Skip to contentSkip to conversation

munlockall: add test case that verifies memory has been unlocked

PATCH Superseded Dennis Brendel 2 participants 5 messages Mar 5, 2024 – Mar 6, 2024

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 munlockall: add test case that verifies memory has been unlocked

Changes to v1:

- use a docparse comment
- use tabs for indentation
- report broken test and exit if any preparation/confirmation fails by using tst_brk(TBROK, ...)
- fix further violations reported by make check

I did not yet replace munlockall01.c because I am not familiar with that (legacy?) syntax and why uclinux needs special handling.

---
 .../kernel/syscalls/munlockall/munlockall02.c | 51 +++++++++++++++++++
 1 file changed, 51 insertions(+)
 create mode 100644 testcases/kernel/syscalls/munlockall/munlockall02.c
testcases/kernel/syscalls/munlockall/munlockall02.c new file +51 -0
new file mode 100644
index 000000000..06f781d86
@@ -0,0 +1,51 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright Red Hat
+ * Author: Dennis Brendel <[email protected]>
+ */
+
+/*\
+ * [Description]
+ *
+ * Verify that munlockall(2) unlocks all previously locked memory
+ */
+
+#include <sys/mman.h>
+
+#include "tst_test.h"
+
+static void verify_munlockall(void)
+{
+	size_t size = 0;
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size != 0UL)
+		tst_brk(TBROK, "Locked memory after init should be 0 but is "
+			       "%ld", size);
+
+	if (mlockall(MCL_CURRENT | MCL_FUTURE) != 0)
+		tst_brk(TBROK, "Could not lock memory using mlockall()");
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size == 0UL)
+		tst_brk(TBROK, "Locked memory after mlockall() should be "
+			       "greater than 0, but is %ld", size);
+
+	if (munlockall() != 0)
+		tst_brk(TBROK, "Could not unlock memory using munlockall()");
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size != 0UL) {
+		tst_res(TFAIL, "Locked memory after munlockall() should be 0 "
+			       "but is %ld", size);
+	} else {
+		tst_res(TPASS, "Test passed");
+	}
+}
+
+static struct tst_test test = {
+	.test_all = verify_munlockall,
+};
Signature · 6 lines
2.44.0



-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
Reviewed-by on v2 1/1
#1 lore raw

Hi Dennis,

Generally LGTM, thanks!

you should also add test into .gitignore and to some file in runtest/ (in this case into runtest/syscalls), if we don't simply replace your test with munlockall01.c as Cyril suggested (I would also vote for replacing it).

FYI you can have look for what we check:

https://github.com/linux-test-project/ltp/wiki/Maintainer-Patch-Review-Checklist

Quoted 9 lines

Changes to v1:

- use a docparse comment
- use tabs for indentation
- report broken test and exit if any preparation/confirmation fails by using tst_brk(TBROK, ...)
- fix further violations reported by make check

I did not yet replace munlockall01.c because I am not familiar with

I guess Cyril meant just to replace munlockall01.c with what you wrote. +1 for this, there is no point to keep original munlockall01.c. And because you replace, you can delete the original copyright and can use GPL-2.0-or-later (original test was GPL v2 only).

Quoted 1 line

that (legacy?) syntax and why uclinux needs special handling.

uclinux is for nommu, we don't support it in new API (and nobody from the community standup for the support), it can be just deleted.

Quoted 26 lines
---
 .../kernel/syscalls/munlockall/munlockall02.c | 51 +++++++++++++++++++
 1 file changed, 51 insertions(+)
 create mode 100644 testcases/kernel/syscalls/munlockall/munlockall02.c
testcases/kernel/syscalls/munlockall/munlockall02.c new file +15 -0
new file mode 100644
index 000000000..06f781d86
@@ -0,0 +1,51 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright Red Hat
+ * Author: Dennis Brendel <[email protected]>
+ */
+
+/*\
+ * [Description]
+ *
+ * Verify that munlockall(2) unlocks all previously locked memory
+ */
+
+#include <sys/mman.h>
+
+#include "tst_test.h"

testcases/kernel/syscalls/munlockall/munlockall02.c line 25

Quoted 10 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +10 -0
+
+static void verify_munlockall(void)
+{
+	size_t size = 0;
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size != 0UL)
+		tst_brk(TBROK, "Locked memory after init should be 0 but is "
+			       "%ld", size);

nit: I would not split string (kernel source code also relaxed to 100 chars instead of 8O).

testcases/kernel/syscalls/munlockall/munlockall02.c line 28

Quoted 3 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +3 -0
+
+	if (mlockall(MCL_CURRENT | MCL_FUTURE) != 0)
+		tst_brk(TBROK, "Could not lock memory using mlockall()");

Maybe use tst_brk(TBROK | TERRNO, to get errno printed?

testcases/kernel/syscalls/munlockall/munlockall02.c line 34

Quoted 6 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +6 -0
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size == 0UL)
+		tst_brk(TBROK, "Locked memory after mlockall() should be "
+			       "greater than 0, but is %ld", size);

I suppose < 0 really means no memory locked, thus really safe to quit before munlockall(), right?

testcases/kernel/syscalls/munlockall/munlockall02.c line 37

Quoted 3 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +3 -0
+
+	if (munlockall() != 0)
+		tst_brk(TBROK, "Could not unlock memory using munlockall()");

Also here use TBROK | TERRNO ?

testcases/kernel/syscalls/munlockall/munlockall02.c line 43

Quoted 6 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +6 -0
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size != 0UL) {
+		tst_res(TFAIL, "Locked memory after munlockall() should be 0 "
+			       "but is %ld", size);

nit: also here I would join string to single line (still below 100 chars).

testcases/kernel/syscalls/munlockall/munlockall02.c line 45

Quoted 2 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +2 -0
+	} else {
+		tst_res(TPASS, "Test passed");

nit maybe:

		tst_res(TPASS, "Memory successfully locked and unlocked");

Kind regards,
Petr

Quoted 6 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +6 -0
+	}
+}
+
+static struct tst_test test = {
+	.test_all = verify_munlockall,
+};
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#2 lore raw

Hi Petr,

Thank you very much for your feedback! I'll send a v3 which considers your suggestions.

Dennis

Quoted 110 lines · Petr Vorel

Hi Dennis,

Generally LGTM, thanks!

you should also add test into .gitignore and to some file in runtest/ (in this case into runtest/syscalls), if we don't simply replace your test with munlockall01.c as Cyril suggested (I would also vote for replacing it).

FYI you can have look for what we check:

https://github.com/linux-test-project/ltp/wiki/Maintainer-Patch-Review-Checklist

Quoted 9 lines

Changes to v1:

- use a docparse comment
- use tabs for indentation
- report broken test and exit if any preparation/confirmation fails by using tst_brk(TBROK, ...)
- fix further violations reported by make check

I did not yet replace munlockall01.c because I am not familiar with

I guess Cyril meant just to replace munlockall01.c with what you wrote. +1 for this, there is no point to keep original munlockall01.c. And because you replace, you can delete the original copyright and can use GPL-2.0-or-later (original test was GPL v2 only).

Quoted 1 line

that (legacy?) syntax and why uclinux needs special handling.

uclinux is for nommu, we don't support it in new API (and nobody from the community standup for the support), it can be just deleted.

Quoted 26 lines
---
 .../kernel/syscalls/munlockall/munlockall02.c | 51 +++++++++++++++++++
 1 file changed, 51 insertions(+)
 create mode 100644 testcases/kernel/syscalls/munlockall/munlockall02.c
testcases/kernel/syscalls/munlockall/munlockall02.c new file +15 -0
new file mode 100644
index 000000000..06f781d86
@@ -0,0 +1,51 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright Red Hat
+ * Author: Dennis Brendel <[email protected]>
+ */
+
+/*\
+ * [Description]
+ *
+ * Verify that munlockall(2) unlocks all previously locked memory
+ */
+
+#include <sys/mman.h>
+
+#include "tst_test.h"

testcases/kernel/syscalls/munlockall/munlockall02.c line 25

Quoted 10 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +10 -0
+
+static void verify_munlockall(void)
+{
+	size_t size = 0;
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size != 0UL)
+		tst_brk(TBROK, "Locked memory after init should be 0 but is "
+			       "%ld", size);

nit: I would not split string (kernel source code also relaxed to 100 chars instead of 8O).

testcases/kernel/syscalls/munlockall/munlockall02.c line 28

Quoted 3 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +3 -0
+
+	if (mlockall(MCL_CURRENT | MCL_FUTURE) != 0)
+		tst_brk(TBROK, "Could not lock memory using mlockall()");

Maybe use tst_brk(TBROK | TERRNO, to get errno printed?

testcases/kernel/syscalls/munlockall/munlockall02.c line 34

Quoted 6 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +6 -0
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size == 0UL)
+		tst_brk(TBROK, "Locked memory after mlockall() should be "
+			       "greater than 0, but is %ld", size);

I suppose < 0 really means no memory locked, thus really safe to quit before munlockall(), right?

testcases/kernel/syscalls/munlockall/munlockall02.c line 37

Quoted 3 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +3 -0
+
+	if (munlockall() != 0)
+		tst_brk(TBROK, "Could not unlock memory using munlockall()");

Also here use TBROK | TERRNO ?

testcases/kernel/syscalls/munlockall/munlockall02.c line 43

Quoted 6 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +6 -0
+
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size != 0UL) {
+		tst_res(TFAIL, "Locked memory after munlockall() should be 0 "
+			       "but is %ld", size);

nit: also here I would join string to single line (still below 100 chars).

testcases/kernel/syscalls/munlockall/munlockall02.c line 45

Quoted 2 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +2 -0
+	} else {
+		tst_res(TPASS, "Test passed");

nit maybe:

		tst_res(TPASS, "Memory successfully locked and unlocked");

Kind regards,
Petr

Quoted 6 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +6 -0
+	}
+}
+
+static struct tst_test test = {
+	.test_all = verify_munlockall,
+};
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#3 lore raw
In reply to Petr Vorel
Quoted 7 lines · Petr Vorel

testcases/kernel/syscalls/munlockall/munlockall02.c line 34

Quoted 5 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +5 -0
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size == 0UL)
+		tst_brk(TBROK, "Locked memory after mlockall() should be "
+			       "greater than 0, but is %ld", size);

I suppose < 0 really means no memory locked, thus really safe to quit before munlockall(), right?

Forgot to reply to that.

< 0 should not happen due to unsigned type.

If the size is 0 then it is safe to quit. In general it should always be safe to exit even if there was memory locked, i.e. when the parsing of /proc/self/status is faulty or when wrong numbers are reported.

Dennis

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#4 lore raw
Quoted 9 lines · Petr Vorel
Quoted 7 lines

testcases/kernel/syscalls/munlockall/munlockall02.c line 34

Quoted 5 lines
testcases/kernel/syscalls/munlockall/munlockall02.c new file +5 -0
+	SAFE_FILE_LINES_SCANF("/proc/self/status", "VmLck: %ld", &size);
+
+	if (size == 0UL)
+		tst_brk(TBROK, "Locked memory after mlockall() should be "
+			       "greater than 0, but is %ld", size);

I suppose < 0 really means no memory locked, thus really safe to quit before munlockall(), right?

Quoted 7 lines

Forgot to reply to that.

< 0 should not happen due to unsigned type.

If the size is 0 then it is safe to quit. In general it should always be safe to exit even if there was memory locked, i.e. when the parsing of /proc/self/status is faulty or when wrong numbers are reported.

Good point, thanks!

Kind regards,
Petr

Quoted 1 line

Dennis

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp