Message-ID: <7b40fc4afa41e382d72f556399ed5e0808b969b5.1681430907.git.ackerleytng@google.com>
Date:   Fri, 14 Apr 2023 00:11:55 +0000
From:   Ackerley Tng <ackerleytng@...gle.com>
To:     kvm@...r.kernel.org, linux-api@...r.kernel.org,
        linux-arch@...r.kernel.org, linux-doc@...r.kernel.org,
        linux-fsdevel@...r.kernel.org, linux-kernel@...r.kernel.org,
        linux-mm@...ck.org, qemu-devel@...gnu.org
Cc:     aarcange@...hat.com, ak@...ux.intel.com, akpm@...ux-foundation.org,
        arnd@...db.de, bfields@...ldses.org, bp@...en8.de,
        chao.p.peng@...ux.intel.com, corbet@....net, dave.hansen@...el.com,
        david@...hat.com, ddutile@...hat.com, dhildenb@...hat.com,
        hpa@...or.com, hughd@...gle.com, jlayton@...nel.org,
        jmattson@...gle.com, joro@...tes.org, jun.nakajima@...el.com,
        kirill.shutemov@...ux.intel.com, linmiaohe@...wei.com,
        luto@...nel.org, mail@...iej.szmigiero.name, mhocko@...e.com,
        michael.roth@....com, mingo@...hat.com, naoya.horiguchi@....com,
        pbonzini@...hat.com, qperret@...gle.com, rppt@...nel.org,
        seanjc@...gle.com, shuah@...nel.org, steven.price@....com,
        tabba@...gle.com, tglx@...utronix.de, vannapurve@...gle.com,
        vbabka@...e.cz, vkuznets@...hat.com, wanpengli@...cent.com,
        wei.w.wang@...el.com, x86@...nel.org, yu.c.zhang@...ux.intel.com,
        muchun.song@...ux.dev, feng.tang@...el.com, brgerst@...il.com,
        rdunlap@...radead.org, masahiroy@...nel.org,
        mailhol.vincent@...adoo.fr, Ackerley Tng <ackerleytng@...gle.com>
Subject: [RFC PATCH 6/6] selftests: mm: Add selftest for memfd_restricted_bind()

This selftest uses memfd_restricted_bind() to set the mempolicy for a
restrictedmem file, and then checks that pages were indeed allocated
according to that policy.

Because restrictedmem pages are never mapped into userspace memory,
the usual ways of checking which NUMA node the page was allocated
on (e.g. /proc/pid/numa_maps) cannot be used.

This selftest adds a small kernel module that overloads the ioctl
syscall on /proc/restrictedmem to request a restrictedmem page and get
the node it was allocated on. The page is freed within the ioctl handler.

Signed-off-by: Ackerley Tng <ackerleytng@...gle.com>
---
 tools/testing/selftests/mm/.gitignore         |   1 +
 tools/testing/selftests/mm/Makefile           |   8 +
 .../selftests/mm/memfd_restricted_bind.c      | 139 ++++++++++++++++++
 .../mm/restrictedmem_testmod/Makefile         |  21 +++
 .../restrictedmem_testmod.c                   |  89 +++++++++++
 tools/testing/selftests/mm/run_vmtests.sh     |   6 +
 6 files changed, 264 insertions(+)
 create mode 100644 tools/testing/selftests/mm/memfd_restricted_bind.c
 create mode 100644 tools/testing/selftests/mm/restrictedmem_testmod/Makefile
 create mode 100644 tools/testing/selftests/mm/restrictedmem_testmod/restrictedmem_testmod.c

