Skip to contentSkip to conversation

chmod02.c: Block mode changes of symlinks

PATCH Superseded Wei Gao via ltp 3 participants 7 messages May 10, 2024 – Sep 17, 2024

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 chmod02.c: Block mode changes of symlinks

---
 runtest/syscalls                           |  1 +
 testcases/kernel/syscalls/chmod/.gitignore |  1 +
 testcases/kernel/syscalls/chmod/chmod02.c  | 68 ++++++++++++++++++++++
 3 files changed, 70 insertions(+)
 create mode 100644 testcases/kernel/syscalls/chmod/chmod02.c
runtest/syscalls +1 -0
index fe9ad0895..b8f1d53d5 100644
@@ -66,6 +66,7 @@ chdir04 chdir04
 
 chmod01 chmod01
 chmod01A symlink01 -T chmod01
+chmod02 chmod02
 chmod03 chmod03
 chmod05 chmod05
 chmod06 chmod06
testcases/kernel/syscalls/chmod/.gitignore +1 -0
index 27ddfce16..9cc923e69 100644
@@ -1,4 +1,5 @@
 /chmod01
+/chmod02
 /chmod03
 /chmod05
 /chmod06
testcases/kernel/syscalls/chmod/chmod02.c new file +68 -0

68 lines not on this page · raw message

Signature · 5 lines
2.35.3


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

Hi Wei,

...

Quoted 1 line
 testcases/kernel/syscalls/chmod/chmod02.c  | 68 ++++++++++++++++++++++

nit: original chmod02.c was removed in eab36ea01, now you are using this filename again. I don't remember if there is any consensus about reusing test name (maybe it's ok), but I would prefer to create new file (i.e. chmod08.c), because reusing names can lead to confusion (people may not notice when looking into git history that these 2 tests are completely unrelated. If nothing, different GPL version can be taken by mistake (e.g. original chmod02.c used GPL 2 only license).

...

Quoted 5 lines

+/*\
+ * [Description]
+ *
+ * Test for kernel commit 5d1f903f75a80daa4dfb3d84e114ec8ecbf29956
+ * "attr: block mode changes of symlinks".

nit: 5d1f903f75a8 ("attr: block mode changes of symlinks")

Quoted 1 line

+ *

nit: please remove this blank line above.

Quoted 25 lines

+ */

testcases/kernel/syscalls/chmod/chmod02.c new file +24 -0
+
+#include "lapi/fcntl.h"
+#include "tst_test.h"
+
+#define MODE 0644
+#define TESTFILE "testfile"
+#define TESTFILE_SYMLINK "testfile_symlink"
+
+static void run(void)
+{
+	struct stat stat_file, stat_sym;
+	int mode = 0;
+	char fd_path[100];
+
+	int fd = SAFE_OPEN(TESTFILE_SYMLINK, O_PATH | O_NOFOLLOW);
+
+	sprintf(fd_path, "/proc/self/fd/%d", fd);
+
+	TST_EXP_FAIL(chmod(fd_path, mode), ENOTSUP, "chmod(%s, %04o)",
+			TESTFILE_SYMLINK, mode);
+
+	SAFE_STAT(TESTFILE, &stat_file);
+	SAFE_LSTAT(TESTFILE_SYMLINK, &stat_sym);
+

testcases/kernel/syscalls/chmod/chmod02.c line 46

Quoted 10 lines
testcases/kernel/syscalls/chmod/chmod02.c new file +10 -0
+	stat_file.st_mode &= ~S_IFREG;
+	stat_sym.st_mode &= ~S_IFLNK;
+
+	if (stat_file.st_mode == (unsigned int)mode) {
+		tst_res(TFAIL, "stat(%s) mode=%04o",
+				TESTFILE, stat_file.st_mode);
+	} else {
+		tst_res(TPASS, "stat(%s) mode=%04o",
+				TESTFILE, stat_file.st_mode);
+	}

Maybe using TST_EXP_EXPR() to not repeat yourself?

testcases/kernel/syscalls/chmod/chmod02.c line 54

Quoted 8 lines
testcases/kernel/syscalls/chmod/chmod02.c new file +8 -0
+
+	if (stat_sym.st_mode == (unsigned int)mode) {
+		tst_res(TFAIL, "stat(%s) mode=%04o",
+				TESTFILE_SYMLINK, stat_sym.st_mode);
+	} else {
+		tst_res(TPASS, "stat(%s) mode=%04o",
+				TESTFILE_SYMLINK, stat_sym.st_mode);
+	}

And here as well?

Missing SAFE_CLOSE(fd); causes EMFILE (Too many open files) failure on high -i (e.g. -i 1100).

testcases/kernel/syscalls/chmod/chmod02.c line 67

Quoted 13 lines
testcases/kernel/syscalls/chmod/chmod02.c new file +13 -0
+}
+
+static void setup(void)
+{
+	SAFE_TOUCH(TESTFILE, MODE, NULL);
+	SAFE_SYMLINK(TESTFILE, TESTFILE_SYMLINK);
+}
+
+static struct tst_test test = {
+	.setup = setup,
+	.test_all = run,
+	.needs_tmpdir = 1,
+	.min_kver = "6.6",

Looking into kernel commit [1] it's in VFS therefore it should be safe to test it on single filesystem which is in TMPDIR (i.e. not using .all_filesystems).

But "If this causes any regressions" and "It's a simple patch that can be easily reverted." suggests that we should think twice when not running it with .all_filesystems (if used, exfat and vfat fails with EPERM, but it works on NTFS).

Kind regards,
Petr

[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5d1f903f75a80daa4dfb3d84e114ec8ecbf29956

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

PATCH v2 1/1 chmod08.c: Block mode changes of symlinks

In reply to Wei Gao via ltp
---
 runtest/syscalls                           |  1 +
 testcases/kernel/syscalls/chmod/.gitignore |  1 +
 testcases/kernel/syscalls/chmod/chmod08.c  | 67 ++++++++++++++++++++++
 3 files changed, 69 insertions(+)
 create mode 100644 testcases/kernel/syscalls/chmod/chmod08.c
runtest/syscalls +1 -0
index fe9ad0895..5754d8195 100644
@@ -70,6 +70,7 @@ chmod03 chmod03
 chmod05 chmod05
 chmod06 chmod06
 chmod07 chmod07
+chmod08 chmod08
 
 chown01 chown01
 chown01_16 chown01_16
testcases/kernel/syscalls/chmod/.gitignore +1 -0
index 27ddfce16..f295f4dcb 100644
@@ -3,3 +3,4 @@
 /chmod05
 /chmod06
 /chmod07
+/chmod08
testcases/kernel/syscalls/chmod/chmod08.c new file +67 -0

67 lines not on this page · raw message

Signature · 5 lines
2.35.3


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#3 lore raw

Re: [LTP] [PATCH v2] chmod08.c: Block mode changes of symlinks

Hi Wei,

there's already a patch doing that. I wrote it long time ago, but it's pending for review. Please take a look if you are interested:
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

Andrea

Quoted 102 lines · Wei Gao via ltp
---
  runtest/syscalls                           |  1 +
  testcases/kernel/syscalls/chmod/.gitignore |  1 +
  testcases/kernel/syscalls/chmod/chmod08.c  | 67 ++++++++++++++++++++++
  3 files changed, 69 insertions(+)

create mode 100644 testcases/kernel/syscalls/chmod/chmod08.c

runtest/syscalls +1 -0
index fe9ad0895..5754d8195 100644
@@ -70,6 +70,7 @@ chmod03 chmod03
  chmod05 chmod05
  chmod06 chmod06
  chmod07 chmod07
+chmod08 chmod08
 
  chown01 chown01
  chown01_16 chown01_16
testcases/kernel/syscalls/chmod/.gitignore +1 -0
index 27ddfce16..f295f4dcb 100644
@@ -3,3 +3,4 @@
  /chmod05
  /chmod06
  /chmod07
+/chmod08
testcases/kernel/syscalls/chmod/chmod08.c new file +67 -0
new file mode 100644
index 000000000..368a1d635
@@ -0,0 +1,67 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (c) 2024 Wei Gao <[email protected]>
+ */
+
+/*\
+ * [Description]
+ *
+ * Test for kernel commit
+ * 5d1f903f75a8 ("attr: block mode changes of symlinks")
+ */
+
+#include "lapi/fcntl.h"
+#include "tst_test.h"
+
+#define MODE 0644
+#define TESTFILE "testfile"
+#define TESTFILE_SYMLINK "testfile_symlink"
+
+static void run(void)
+{
+	struct stat stat_file, stat_sym;
+	int mode = 0;
+	char fd_path[100];
+
+	int fd = SAFE_OPEN(TESTFILE_SYMLINK, O_PATH | O_NOFOLLOW);
+
+	sprintf(fd_path, "/proc/self/fd/%d", fd);
+
+	TST_EXP_FAIL(chmod(fd_path, mode), ENOTSUP, "chmod(%s, %04o)",
+			TESTFILE_SYMLINK, mode);
+
+	SAFE_STAT(TESTFILE, &stat_file);
+	SAFE_LSTAT(TESTFILE_SYMLINK, &stat_sym);
+
+	stat_file.st_mode &= ~S_IFREG;
+	stat_sym.st_mode &= ~S_IFLNK;
+
+	TST_EXP_EXPR(stat_file.st_mode != (unsigned int)mode,
+			"stat(%s) mode=%04o", TESTFILE, stat_file.st_mode);
+
+	TST_EXP_EXPR(stat_sym.st_mode != (unsigned int)mode,
+			"stat(%s) mode=%04o", TESTFILE, stat_sym.st_mode);
+
+	SAFE_CLOSE(fd);
+}
+
+static void setup(void)
+{
+	SAFE_TOUCH(TESTFILE, MODE, NULL);
+	SAFE_SYMLINK(TESTFILE, TESTFILE_SYMLINK);
+}
+
+static void cleanup(void)
+{
+	remove(TESTFILE);
+	remove(TESTFILE_SYMLINK);
+}
+
+static struct tst_test test = {
+	.setup = setup,
+	.cleanup = cleanup,
+	.test_all = run,
+	.min_kver = "6.6",
+	.mntpoint = "mntpoint",
+	.all_filesystems = 1
+};
Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#4 lore raw

Re: [LTP] [PATCH v2] chmod08.c: Block mode changes of symlinks

Hi!

Quoted 3 lines

there's already a patch doing that. I wrote it long time ago, but it's pending for review. Please take a look if you are interested:
https://patchwork.ozlabs.org/project/ltp/patch/[email protected]/

This is a different test though, it checks that we can't change mode for the file via the /proc/self/fd/fd* symlinks.

Signature · 5 lines
Cyril Hrubis
[email protected]

-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#5 lore raw

PATCH v3 1/1 chmod09.c: Block mode changes of symlinks

In reply to Wei Gao via ltp
---
 runtest/syscalls                           |  1 +
 testcases/kernel/syscalls/chmod/.gitignore |  1 +
 testcases/kernel/syscalls/chmod/chmod09.c  | 67 ++++++++++++++++++++++
 3 files changed, 69 insertions(+)
 create mode 100644 testcases/kernel/syscalls/chmod/chmod09.c
runtest/syscalls +1 -0
index fe9ad0895..c8a7b3ecb 100644
@@ -70,6 +70,7 @@ chmod03 chmod03
 chmod05 chmod05
 chmod06 chmod06
 chmod07 chmod07
+chmod09 chmod09
 
 chown01 chown01
 chown01_16 chown01_16
testcases/kernel/syscalls/chmod/.gitignore +1 -0
index 27ddfce16..10249b089 100644
@@ -3,3 +3,4 @@
 /chmod05
 /chmod06
 /chmod07
+/chmod09
testcases/kernel/syscalls/chmod/chmod09.c new file +67 -0

67 lines not on this page · raw message

Signature · 5 lines
2.35.3


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#6 lore raw

Re: [LTP] [PATCH v3] chmod09.c: Block mode changes of symlinks

Hi!

Quoted 5 lines

+/*\
+ * [Description]
+ *
+ * Test for kernel commit
+ * 5d1f903f75a8 ("attr: block mode changes of symlinks")

I've added this to the tst_test tags as well and pushed, thanks.

Signature · 5 lines
Cyril Hrubis
[email protected]

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