Skip to contentSkip to conversation

splice07.c: Skip invalid splice() tests involving memfd secret

RFC PATCH Applied Jan Polensky 2 participants 5 messages Jul 9, 2025 – Jul 15, 2025

lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 splice07.c: Skip invalid splice() tests involving memfd secret

Linux commit cbe4134ea4bc ("fs: export anon_inode_make_secure_inode() and fix secretmem LSM bypass") prevents any access to secret memory pages from other kernel subsystems.

Splice operations involving memfd secret are no longer valid and return EACCES. These test cases are skipped accordingly.

This avoids false negatives in splice07 test:

    [skip]
    splice07.c:54: TFAIL: splice() on pipe read end -> memfd secret expected EBADF, EINVAL: EACCES (13)
    [skip]
    splice07.c:54: TFAIL: splice() on memfd secret -> pipe write end expected EBADF, EINVAL: EACCES (13)
    [skip]

Reference: cbe4134ea4bc ("fs: export anon_inode_make_secure_inode() and fix secretmem LSM bypass")

---
 testcases/kernel/syscalls/splice/splice07.c | 2 ++
 1 file changed, 2 insertions(+)
testcases/kernel/syscalls/splice/splice07.c +2 -0
index 2228e2f269de..c750a14bf439 100644
@@ -27,6 +27,7 @@ static void check_splice(struct tst_fd *fd_in, struct tst_fd *fd_out)
 		case TST_FD_UNIX_SOCK:
 		case TST_FD_INET_SOCK:
 		case TST_FD_MEMFD:
+		case TST_FD_MEMFD_SECRET:
 			return;
 		default:
 		break;
@@ -40,6 +41,7 @@ static void check_splice(struct tst_fd *fd_in, struct tst_fd *fd_out)
 		case TST_FD_FILE:
 		case TST_FD_PROC_MAPS:
 		case TST_FD_MEMFD:
+		case TST_FD_MEMFD_SECRET:
 			return;
 		/* And this complains about socket not being connected */
 		case TST_FD_INET_SOCK:
Signature · 5 lines
2.50.0


-- 
Mailing list info: https://lists.linux.it/listinfo/ltp
#1 lore raw
Quoted 34 lines · Jan Polensky

Linux commit cbe4134ea4bc ("fs: export anon_inode_make_secure_inode() and fix secretmem LSM bypass") prevents any access to secret memory pages from other kernel subsystems.

Splice operations involving memfd secret are no longer valid and return EACCES. These test cases are skipped accordingly.

This avoids false negatives in splice07 test:

    [skip]
    splice07.c:54: TFAIL: splice() on pipe read end -> memfd secret expected EBADF, EINVAL: EACCES (13)
    [skip]
    splice07.c:54: TFAIL: splice() on memfd secret -> pipe write end expected EBADF, EINVAL: EACCES (13)
    [skip]

Reference: cbe4134ea4bc ("fs: export anon_inode_make_secure_inode() and fix secretmem LSM bypass")

---
 testcases/kernel/syscalls/splice/splice07.c | 2 ++
 1 file changed, 2 insertions(+)
testcases/kernel/syscalls/splice/splice07.c +1 -0
index 2228e2f269de..c750a14bf439 100644
@@ -27,6 +27,7 @@ static void check_splice(struct tst_fd *fd_in, struct tst_fd *fd_out)
 		case TST_FD_UNIX_SOCK:
 		case TST_FD_INET_SOCK:
 		case TST_FD_MEMFD:
+		case TST_FD_MEMFD_SECRET:
 			return;
 		default:
 		break;

testcases/kernel/syscalls/splice/splice07.c static void check_splice(struct tst_fd *fd_in, struct tst_fd *fd_out) · line 47

Quoted 7 lines
testcases/kernel/syscalls/splice/splice07.c +1 -0
@@ -40,6 +41,7 @@ static void check_splice(struct tst_fd *fd_in, struct tst_fd *fd_out)
 		case TST_FD_FILE:
 		case TST_FD_PROC_MAPS:
 		case TST_FD_MEMFD:
+		case TST_FD_MEMFD_SECRET:
 			return;
 		/* And this complains about socket not being connected */
 		case TST_FD_INET_SOCK:

Another option maybe we can add this error into exp_errnos, i guess can fix your issue?

for example:

        const int exp_errnos[] = {EBADF, EINVAL};

+ const int exp_errnos[] = {EBADF, EINVAL, EACCES};

Quoted 6 lines
Signature · 5 lines
2.50.0


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

Hi Wei,

Quoted 2 lines · Wei Gao
Quoted 1 line · Jan Polensky

Linux commit cbe4134ea4bc ("fs: export anon_inode_make_secure_inode() and fix

[skip]

Quoted 9 lines
Quoted 1 line

case TST_FD_INET_SOCK:

Another option maybe we can add this error into exp_errnos, i guess can fix your issue?

for example:

        const int exp_errnos[] = {EBADF, EINVAL};

+ const int exp_errnos[] = {EBADF, EINVAL, EACCES};

I’m concerned that this approach might make the tests too unspecific.

Associating specific combinations with corresponding error codes would provide much more precise results. However, given the vast number of possible combinations, such an effort seems hardly practical.

Quoted 6 lines
Quoted 6 lines

--
2.50.0

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

Signature · 1 line
Mailing list info: https://lists.linux.it/listinfo/ltp
#3 lore raw
Quoted 7 lines · Jan Polensky
Quoted 7 lines

Another option maybe we can add this error into exp_errnos, i guess can fix your issue?

for example:

         const int exp_errnos[] = {EBADF, EINVAL};

+ const int exp_errnos[] = {EBADF, EINVAL, EACCES};

Quoted 5 lines

I’m concerned that this approach might make the tests too unspecific.

Associating specific combinations with corresponding error codes would provide much more precise results. However, given the vast number of possible combinations, such an effort seems hardly practical.

A specific approach for the TST_FD_MEMFD_SECRET is probably required, because we don't really want to test EACCESS for all the other file descriptions. That might hide a bug in the kernel. We need to create a new test, since this one is only checking for EBADF/EINVAL according to the description.

- Andrea

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