[PATCH 0/3] selftests: cgroup: add zswap test program

[PATCH 0/3] selftests: cgroup: add zswap test program

[Domenico Cerasuolo]

This series adds 2 zswap related selftests that verify known and fixed
issues. A new dedicated test program (test_zswap) is proposed since
the test cases are specific to zswap and hosts specific helpers.

The first patch adds the (empty) test program, while the other 2 add an
actual test function each.

Domenico Cerasuolo (3):
  selftests: cgroup: add test_zswap program
  selftests: cgroup: add test_zswap with no kmem bypass test
  selftests: cgroup: add zswap-memcg unwanted writeback test

 tools/testing/selftests/cgroup/.gitignore   |   1 +
 tools/testing/selftests/cgroup/Makefile     |   2 +
 tools/testing/selftests/cgroup/test_zswap.c | 286 ++++++++++++++++++++
 3 files changed, 289 insertions(+)
 create mode 100644 tools/testing/selftests/cgroup/test_zswap.c

-- 
2.34.1

[PATCH 1/3] selftests: cgroup: add test_zswap program

[Domenico Cerasuolo]

Add empty cgroup-zswap self test scaffold program, test functions to be
added in the next commits.

Signed-off-by: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
---
 tools/testing/selftests/cgroup/.gitignore   |  1 +
 tools/testing/selftests/cgroup/Makefile     |  2 +
 tools/testing/selftests/cgroup/test_zswap.c | 62 +++++++++++++++++++++
 3 files changed, 65 insertions(+)
 create mode 100644 tools/testing/selftests/cgroup/test_zswap.c

diff --git a/tools/testing/selftests/cgroup/.gitignore b/tools/testing/selftests/cgroup/.gitignore
index c4a57e69f749..4d556df4f77b 100644
--- a/tools/testing/selftests/cgroup/.gitignore
+++ b/tools/testing/selftests/cgroup/.gitignore
@@ -5,4 +5,5 @@ test_freezer
 test_kmem
 test_kill
 test_cpu
+test_zswap
 wait_inotify
diff --git a/tools/testing/selftests/cgroup/Makefile b/tools/testing/selftests/cgroup/Makefile
index 3d263747d2ad..27dbdd7bb4bb 100644
--- a/tools/testing/selftests/cgroup/Makefile
+++ b/tools/testing/selftests/cgroup/Makefile
@@ -12,6 +12,7 @@ TEST_GEN_PROGS += test_core
 TEST_GEN_PROGS += test_freezer
 TEST_GEN_PROGS += test_kill
 TEST_GEN_PROGS += test_cpu
+TEST_GEN_PROGS += test_zswap
 
