[PATCH] selftests/seccomp: More closely track fds being assigned

[PATCH] selftests/seccomp: More closely track fds being assigned

[Kees Cook]

Since the open fds might not always start at "4" (especially when
running under kselftest, etc), start counting from the first assigned
fd, rather than using the more permissive EXPECT_GE(fd, 0).

Signed-off-by: Kees Cook <keescook@chromium.org>
---
 tools/testing/selftests/seccomp/seccomp_bpf.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/tools/testing/selftests/seccomp/seccomp_bpf.c b/tools/testing/selftests/seccomp/seccomp_bpf.c
index e2ba7adc2694..03b37e660965 100644
--- a/tools/testing/selftests/seccomp/seccomp_bpf.c
+++ b/tools/testing/selftests/seccomp/seccomp_bpf.c
@@ -3954,7 +3954,7 @@ TEST(user_notification_addfd)
 {
 	pid_t pid;
 	long ret;
-	int status, listener, memfd, fd;
+	int status, listener, memfd, fd, nextfd;
 	struct seccomp_notif_addfd addfd = {};
 	struct seccomp_notif_addfd_small small = {};
 	struct seccomp_notif_addfd_big big = {};
@@ -3963,18 +3963,21 @@ TEST(user_notification_addfd)
 	/* 100 ms */
 	struct timespec delay = { .tv_nsec = 100000000 };
 
+	/* There may be arbitrary already-open fds at test start. */
 	memfd = memfd_create("test", 0);
 	ASSERT_GE(memfd, 0);
+	nextfd = memfd + 1;
 
 	ret = prctl(PR_SET_NO_NEW_PRIVS, 1, 0, 0, 0);
 	ASSERT_EQ(0, ret) {
 		TH_LOG("Kernel does not support PR_SET_NO_NEW_PRIVS!");
 	}
 
+	/* fd: 4 */
 	/* Check that the basic notification machinery works */
 	listener = user_notif_syscall(__NR_getppid,
 				      SECCOMP_FILTER_FLAG_NEW_LISTENER);
-	ASSERT_GE(listener, 0);
+	ASSERT_EQ(listener, nextfd++);
 
 	pid = fork();
 	ASSERT_GE(pid, 0);
@@ -4029,14 +4032,14 @@ TEST(user_notification_addfd)
 
 	/* Verify we can set an arbitrary remote fd */
 	fd = ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd);
-	EXPECT_GE(fd, 0);
+	EXPECT_EQ(fd, nextfd++);
 	EXPECT_EQ(filecmp(getpid(), pid, memfd, fd), 0);
 
 	/* Verify we can set an arbitrary remote fd with large size */
 	memset(&big, 0x0, sizeof(big));
 	big.addfd = addfd;
 	fd = ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD_BIG, &big);
-	EXPECT_GE(fd, 0);
+	EXPECT_EQ(fd, nextfd++);
 
 	/* Verify we can set a specific remote fd */
 	addfd.newfd = 42;
@@ -4070,9 +4073,11 @@ TEST(user_notification_addfd)
 	addfd.newfd = 0;
 	addfd.flags = SECCOMP_ADDFD_FLAG_SEND;
 	fd = ioctl(listener, SECCOMP_IOCTL_NOTIF_ADDFD, &addfd);
-
-	/* Child has fds 0-6 and 42 used, we expect the lower fd available: 7 */
-	EXPECT_EQ(fd, 7);
+	/*
+	 * Child has earlier "low" fds and now 42, so we expect the next
+	 * lowest available fd to be assigned here.
+	 */
+	EXPECT_EQ(fd, nextfd++);
 	EXPECT_EQ(filecmp(getpid(), pid, memfd, fd), 0);
 
 	/*
-- 
2.25.1

Re: [PATCH] selftests/seccomp: More closely track fds being assigned

[Rodrigo Campos]

On Thu, May 27, 2021 at 5:29 AM Kees Cook <keescook@chromium.org> wrote:
>
> Since the open fds might not always start at "4" (especially when
> running under kselftest, etc), start counting from the first assigned
> fd, rather than using the more permissive EXPECT_GE(fd, 0).
>
> Signed-off-by: Kees Cook <keescook@chromium.org>

Nice cleanup, thanks! Just in case, tested it here, works fine. Feel
free to add:

Reviewed-by: Rodrigo Campos <rodrigo@kinvolk.io>

I can improve the selftest to test the new addfd flag we just added
also in combination existing flags (like setting the fd number to
use), and maybe also split the big chunk test, if you think that is
valuable.

Best,
Rodrigo

Re: [PATCH] selftests/seccomp: More closely track fds being assigned

[Christian Brauner]

On Wed, May 26, 2021 at 08:29:48PM -0700, Kees Cook wrote:
> Since the open fds might not always start at "4" (especially when
> running under kselftest, etc), start counting from the first assigned
> fd, rather than using the more permissive EXPECT_GE(fd, 0).
> 
> Signed-off-by: Kees Cook <keescook@chromium.org>
> ---

Looks good,
Acked-by: Christian Brauner <christian.brauner@ubuntu.com>

Re: [PATCH] selftests/seccomp: More closely track fds being assigned

[Kees Cook]

On Thu, May 27, 2021 at 02:45:26PM +0200, Rodrigo Campos wrote:
> On Thu, May 27, 2021 at 5:29 AM Kees Cook <keescook@chromium.org> wrote:
> >
> > Since the open fds might not always start at "4" (especially when
> > running under kselftest, etc), start counting from the first assigned
> > fd, rather than using the more permissive EXPECT_GE(fd, 0).
> >
> > Signed-off-by: Kees Cook <keescook@chromium.org>
> 
> Nice cleanup, thanks! Just in case, tested it here, works fine. Feel
> free to add:
> 
> Reviewed-by: Rodrigo Campos <rodrigo@kinvolk.io>

Thanks!

> I can improve the selftest to test the new addfd flag we just added
> also in combination existing flags (like setting the fd number to
> use), and maybe also split the big chunk test, if you think that is
> valuable.

Yeah, I was pondering splitting the test up, but I think it's okay how
it is for now.

-- 
Kees Cook