* [PATCH v6 1/3] selftests: cgroup: Add dmem_selftest helper module
2026-08-31 13:25 [PATCH v6 0/3] cgroup: dmem: add selftest helper and coverage Albert Esteve
@ 2026-08-31 13:25 ` Albert Esteve
2026-08-31 13:25 ` [PATCH v6 2/3] selftests: cgroup: Add KB() helper Albert Esteve
2026-08-31 13:25 ` [PATCH v6 3/3] selftests: cgroup: Add dmem selftest coverage Albert Esteve
2 siblings, 0 replies; 4+ messages in thread
From: Albert Esteve @ 2026-08-31 13:25 UTC (permalink / raw)
To: Tejun Heo, Johannes Weiner, Michal Koutný, Shuah Khan
Cc: linux-kernel, cgroups, linux-kselftest, Albert Esteve
Currently, dmem charging is driver-driven through direct
calls to dmem_cgroup_try_charge() (for example TTM).
That does not give cgroup selftests a generic way to
trigger alloc/free paths from userspace in a chosen
cgroup, so coverage of nested dmem.max still
needs a synthetic region.
Add tools/testing/selftests/cgroup/test_modules/dmem_selftest.c
as an OOT helper, built like other test_modules (no extra
Kconfig needed). Tests can insmod the .ko module to register a
"dmem_selftest" region.
Module parameters drive a single outstanding allocation against
the calling task's cgroup:
/sys/module/dmem_selftest/parameters/alloc
/sys/module/dmem_selftest/parameters/free
Writing a size to alloc triggers dmem_cgroup_try_charge() for
the calling task's cgroup. Writing to free with no argument
releases the outstanding charge via dmem_cgroup_uncharge().
Only a single outstanding charge is supported.
This provides a deterministic, driver-independent mechanism
for exercising dmem accounting paths in selftests.
Signed-off-by: Albert Esteve <aesteve@redhat.com>
---
tools/testing/selftests/cgroup/Makefile | 1 +
tools/testing/selftests/cgroup/config | 1 +
.../testing/selftests/cgroup/test_modules/Makefile | 17 +++
.../selftests/cgroup/test_modules/dmem_selftest.c | 146 +++++++++++++++++++++
4 files changed, 165 insertions(+)
diff --git a/tools/testing/selftests/cgroup/Makefile b/tools/testing/selftests/cgroup/Makefile
index e01584c2189a..a08661320a07 100644
--- a/tools/testing/selftests/cgroup/Makefile
+++ b/tools/testing/selftests/cgroup/Makefile
@@ -6,6 +6,7 @@ all: ${HELPER_PROGS}
TEST_FILES := with_stress.sh
TEST_PROGS := test_stress.sh test_cpuset_prs.sh test_cpuset_v1_hp.sh
TEST_GEN_FILES := wait_inotify
+TEST_GEN_MODS_DIR := test_modules
# Keep the lists lexicographically sorted
TEST_GEN_PROGS = test_core
TEST_GEN_PROGS += test_cpu
diff --git a/tools/testing/selftests/cgroup/config b/tools/testing/selftests/cgroup/config
index 39f979690dd3..2ee0488c3d65 100644
--- a/tools/testing/selftests/cgroup/config
+++ b/tools/testing/selftests/cgroup/config
@@ -1,5 +1,6 @@
CONFIG_CGROUPS=y
CONFIG_CGROUP_CPUACCT=y
+CONFIG_CGROUP_DMEM=y
CONFIG_CGROUP_FREEZER=y
CONFIG_CGROUP_SCHED=y
CONFIG_MEMCG=y
diff --git a/tools/testing/selftests/cgroup/test_modules/Makefile b/tools/testing/selftests/cgroup/test_modules/Makefile
new file mode 100644
index 000000000000..dd8a17313cec
--- /dev/null
+++ b/tools/testing/selftests/cgroup/test_modules/Makefile
@@ -0,0 +1,17 @@
+# SPDX-License-Identifier: GPL-2.0
+TESTMODS_DIR := $(realpath $(dir $(abspath $(lastword $(MAKEFILE_LIST)))))
+KDIR ?= /lib/modules/$(shell uname -r)/build
+
+obj-m += dmem_selftest.o
+
+# Ensure that KDIR exists, otherwise skip the compilation
+modules:
+ifneq ("$(wildcard $(KDIR))", "")
+ $(Q)$(MAKE) -C $(KDIR) modules KBUILD_EXTMOD=$(TESTMODS_DIR)
+endif
+
+# Ensure that KDIR exists, otherwise skip the clean target
+clean:
+ifneq ("$(wildcard $(KDIR))", "")
+ $(Q)$(MAKE) -C $(KDIR) clean KBUILD_EXTMOD=$(TESTMODS_DIR)
+endif
diff --git a/tools/testing/selftests/cgroup/test_modules/dmem_selftest.c b/tools/testing/selftests/cgroup/test_modules/dmem_selftest.c
new file mode 100644
index 000000000000..a763a156f13f
--- /dev/null
+++ b/tools/testing/selftests/cgroup/test_modules/dmem_selftest.c
@@ -0,0 +1,146 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Kselftest helper for the dmem cgroup controller.