 LOCAL_HDRS += $(selfdir)/clone3/clone3_selftests.h $(selfdir)/pidfd/pidfd.h
 
@@ -23,3 +24,4 @@ $(OUTPUT)/test_core: cgroup_util.c
 $(OUTPUT)/test_freezer: cgroup_util.c
 $(OUTPUT)/test_kill: cgroup_util.c
 $(OUTPUT)/test_cpu: cgroup_util.c
+$(OUTPUT)/test_zswap: cgroup_util.c
diff --git a/tools/testing/selftests/cgroup/test_zswap.c b/tools/testing/selftests/cgroup/test_zswap.c
new file mode 100644
index 000000000000..001533667322
--- /dev/null
+++ b/tools/testing/selftests/cgroup/test_zswap.c
@@ -0,0 +1,62 @@
+// SPDX-License-Identifier: GPL-2.0
+#define _GNU_SOURCE
+
+#include <linux/limits.h>
+#include <unistd.h>
+#include <stdio.h>
+
+#include "../kselftest.h"
+#include "cgroup_util.h"
+
+#define T(x) { x, #x }
+struct zswap_test {
+	int (*fn)(const char *root);
+	const char *name;
+} tests[] = {
+};
+#undef T
+
+static bool zswap_configured(void)
+{
+	return access("/sys/module/zswap", F_OK) == 0;
+}
+
+int main(int argc, char **argv)
+{
+	char root[PATH_MAX];
+	int i, ret = EXIT_SUCCESS;
+
+	if (cg_find_unified_root(root, sizeof(root)))
+		ksft_exit_skip("cgroup v2 isn't mounted\n");
+
+	if (!zswap_configured())
+		ksft_exit_skip("zswap isn't configured\n");
+
+	/*
+	 * Check that memory controller is available:
+	 * memory is listed in cgroup.controllers
+	 */
+	if (cg_read_strstr(root, "cgroup.controllers", "memory"))
+		ksft_exit_skip("memory controller isn't available\n");
+
+	if (cg_read_strstr(root, "cgroup.subtree_control", "memory"))
+		if (cg_write(root, "cgroup.subtree_control", "+memory"))
+			ksft_exit_skip("Failed to set memory controller\n");
+
+	for (i = 0; i < ARRAY_SIZE(tests); i++) {
+		switch (tests[i].fn(root)) {
+		case KSFT_PASS:
+			ksft_test_result_pass("%s\n", tests[i].name);
+			break;
+		case KSFT_SKIP:
+			ksft_test_result_skip("%s\n", tests[i].name);
+			break;
+		default:
+			ret = EXIT_FAILURE;
+			ksft_test_result_fail("%s\n", tests[i].name);
+			break;
+		}
+	}
+
+	return ret;
+}
-- 
2.34.1

[PATCH 2/3] selftests: cgroup: add test_zswap with no kmem bypass test

[Domenico Cerasuolo]

Add a cgroup selftest that verifies memcg charging in zswap.
The original issue was that kmem bypass was applied to pages swapped out
to zswap by kswapd, resulting in zswapped memory not being charged.
It was fixed by commit cd08d80ecdac("mm: correctly charge compressed
memory to its memcg").

Signed-off-by: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
---
 tools/testing/selftests/cgroup/test_zswap.c | 163 ++++++++++++++++++++
 1 file changed, 163 insertions(+)

diff --git a/tools/testing/selftests/cgroup/test_zswap.c b/tools/testing/selftests/cgroup/test_zswap.c
index 001533667322..e859fecd310b 100644
--- a/tools/testing/selftests/cgroup/test_zswap.c
+++ b/tools/testing/selftests/cgroup/test_zswap.c
@@ -4,15 +4,178 @@
 #include <linux/limits.h>
 #include <unistd.h>
 #include <stdio.h>
+#include <signal.h>
+#include <sys/sysinfo.h>
+#include <string.h>
+#include <sys/wait.h>
+#include <sys/mman.h>
 
 #include "../kselftest.h"
 #include "cgroup_util.h"
 
+static int read_int(const char *path, size_t *value)
+{
+	FILE *file;
+	int ret = 0;
+
+	file = fopen(path, "r");
+	if (!file)
+		return -1;
+	if (fscanf(file, "%ld", value) != 1)
+		ret = -1;
+	fclose(file);
+	return ret;
+}
+
+static int set_min_free_kb(size_t value)
+{
+	FILE *file;
+	int ret;
+
+	file = fopen("/proc/sys/vm/min_free_kbytes", "w");
+	if (!file)
+		return -1;
+	ret = fprintf(file, "%ld\n", value);
+	fclose(file);
+	return ret;
+}
+
+static int read_min_free_kb(size_t *value)
+{
+	return read_int("/proc/sys/vm/min_free_kbytes", value);
+}
+
+static int get_zswap_stored_pages(size_t *value)
+{
+	return read_int("/sys/kernel/debug/zswap/stored_pages", value);
+}
+
+struct no_kmem_bypass_child_args {
+	size_t target_alloc_bytes;
+	size_t child_allocated;
+};
+
+static int no_kmem_bypass_child(const char *cgroup, void *arg)
+{
+	struct no_kmem_bypass_child_args *values = arg;
+	void *allocation;
+
+	allocation = malloc(values->target_alloc_bytes);
+	if (!allocation) {
+		values->child_allocated = true;
+		return -1;
+	}
+	for (long i = 0; i < values->target_alloc_bytes; i += 4095)
+		((char *)allocation)[i] = 'a';
+	values->child_allocated = true;
+	pause();
+	free(allocation);
+	return 0;
+}
+
+/*
+ * When pages owned by a memcg are pushed to zswap by kswapd, they should be
+ * charged to that cgroup. This wasn't the case before commit
+ * cd08d80ecdac("mm: correctly charge compressed memory to its memcg").
+ *
+ * The test first allocates memory in a memcg, then raises min_free_kbytes to
+ * a very high value so that the allocation falls below low wm, then makes
+ * another allocation to trigger kswapd that should push the memcg-owned pages
+ * to zswap and verifies that the zswap pages are correctly charged.
+ *
+ * To be run on a VM with at most 4G of memory.
+ */
+static int test_no_kmem_bypass(const char *root)
+{
+	size_t min_free_kb_high, min_free_kb_low, min_free_kb_original;
+	struct no_kmem_bypass_child_args *values;
+	size_t trigger_allocation_size;
+	int wait_child_iteration = 0;
+	long stored_pages_threshold;
+	struct sysinfo sys_info;
+	int ret = KSFT_FAIL;
+	int child_status;
+	char *test_group;
+	pid_t child_pid;
+
+	/* Read sys info and compute test values accordingly */
+	if (sysinfo(&sys_info) != 0)
+		return KSFT_FAIL;
+	if (sys_info.totalram > 5000000000)
+		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)
+		return KSFT_FAIL;
+	if (read_min_free_kb(&min_free_kb_original))
+		return KSFT_FAIL;
+	min_free_kb_high = sys_info.totalram / 2000;
+	min_free_kb_low = sys_info.totalram / 500000;
+	values->target_alloc_bytes = (sys_info.totalram - min_free_kb_high * 1000) +
+		sys_info.totalram * 5 / 100;
+	stored_pages_threshold = sys_info.totalram / 5 / 4096;
+	trigger_allocation_size = sys_info.totalram / 20;
+
+	/* Set up test memcg */
+	if (cg_write(root, "cgroup.subtree_control", "+memory"))
+		goto out;
+	test_group = cg_name(root, "kmem_bypass_test");
+	if (!test_group)
+		goto out;
+
+	/* Spawn memcg child and wait for it to allocate */
+	set_min_free_kb(min_free_kb_low);
+	if (cg_create(test_group))
+		goto out;
+	values->child_allocated = false;
+	child_pid = cg_run_nowait(test_group, no_kmem_bypass_child, values);
+	if (child_pid < 0)
+		goto out;
+	while (!values->child_allocated && wait_child_iteration++ < 10000)
+		usleep(1000);
+
+	/* Try to wakeup kswapd and let it push child memory to zswap */
+	set_min_free_kb(min_free_kb_high);
+	for (int i = 0; i < 20; i++) {
+		size_t stored_pages;
+		char *trigger_allocation = malloc(trigger_allocation_size);
+
+		if (!trigger_allocation)
+			break;
+		for (int i = 0; i < trigger_allocation_size; i += 4095)
+			trigger_allocation[i] = 'b';
+		usleep(100000);
+		free(trigger_allocation);
+		if (get_zswap_stored_pages(&stored_pages))
+			break;
+		if (stored_pages < 0)
+			break;
+		/* If memory was pushed to zswap, verify it belongs to memcg */
+		if (stored_pages > stored_pages_threshold) {
+			int zswapped = cg_read_key_long(test_group, "memory.stat", "zswapped ");
+			int delta = stored_pages * 4096 - zswapped;
+			int result_ok = delta < stored_pages * 4096 / 4;
+
+			ret = result_ok ? KSFT_PASS : KSFT_FAIL;
+			break;
+		}
+	}
+
+	kill(child_pid, SIGTERM);
+	waitpid(child_pid, &child_status, 0);
+out:
+	set_min_free_kb(min_free_kb_original);
+	cg_destroy(test_group);
+	free(test_group);
+	return ret;
+}
+
 #define T(x) { x, #x }
 struct zswap_test {
 	int (*fn)(const char *root);
 	const char *name;
 } tests[] = {
+	T(test_no_kmem_bypass),
 };
 #undef T
 
