for disabling THPs except for madvise
Date: Wed, 13 Aug 2025 16:13:59 +0100 [thread overview]
Message-ID: <13220ee2-d767-4133-9ef8-780fa165bbeb@lucifer.local> (raw)
In-Reply-To: <20250813135642.1986480-8-usamaarif642@gmail.com>
On Wed, Aug 13, 2025 at 02:55:42PM +0100, Usama Arif wrote:
> The test will set the global system THP setting to never, madvise
> or always depending on the fixture variant and the 2M setting to
> inherit before it starts (and reset to original at teardown).
> The fixture setup will also test if PR_SET_THP_DISABLE prctl call can
> be made with PR_THP_DISABLE_EXCEPT_ADVISED and skip if it fails.
>
> This tests if the process can:
> - successfully get the policy to disable THPs expect for madvise.
> - get hugepages only on MADV_HUGE and MADV_COLLAPSE if the global policy
> is madvise/always and only with MADV_COLLAPSE if the global policy is
> never.
> - successfully reset the policy of the process.
> - after reset, only get hugepages with:
> - MADV_COLLAPSE when policy is set to never.
> - MADV_HUGE and MADV_COLLAPSE when policy is set to madvise.
> - always when policy is set to "always".
> - repeat the above tests in a forked process to make sure the policy is
> carried across forks.
>
> Test results:
> ./prctl_thp_disable
> TAP version 13
> 1..12
> ok 1 prctl_thp_disable_completely.never.nofork
> ok 2 prctl_thp_disable_completely.never.fork
> ok 3 prctl_thp_disable_completely.madvise.nofork
> ok 4 prctl_thp_disable_completely.madvise.fork
> ok 5 prctl_thp_disable_completely.always.nofork
> ok 6 prctl_thp_disable_completely.always.fork
> ok 7 prctl_thp_disable_except_madvise.never.nofork
> ok 8 prctl_thp_disable_except_madvise.never.fork
> ok 9 prctl_thp_disable_except_madvise.madvise.nofork
> ok 10 prctl_thp_disable_except_madvise.madvise.fork
> ok 11 prctl_thp_disable_except_madvise.always.nofork
> ok 12 prctl_thp_disable_except_madvise.always.fork
>
> Signed-off-by: Usama Arif <usamaarif642@gmail.com>
I don't see any tests asserting VM_NOHUGEPAGE behaviour could you expand
this test to add that?
Other than that this is looking good.
Thanks!
> ---
> .../testing/selftests/mm/prctl_thp_disable.c | 107 ++++++++++++++++++
> 1 file changed, 107 insertions(+)
>
> diff --git a/tools/testing/selftests/mm/prctl_thp_disable.c b/tools/testing/selftests/mm/prctl_thp_disable.c
> index 8845e9f414560..9bfed4598a1a6 100644
> --- a/tools/testing/selftests/mm/prctl_thp_disable.c
> +++ b/tools/testing/selftests/mm/prctl_thp_disable.c
> @@ -16,6 +16,10 @@
> #include "thp_settings.h"
> #include "vm_util.h"
>
> +#ifndef PR_THP_DISABLE_EXCEPT_ADVISED
> +#define PR_THP_DISABLE_EXCEPT_ADVISED (1 << 1)
> +#endif
> +
> enum thp_collapse_type {
> THP_COLLAPSE_NONE,
> THP_COLLAPSE_MADV_HUGEPAGE, /* MADV_HUGEPAGE before access */
> @@ -165,4 +169,107 @@ TEST_F(prctl_thp_disable_completely, fork)
> ASSERT_EQ(ret, 0);
> }
>
> +static void prctl_thp_disable_except_madvise_test(struct __test_metadata *const _metadata,
> + size_t pmdsize,
> + enum thp_enabled thp_policy)
> +{
> + ASSERT_EQ(prctl(PR_GET_THP_DISABLE, NULL, NULL, NULL, NULL), 3);
> +
> + /* tests after prctl overrides global policy */
> + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_NONE, pmdsize), 0);
> +
> + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_HUGEPAGE, pmdsize),
> + thp_policy == THP_NEVER ? 0 : 1);
> +
> + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_COLLAPSE, pmdsize), 1);
> +
> + /* Reset to global policy */
> + ASSERT_EQ(prctl(PR_SET_THP_DISABLE, 0, NULL, NULL, NULL), 0);
> +
> + /* tests after prctl is cleared, and only global policy is effective */
> + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_NONE, pmdsize),
> + thp_policy == THP_ALWAYS ? 1 : 0);
> +
> + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_HUGEPAGE, pmdsize),
> + thp_policy == THP_NEVER ? 0 : 1);
> +
> + ASSERT_EQ(test_mmap_thp(THP_COLLAPSE_MADV_COLLAPSE, pmdsize), 1);
> +}
> +
> +FIXTURE(prctl_thp_disable_except_madvise)
> +{
> + struct thp_settings settings;
> + size_t pmdsize;
> +};
> +
> +FIXTURE_VARIANT(prctl_thp_disable_except_madvise)
> +{
> + enum thp_enabled thp_policy;
> +};
> +
> +FIXTURE_VARIANT_ADD(prctl_thp_disable_except_madvise, never)
> +{
> + .thp_policy = THP_NEVER,
> +};
> +
> +FIXTURE_VARIANT_ADD(prctl_thp_disable_except_madvise, madvise)
> +{
> + .thp_policy = THP_MADVISE,
> +};
> +
> +FIXTURE_VARIANT_ADD(prctl_thp_disable_except_madvise, always)
> +{
> + .thp_policy = THP_ALWAYS,
> +};
again the kselftest_harness stuff is really useful!
> +
> +FIXTURE_SETUP(prctl_thp_disable_except_madvise)
> +{
> + if (!thp_available())
> + SKIP(return, "Transparent Hugepages not available\n");
> +
> + self->pmdsize = read_pmd_pagesize();
> + if (!self->pmdsize)
> + SKIP(return, "Unable to read PMD size\n");
> +
> + if (prctl(PR_SET_THP_DISABLE, 1, PR_THP_DISABLE_EXCEPT_ADVISED, NULL, NULL))
> + SKIP(return, "Unable to set PR_THP_DISABLE_EXCEPT_ADVISED\n");
This should be a test fail I think, as the only ways this could fail are
invalid flags, or failure to obtain an mmap write lock.
> +
> + thp_save_settings();
> + thp_read_settings(&self->settings);
> + self->settings.thp_enabled = variant->thp_policy;
> + self->settings.hugepages[sz2ord(self->pmdsize, getpagesize())].enabled = THP_INHERIT;
> + thp_write_settings(&self->settings);
> +}
> +
> +FIXTURE_TEARDOWN(prctl_thp_disable_except_madvise)
> +{
> + thp_restore_settings();
> +}
> +
> +TEST_F(prctl_thp_disable_except_madvise, nofork)
> +{
> + prctl_thp_disable_except_madvise_test(_metadata, self->pmdsize, variant->thp_policy);
> +}
> +
> +TEST_F(prctl_thp_disable_except_madvise, fork)
> +{
> + int ret = 0;
> + pid_t pid;
> +
> + /* Make sure prctl changes are carried across fork */
> + pid = fork();
> + ASSERT_GE(pid, 0);
> +
> + if (!pid)
> + prctl_thp_disable_except_madvise_test(_metadata, self->pmdsize,
> + variant->thp_policy);
> +
> + wait(&ret);
> + if (WIFEXITED(ret))
> + ret = WEXITSTATUS(ret);
> + else
> + ret = -EINVAL;
> + ASSERT_EQ(ret, 0);
> +}
> +
> TEST_HARNESS_MAIN
> --
> 2.47.3
>
next prev parent reply other threads:[~2025-08-13 15:15 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-08-13 13:55 [PATCH v4 0/7] prctl: extend PR_SET_THP_DISABLE to only provide THPs when advised Usama Arif
2025-08-13 13:55 ` [PATCH v4 1/7] prctl: extend PR_SET_THP_DISABLE to optionally exclude VM_HUGEPAGE Usama Arif
2025-08-13 13:55 ` [PATCH v4 2/7] mm/huge_memory: convert "tva_flags" to "enum tva_type" Usama Arif
2025-08-14 3:07 ` Yafang Shao
2025-08-14 10:43 ` Usama Arif
2025-08-15 1:11 ` Andrew Morton
2025-08-15 9:29 ` Usama Arif
2025-08-14 14:59 ` Zi Yan
2025-08-13 13:55 ` [PATCH v4 3/7] mm/huge_memory: respect MADV_COLLAPSE with PR_THP_DISABLE_EXCEPT_ADVISED Usama Arif
2025-08-14 15:14 ` Zi Yan
2025-08-13 13:55 ` [PATCH v4 4/7] docs: transhuge: document process level THP controls Usama Arif
2025-08-13 14:30 ` Lorenzo Stoakes
2025-08-14 15:47 ` Zi Yan
2025-08-13 13:55 ` [PATCH v4 5/7] selftest/mm: Extract sz2ord function into vm_util.h Usama Arif
2025-08-13 14:31 ` Lorenzo Stoakes
2025-08-14 15:52 ` Zi Yan
2025-08-13 13:55 ` [PATCH v4 6/7] selftests: prctl: introduce tests for disabling THPs completely Usama Arif
2025-08-13 14:54 ` Lorenzo Stoakes
2025-08-13 13:55 ` [PATCH v4 7/7] selftests: prctl: introduce tests for disabling THPs except for madvise Usama Arif
2025-08-13 15:13 ` Lorenzo Stoakes [this message]
2025-08-13 16:24 ` David Hildenbrand
2025-08-13 18:52 ` Lorenzo Stoakes
2025-08-14 9:32 ` David Hildenbrand
2025-08-14 10:49 ` Lorenzo Stoakes
2025-08-14 11:45 ` Mark Brown
2025-08-14 12:00 ` David Hildenbrand
2025-08-14 12:09 ` Mark Brown
2025-08-14 12:59 ` David Hildenbrand
2025-08-14 13:08 ` Mark Brown
2025-08-14 15:02 ` Lorenzo Stoakes
2025-08-14 15:41 ` Usama Arif
2025-08-14 10:36 ` Usama Arif
2025-08-14 10:53 ` Lorenzo Stoakes
2025-08-14 11:51 ` Usama Arif
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=13220ee2-d767-4133-9ef8-780fa165bbeb@lucifer.local \
--to=lorenzo.stoakes@oracle.com \
--cc=Liam.Howlett@oracle.com \
--cc=akpm@linux-foundation.org \
--cc=arnd@arndb.de \
--cc=baohua@kernel.org \
--cc=baolin.wang@linux.alibaba.com \
--cc=corbet@lwn.net \
--cc=david@redhat.com \
--cc=dev.jain@arm.com \
--cc=hannes@cmpxchg.org \
--cc=jannh@google.com \
--cc=kernel-team@meta.com \
--cc=laoar.shao@gmail.com \
--cc=linux-doc@vger.kernel.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=mhocko@suse.com \
--cc=npache@redhat.com \
--cc=riel@surriel.com \
--cc=rppt@kernel.org \
--cc=ryan.roberts@arm.com \
--cc=shakeel.butt@linux.dev \
--cc=sj@kernel.org \
--cc=surenb@google.com \
--cc=usamaarif642@gmail.com \
--cc=vbabka@suse.cz \
--cc=ziy@nvidia.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).