+ *
+ * Registers a synthetic dmem region so tests can trigger allocations
+ * from the calling task's cgroup via module parameters:
+ * /sys/module/dmem_selftest/parameters/alloc
+ * /sys/module/dmem_selftest/parameters/free
+ */
+
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/cgroup_dmem.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/moduleparam.h>
+#include <linux/mutex.h>
+#include <linux/panic.h>
+#include <linux/sysfs.h>
+
+#define DM_SELFTEST_REGION_NAME "dmem_selftest"
+#define DM_SELFTEST_REGION_SIZE (256ULL * 1024 * 1024)
+
+static struct dmem_cgroup_region *selftest_region;
+static struct dmem_cgroup_pool_state *alloc_pool;
+static u64 alloc_size;
+static DEFINE_MUTEX(alloc_lock);
+
+static int param_set_alloc(const char *val, const struct kernel_param *kp)
+{
+ struct dmem_cgroup_pool_state *pool = NULL, *limit = NULL;
+ u64 size;
+ int ret;
+
+ if (!selftest_region)
+ return -ENODEV;
+
+ ret = kstrtou64(val, 0, &size);
+ if (ret)
+ return ret;
+ if (!size || size > DM_SELFTEST_REGION_SIZE)
+ return -EINVAL;
+
+ mutex_lock(&alloc_lock);
+ if (alloc_pool) {
+ mutex_unlock(&alloc_lock);
+ return -EBUSY;
+ }
+
+ ret = dmem_cgroup_try_charge(selftest_region, size, &pool, &limit);
+ if (ret == -EAGAIN && limit)
+ dmem_cgroup_pool_state_put(limit);
+ if (ret) {
+ mutex_unlock(&alloc_lock);
+ return ret;
+ }
+
+ alloc_pool = pool;
+ alloc_size = size;
+ mutex_unlock(&alloc_lock);
+ return 0;
+}
+
+static int param_get_alloc(char *buffer, const struct kernel_param *kp)
+{
+ u64 size;
+
+ mutex_lock(&alloc_lock);
+ size = alloc_size;
+ mutex_unlock(&alloc_lock);
+ return sysfs_emit(buffer, "%llu\n", size);
+}
+
+static const struct kernel_param_ops alloc_ops = {
+ .set = param_set_alloc,
+ .get = param_get_alloc,
+};
+
+module_param_cb(alloc, &alloc_ops, NULL, 0644);
+MODULE_PARM_DESC(alloc, "Allocate (charge) SIZE bytes against the calling task's dmem cgroup");
+
+static int param_set_free(const char *val, const struct kernel_param *kp)
+{
+ mutex_lock(&alloc_lock);
+ if (!alloc_pool) {
+ mutex_unlock(&alloc_lock);
+ return -EINVAL;
+ }
+
+ dmem_cgroup_uncharge(alloc_pool, alloc_size);
+ alloc_pool = NULL;
+ alloc_size = 0;
+ mutex_unlock(&alloc_lock);
+ return 0;
+}
+
+static const struct kernel_param_ops free_ops = {
+ .flags = KERNEL_PARAM_OPS_FL_NOARG,
+ .set = param_set_free,
+};
+
+module_param_cb(free, &free_ops, NULL, 0200);
+MODULE_PARM_DESC(free, "Free the outstanding dmem selftest allocation");
+
+static int __init dmem_selftest_init(void)