diff --git a/tools/testing/selftests/mm/.gitignore b/tools/testing/selftests/mm/.gitignore
index fb6e4233374d..10c5701b9645 100644
--- a/tools/testing/selftests/mm/.gitignore
+++ b/tools/testing/selftests/mm/.gitignore
@@ -31,6 +31,7 @@ map_fixed_noreplace
 write_to_hugetlbfs
 hmm-tests
 memfd_restricted
+memfd_restricted_bind
 memfd_secret
 soft-dirty
 split_huge_page_test
diff --git a/tools/testing/selftests/mm/Makefile b/tools/testing/selftests/mm/Makefile
index 5ec338ea1fed..4a6cf922db45 100644
--- a/tools/testing/selftests/mm/Makefile
+++ b/tools/testing/selftests/mm/Makefile
@@ -46,6 +46,8 @@ TEST_GEN_FILES += map_fixed_noreplace
 TEST_GEN_FILES += map_hugetlb
 TEST_GEN_FILES += map_populate
 TEST_GEN_FILES += memfd_restricted
+TEST_GEN_FILES += memfd_restricted_bind
+TEST_GEN_FILES += restrictedmem_testmod.ko
 TEST_GEN_FILES += memfd_secret
 TEST_GEN_FILES += migration
 TEST_GEN_FILES += mlock-random-test
@@ -171,6 +173,12 @@ $(OUTPUT)/ksm_tests: LDLIBS += -lnuma
 
 $(OUTPUT)/migration: LDLIBS += -lnuma
 
+$(OUTPUT)/memfd_restricted_bind: LDLIBS += -lnuma
+$(OUTPUT)/restrictedmem_testmod.ko: $(wildcard restrictedmem_testmod/Makefile restrictedmem_testmod/*.[ch])
+	$(call msg,MOD,,$@)
+	$(Q)$(MAKE) -C restrictedmem_testmod
+	$(Q)cp restrictedmem_testmod/restrictedmem_testmod.ko $@
+
 local_config.mk local_config.h: check_config.sh
 	/bin/sh ./check_config.sh $(CC)
 
diff --git a/tools/testing/selftests/mm/memfd_restricted_bind.c b/tools/testing/selftests/mm/memfd_restricted_bind.c
new file mode 100644
index 000000000000..64aa44c72d09
--- /dev/null
+++ b/tools/testing/selftests/mm/memfd_restricted_bind.c
@@ -0,0 +1,139 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include <fcntl.h>
+#include <linux/mempolicy.h>
+#include <numa.h>
+#include <sys/ioctl.h>
+#include <sys/stat.h>
+#include <sys/syscall.h>
+#include <unistd.h>
+
+#include "../kselftest_harness.h"
+
+int memfd_restricted(int flags, int fd)
+{
+	return syscall(__NR_memfd_restricted, flags, fd);
+}
+
+int memfd_restricted_bind(
+	int fd, loff_t offset, unsigned long len, unsigned long mode,
+	const unsigned long *nmask, unsigned long maxnode, unsigned int flags)
+{
+	struct file_range range = {
+		.offset = offset,
+		.len = len,
+	};
+
+	return syscall(__NR_memfd_restricted_bind, fd, &range, mode, nmask, maxnode, flags);
+}
+
+int memfd_restricted_bind_node(
+	int fd, loff_t offset, unsigned long len,
+	unsigned long mode, int node, unsigned int flags)
+{
+	int ret;
+	struct bitmask *mask = numa_allocate_nodemask();
+
+	numa_bitmask_setbit(mask, node);
+
+	ret = memfd_restricted_bind(fd, offset, len, mode, mask->maskp, mask->size, flags);
+
+	numa_free_nodemask(mask);
+
+	return ret;
+}
+
+/**
+ * Allocates a page in restrictedmem_fd, reads the node that the page was
+ * allocated it and returns it. Returns -1 on error.
+ */
+int read_node(int restrictedmem_fd, unsigned long offset)
+{
+	int ret;
+	int fd;
+
+	fd = open("/proc/restrictedmem", O_RDWR);
+	if (!fd)
+		return -ENOTSUP;
+
+	ret = ioctl(fd, restrictedmem_fd, offset);
+
+	close(fd);
+
+	return ret;
+}
+
+bool restrictedmem_testmod_loaded(void)
+{
+	struct stat buf;
+
+	return stat("/proc/restrictedmem", &buf) == 0;
+}
+
+FIXTURE(restrictedmem_file)
+{
+	int fd;
+	size_t page_size;
+};
+
+FIXTURE_SETUP(restrictedmem_file)
+{
+	int fd;
+	int ret;
+	struct stat stat;
+
+	fd = memfd_restricted(0, -1);
+	ASSERT_GT(fd, 0);
+
+#define RESTRICTEDMEM_TEST_NPAGES 16
+	ret = ftruncate(fd, getpagesize() * RESTRICTEDMEM_TEST_NPAGES);
+	ASSERT_EQ(ret, 0);
+
+	ret = fstat(fd, &stat);
+	ASSERT_EQ(ret, 0);
+
+	self->fd = fd;
+	self->page_size = stat.st_blksize;
+};
+
+FIXTURE_TEARDOWN(restrictedmem_file)
+{
+	int ret;
+
+	ret = close(self->fd);
+	EXPECT_EQ(ret, 0);
+}
+
+#define ASSERT_REQUIREMENTS()					\
+	do {							\
+		struct bitmask *mask = numa_get_membind();	\
+		ASSERT_GT(numa_num_configured_nodes(), 1);	\
+		ASSERT_TRUE(numa_bitmask_isbitset(mask, 0));	\
+		ASSERT_TRUE(numa_bitmask_isbitset(mask, 1));	\
+		numa_bitmask_free(mask);			\
+		ASSERT_TRUE(restrictedmem_testmod_loaded());	\
+	} while (0)
+
+TEST_F(restrictedmem_file, memfd_restricted_bind_works_as_expected)
+{
+	int ret;
+	int node;
+	int i;
+	int node_bindings[] = { 1, 0, 1, 0, 1, 1, 0, 1 };
+
+	ASSERT_REQUIREMENTS();
+
+	for (i = 0; i < ARRAY_SIZE(node_bindings); i++) {
+		ret = memfd_restricted_bind_node(
+			self->fd, i * self->page_size, self->page_size,
+			MPOL_BIND, node_bindings[i], 0);
+		ASSERT_EQ(ret, 0);
+	}
+
+	for (i = 0; i < ARRAY_SIZE(node_bindings); i++) {
+		node = read_node(self->fd, i * self->page_size);
+		ASSERT_EQ(node, node_bindings[i]);
+	}
+}
+
+TEST_HARNESS_MAIN
diff --git a/tools/testing/selftests/mm/restrictedmem_testmod/Makefile b/tools/testing/selftests/mm/restrictedmem_testmod/Makefile
new file mode 100644
index 000000000000..11b1d5d15e3c
--- /dev/null
+++ b/tools/testing/selftests/mm/restrictedmem_testmod/Makefile
@@ -0,0 +1,21 @@
+# SPDX-License-Identifier: GPL-2.0-only
+
+RESTRICTEDMEM_TESTMOD_DIR := $(realpath $(dir $(abspath $(lastword $(MAKEFILE_LIST)))))
+KDIR ?= $(abspath $(RESTRICTEDMEM_TESTMOD_DIR)/../../../../..)
+
+ifeq ($(V),1)
+Q =
+else
+Q = @
+endif
+
+MODULES = restrictedmem_testmod.ko
+
+obj-m += restrictedmem_testmod.o
+CFLAGS_restrictedmem_testmod.o = -I$(src)
+
+all:
+	+$(Q)make -C $(KDIR) M=$(RESTRICTEDMEM_TESTMOD_DIR) modules
+
+clean:
+	+$(Q)make -C $(KDIR) M=$(RESTRICTEDMEM_TESTMOD_DIR) clean
diff --git a/tools/testing/selftests/mm/restrictedmem_testmod/restrictedmem_testmod.c b/tools/testing/selftests/mm/restrictedmem_testmod/restrictedmem_testmod.c
new file mode 100644
index 000000000000..d35f55d26408
--- /dev/null
+++ b/tools/testing/selftests/mm/restrictedmem_testmod/restrictedmem_testmod.c
@@ -0,0 +1,89 @@
+// SPDX-License-Identifier: GPL-2.0-only
+
+#include "linux/printk.h"
+#include "linux/types.h"
+#include <linux/file.h>
+#include <linux/fs.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/module.h>
+#include <linux/proc_fs.h>
+#include <linux/restrictedmem.h>
+
+MODULE_DESCRIPTION("A kernel module to support restrictedmem testing");
+MODULE_AUTHOR("ackerleytng@...gle.com");
+MODULE_LICENSE("GPL");
+
+void dummy_op(struct restrictedmem_notifier *notifier, pgoff_t start, pgoff_t end)
+{
+}
+
+static const struct restrictedmem_notifier_ops dummy_notifier_ops = {
+	.invalidate_start = dummy_op,
+	.invalidate_end = dummy_op,
+	.error = dummy_op,
+};
+
+static struct restrictedmem_notifier dummy_notifier = {
+	.ops = &dummy_notifier_ops,
+};
+
+static long restrictedmem_testmod_ioctl(
+	struct file *file, unsigned int cmd, unsigned long offset)
+{
+	long ret;
+	struct fd f;
+	struct page *page;
+	pgoff_t start = offset >> PAGE_SHIFT;
+
+	f = fdget(cmd);
+	if (!f.file)
+		return -EBADF;
+
+	ret = -EINVAL;
+	if (!file_is_restrictedmem(f.file))
+		goto out;
+
+
+	ret = restrictedmem_bind(f.file, start, start + 1, &dummy_notifier, true);
+	if (ret)
+		goto out;
+
+	ret = restrictedmem_get_page(f.file, (unsigned long)start, &page, NULL);
+	if (ret)
+		goto out;
+
+	ret = page_to_nid(page);
+
+	folio_put(page_folio(page));
+
+	restrictedmem_unbind(f.file, start, start + 1, &dummy_notifier);
+
+out:
+	fdput(f);
+
+	return ret;
+}
+
+static const struct proc_ops restrictedmem_testmod_ops = {
+	.proc_ioctl = restrictedmem_testmod_ioctl,
+};
+
+static struct proc_dir_entry *restrictedmem_testmod_entry;
+
+static int restrictedmem_testmod_init(void)
+{
+	restrictedmem_testmod_entry = proc_create(
+		"restrictedmem", 0660, NULL, &restrictedmem_testmod_ops);
+
+	return 0;
+}
+
+static void restrictedmem_testmod_exit(void)
+{
+	proc_remove(restrictedmem_testmod_entry);
+}
+
+module_init(restrictedmem_testmod_init);
+module_exit(restrictedmem_testmod_exit);
diff --git a/tools/testing/selftests/mm/run_vmtests.sh b/tools/testing/selftests/mm/run_vmtests.sh
index 53de84e3ec2c..bdc853d6afe4 100644
--- a/tools/testing/selftests/mm/run_vmtests.sh
+++ b/tools/testing/selftests/mm/run_vmtests.sh
@@ -40,6 +40,8 @@ separated by spaces:
 	test memadvise(2) MADV_POPULATE_{READ,WRITE} options
 - memfd_restricted_
 	test memfd_restricted(2)
+- memfd_restricted_bind
+	test memfd_restricted_bind(2)
 - memfd_secret
 	test memfd_secret(2)
 - process_mrelease
@@ -240,6 +242,10 @@ CATEGORY="madv_populate" run_test ./madv_populate
 
 CATEGORY="memfd_restricted" run_test ./memfd_restricted
 
+test_selected "memfd_restricted_bind" && insmod ./restrictedmem_testmod.ko && \
+  CATEGORY="memfd_restricted_bind" run_test ./memfd_restricted_bind && \
+  rmmod restrictedmem_testmod > /dev/null
+
 CATEGORY="memfd_secret" run_test ./memfd_secret
 
 # KSM MADV_MERGEABLE test with 10 identical pages
-- 
2.40.0.634.g4ca3ef3211-goog

Powered by blists - more mailing lists