-- 
2.34.1

[PATCH 3/3] selftests: cgroup: add zswap-memcg unwanted writeback test

[Domenico Cerasuolo]

Add a test to verify that when a memcg hits its limit in zswap, it
doesn't trigger an unwanted writeback that would result in pages not
owned by that memcg to be sent to disk, even if zswap isn't full.
This was fixed by commit 0bdf0efa180a("zswap: do not shrink if cgroup
may not zswap").

Signed-off-by: Domenico Cerasuolo <cerasuolodomenico@gmail.com>
---
 tools/testing/selftests/cgroup/test_zswap.c | 61 +++++++++++++++++++++
 1 file changed, 61 insertions(+)

diff --git a/tools/testing/selftests/cgroup/test_zswap.c b/tools/testing/selftests/cgroup/test_zswap.c
index e859fecd310b..49def87a909b 100644
--- a/tools/testing/selftests/cgroup/test_zswap.c
+++ b/tools/testing/selftests/cgroup/test_zswap.c
@@ -50,6 +50,66 @@ static int get_zswap_stored_pages(size_t *value)
 	return read_int("/sys/kernel/debug/zswap/stored_pages", value);
 }
 
+static int get_zswap_written_back_pages(size_t *value)
+{
+	return read_int("/sys/kernel/debug/zswap/written_back_pages", value);
+}
+
+static int allocate_bytes(const char *cgroup, void *arg)
+{
+	size_t size = (size_t)arg;
+	char *mem = (char *)malloc(size);
+
+	if (!mem)
+		return -1;
+	for (int i = 0; i < size; i += 4095)
+		mem[i] = 'a';
+	free(mem);
+	return 0;
+}
+
+/*
+ * When trying to store a memcg page in zswap, if the memcg hits its memory
+ * limit in zswap, writeback should not be triggered.
+ *
+ * This was fixed with commit 0bdf0efa180a("zswap: do not shrink if cgroup may
+ * not zswap"). Needs to be revised when a per memcg writeback mechanism is
+ * implemented.
+ */
+static int test_no_invasive_cgroup_shrink(const char *root)
+{
+	size_t written_back_before, written_back_after;
+	int ret = KSFT_FAIL;
+	char *test_group;
+
+	/* Set up */
+	test_group = cg_name(root, "no_shrink_test");
+	if (!test_group)
+		goto out;
+	if (cg_create(test_group))
+		goto out;
+	if (cg_write(test_group, "memory.max", "1M"))
+		goto out;
+	if (cg_write(test_group, "memory.zswap.max", "10K"))
+		goto out;
+	if (get_zswap_written_back_pages(&written_back_before))
+		goto out;
+
+	/* Allocate 10x memory.max to push memory into zswap */
+	if (cg_run(test_group, allocate_bytes, (void *)MB(10)))
+		goto out;
+
+	/* Verify that no writeback happened because of the memcg allocation */
+	if (get_zswap_written_back_pages(&written_back_after))
+		goto out;
+	if (written_back_after == written_back_before)
+		ret = KSFT_PASS;
+out:
+	cg_destroy(test_group);
+	free(test_group);
+	return ret;
+}
+
 struct no_kmem_bypass_child_args {
 	size_t target_alloc_bytes;
 	size_t child_allocated;
@@ -176,6 +236,7 @@ struct zswap_test {
 	const char *name;
 } tests[] = {
 	T(test_no_kmem_bypass),
+	T(test_no_invasive_cgroup_shrink),
 };
 #undef T
 
-- 
2.34.1