+{
+ static const struct dmem_cgroup_init init = {
+ .size = DM_SELFTEST_REGION_SIZE,
+ };
+
+ selftest_region = dmem_cgroup_register_region(&init, DM_SELFTEST_REGION_NAME);
+ if (IS_ERR(selftest_region))
+ return PTR_ERR(selftest_region);
+ if (!selftest_region)
+ return -EINVAL;
+
+ add_taint(TAINT_TEST, LOCKDEP_STILL_OK);
+ pr_info("region '%s' registered; parameters alloc/free\n",
+ DM_SELFTEST_REGION_NAME);
+ return 0;
+}
+
+static void __exit dmem_selftest_exit(void)
+{
+ mutex_lock(&alloc_lock);
+ if (alloc_pool) {
+ dmem_cgroup_uncharge(alloc_pool, alloc_size);
+ alloc_pool = NULL;
+ alloc_size = 0;
+ }
+ mutex_unlock(&alloc_lock);
+
+ if (selftest_region) {
+ dmem_cgroup_unregister_region(selftest_region);
+ selftest_region = NULL;
+ }
+ pr_info("unloaded.\n");
+}
+
+module_init(dmem_selftest_init);
+module_exit(dmem_selftest_exit);
+
+MODULE_AUTHOR("Albert Esteve <aesteve@redhat.com>");
+MODULE_DESCRIPTION("Kselftest helper for cgroup dmem controller");
+MODULE_LICENSE("GPL");
+MODULE_INFO(test, "Y");
--
2.55.0
^ permalink raw reply related [flat|nested] 4+ messages in thread* [PATCH v6 3/3] selftests: cgroup: Add dmem selftest coverage
2026-08-31 13:25 [PATCH v6 0/3] cgroup: dmem: add selftest helper and coverage Albert Esteve
2026-08-31 13:25 ` [PATCH v6 1/3] selftests: cgroup: Add dmem_selftest helper module Albert Esteve
2026-08-31 13:25 ` [PATCH v6 2/3] selftests: cgroup: Add KB() helper Albert Esteve
@ 2026-08-31 13:25 ` Albert Esteve
2 siblings, 0 replies; 4+ messages in thread
From: Albert Esteve @ 2026-08-31 13:25 UTC (permalink / raw)
To: Tejun Heo, Johannes Weiner, Michal Koutný, Shuah Khan
Cc: linux-kernel, cgroups, linux-kselftest, Albert Esteve
Currently, tools/testing/selftests/cgroup/ has no
dmem-specific test. This leaves dmem accounting and
dmem.max enforcement largely unvalidated without a
production driver.
Add test_dmem, driven by the dmem_selftest helper module
parameters (alloc/free) on the dmem_selftest region.
insmod the helper before running the tests.
The tests cover:
- test_dmem_max: nested hierarchy with per-leaf dmem.max
values and verifies that over-limit charges fail while
in-limit charges succeed in dmem.current.
- test_dmem_alloc_byte_granularity: non-page-aligned alloc
sizes, dmem.current equals the requested byte count, and
free-to-zero behavior.
Limits are read with cg_read_key_long(). The helper is stat'd
in main() so a missing module skips the whole test instead
of each subtest.
Signed-off-by: Albert Esteve <aesteve@redhat.com>
---
tools/testing/selftests/cgroup/.gitignore | 1 +
tools/testing/selftests/cgroup/Makefile | 2 +
tools/testing/selftests/cgroup/test_dmem.c | 314 +++++++++++++++++++++++++++++
3 files changed, 317 insertions(+)
diff --git a/tools/testing/selftests/cgroup/.gitignore b/tools/testing/selftests/cgroup/.gitignore
index 952e4448bf07..ea2322598217 100644
--- a/tools/testing/selftests/cgroup/.gitignore
+++ b/tools/testing/selftests/cgroup/.gitignore
@@ -2,6 +2,7 @@
test_core
test_cpu
test_cpuset
+test_dmem
test_freezer
test_hugetlb_memcg
test_kill
diff --git a/tools/testing/selftests/cgroup/Makefile b/tools/testing/selftests/cgroup/Makefile
index a08661320a07..d01e8eb1ead2 100644
--- a/tools/testing/selftests/cgroup/Makefile
+++ b/tools/testing/selftests/cgroup/Makefile
@@ -11,6 +11,7 @@ TEST_GEN_MODS_DIR := test_modules
TEST_GEN_PROGS = test_core
TEST_GEN_PROGS += test_cpu
TEST_GEN_PROGS += test_cpuset
+TEST_GEN_PROGS += test_dmem
TEST_GEN_PROGS += test_freezer
TEST_GEN_PROGS += test_hugetlb_memcg
TEST_GEN_PROGS += test_kill
@@ -27,6 +28,7 @@ include lib/libcgroup.mk
$(OUTPUT)/test_core: $(LIBCGROUP_O)
$(OUTPUT)/test_cpu: $(LIBCGROUP_O)
$(OUTPUT)/test_cpuset: $(LIBCGROUP_O)
+$(OUTPUT)/test_dmem: $(LIBCGROUP_O)
$(OUTPUT)/test_freezer: $(LIBCGROUP_O)
$(OUTPUT)/test_hugetlb_memcg: $(LIBCGROUP_O)
$(OUTPUT)/test_kill: $(LIBCGROUP_O)
diff --git a/tools/testing/selftests/cgroup/test_dmem.c b/tools/testing/selftests/cgroup/test_dmem.c
new file mode 100644
index 000000000000..3793664678b9
--- /dev/null
+++ b/tools/testing/selftests/cgroup/test_dmem.c
@@ -0,0 +1,314 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Test the dmem (device memory) cgroup controller.
+ *
+ * Depends on the dmem_selftest helper module
+ * (tools/testing/selftests/cgroup/test_modules/dmem_selftest.c).
+ */
+
+#define _GNU_SOURCE
+
+#include <linux/limits.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <unistd.h>
+
+#include "kselftest.h"
+#include "cgroup_util.h"
+
+#define DM_SELFTEST_REGION "dmem_selftest"
+#define DM_SELFTEST_ALLOC "/sys/module/dmem_selftest/parameters/alloc"
+#define DM_SELFTEST_FREE "/sys/module/dmem_selftest/parameters/free"
+
+static long dmem_read_limit(const char *cgroup, const char *ctrl)
+{
+ return cg_read_key_long(cgroup, ctrl, DM_SELFTEST_REGION " ");
+}
+
+static int dmem_write_limit(const char *cgroup, const char *ctrl, long val)
+{
+ char wr[64];
+
+ snprintf(wr, sizeof(wr), "%s %ld", DM_SELFTEST_REGION, val);
+ return cg_write(cgroup, ctrl, wr);
+}
+
+static int dmem_selftest_alloc(long bytes)
+{
+ char wr[32];
+
+ snprintf(wr, sizeof(wr), "%ld", bytes);
+ return write_text(DM_SELFTEST_ALLOC, wr, strlen(wr));
+}
+
+static int dmem_selftest_free(void)
+{
+ char wr[] = "1";
+
+ return write_text(DM_SELFTEST_FREE, wr, strlen(wr));
+}
+
+/*
+ * First, this test creates the following hierarchy:
+ * A
+ * A/B dmem.max=8M
+ * A/B/C dmem.max=2M
+ * A/B/D dmem.max=1M
+ * A/B/E dmem.max=75K
+ * A/B/F dmem.max=25K
+ * A/B/G dmem.max=8K
+ * A/B/H dmem.max=0
+ *
+ * Then for each leaf cgroup it tries to alloc above dmem.max
+ * and expects the request to fail and dmem.current to remain
+ * unchanged.
