* [LTP] [PATCH v2] mknod02.c: Simplify and convert to new LTP API
@ 2023-03-15 15:56 Avinesh Kumar
2023-05-16 11:58 ` Petr Vorel
0 siblings, 1 reply; 4+ messages in thread
From: Avinesh Kumar @ 2023-03-15 15:56 UTC (permalink / raw)
To: ltp
Simply test when parent directory does not have set-group-ID bit set,
new node gets GID from effective GID of the process and does not inherit
the group ownership from its parent directory.
Signed-off-by: Avinesh Kumar <akumar@suse.de>
---
testcases/kernel/syscalls/mknod/mknod02.c | 316 +++-------------------
1 file changed, 36 insertions(+), 280 deletions(-)
diff --git a/testcases/kernel/syscalls/mknod/mknod02.c b/testcases/kernel/syscalls/mknod/mknod02.c
index 594473e24..b1885fed1 100644
--- a/testcases/kernel/syscalls/mknod/mknod02.c
+++ b/testcases/kernel/syscalls/mknod/mknod02.c
@@ -1,301 +1,57 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
/*
- *
* Copyright (c) International Business Machines Corp., 2001
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
- * the GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, write to the Free Software
- * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ * 07/2001 Ported by Wayne Boyer
+ * Copyright (c) 2023 SUSE LLC Avinesh Kumar <avinesh.kumar@suse.com>
*/
-/*
- * Test Name: mknod02
- *
- * Test Description:
- * Verify that mknod(2) succeeds when used to create a filesystem
- * node with set group-ID bit set on a directory without set group-ID bit set.
- * The node created should have set group-ID bit set and its gid should be
- * equal to that of its parent directory.
- *
- * Expected Result:
- * mknod() should return value 0 on success and node created should have
- * set group-ID bit set, its gid should be equal to that of its parent
- * directory.
- *
- * Algorithm:
- * Setup:
- * Setup signal handling.
- * Create temporary directory.
- * Pause for SIGUSR1 if option specified.
- *
- * Test:
- * Loop if the proper options are given.
- * Execute system call
- * Check return code, if system call failed (return=-1)
- * Log the errno and Issue a FAIL message.
- * Otherwise,
- * Verify the Functionality of system call
- * if successful,
- * Issue Functionality-Pass message.
- * Otherwise,
- * Issue Functionality-Fail message.
- * Cleanup:
- * Print errno log and/or timing stats if options given
- * Delete the temporary directory created.
- *
- * Usage: <for command-line>
- * mknod02 [-c n] [-f] [-i n] [-I x] [-P x] [-t]
- * where, -c n : Run n copies concurrently.
- * -f : Turn off functionality Testing.
- * -i n : Execute test n times.
- * -I x : Execute test for x seconds.
- * -P x : Pause for x seconds between iterations.
- * -t : Turn on syscall timing.
- *
- * HISTORY
- * 07/2001 Ported by Wayne Boyer
- *
- * RESTRICTIONS:
- * This test should be run by 'super-user' (root) only.
+/*\
+ * [Description]
*
+ * Verify that if mknod(2) creates a filesystem node in a directory which
+ * does not have the set-group-ID bit set, new node will not inherit the
+ * group ownership from its parent directory and its group ID will be the
+ * effective group ID of the process.
*/
-#include <stdio.h>
-#include <stdlib.h>
-#include <unistd.h>
-#include <errno.h>
-#include <string.h>
-#include <signal.h>
#include <pwd.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-
-#include "test.h"
-#include "safe_macros.h"
+#include "tst_test.h"
-#define LTPUSER "nobody"
-#define MODE_RWX S_IFIFO | S_IRWXU | S_IRWXG | S_IRWXO
-#define MODE_SGID S_IFIFO | S_ISGID | S_IRWXU | S_IRWXG | S_IRWXO
-#define DIR_TEMP "testdir_2"
-#define TNODE "tnode_%d"
+#define MODE_DIR 0777
+#define MODE1 0010777
+#define MODE_SGID 02000
-struct stat buf; /* struct. to hold stat(2) o/p contents */
-struct passwd *user1; /* struct. to hold getpwnam(3) o/p contents */
+#define TEMP_DIR "testdir"
+#define TEMP_NODE "testnode"
-char *TCID = "mknod02";
-int TST_TOTAL = 1;
-char node_name[PATH_MAX]; /* buffer to hold node name created */
+static struct stat buf;
+static struct passwd *user_nobody;
+static gid_t gid_nobody;
-gid_t group1_gid, group2_gid, mygid; /* user and process group id's */
-uid_t save_myuid, user1_uid; /* user and process user id's */
-pid_t mypid; /* process id */
-
-void setup(); /* setup function for the test */
-void cleanup(); /* cleanup function for the test */
-
-int main(int ac, char **av)
+static void setup(void)
{
- int lc;
- int fflag;
-
- tst_parse_opts(ac, av, NULL, NULL);
-
- setup();
-
- for (lc = 0; TEST_LOOPING(lc); lc++) {
-
- tst_count = 0;
-
- /*
- * Attempt to create a filesystem node with group-id bit set
- * on a directory without group id bit set such that,
- * the node created by mknod(2) should have group-id (sgid)
- * bit set and node's gid should be equal to that of its
- * parent directory.
- */
- TEST(mknod(node_name, MODE_SGID, 0));
-
- /* Check return code from mknod(2) */
- if (TEST_RETURN == -1) {
- tst_resm(TFAIL,
- "mknod(%s, %#o, 0) failed, errno=%d : %s",
- node_name, MODE_SGID, TEST_ERRNO,
- strerror(TEST_ERRNO));
- continue;
- }
- /* Set the functionality flag */
- fflag = 1;
-
- /* Check for node's creation */
- if (stat(node_name, &buf) < 0) {
- tst_resm(TFAIL, "stat() of %s failed, errno:%d",
- node_name, TEST_ERRNO);
-
- /* unset functionality flag */
- fflag = 0;
- }
-
- /* Verify mode permissions of node */
- if (!(buf.st_mode & S_ISGID)) {
- tst_resm(TFAIL, "%s: Incorrect modes, setgid "
- "bit not set", node_name);
- /* unset flag as functionality fails */
- fflag = 0;
- }
-
- /* Verify group ID of node */
- if (buf.st_gid != mygid) {
- tst_resm(TFAIL, "%s: Incorrect group",
- node_name);
- /* unset flag as functionality fails */
- fflag = 0;
- }
- if (fflag) {
- tst_resm(TPASS, "Functionality of mknod(%s, "
- "%#o, 0) successful",
- node_name, MODE_SGID);
- }
-
- /* Remove the node for the next go `round */
- if (unlink(node_name) == -1) {
- tst_resm(TWARN, "unlink(%s) failed, errno:%d %s",
- node_name, errno, strerror(errno));
- }
- }
+ user_nobody = SAFE_GETPWNAM("nobody");
+ gid_nobody = user_nobody->pw_gid;
- /* Change the directory back to temporary directory */
- SAFE_CHDIR(cleanup, "..");
-
- /*
- * Invoke cleanup() to delete the test directories created
- * in the setup() and exit main().
- */
- cleanup();
-
- tst_exit();
+ SAFE_MKDIR(TEMP_DIR, MODE_DIR);
+ SAFE_CHOWN(TEMP_DIR, -1, gid_nobody);
}
-/*
- * setup(void) - performs all ONE TIME setup for this test.
- * Exit the test program on receipt of unexpected signals.
- * Create a temporary directory used to hold test directories created
- * and change the directory to it.
- * Verify that pid of process executing the test is root.
- * Create a test directory on temporary directory and set the ownership
- * of test directory to ltp user and process.
- * Set the effective uid/gid of the process to that of ltp user.
- */
-void setup(void)
+static void run(void)
{
- tst_require_root();
-
- /* Capture unexpected signals */
- tst_sig(NOFORK, DEF_HANDLER, cleanup);
-
- TEST_PAUSE;
-
- /* Make a temp dir and cd to it */
- tst_tmpdir();
-
- /* fix permissions on the tmpdir */
- if (chmod(".", 0711) != 0) {
- tst_brkm(TBROK, cleanup, "chmod() failed");
- }
-
- /* Save the real user id of the current test process */
- save_myuid = getuid();
-
- /* Save the process id of the current test process */
- mypid = getpid();
-
- /* Get the node name to be created in the test */
- sprintf(node_name, TNODE, mypid);
-
- /* Get the uid/gid of ltpuser */
- if ((user1 = getpwnam(LTPUSER)) == NULL) {
- tst_brkm(TBROK | TERRNO, cleanup,
- "Couldn't determine if %s was in /etc/passwd",
- LTPUSER);
- }
- user1_uid = user1->pw_uid;
- group1_gid = user1->pw_gid;
-
- /* Get the effective group id of the test process */
- group2_gid = getegid();
+ SAFE_CHDIR(TEMP_DIR);
+ TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0), "mknod(%s, %o, 0)", TEMP_NODE, MODE1);
- /*
- * Create a test directory under temporary directory with the
- * specified mode permissions, with uid/gid set to that of guest
- * user and the test process.
- */
- SAFE_MKDIR(cleanup, DIR_TEMP, MODE_RWX);
- SAFE_CHOWN(cleanup, DIR_TEMP, user1_uid, group2_gid);
+ SAFE_STAT(TEMP_NODE, &buf);
+ TST_EXP_EQ_LI(buf.st_gid, 0);
- /*
- * Verify that test directory created with expected permission modes
- * and ownerships.
- */
- SAFE_STAT(cleanup, DIR_TEMP, &buf);
-
- /* Verify modes of test directory */
- if (buf.st_mode & S_ISGID) {
- tst_brkm(TBROK, cleanup,
- "%s: Incorrect modes, setgid bit set", DIR_TEMP);
- }
-
- /* Verify group ID of test directory */
- if (buf.st_gid != group2_gid) {
- tst_brkm(TBROK, cleanup, "%s: Incorrect group", DIR_TEMP);
- }
-
- /*
- * Set the effective group id and user id of the test process
- * to that of guest user.
- */
- SAFE_SETGID(cleanup, group1_gid);
- if (setreuid(-1, user1_uid) < 0) {
- tst_brkm(TBROK, cleanup,
- "Unable to set process uid to that of ltp user");
- }
-
- /* Save the real group ID of the current process */
- mygid = getgid();
-
- /* Change directory to DIR_TEMP */
- SAFE_CHDIR(cleanup, DIR_TEMP);
+ SAFE_UNLINK(TEMP_NODE);
+ SAFE_CHDIR("..");
}
-/*
- * cleanup() - Performs all ONE TIME cleanup for this test at
- * completion or premature exit.
- * Print test timing stats and errno log if test executed with options.
- * Restore the real/effective user id of the process changed during
- * setup().
- * Remove temporary directory and sub-directories/files under it
- * created during setup().
- * Exit the test program with normal exit code.
- */
-void cleanup(void)
-{
-
- /*
- * Restore the effective uid of the process changed in the
- * setup().
- */
- if (setreuid(-1, save_myuid) < 0) {
- tst_brkm(TBROK, NULL,
- "resetting process real/effective uid failed");
- }
-
- tst_rmdir();
-
-}
+static struct tst_test test = {
+ .setup = setup,
+ .test_all = run,
+ .needs_root = 1,
+ .needs_tmpdir = 1
+};
--
2.39.2
--
Mailing list info: https://lists.linux.it/listinfo/ltp
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [LTP] [PATCH v2] mknod02.c: Simplify and convert to new LTP API
2023-03-15 15:56 [LTP] [PATCH v2] mknod02.c: Simplify and convert to new LTP API Avinesh Kumar
@ 2023-05-16 11:58 ` Petr Vorel
2023-05-17 15:07 ` Avinesh Kumar
2023-08-23 9:21 ` Richard Palethorpe
0 siblings, 2 replies; 4+ messages in thread
From: Petr Vorel @ 2023-05-16 11:58 UTC (permalink / raw)
To: Avinesh Kumar; +Cc: ltp
Hi Avinesh,
> Simply test when parent directory does not have set-group-ID bit set,
> new node gets GID from effective GID of the process and does not inherit
> the group ownership from its parent directory.
Reviewed-by: Petr Vorel <pvorel@suse.cz>
Few comments below.
> testcases/kernel/syscalls/mknod/mknod02.c | 316 +++-------------------
> 1 file changed, 36 insertions(+), 280 deletions(-)
...
> +/*\
> + * [Description]
> *
> + * Verify that if mknod(2) creates a filesystem node in a directory which
> + * does not have the set-group-ID bit set, new node will not inherit the
> + * group ownership from its parent directory and its group ID will be the
> + * effective group ID of the process.
@Cyril I wonder if it'd be good to test this on all_filesystems. Are we trying
to use use all_filesystems = 1 when subject of testing is using VFS or the
opposite? (kernel docs mentions "VFS system calls open(2), stat(2), read(2),
write(2), chmod(2)". It also mentions locking [2]).
BTW looking what has mknod in vfs, it's just nfsd and 9p (none of them are used
in all_filesystems):
$ git grep mknod $(git ls-files fs/|grep -i vfs)
fs/9p/vfs_inode.c: * for mknod(2).
fs/9p/vfs_inode.c: * v9fs_vfs_mknod - create a special file
fs/9p/vfs_inode.c:v9fs_vfs_mknod(struct mnt_idmap *idmap, struct inode *dir,
fs/9p/vfs_inode.c: .mknod = v9fs_vfs_mknod,
fs/9p/vfs_inode.c: .mknod = v9fs_vfs_mknod,
fs/9p/vfs_inode_dotl.c:v9fs_vfs_mknod_dotl(struct mnt_idmap *idmap, struct inode *dir,
fs/9p/vfs_inode_dotl.c: return v9fs_vfs_mknod_dotl(idmap, dir, dentry, omode, 0);
fs/9p/vfs_inode_dotl.c: * v9fs_vfs_mknod_dotl - create a special file
fs/9p/vfs_inode_dotl.c:v9fs_vfs_mknod_dotl(struct mnt_idmap *idmap, struct inode *dir,
fs/9p/vfs_inode_dotl.c: p9_debug(P9_DEBUG_VFS, "Failed to get acl values in mknod %d\n",
fs/9p/vfs_inode_dotl.c: err = p9_client_mknod_dotl(dfid, name, mode, rdev, gid, &qid);
fs/9p/vfs_inode_dotl.c: .mknod = v9fs_vfs_mknod_dotl,
fs/nfsd/vfs.c: host_err = vfs_mknod(&nop_mnt_idmap, dirp, dchild,
> */
...
> +static void run(void)
> {
...
> + SAFE_CHDIR(TEMP_DIR);
> + TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0), "mknod(%s, %o, 0)", TEMP_NODE, MODE1);
IMHO this simple form will print the same info, I suggest to use it.
TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0));
> - /*
> - * Create a test directory under temporary directory with the
> - * specified mode permissions, with uid/gid set to that of guest
> - * user and the test process.
> - */
> - SAFE_MKDIR(cleanup, DIR_TEMP, MODE_RWX);
> - SAFE_CHOWN(cleanup, DIR_TEMP, user1_uid, group2_gid);
> + SAFE_STAT(TEMP_NODE, &buf);
> + TST_EXP_EQ_LI(buf.st_gid, 0);
...
Diff for using all_filesystems (I'm not sure myself).
Kind regards,
Petr
[1] https://www.kernel.org/doc/html/next/filesystems/vfs.html
[2] https://www.kernel.org/doc/html/next/filesystems/locking.html
+++ testcases/kernel/syscalls/mknod/mknod02.c
@@ -24,6 +24,8 @@
#define TEMP_DIR "testdir"
#define TEMP_NODE "testnode"
+#define MNTPOINT "mntpoint"
+
static struct stat buf;
static struct passwd *user_nobody;
static gid_t gid_nobody;
@@ -40,18 +42,28 @@ static void setup(void)
static void run(void)
{
SAFE_CHDIR(TEMP_DIR);
- TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0), "mknod(%s, %o, 0)", TEMP_NODE, MODE1);
+ TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0));
SAFE_STAT(TEMP_NODE, &buf);
TST_EXP_EQ_LI(buf.st_gid, 0);
SAFE_UNLINK(TEMP_NODE);
SAFE_CHDIR("..");
+
+}
+
+static void cleanup(void)
+{
+ SAFE_RMDIR(TEMP_DIR);
}
static struct tst_test test = {
.setup = setup,
+ .cleanup = cleanup,
.test_all = run,
.needs_root = 1,
- .needs_tmpdir = 1
+ .needs_tmpdir = 1,
+ .mount_device = 1,
+ .mntpoint = MNTPOINT,
+ .all_filesystems = 1,
};
--
Mailing list info: https://lists.linux.it/listinfo/ltp
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [LTP] [PATCH v2] mknod02.c: Simplify and convert to new LTP API
2023-05-16 11:58 ` Petr Vorel
@ 2023-05-17 15:07 ` Avinesh Kumar
2023-08-23 9:21 ` Richard Palethorpe
1 sibling, 0 replies; 4+ messages in thread
From: Avinesh Kumar @ 2023-05-17 15:07 UTC (permalink / raw)
To: Petr Vorel; +Cc: ltp
Hi Petr,
On Tuesday, May 16, 2023 5:28:41 PM IST Petr Vorel wrote:
> Hi Avinesh,
>
> > Simply test when parent directory does not have set-group-ID bit set,
> > new node gets GID from effective GID of the process and does not inherit
> > the group ownership from its parent directory.
>
> Reviewed-by: Petr Vorel <pvorel@suse.cz>
>
> Few comments below.
>
> > testcases/kernel/syscalls/mknod/mknod02.c | 316 +++-------------------
> > 1 file changed, 36 insertions(+), 280 deletions(-)
>
> ...
>
> > +/*\
> > + * [Description]
> >
> > *
> >
> > + * Verify that if mknod(2) creates a filesystem node in a directory which
> > + * does not have the set-group-ID bit set, new node will not inherit the
> > + * group ownership from its parent directory and its group ID will be the
> > + * effective group ID of the process.
>
> @Cyril I wonder if it'd be good to test this on all_filesystems. Are we
> trying to use use all_filesystems = 1 when subject of testing is using VFS
> or the opposite? (kernel docs mentions "VFS system calls open(2), stat(2),
> read(2), write(2), chmod(2)". It also mentions locking [2]).
>
> BTW looking what has mknod in vfs, it's just nfsd and 9p (none of them are
> used in all_filesystems):
>
> $ git grep mknod $(git ls-files fs/|grep -i vfs)
> fs/9p/vfs_inode.c: * for mknod(2).
> fs/9p/vfs_inode.c: * v9fs_vfs_mknod - create a special file
> fs/9p/vfs_inode.c:v9fs_vfs_mknod(struct mnt_idmap *idmap, struct inode *dir,
> fs/9p/vfs_inode.c: .mknod = v9fs_vfs_mknod,
> fs/9p/vfs_inode.c: .mknod = v9fs_vfs_mknod,
> fs/9p/vfs_inode_dotl.c:v9fs_vfs_mknod_dotl(struct mnt_idmap *idmap, struct
> inode *dir, fs/9p/vfs_inode_dotl.c: return v9fs_vfs_mknod_dotl(idmap, dir,
> dentry, omode, 0); fs/9p/vfs_inode_dotl.c: * v9fs_vfs_mknod_dotl - create a
> special file fs/9p/vfs_inode_dotl.c:v9fs_vfs_mknod_dotl(struct mnt_idmap
> *idmap, struct inode *dir, fs/9p/vfs_inode_dotl.c:
> p9_debug(P9_DEBUG_VFS, "Failed to get acl values in mknod %d\n",
> fs/9p/vfs_inode_dotl.c: err = p9_client_mknod_dotl(dfid, name, mode, rdev,
> gid, &qid); fs/9p/vfs_inode_dotl.c: .mknod = v9fs_vfs_mknod_dotl,
> fs/nfsd/vfs.c: host_err = vfs_mknod(&nop_mnt_idmap, dirp, dchild,
>
> > */
>
> ...
>
> > +static void run(void)
> >
> > {
>
> ...
>
> > + SAFE_CHDIR(TEMP_DIR);
> > + TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0), "mknod(%s, %o, 0)",
TEMP_NODE,
> > MODE1);
> IMHO this simple form will print the same info, I suggest to use it.
> TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0));
>
> > - /*
> > - * Create a test directory under temporary directory with the
> > - * specified mode permissions, with uid/gid set to that of guest
> > - * user and the test process.
> > - */
> > - SAFE_MKDIR(cleanup, DIR_TEMP, MODE_RWX);
> > - SAFE_CHOWN(cleanup, DIR_TEMP, user1_uid, group2_gid);
> > + SAFE_STAT(TEMP_NODE, &buf);
> > + TST_EXP_EQ_LI(buf.st_gid, 0);
>
> ...
>
> Diff for using all_filesystems (I'm not sure myself).
>
> Kind regards,
> Petr
Thanks for reviewing, I agree with your proposed changes if we are going to
enable if for all filesystems, test executes fine with all_filesystems=1 in my
setup.
>
> [1] https://www.kernel.org/doc/html/next/filesystems/vfs.html
> [2] https://www.kernel.org/doc/html/next/filesystems/locking.html
>
> +++ testcases/kernel/syscalls/mknod/mknod02.c
> @@ -24,6 +24,8 @@
> #define TEMP_DIR "testdir"
> #define TEMP_NODE "testnode"
>
> +#define MNTPOINT "mntpoint"
> +
> static struct stat buf;
> static struct passwd *user_nobody;
> static gid_t gid_nobody;
> @@ -40,18 +42,28 @@ static void setup(void)
> static void run(void)
> {
> SAFE_CHDIR(TEMP_DIR);
> - TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0), "mknod(%s, %o, 0)",
TEMP_NODE,
> MODE1); + TST_EXP_PASS(mknod(TEMP_NODE, MODE1, 0));
>
> SAFE_STAT(TEMP_NODE, &buf);
> TST_EXP_EQ_LI(buf.st_gid, 0);
>
> SAFE_UNLINK(TEMP_NODE);
> SAFE_CHDIR("..");
> +
> +}
> +
> +static void cleanup(void)
> +{
> + SAFE_RMDIR(TEMP_DIR);
> }
>
> static struct tst_test test = {
> .setup = setup,
> + .cleanup = cleanup,
> .test_all = run,
> .needs_root = 1,
> - .needs_tmpdir = 1
> + .needs_tmpdir = 1,
> + .mount_device = 1,
> + .mntpoint = MNTPOINT,
> + .all_filesystems = 1,
> };
Regards,
Avinesh
--
Mailing list info: https://lists.linux.it/listinfo/ltp
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [LTP] [PATCH v2] mknod02.c: Simplify and convert to new LTP API
2023-05-16 11:58 ` Petr Vorel
2023-05-17 15:07 ` Avinesh Kumar
@ 2023-08-23 9:21 ` Richard Palethorpe
1 sibling, 0 replies; 4+ messages in thread
From: Richard Palethorpe @ 2023-08-23 9:21 UTC (permalink / raw)
To: Petr Vorel; +Cc: ltp
Hello,
Petr Vorel <pvorel@suse.cz> writes:
> Hi Avinesh,
>
>> Simply test when parent directory does not have set-group-ID bit set,
>> new node gets GID from effective GID of the process and does not inherit
>> the group ownership from its parent directory.
>
> Reviewed-by: Petr Vorel <pvorel@suse.cz>
>
> Few comments below.
>
>> testcases/kernel/syscalls/mknod/mknod02.c | 316 +++-------------------
>> 1 file changed, 36 insertions(+), 280 deletions(-)
>
> ...
>> +/*\
>> + * [Description]
>> *
>> + * Verify that if mknod(2) creates a filesystem node in a directory which
>> + * does not have the set-group-ID bit set, new node will not inherit the
>> + * group ownership from its parent directory and its group ID will be the
>> + * effective group ID of the process.
>
> @Cyril I wonder if it'd be good to test this on all_filesystems. Are we trying
> to use use all_filesystems = 1 when subject of testing is using VFS or the
> opposite? (kernel docs mentions "VFS system calls open(2), stat(2), read(2),
> write(2), chmod(2)". It also mentions locking [2]).
>
> BTW looking what has mknod in vfs, it's just nfsd and 9p (none of them are used
> in all_filesystems):
>
> $ git grep mknod $(git ls-files fs/|grep -i vfs)
> fs/9p/vfs_inode.c: * for mknod(2).
> fs/9p/vfs_inode.c: * v9fs_vfs_mknod - create a special file
> fs/9p/vfs_inode.c:v9fs_vfs_mknod(struct mnt_idmap *idmap, struct inode *dir,
> fs/9p/vfs_inode.c: .mknod = v9fs_vfs_mknod,
> fs/9p/vfs_inode.c: .mknod = v9fs_vfs_mknod,
> fs/9p/vfs_inode_dotl.c:v9fs_vfs_mknod_dotl(struct mnt_idmap *idmap, struct inode *dir,
> fs/9p/vfs_inode_dotl.c: return v9fs_vfs_mknod_dotl(idmap, dir, dentry, omode, 0);
> fs/9p/vfs_inode_dotl.c: * v9fs_vfs_mknod_dotl - create a special file
> fs/9p/vfs_inode_dotl.c:v9fs_vfs_mknod_dotl(struct mnt_idmap *idmap, struct inode *dir,
> fs/9p/vfs_inode_dotl.c: p9_debug(P9_DEBUG_VFS, "Failed to get acl values in mknod %d\n",
> fs/9p/vfs_inode_dotl.c: err = p9_client_mknod_dotl(dfid, name, mode, rdev, gid, &qid);
> fs/9p/vfs_inode_dotl.c: .mknod = v9fs_vfs_mknod_dotl,
> fs/nfsd/vfs.c: host_err = vfs_mknod(&nop_mnt_idmap, dirp,
> dchild,
It is defined for all major file systems on the directory
inode_operations struct (e.g. search btrfs_mknod, xfs_vn_mknod, ext4_mknod).
If you setup clangd and provide a reasonabl compile_commands.json
(i.e. relatively large kernel config) you can search for places where
struct inode_operations.mk_nod is set.
So I'll go ahead and try to merge it.
--
Thank you,
Richard.
--
Mailing list info: https://lists.linux.it/listinfo/ltp
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2023-08-23 9:37 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-03-15 15:56 [LTP] [PATCH v2] mknod02.c: Simplify and convert to new LTP API Avinesh Kumar
2023-05-16 11:58 ` Petr Vorel
2023-05-17 15:07 ` Avinesh Kumar
2023-08-23 9:21 ` Richard Palethorpe
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).