[LTP] [PATCH v2 0/3] new test for madvise(MADV_DONTNEED)

[LTP] [PATCH v2 0/3] new test for madvise(MADV_DONTNEED)

[Zhao Gongyi via ltp]

1. verify that after a successful MADV_DONTNEED operation, it will
   result in zero-fill-on-demand pages for anonymous private mappings.
2. Test cases for madvise(2) system call, verify that MADV_DONTNEED
   cannot be applied to Huge TLB pages.
3. Verify that when MADV_DONTNEED applied to shared mappings, it will
   lead to the resident set size(RSS) of the calling process reduced
   immediately.

Changes in v2:
  - Modify the test adn split the test into two as suggestion
  - Add a new test for madvise

Zhao Gongyi (3):
  syscalls/madvise03: new test for madvise(MADV_DONTNEED)
  syscalls/madvise04: new test for madvise(MADV_DONTNEED)
  syscalls/madvise11: new test for madvise(MADV_DONTNEED)

 runtest/syscalls                              |  3 +
 testcases/kernel/syscalls/madvise/.gitignore  |  3 +
 testcases/kernel/syscalls/madvise/madvise03.c | 69 ++++++++++++++++
 testcases/kernel/syscalls/madvise/madvise04.c | 48 ++++++++++++
 testcases/kernel/syscalls/madvise/madvise11.c | 78 +++++++++++++++++++
 5 files changed, 201 insertions(+)
 create mode 100644 testcases/kernel/syscalls/madvise/madvise03.c
 create mode 100644 testcases/kernel/syscalls/madvise/madvise04.c
 create mode 100644 testcases/kernel/syscalls/madvise/madvise11.c

--
2.17.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v2 1/3] syscalls/madvise03: new test for madvise(MADV_DONTNEED)

[Zhao Gongyi via ltp]

Test cases for madvise(2) system call, verify that after a successful
MADV_DONTNEED operation, it will result in zero-fill-on-demand pages
for anonymous private mappings.

Signed-off-by: Zhao Gongyi <zhaogongyi@huawei.com>
---
 runtest/syscalls                              |  1 +
 testcases/kernel/syscalls/madvise/.gitignore  |  1 +
 testcases/kernel/syscalls/madvise/madvise03.c | 69 +++++++++++++++++++
 3 files changed, 71 insertions(+)
 create mode 100644 testcases/kernel/syscalls/madvise/madvise03.c

diff --git a/runtest/syscalls b/runtest/syscalls
index 51de0a614..c81764df4 100644
--- a/runtest/syscalls
+++ b/runtest/syscalls
@@ -940,6 +940,7 @@ mincore04 mincore04

 madvise01 madvise01
 madvise02 madvise02
+madvise03 madvise03
 madvise05 madvise05
 madvise06 madvise06
 madvise07 madvise07
diff --git a/testcases/kernel/syscalls/madvise/.gitignore b/testcases/kernel/syscalls/madvise/.gitignore
index 002d8e5d9..f4bfdfefe 100644
--- a/testcases/kernel/syscalls/madvise/.gitignore
+++ b/testcases/kernel/syscalls/madvise/.gitignore
@@ -1,5 +1,6 @@
 /madvise01
 /madvise02
+/madvise03
 /madvise05
 /madvise06
 /madvise07
diff --git a/testcases/kernel/syscalls/madvise/madvise03.c b/testcases/kernel/syscalls/madvise/madvise03.c
new file mode 100644
index 000000000..70330060e
--- /dev/null
+++ b/testcases/kernel/syscalls/madvise/madvise03.c
@@ -0,0 +1,69 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2022. All rights reserved.
+ * Author: Zhao Gongyi <zhaogongyi@huawei.com>
+ */
+
+/*\
+ * [Description]
+ *
+ * Test cases for madvise(2) system call, advise value as "MADV_MADV_DONTNEED":
+ *   After a successful MADV_DONTNEED operation, it will result in
+ *   zero-fill-on-demand pages for anonymous private mappings.
+ */
+
+#include "tst_test.h"
+
+#define MAP_SIZE (8 * 1024)
+
+static char *addr;
+
+static void run(void)
+{
+	int i;
+
+	TEST(madvise(addr, MAP_SIZE, MADV_DONTNEED));
+	if (TST_RET == -1) {
+		tst_brk(TBROK | TTERRNO, "madvise(%p, %d, 0x%x) failed",
+			addr, MAP_SIZE, MADV_DONTNEED);
+	}
+
+	for (i = 0; i < MAP_SIZE; i++) {
+		if (addr[i]) {
+			tst_res(TFAIL,
+				"There are no zero-fill-on-demand pages "
+				"for anonymous private mappings");
+			return;
+		}
+	}
+
+	if (i == MAP_SIZE) {
+		tst_res(TPASS,
+			"There are zero-fill-on-demand pages "
+			"for anonymous private mappings");
+	}
+}
+
+static void setup(void)
+{
+	addr = SAFE_MMAP(NULL, MAP_SIZE,
+			PROT_READ | PROT_WRITE,
+			MAP_PRIVATE | MAP_ANONYMOUS,
+			-1, 0);
+	memset(addr, 1, MAP_SIZE);
+}
+
+static void cleanup(void)
+{
+	if (addr)
+		SAFE_MUNMAP(addr, MAP_SIZE);
+}
+
+
+
+static struct tst_test test = {
+	.test_all = run,
+	.setup = setup,
+	.cleanup = cleanup,
+};
+
--
2.17.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v2 2/3] syscalls/madvise04: new test for madvise(MADV_DONTNEED)