+ *
+ * For leaves with non-zero dmem.max, it additionally allocs a
+ * smaller amount and verifies dmem.current matches the requested
+ * size exactly (the controller accounts in bytes, with no page
+ * rounding), then frees and verifies dmem.current returns
+ * to the previous value.
+ */
+static int test_dmem_max(const char *root)
+{
+ static const long leaf_max[] = {
+ MB(2), MB(1), KB(75), KB(25), KB(8), 0
+ };
+ static const long pass_sz[] = {
+ MB(1), MB(1), KB(4), KB(4), KB(4), 0
+ };
+ char *parent[2] = {NULL};
+ char *children[ARRAY_SIZE(leaf_max)] = {NULL};
+
+ _Static_assert(ARRAY_SIZE(pass_sz) == ARRAY_SIZE(leaf_max),
+ "pass_sz doesn't match leaf_max length");
+ long cur_before, cur_after;
+ int ret = KSFT_FAIL;
+ int charged = 0;
+ int in_child = 0;
+ long v;
+ int i;
+
+ parent[0] = cg_name(root, "dmem_prot_0");
+ if (!parent[0])
+ goto cleanup;
+
+ parent[1] = cg_name(parent[0], "dmem_prot_1");
+ if (!parent[1])
+ goto cleanup;
+
+ if (cg_create(parent[0]))
+ goto cleanup;
+
+ if (cg_write(parent[0], "cgroup.subtree_control", "+dmem"))
+ goto cleanup;
+
+ if (cg_create(parent[1]))
+ goto cleanup;
+
+ if (cg_write(parent[1], "cgroup.subtree_control", "+dmem"))
+ goto cleanup;
+
+ for (i = 0; i < ARRAY_SIZE(children); i++) {
+ children[i] = cg_name_indexed(parent[1], "dmem_child", i);
+ if (!children[i])
+ goto cleanup;
+ if (cg_create(children[i]))
+ goto cleanup;
+ }
+
+ if (dmem_write_limit(parent[1], "dmem.max", MB(8)))
+ goto cleanup;
+ for (i = 0; i < ARRAY_SIZE(children); i++)
+ if (dmem_write_limit(children[i], "dmem.max", leaf_max[i]))
+ goto cleanup;
+
+ v = dmem_read_limit(parent[1], "dmem.max");
+ if (v != MB(8))
+ goto cleanup;
+ for (i = 0; i < ARRAY_SIZE(children); i++) {
+ v = dmem_read_limit(children[i], "dmem.max");
+ if (v != leaf_max[i])
+ goto cleanup;
+ }
+
+ for (i = 0; i < ARRAY_SIZE(children); i++) {
+ if (cg_enter_current(children[i]))
+ goto cleanup;
+ in_child = 1;
+
+ cur_before = dmem_read_limit(children[i], "dmem.current");
+ if (cur_before < 0)
+ goto cleanup;
+
+ if (dmem_selftest_alloc(leaf_max[i] + 1) >= 0) {
+ charged = 1;
+ goto cleanup;
+ }
+
+ cur_after = dmem_read_limit(children[i], "dmem.current");
+ if (cur_after != cur_before)
+ goto cleanup;
+
+ if (pass_sz[i] > 0) {
+ if (dmem_selftest_alloc(pass_sz[i]) < 0)
+ goto cleanup;
+ charged = 1;
+
+ cur_after = dmem_read_limit(children[i], "dmem.current");
+ if (cur_after != cur_before + pass_sz[i])
+ goto cleanup;
+
+ if (dmem_selftest_free() < 0)
+ goto cleanup;
+ charged = 0;
+
+ cur_after = dmem_read_limit(children[i], "dmem.current");
+ if (cur_after != cur_before)
+ goto cleanup;
+ }
+
+ if (cg_enter_current(root))
+ goto cleanup;
+ in_child = 0;
+ }
+
+ ret = KSFT_PASS;
+
+cleanup:
+ if (charged)
+ dmem_selftest_free();
+ if (in_child)
+ cg_enter_current(root);
+ for (i = ARRAY_SIZE(children) - 1; i >= 0; i--) {
+ if (!children[i])
+ continue;
+ cg_destroy(children[i]);
+ free(children[i]);
+ }
+ for (i = ARRAY_SIZE(parent) - 1; i >= 0; i--) {
+ if (!parent[i])
+ continue;
+ cg_destroy(parent[i]);
+ free(parent[i]);
+ }
+ return ret;
+}
+
+/*
+ * Alloc non-page-aligned byte sizes and verify dmem.current matches
+ * the requested size exactly. The controller charges the byte count
+ * passed to dmem_cgroup_try_charge() with no page rounding, for this
+ * helper and for production drivers. Then free must return usage to 0.
+ */
+static int test_dmem_alloc_byte_granularity(const char *root)
+{
+ static const long sizes[] = {
+ 1, 4095, 4097, KB(75) + 1, MB(1), MB(1) + 1
+ };
+ char *cg = NULL;
+ long cur;
+ int ret = KSFT_FAIL;
+ int charged = 0;
+ int in_child = 0;
+ size_t i;
+
+ cg = cg_name(root, "dmem_dbg_byte_gran");
+ if (!cg)
+ goto cleanup;
+
+ if (cg_create(cg))
+ goto cleanup;
+
+ if (dmem_write_limit(cg, "dmem.max", MB(16)))
+ goto cleanup;
+
+ if (cg_enter_current(cg))
+ goto cleanup;
+ in_child = 1;
+
+ for (i = 0; i < ARRAY_SIZE(sizes); i++) {
+ if (dmem_selftest_alloc(sizes[i]) < 0)
+ goto cleanup;
+ charged = 1;
+
+ cur = dmem_read_limit(cg, "dmem.current");
+ if (cur != sizes[i])
+ goto cleanup;
+
+ if (dmem_selftest_free() < 0)
+ goto cleanup;
+ charged = 0;
+
+ cur = dmem_read_limit(cg, "dmem.current");
+ if (cur != 0)
+ goto cleanup;
+ }
+
+ ret = KSFT_PASS;
+
+cleanup:
+ if (charged)
+ dmem_selftest_free();
+ if (in_child)
+ cg_enter_current(root);
+ if (cg) {
+ cg_destroy(cg);
+ free(cg);
+ }
+ return ret;
+}
+
+#define T(x) { x, #x }
+struct dmem_test {
+ int (*fn)(const char *root);
+ const char *name;
+} tests[] = {
+ T(test_dmem_max),
+ T(test_dmem_alloc_byte_granularity),
+};
+#undef T
+
+int main(int argc, char **argv)
+{
+ char root[PATH_MAX];
+ struct stat st;
+ int i;
+
+ ksft_print_header();
+
+ if (cg_find_unified_root(root, sizeof(root), NULL))
+ ksft_exit_skip("cgroup v2 isn't mounted\n");
+
+ if (cg_read_strstr(root, "cgroup.controllers", "dmem"))
+ ksft_exit_skip("dmem controller isn't available (CONFIG_CGROUP_DMEM?)\n");
+
+ if (cg_read_strstr(root, "cgroup.subtree_control", "dmem"))
+ if (cg_write(root, "cgroup.subtree_control", "+dmem"))
+ ksft_exit_skip("Failed to enable dmem controller\n");
+
+ if (stat(DM_SELFTEST_ALLOC, &st) < 0)
+ ksft_exit_skip(
+ "dmem_selftest helper not loaded (insmod test_modules/dmem_selftest.ko)\n");
+
+ if (dmem_read_limit(root, "dmem.capacity") < 0)
+ ksft_exit_skip("dmem_selftest region not registered\n");
+
+ ksft_set_plan(ARRAY_SIZE(tests));
+
+ 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;
+ default:
+ ksft_test_result_fail("%s\n", tests[i].name);
+ break;
+ }
+ }
+
+ ksft_finished();
+}
--
2.55.0
^ permalink raw reply related [flat|nested] 4+ messages in thread