[LTP] [PATCH v3 0/2] mount03: Convert to new API
xuyang2018.jy@fujitsu.com
xuyang2018.jy@fujitsu.com
Mon Aug 15 11:57:17 CEST 2022
Hi Petr
> Hi Xu,
>
> ...
>>>>> - SAFE_READ(0, otfd, readbuf, sizeof(readbuf));
>>>>> + nobody_uid = ltpuser->pw_uid;
>>>>> + nobody_gid = ltpuser->pw_gid;
>
>>>>> - SAFE_FSTAT(otfd, &file_stat);
>>>>> + snprintf(file, PATH_MAX, "%s/%s", MNTPOINT, TESTBIN);
>>>>> + TST_RESOURCE_COPY(NULL, TESTBIN, file);
>
>>>> In fact, old test case copy resource file when mount fileystem, but now,
>>>> you change this. So in test_nosuid function, you test nosuid behaviour
>>>> in tmpdir instead of different filesystems.
>
>>> old code in setup:
>>> fs_type = tst_dev_fs_type();
>>> device = tst_acquire_device(cleanup);
>
>>> if (!device)
>>> tst_brkm(TCONF, cleanup, "Failed to obtain block device");
>
>>> tst_mkfs(cleanup, device, fs_type, NULL, NULL);
>
>>> SAFE_MKDIR(cleanup, mntpoint, DIR_MODE);
>
>>> SAFE_MOUNT(cleanup, device, mntpoint, fs_type, 0, NULL);
>>> TST_RESOURCE_COPY(cleanup, "mount03_setuid_test", path_name);
>
>>> new code:
>>> snprintf(file, PATH_MAX, "%s/%s", MNTPOINT, TESTBIN);
>>> SAFE_CP(TESTBIN, file);
>
>>> Well, Li in his v2 removed the code because there is .mntpoint = MNTPOINT, in
>>> struct tst_test, therefore MNTPOINT is mounted in the filesystem, right?
>
>>> But he also did SAFE_STAT and SAFE_CHMOD on MNTPOINT, which is IMHO wrong
>>> (or at least different from the old code).
>
>> Yes, it is wrong. I guess Chen misundertand mntpoint usage(it just
>> create mntpoint instead mount dev to a moutpoint).
>
>> So do you will fix this?
>
> Yes, see the diff below. I'm waiting little longer if anybody else has some
> comments before merging it.
>
> Thanks for your review.
>
>> Best Regards
>> Yang Xu
>
> diff --git testcases/kernel/syscalls/mount/mount03.c testcases/kernel/syscalls/mount/mount03.c
> index 74b018d78..9c58783d7 100644
> --- testcases/kernel/syscalls/mount/mount03.c
> +++ testcases/kernel/syscalls/mount/mount03.c
> @@ -15,7 +15,6 @@
> #include <sys/types.h>
> #include <sys/wait.h>
> #include <pwd.h>
> -#include "old_resource.h"
> #include "tst_test.h"
> #include "lapi/mount.h"
>
> @@ -145,7 +144,7 @@ static void setup(void)
> nobody_gid = ltpuser->pw_gid;
>
> snprintf(file, PATH_MAX, "%s/%s", MNTPOINT, TESTBIN);
> - TST_RESOURCE_COPY(NULL, TESTBIN, file);
> + SAFE_CP(TESTBIN, file);
I still think we should test nosuid behaviour on different filesystem
like other test function because we have expand it to all filesystems.
Also include tmpfs, so SAFE_CP should be in test_nosuid function
otherwise may hit ENOENT problem.
different code as below:
[root@localhost mount]# git diff .
diff --git a/testcases/kernel/syscalls/mount/mount03.c
b/testcases/kernel/syscalls/mount/mount03.c
index 74b018d78..b0582c76b 100644
--- a/testcases/kernel/syscalls/mount/mount03.c
+++ b/testcases/kernel/syscalls/mount/mount03.c
@@ -21,6 +21,7 @@
#define MNTPOINT "mntpoint"
#define TESTBIN "mount03_setuid_test"
+#define BIN_PATH MNTPOINT"/"TESTBIN
#define TEST_STR "abcdefghijklmnopqrstuvwxyz"
#define FILE_MODE 0644
#define SUID_MODE 0511
@@ -75,12 +76,19 @@ static void test_nosuid(void)
{
pid_t pid;
int status;
+ struct stat st;
+
+ snprintf(file, PATH_MAX, "%s/%s", MNTPOINT, TESTBIN);
+ SAFE_CP(TESTBIN, file);
+ SAFE_STAT(file, &st);
+ if (st.st_mode != SUID_MODE)
+ SAFE_CHMOD(file, SUID_MODE);
pid = SAFE_FORK();
if (!pid) {
SAFE_SETGID(nobody_gid);
SAFE_SETREUID(-1, nobody_uid);
- SAFE_EXECLP(TESTBIN, TESTBIN, NULL);
+ SAFE_EXECLP(BIN_PATH, TESTBIN, NULL);
}
SAFE_WAITPID(pid, &status, 0);
@@ -138,18 +146,10 @@ static struct tcase {
static void setup(void)
{
- struct stat st;
struct passwd *ltpuser = SAFE_GETPWNAM("nobody");
nobody_uid = ltpuser->pw_uid;
nobody_gid = ltpuser->pw_gid;
-
- snprintf(file, PATH_MAX, "%s/%s", MNTPOINT, TESTBIN);
- TST_RESOURCE_COPY(NULL, TESTBIN, file);
-
- SAFE_STAT(file, &st);
- if (st.st_mode != SUID_MODE)
- SAFE_CHMOD(file, SUID_MODE);
}
static void cleanup(void)
[root@localhost mount]#
Best Regards
Yang Xu
>
> SAFE_STAT(file, &st);
> if (st.st_mode != SUID_MODE)
More information about the ltp
mailing list