* [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable
@ 2026-08-11 5:14 Wilson Felipe Pereira
2026-08-11 16:41 ` Yosry Ahmed
` (2 more replies)
0 siblings, 3 replies; 6+ messages in thread
From: Wilson Felipe Pereira @ 2026-08-11 5:14 UTC (permalink / raw)
To: Johannes Weiner, Yosry Ahmed, Nhat Pham, Andrew Morton,
Shuah Khan, Anshuman Khandual
Cc: Tejun Heo, Michal Koutný, Chengming Zhou, linux-mm, cgroups,
linux-kselftest, Wilson Felipe Pereira
test_no_kmem_bypass() needs to read
/sys/kernel/debug/zswap/stored_pages via get_zswap_stored_pages() to
verify that compressed pages are charged to the memcg.
When running in an environment where debugfs is not mounted or
CONFIG_DEBUG_FS is disabled, get_zswap_stored_pages() fails, causing the
loop to terminate early and report a false negative (KSFT_FAIL).
Selftests should not fail if debugfs is unavailable, and it should print
a message when it is skipped.
While I'm here, also add a warning message if the test is being skipped
due to totalram size and make the check for totalram more readable.
Signed-off-by: Wilson Felipe Pereira <wfelipe@google.com>
---
v1 -> v2:
- Check for debugfs availability using access("/sys/kernel/debug/zswap/stored_pages", R_OK)
instead of calling get_zswap_stored_pages() (Yosry Ahmed).
- Update debugfs print message to exactly match gup_test.c (Anshuman Khandual).
- Add GB(x) macro to cgroup_util.h (Yosry Ahmed). Cast to unsigned long long
to prevent 32-bit truncation without altering existing MB(x) signedness.
- Update RAM check to GB(4) to match comments and format totalram output
with 1 decimal place (%.1fGB) (Yosry Ahmed).
- Fix commit message typo (Yosry Ahmed).
- Moved stored_pages back inside the for loop.
v1: https://lore.kernel.org/all/20260801041255.752039-1-wfelipe@google.com/
.../selftests/cgroup/lib/include/cgroup_util.h | 1 +
tools/testing/selftests/cgroup/test_zswap.c | 13 +++++++++++--
2 files changed, 12 insertions(+), 2 deletions(-)
diff --git a/tools/testing/selftests/cgroup/lib/include/cgroup_util.h b/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
index febc1723d0903..c0f07226b222f 100644
--- a/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
+++ b/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
@@ -7,6 +7,7 @@
#endif
#define MB(x) (x << 20)
+#define GB(x) ((unsigned long long)(x) << 30)
#define USEC_PER_SEC 1000000L
#define NSEC_PER_SEC 1000000000L
diff --git a/tools/testing/selftests/cgroup/test_zswap.c b/tools/testing/selftests/cgroup/test_zswap.c
index 49b36ee791606..a836337440040 100644
--- a/tools/testing/selftests/cgroup/test_zswap.c
+++ b/tools/testing/selftests/cgroup/test_zswap.c
@@ -20,6 +20,7 @@ static int page_size;
#define PATH_ZSWAP "/sys/module/zswap"
#define PATH_ZSWAP_ENABLED "/sys/module/zswap/parameters/enabled"
+#define PATH_ZSWAP_STORED_PAGES "/sys/kernel/debug/zswap/stored_pages"
static int read_int(const char *path, size_t *value)
{
@@ -55,7 +56,7 @@ static int read_min_free_kb(size_t *value)
static int get_zswap_stored_pages(size_t *value)
{
- return read_int("/sys/kernel/debug/zswap/stored_pages", value);
+ return read_int(PATH_ZSWAP_STORED_PAGES, value);
}
static long get_cg_wb_count(const char *cg)
@@ -570,8 +571,16 @@ static int test_no_kmem_bypass(const char *root)
/* Read sys info and compute test values accordingly */
if (sysinfo(&sys_info) != 0)
return KSFT_FAIL;
- if (sys_info.totalram > 5000000000)
+ if (sys_info.totalram > GB(4)) {
+ ksft_print_msg(
+ "requires less than 4GB total ram, sys_info.totalram: %.1fGB\n",
+ (double)sys_info.totalram / GB(1));
return KSFT_SKIP;
+ }
+ if (access(PATH_ZSWAP_STORED_PAGES, R_OK)) {
+ ksft_print_msg("mount debugfs at /sys/kernel/debug\n");
+ return KSFT_SKIP;
+ }
values = mmap(0, sizeof(struct no_kmem_bypass_child_args), PROT_READ |
PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0);
if (values == MAP_FAILED)
--
2.55.0.679.g6767b8d81c-goog
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable
2026-08-11 5:14 [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable Wilson Felipe Pereira
@ 2026-08-11 16:41 ` Yosry Ahmed
2026-08-11 19:20 ` Nhat Pham
2026-08-12 2:06 ` Andrew Morton
2 siblings, 0 replies; 6+ messages in thread
From: Yosry Ahmed @ 2026-08-11 16:41 UTC (permalink / raw)
To: Wilson Felipe Pereira
Cc: Johannes Weiner, Nhat Pham, Andrew Morton, Shuah Khan,
Anshuman Khandual, Tejun Heo, Michal Koutný, Chengming Zhou,
linux-mm, cgroups, linux-kselftest
On Mon, Aug 10, 2026 at 10:14 PM Wilson Felipe Pereira
<wfelipe@google.com> wrote:
>
> test_no_kmem_bypass() needs to read
> /sys/kernel/debug/zswap/stored_pages via get_zswap_stored_pages() to
> verify that compressed pages are charged to the memcg.
>
> When running in an environment where debugfs is not mounted or
> CONFIG_DEBUG_FS is disabled, get_zswap_stored_pages() fails, causing the
> loop to terminate early and report a false negative (KSFT_FAIL).
>
> Selftests should not fail if debugfs is unavailable, and it should print
> a message when it is skipped.
>
> While I'm here, also add a warning message if the test is being skipped
> due to totalram size and make the check for totalram more readable.
>
> Signed-off-by: Wilson Felipe Pereira <wfelipe@google.com>
Acked-by: Yosry Ahmed <yosry@kernel.org>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable
2026-08-11 5:14 [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable Wilson Felipe Pereira
2026-08-11 16:41 ` Yosry Ahmed
@ 2026-08-11 19:20 ` Nhat Pham
2026-08-12 2:06 ` Andrew Morton
2 siblings, 0 replies; 6+ messages in thread
From: Nhat Pham @ 2026-08-11 19:20 UTC (permalink / raw)
To: Wilson Felipe Pereira
Cc: Johannes Weiner, Yosry Ahmed, Andrew Morton, Shuah Khan,
Anshuman Khandual, Tejun Heo, Michal Koutný, Chengming Zhou,
linux-mm, cgroups, linux-kselftest
On Mon, Aug 10, 2026 at 10:14 PM Wilson Felipe Pereira
<wfelipe@google.com> wrote:
>
> test_no_kmem_bypass() needs to read
> /sys/kernel/debug/zswap/stored_pages via get_zswap_stored_pages() to
> verify that compressed pages are charged to the memcg.
>
> When running in an environment where debugfs is not mounted or
> CONFIG_DEBUG_FS is disabled, get_zswap_stored_pages() fails, causing the
> loop to terminate early and report a false negative (KSFT_FAIL).
>
> Selftests should not fail if debugfs is unavailable, and it should print
> a message when it is skipped.
>
> While I'm here, also add a warning message if the test is being skipped
> due to totalram size and make the check for totalram more readable.
>
> Signed-off-by: Wilson Felipe Pereira <wfelipe@google.com>
Acked-by: Nhat Pham <nphamcs@gmail.com>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable
2026-08-11 5:14 [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable Wilson Felipe Pereira
2026-08-11 16:41 ` Yosry Ahmed
2026-08-11 19:20 ` Nhat Pham
@ 2026-08-12 2:06 ` Andrew Morton
2026-08-12 2:33 ` Anshuman Khandual
2026-08-12 2:45 ` SJ Park
2 siblings, 2 replies; 6+ messages in thread
From: Andrew Morton @ 2026-08-12 2:06 UTC (permalink / raw)
To: Wilson Felipe Pereira
Cc: Johannes Weiner, Yosry Ahmed, Nhat Pham, Shuah Khan,
Anshuman Khandual, Tejun Heo, Michal Koutný ,
Chengming Zhou, linux-mm, cgroups, linux-kselftest
On Tue, 11 Aug 2026 05:14:11 +0000 Wilson Felipe Pereira <wfelipe@google.com> wrote:
> test_no_kmem_bypass() needs to read
> /sys/kernel/debug/zswap/stored_pages via get_zswap_stored_pages() to
> verify that compressed pages are charged to the memcg.
>
> When running in an environment where debugfs is not mounted or
> CONFIG_DEBUG_FS is disabled, get_zswap_stored_pages() fails, causing the
> loop to terminate early and report a false negative (KSFT_FAIL).
>
> Selftests should not fail if debugfs is unavailable, and it should print
> a message when it is skipped.
>
> While I'm here, also add a warning message if the test is being skipped
> due to totalram size and make the check for totalram more readable.
>
> ...
>
> --- a/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
> +++ b/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
>
> ...
>
> @@ -570,8 +571,16 @@ static int test_no_kmem_bypass(const char *root)
> /* Read sys info and compute test values accordingly */
> if (sysinfo(&sys_info) != 0)
> return KSFT_FAIL;
> - if (sys_info.totalram > 5000000000)
> + if (sys_info.totalram > GB(4)) {
> + ksft_print_msg(
> + "requires less than 4GB total ram, sys_info.totalram: %.1fGB\n",
> + (double)sys_info.totalram / GB(1));
> return KSFT_SKIP;
> + }
> + if (access(PATH_ZSWAP_STORED_PAGES, R_OK)) {
> + ksft_print_msg("mount debugfs at /sys/kernel/debug\n");
Super nittynit: "mount debugfs at /sys/kernel/debug" sounds like a
progress message. "I just mounted debugfs at ...". I think the below is
more clear?
--- a/tools/testing/selftests/cgroup/test_zswap.c~selftests-cgroup-test_zswap-skip-test_no_kmem_bypass-if-debugfs-is-unavailable-fix
+++ a/tools/testing/selftests/cgroup/test_zswap.c
@@ -578,7 +578,7 @@ static int test_no_kmem_bypass(const cha
return KSFT_SKIP;
}
if (access(PATH_ZSWAP_STORED_PAGES, R_OK)) {
- ksft_print_msg("mount debugfs at /sys/kernel/debug\n");
+ ksft_print_msg("debugfs not mounted at /sys/kernel/debug\n");
return KSFT_SKIP;
}
values = mmap(0, sizeof(struct no_kmem_bypass_child_args), PROT_READ |
_
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable
2026-08-12 2:06 ` Andrew Morton
@ 2026-08-12 2:33 ` Anshuman Khandual
2026-08-12 2:45 ` SJ Park
1 sibling, 0 replies; 6+ messages in thread
From: Anshuman Khandual @ 2026-08-12 2:33 UTC (permalink / raw)
To: Andrew Morton
Cc: Wilson Felipe Pereira, Johannes Weiner, Yosry Ahmed, Nhat Pham,
Shuah Khan, Tejun Heo, Michal Koutný, Chengming Zhou,
linux-mm, cgroups, linux-kselftest
On Tue, Aug 11, 2026 at 07:06:11PM -0700, Andrew Morton wrote:
> On Tue, 11 Aug 2026 05:14:11 +0000 Wilson Felipe Pereira <wfelipe@google.com> wrote:
>
> > test_no_kmem_bypass() needs to read
> > /sys/kernel/debug/zswap/stored_pages via get_zswap_stored_pages() to
> > verify that compressed pages are charged to the memcg.
> >
> > When running in an environment where debugfs is not mounted or
> > CONFIG_DEBUG_FS is disabled, get_zswap_stored_pages() fails, causing the
> > loop to terminate early and report a false negative (KSFT_FAIL).
> >
> > Selftests should not fail if debugfs is unavailable, and it should print
> > a message when it is skipped.
> >
> > While I'm here, also add a warning message if the test is being skipped
> > due to totalram size and make the check for totalram more readable.
> >
> > ...
> >
> > --- a/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
> > +++ b/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
> >
> > ...
> >
> > @@ -570,8 +571,16 @@ static int test_no_kmem_bypass(const char *root)
> > /* Read sys info and compute test values accordingly */
> > if (sysinfo(&sys_info) != 0)
> > return KSFT_FAIL;
> > - if (sys_info.totalram > 5000000000)
> > + if (sys_info.totalram > GB(4)) {
> > + ksft_print_msg(
> > + "requires less than 4GB total ram, sys_info.totalram: %.1fGB\n",
> > + (double)sys_info.totalram / GB(1));
> > return KSFT_SKIP;
> > + }
> > + if (access(PATH_ZSWAP_STORED_PAGES, R_OK)) {
> > + ksft_print_msg("mount debugfs at /sys/kernel/debug\n");
>
> Super nittynit: "mount debugfs at /sys/kernel/debug" sounds like a
> progress message. "I just mounted debugfs at ...". I think the below is
> more clear?
>
> --- a/tools/testing/selftests/cgroup/test_zswap.c~selftests-cgroup-test_zswap-skip-test_no_kmem_bypass-if-debugfs-is-unavailable-fix
> +++ a/tools/testing/selftests/cgroup/test_zswap.c
> @@ -578,7 +578,7 @@ static int test_no_kmem_bypass(const cha
> return KSFT_SKIP;
> }
> if (access(PATH_ZSWAP_STORED_PAGES, R_OK)) {
> - ksft_print_msg("mount debugfs at /sys/kernel/debug\n");
> + ksft_print_msg("debugfs not mounted at /sys/kernel/debug\n");
This is definitely more accurate.
> return KSFT_SKIP;
> }
> values = mmap(0, sizeof(struct no_kmem_bypass_child_args), PROT_READ |
> _
>
I assume you would change this while applying the patch.
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable
2026-08-12 2:06 ` Andrew Morton
2026-08-12 2:33 ` Anshuman Khandual
@ 2026-08-12 2:45 ` SJ Park
1 sibling, 0 replies; 6+ messages in thread
From: SJ Park @ 2026-08-12 2:45 UTC (permalink / raw)
To: Andrew Morton
Cc: SJ Park, Wilson Felipe Pereira, Johannes Weiner, Yosry Ahmed,
Nhat Pham, Shuah Khan, Anshuman Khandual, Tejun Heo,
Michal Koutný , Chengming Zhou, linux-mm, cgroups,
linux-kselftest
On Tue, 11 Aug 2026 19:06:11 -0700 Andrew Morton <akpm@linux-foundation.org> wrote:
> On Tue, 11 Aug 2026 05:14:11 +0000 Wilson Felipe Pereira <wfelipe@google.com> wrote:
>
> > test_no_kmem_bypass() needs to read
> > /sys/kernel/debug/zswap/stored_pages via get_zswap_stored_pages() to
> > verify that compressed pages are charged to the memcg.
> >
> > When running in an environment where debugfs is not mounted or
> > CONFIG_DEBUG_FS is disabled, get_zswap_stored_pages() fails, causing the
> > loop to terminate early and report a false negative (KSFT_FAIL).
> >
> > Selftests should not fail if debugfs is unavailable, and it should print
> > a message when it is skipped.
> >
> > While I'm here, also add a warning message if the test is being skipped
> > due to totalram size and make the check for totalram more readable.
> >
> > ...
> >
> > --- a/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
> > +++ b/tools/testing/selftests/cgroup/lib/include/cgroup_util.h
> >
> > ...
> >
> > @@ -570,8 +571,16 @@ static int test_no_kmem_bypass(const char *root)
> > /* Read sys info and compute test values accordingly */
> > if (sysinfo(&sys_info) != 0)
> > return KSFT_FAIL;
> > - if (sys_info.totalram > 5000000000)
> > + if (sys_info.totalram > GB(4)) {
> > + ksft_print_msg(
> > + "requires less than 4GB total ram, sys_info.totalram: %.1fGB\n",
> > + (double)sys_info.totalram / GB(1));
> > return KSFT_SKIP;
> > + }
> > + if (access(PATH_ZSWAP_STORED_PAGES, R_OK)) {
> > + ksft_print_msg("mount debugfs at /sys/kernel/debug\n");
>
> Super nittynit: "mount debugfs at /sys/kernel/debug" sounds like a
> progress message. "I just mounted debugfs at ...". I think the below is
> more clear?
+1. I was feeling similar.
>
> --- a/tools/testing/selftests/cgroup/test_zswap.c~selftests-cgroup-test_zswap-skip-test_no_kmem_bypass-if-debugfs-is-unavailable-fix
> +++ a/tools/testing/selftests/cgroup/test_zswap.c
> @@ -578,7 +578,7 @@ static int test_no_kmem_bypass(const cha
> return KSFT_SKIP;
> }
> if (access(PATH_ZSWAP_STORED_PAGES, R_OK)) {
> - ksft_print_msg("mount debugfs at /sys/kernel/debug\n");
> + ksft_print_msg("debugfs not mounted at /sys/kernel/debug\n");
This reads better for me. I also find this is more consistent to similar skip
messages in main() and check_zswap_enabled().
> return KSFT_SKIP;
> }
> values = mmap(0, sizeof(struct no_kmem_bypass_child_args), PROT_READ |
> _
Thanks,
SJ
^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2026-08-12 2:46 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-08-11 5:14 [PATCH v2] selftests/cgroup: test_zswap: skip test_no_kmem_bypass if debugfs is unavailable Wilson Felipe Pereira
2026-08-11 16:41 ` Yosry Ahmed
2026-08-11 19:20 ` Nhat Pham
2026-08-12 2:06 ` Andrew Morton
2026-08-12 2:33 ` Anshuman Khandual
2026-08-12 2:45 ` SJ Park
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox