[LTP] [PATCH v2 3/8] Rewrite userns03.c using new LTP API
Petr Vorel
pvorel@suse.cz
Thu Mar 24 21:40:47 CET 2022
Hi Andrea,
generally LGTM, few notes below.
Reviewed-by: Petr Vorel <pvorel@suse.cz>
> testcases/kernel/containers/userns/userns03.c | 266 ++++++++----------
> +/*\
> + * [Description]
> + *
> + * Verify that /proc/PID/uid_map and /proc/PID/gid_map contains three values
> + * separated by white space:
> * ID-inside-ns ID-outside-ns length
> *
> * ID-outside-ns is interpreted according to which process is opening the file.
> @@ -26,29 +21,23 @@
> * The string "deny" would be written to /proc/self/setgroups before GID
> * check if setgroups is allowed, see kernel commits:
> *
> - * commit 9cc46516ddf497ea16e8d7cb986ae03a0f6b92f8
> - * Author: Eric W. Biederman <ebiederm@xmission.com>
> - * Date: Tue Dec 2 12:27:26 2014 -0600
> - * userns: Add a knob to disable setgroups on a per user namespace basis
> - *
> - * commit 66d2f338ee4c449396b6f99f5e75cd18eb6df272
> - * Author: Eric W. Biederman <ebiederm@xmission.com>
> - * Date: Fri Dec 5 19:36:04 2014 -0600
> - * userns: Allow setting gid_maps without privilege when setgroups is disabled
> + * commit 9cc46516ddf497ea16e8d7cb986ae03a0f6b92f8
> + * Author: Eric W. Biederman <ebiederm@xmission.com>
> + * Date: Tue Dec 2 12:27:26 2014 -0600
> + * userns: Add a knob to disable setgroups on a per user namespace basis
> *
> + * commit 66d2f338ee4c449396b6f99f5e75cd18eb6df272
> + * Author: Eric W. Biederman <ebiederm@xmission.com>
> + * Date: Fri Dec 5 19:36:04 2014 -0600
> + * userns: Allow setting gid_maps without privilege when setgroups is disabled
Commits like these two will be very badly formatted in html/pdf output.
I'd also add some blank lines so that text will be split in paragraphs.
Thus I suggest:
/*\
* [Description]
*
* Verify that /proc/PID/uid_map and /proc/PID/gid_map contains three values
* separated by white space:
*
* ID-inside-ns ID-outside-ns length
*
* ID-outside-ns is interpreted according to which process is opening the file.
*
* If the process opening the file is in the same user namespace as the process
* PID, then ID-outside-ns is defined with respect to the parent user namespace.
*
* If the process opening the file is in a different user namespace, then
* ID-outside-ns is defined with respect to the user namespace of the process
* opening the file.
*
* The string "deny" would be written to /proc/self/setgroups before GID
* check if setgroups is allowed, see kernel commits:
*
* * 9cc46516ddf4 ("userns: Add a knob to disable setgroups on a per user namespace basis")
* * 66d2f338ee4c ("userns: Allow setting gid_maps without privilege when setgroups is disabled")
*/
> @@ -75,161 +64,148 @@ static int child_fn1(void)
...
> /* map file format:ID-inside-ns ID-outside-ns length
> - If the process opening the file is in the same user namespace as
> - the process PID, then ID-outside-ns is defined with respect to the
> - parent user namespace.*/
> + * If the process opening the file is in the same user namespace as
> + * the process PID, then ID-outside-ns is defined with respect to the
> + * parent user namespace
> + */
> if (idinsidens != CHILD2UID || idoutsidens != parentuid) {
> - printf("child_fn2 checks /proc/cpid2/uid_map:\n");
> - printf("unexpected: idinsidens=%d idoutsidens=%d\n",
> - idinsidens, idoutsidens);
> - exit_val = 1;
> + tst_res(TINFO, "child2 checks /proc/cpid2/uid_map");
nit: I'd put this TINFO before if, so that it's printed also for TPASS.
Also I'd wrote CPID2 to be obvious it's supposed to be a number, or even put
pid value.
> + tst_res(TFAIL, "unexpected: namespace ID inside=%d outside=%d", idinsidens, idoutsidens);
> + } else {
> + tst_res(TPASS, "expected namespaces IDs");
> }
> sprintf(cpid1uidpath, "/proc/%d/uid_map", cpid1);
> - SAFE_FILE_SCANF(NULL, cpid1uidpath, "%d %d %d", &idinsidens,
> - &idoutsidens, &length);
> + SAFE_FILE_SCANF(cpid1uidpath, "%d %d %d", &idinsidens, &idoutsidens, &length);
> /* If the process opening the file is in a different user namespace,
> - then ID-outside-ns is defined with respect to the user namespace
> - of the process opening the file.*/
> + * then ID-outside-ns is defined with respect to the user namespace
> + * of the process opening the file
> + */
> if (idinsidens != CHILD1UID || idoutsidens != CHILD2UID) {
> - printf("child_fn2 checks /proc/cpid1/uid_map:\n");
> - printf("unexpected: idinsidens=%d idoutsidens=%d\n",
> - idinsidens, idoutsidens);
> - exit_val = 1;
> + tst_res(TINFO, "child2 checks /proc/cpid1/uid_map");
And here as well.
> + tst_res(TFAIL, "unexpected: namespace ID inside=%d outside=%d", idinsidens, idoutsidens);
> + } else {
> + tst_res(TPASS, "expected namespaces IDs");
> }
> sprintf(cpid1gidpath, "/proc/%d/gid_map", cpid1);
> - SAFE_FILE_SCANF(NULL, "/proc/self/gid_map", "%d %d %d",
> - &idinsidens, &idoutsidens, &length);
> + SAFE_FILE_SCANF("/proc/self/gid_map", "%d %d %d", &idinsidens, &idoutsidens, &length);
> if (idinsidens != CHILD2GID || idoutsidens != parentgid) {
> - printf("child_fn2 checks /proc/cpid2/gid_map:\n");
> - printf("unexpected: idinsidens=%d idoutsidens=%d\n",
> - idinsidens, idoutsidens);
> - exit_val = 1;
> + tst_res(TINFO, "child2 checks /proc/cpid2/gid_map");
And here.
> + tst_res(TFAIL, "unexpected: namespace ID inside=%d outside=%d", idinsidens, idoutsidens);
> + } else {
> + tst_res(TPASS, "expected namespaces IDs");
> }
> - SAFE_FILE_SCANF(NULL, cpid1gidpath, "%d %d %d", &idinsidens,
> - &idoutsidens, &length);
> + SAFE_FILE_SCANF(cpid1gidpath, "%d %d %d", &idinsidens, &idoutsidens, &length);
> if (idinsidens != CHILD1GID || idoutsidens != CHILD2GID) {
> - printf("child_fn1 checks /proc/cpid1/gid_map:\n");
> - printf("unexpected: idinsidens=%d idoutsidens=%d\n",
> - idinsidens, idoutsidens);
> - exit_val = 1;
> + tst_res(TINFO, "child1 checks /proc/cpid1/gid_map");
And here.
> + tst_res(TFAIL, "unexpected: namespace ID inside=%d outside=%d", idinsidens, idoutsidens);
> + } else {
> + tst_res(TPASS, "expected namespaces IDs");
> }
> -int main(int argc, char *argv[])
> +static void run(void)
> {
> + parentuid = geteuid();
> + parentgid = getegid();
> +
> + cpid1 = ltp_clone_quick(CLONE_NEWUSER | SIGCHLD, (void *)child_fn1, NULL);
> + if (cpid1 < 0)
> + tst_brk(TBROK | TTERRNO, "cpid1 clone failed");
> +
> + cpid2 = ltp_clone_quick(CLONE_NEWUSER | SIGCHLD, (void *)child_fn2, NULL);
> + if (cpid2 < 0)
> + tst_brk(TBROK | TTERRNO, "cpid2 clone failed");
> +
> + if (access("/proc/self/setgroups", F_OK) == 0) {
> + sprintf(path, "/proc/%d/setgroups", cpid1);
> +
> + fd = SAFE_OPEN(path, O_WRONLY, 0644);
> + SAFE_WRITE(1, fd, "deny", 4);
> + SAFE_CLOSE(fd);
> +
> + /* If the setgroups file has the value "deny",
> + * then the setgroups(2) system call can't
> + * subsequently be reenabled (by writing "allow" to
> + * the file) in this user namespace. (Attempts to
> + * do so will fail with the error EPERM.)
> + */
> +
> + /* test that setgroups can't be re-enabled */
> + fd = SAFE_OPEN(path, O_WRONLY, 0644);
> + ret = write(fd, "allow", 5);
> +
> + if (ret >= 0)
nit: Maybe keep ret != -1 (as was in original?)
> + tst_brk(TBROK, "write action should fail");
> + else if (errno != EPERM)
> + tst_brk(TBROK | TTERRNO, "unexpected error");
> +
> + SAFE_CLOSE(fd);
> +
> + tst_res(TPASS, "setgroups can't be re-enabled");
> +
> + sprintf(path, "/proc/%d/setgroups", cpid2);
> +
> + fd = SAFE_OPEN(path, O_WRONLY, 0644);
> + SAFE_WRITE(1, fd, "deny", 4);
> + SAFE_CLOSE(fd);
...
Kind regards,
Petr
More information about the ltp
mailing list