[Zhao Gongyi via ltp]

Test cases for madvise(2) system call, verify that MADV_DONTNEED
cannot be applied to Huge TLB pages.

Signed-off-by: Zhao Gongyi <zhaogongyi@huawei.com>
---
 runtest/syscalls                              |  1 +
 testcases/kernel/syscalls/madvise/.gitignore  |  1 +
 testcases/kernel/syscalls/madvise/madvise04.c | 48 +++++++++++++++++++
 3 files changed, 50 insertions(+)
 create mode 100644 testcases/kernel/syscalls/madvise/madvise04.c

diff --git a/runtest/syscalls b/runtest/syscalls
index c81764df4..eb1910cec 100644
--- a/runtest/syscalls
+++ b/runtest/syscalls
@@ -941,6 +941,7 @@ mincore04 mincore04
 madvise01 madvise01
 madvise02 madvise02
 madvise03 madvise03
+madvise04 madvise04
 madvise05 madvise05
 madvise06 madvise06
 madvise07 madvise07
diff --git a/testcases/kernel/syscalls/madvise/.gitignore b/testcases/kernel/syscalls/madvise/.gitignore
index f4bfdfefe..db8ce47c1 100644
--- a/testcases/kernel/syscalls/madvise/.gitignore
+++ b/testcases/kernel/syscalls/madvise/.gitignore
@@ -1,6 +1,7 @@
 /madvise01
 /madvise02
 /madvise03
+/madvise04
 /madvise05
 /madvise06
 /madvise07
diff --git a/testcases/kernel/syscalls/madvise/madvise04.c b/testcases/kernel/syscalls/madvise/madvise04.c
new file mode 100644
index 000000000..d7ccaf97e
--- /dev/null
+++ b/testcases/kernel/syscalls/madvise/madvise04.c
@@ -0,0 +1,48 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2022. All rights reserved.
+ * Author: Zhao Gongyi <zhaogongyi@huawei.com>
+ */
+
+/*\
+ * [Description]
+ *
+ * Test cases for madvise(2) system call, advise value as "MADV_MADV_DONTNEED":
+ *   MADV_DONTNEED cannot be applied to Huge TLB pages.
+ */
+
+#include "tst_test.h"
+
+#define MAP_SIZE (8 * 1024)
+
+static char *addr;
+static int mapsz;
+
+static void run(void)
+{
+	TST_EXP_FAIL(madvise(addr, mapsz, MADV_DONTNEED), EINVAL);
+}
+
+static void setup(void)
+{
+	mapsz = tst_get_hugepage_size();
+	addr = SAFE_MMAP(NULL, mapsz,
+			PROT_READ | PROT_WRITE,
+			MAP_PRIVATE | MAP_ANONYMOUS | MAP_HUGETLB,
+			-1, 0);
+}
+
+static void cleanup(void)
+{
+	if (addr)
+		SAFE_MUNMAP(addr, mapsz);
+}
+
+static struct tst_test test = {
+	.test_all = run,
+	.setup = setup,
+	.cleanup = cleanup,
+	.needs_root = 1,
+	.hugepages = {1, TST_NEEDS},
+};
+
--
2.17.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

[LTP] [PATCH v2 3/3] syscalls/madvise11: new test for madvise(MADV_DONTNEED)

[Zhao Gongyi via ltp]

Test cases for madvise(2) system call, verify that when MADV_DONTNEED
applied to shared mappings, it will lead to the resident set size(RSS)
of the calling process reduced immediately.

Signed-off-by: Zhao Gongyi <zhaogongyi@huawei.com>
---
 runtest/syscalls                              |  1 +
 testcases/kernel/syscalls/madvise/.gitignore  |  1 +
 testcases/kernel/syscalls/madvise/madvise11.c | 78 +++++++++++++++++++
 3 files changed, 80 insertions(+)
 create mode 100644 testcases/kernel/syscalls/madvise/madvise11.c

diff --git a/runtest/syscalls b/runtest/syscalls
index eb1910cec..296af9f9d 100644
--- a/runtest/syscalls
+++ b/runtest/syscalls
@@ -948,6 +948,7 @@ madvise07 madvise07
 madvise08 madvise08
 madvise09 madvise09
 madvise10 madvise10
+madvise11 madvise11

 newuname01 newuname01

diff --git a/testcases/kernel/syscalls/madvise/.gitignore b/testcases/kernel/syscalls/madvise/.gitignore
index db8ce47c1..ffd8823d1 100644
--- a/testcases/kernel/syscalls/madvise/.gitignore
+++ b/testcases/kernel/syscalls/madvise/.gitignore
@@ -8,3 +8,4 @@
 /madvise08
 /madvise09
 /madvise10
+/madvise11
diff --git a/testcases/kernel/syscalls/madvise/madvise11.c b/testcases/kernel/syscalls/madvise/madvise11.c
new file mode 100644
index 000000000..77f0b91ac
--- /dev/null
+++ b/testcases/kernel/syscalls/madvise/madvise11.c
@@ -0,0 +1,78 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) Huawei Technologies Co., Ltd. 2022. All rights reserved.
+ * Author: Zhao Gongyi <zhaogongyi@huawei.com>
+ */
+
+/*\
+ * [Description]
+ *
+ * Test cases for madvise(2) system call, advise value as "MADV_MADV_DONTNEED":
+ *   When MADV_DONTNEED applied to shared mappings, it will lead to the
+ *   resident set size(RSS) of the calling process reduced immediately.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include "tst_test.h"
+
+#define MAP_SIZE (8 * 1024)
+#define BUF_SIZE 1024
+
+static FILE *fp;
+static char *addr;
+
+static void run(void)
+{
+	char cmd[BUF_SIZE];
+	char line[BUF_SIZE];
+	char vm_area_addr[128];
+
+	TEST(madvise(addr, MAP_SIZE, MADV_DONTNEED));
+	if (TST_RET == -1) {
+		tst_brk(TBROK | TTERRNO, "madvise(%p, %d, 0x%x)",
+			addr, MAP_SIZE, MADV_DONTNEED);
+	}
+
+	sprintf(vm_area_addr, "%p", addr);
+	sprintf(cmd,
+		"cat /proc/%d/smaps | grep %s -A 4 | grep Rss: | grep '0 kB'",
+		getpid(), &(vm_area_addr[2]));
+
+	fp = popen(cmd, "r");
+	if (!fp)
+		tst_brk(TBROK, "popen failed");
+
+	if (fgets(line, sizeof(line), fp) != NULL) {
+		if (strstr(line, " 0 kB"))
+			tst_res(TPASS, "RSS is released");
+		else
+			tst_res(TFAIL, "RSS is not released");
+
+	} else
+		tst_brk(TBROK, "There is no 'Rss:' in smaps?");
+}
+
+static void setup(void)
+{
+	addr = SAFE_MMAP(NULL, MAP_SIZE,
+			PROT_READ | PROT_WRITE,
+			MAP_PRIVATE | MAP_ANONYMOUS,
+			-1, 0);
+	memset(addr, 1, MAP_SIZE);
+}
+
+static void cleanup(void)
+{
+	if (addr)
+		SAFE_MUNMAP(addr, MAP_SIZE);
+	if (fp)
+		pclose(fp);
+}
+
+static struct tst_test test = {
+	.test_all = run,
+	.setup = setup,
+	.cleanup = cleanup,
+};
+
--
2.17.1


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 1/3] syscalls/madvise03: new test for madvise(MADV_DONTNEED)

[Cyril Hrubis]

Hi!
Pushed with two minor changes, thanks.

* Moved the memset() to the start of the run() function, so that the
  pages are populated again in case that we run with the -i 10 option

* Rewrote the documentation comment so that it renders nicely in the
  exported documentation

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 2/3] syscalls/madvise04: new test for madvise(MADV_DONTNEED)

[Cyril Hrubis]

Hi!
The MADV_DONTNEED has been implemented for hugepages since:

commit 90e7e7f5ef3f712da992d505aee2d921797c3f96
Author: Mike Kravetz <mike.kravetz@oracle.com>
Date:   Thu Mar 24 18:13:18 2022 -0700

    mm: enable MADV_DONTNEED for hugetlb mappings

So for newer kernels this actually works fine. I guess that we should
change the test to expect the call to pass for v5.18 and newer.

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 3/3] syscalls/madvise11: new test for madvise(MADV_DONTNEED)

[Cyril Hrubis]

Hi!
> +static void run(void)
> +{
> +	char cmd[BUF_SIZE];
> +	char line[BUF_SIZE];
> +	char vm_area_addr[128];
> +
> +	TEST(madvise(addr, MAP_SIZE, MADV_DONTNEED));
> +	if (TST_RET == -1) {
> +		tst_brk(TBROK | TTERRNO, "madvise(%p, %d, 0x%x)",
> +			addr, MAP_SIZE, MADV_DONTNEED);
> +	}
> +
> +	sprintf(vm_area_addr, "%p", addr);
> +	sprintf(cmd,
> +		"cat /proc/%d/smaps | grep %s -A 4 | grep Rss: | grep '0 kB'",
> +		getpid(), &(vm_area_addr[2]));

This is way too ugly and may break easily too.

If we are going to parse the file we should do it properly in C instead.
Why can't we just read the file line by line until we find the right
address at the start of the line and once we do look for the Rss?

> +	fp = popen(cmd, "r");
> +	if (!fp)
> +		tst_brk(TBROK, "popen failed");
> +
> +	if (fgets(line, sizeof(line), fp) != NULL) {
> +		if (strstr(line, " 0 kB"))
> +			tst_res(TPASS, "RSS is released");
> +		else
> +			tst_res(TFAIL, "RSS is not released");
> +
> +	} else
> +		tst_brk(TBROK, "There is no 'Rss:' in smaps?");
> +}
> +
> +static void setup(void)
> +{
> +	addr = SAFE_MMAP(NULL, MAP_SIZE,
> +			PROT_READ | PROT_WRITE,
> +			MAP_PRIVATE | MAP_ANONYMOUS,
> +			-1, 0);
> +	memset(addr, 1, MAP_SIZE);
> +}
> +
> +static void cleanup(void)
> +{
> +	if (addr)
> +		SAFE_MUNMAP(addr, MAP_SIZE);
> +	if (fp)
> +		pclose(fp);
> +}
> +
> +static struct tst_test test = {
> +	.test_all = run,
> +	.setup = setup,
> +	.cleanup = cleanup,
> +};
> +
> --
> 2.17.1
> 

-- 
Cyril Hrubis
chrubis@suse.cz

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp

Re: [LTP] [PATCH v2 3/3] syscalls/madvise11: new test for madvise(MADV_DONTNEED)

[zhaogongyi via ltp]

Hi,

> 
> Hi!
> > +static void run(void)
> > +{
> > +	char cmd[BUF_SIZE];
> > +	char line[BUF_SIZE];
> > +	char vm_area_addr[128];
> > +
> > +	TEST(madvise(addr, MAP_SIZE, MADV_DONTNEED));
> > +	if (TST_RET == -1) {
> > +		tst_brk(TBROK | TTERRNO, "madvise(%p, %d, 0x%x)",
> > +			addr, MAP_SIZE, MADV_DONTNEED);
> > +	}
> > +
> > +	sprintf(vm_area_addr, "%p", addr);
> > +	sprintf(cmd,
> > +		"cat /proc/%d/smaps | grep %s -A 4 | grep Rss: | grep '0 kB'",
> > +		getpid(), &(vm_area_addr[2]));
> 
> This is way too ugly and may break easily too.
> 
> If we are going to parse the file we should do it properly in C instead.
> Why can't we just read the file line by line until we find the right address
> at the start of the line and once we do look for the Rss?


I have do it with C at first, but it seems not intuitive. I will change back on next version of the patch.

Regards,
Gongyi 


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp