[PATCH bpf-next v3 0/2] bpf: copy BPF token from main program to subprograms

[PATCH bpf-next v3 0/2] bpf: copy BPF token from main program to subprograms

[Eduard Zingerman]

bpf_jit_subprogs() omits aux->token when it creates a struct
bpf_prog_aux instances for a subprograms.
This means that for programs loaded via BPF token (i.e., from a
non-init user namespace), subprograms fail the bpf_token_capable()
check in bpf_prog_kallsyms_add() and don't appear in /proc/kallsyms.
Which in-turn makes it impossible to freplace such subprograms.

Changelog:
v2 -> v3:
- mark selftest as serial (sashiko).
v1 -> v2:
- target bpf-next tree (fixups.c) instead of bpf tree (verifier.c).

v1: https://lore.kernel.org/bpf/20260414-subprog-token-fix-v1-0-5b1a38e01546@gmail.com/T/
v2: https://lore.kernel.org/bpf/20260414-subprog-token-fix-v2-0-59146c31f6f1@gmail.com/T/
---
Eduard Zingerman (2):
      bpf: copy BPF token from main program to subprograms
      selftests/bpf: verify kallsyms entries for token-loaded subprograms

 kernel/bpf/fixups.c                                |  1 +
 tools/testing/selftests/bpf/Makefile               |  1 +
 tools/testing/selftests/bpf/prog_tests/token.c     | 83 +++++++++++++++++++++-
 .../selftests/bpf/prog_tests/unpriv_bpf_disabled.c | 21 +-----
 tools/testing/selftests/bpf/progs/token_kallsyms.c | 19 +++++
 tools/testing/selftests/bpf/sysctl_helpers.c       | 26 +++++++
 tools/testing/selftests/bpf/sysctl_helpers.h       |  7 ++
 7 files changed, 135 insertions(+), 23 deletions(-)
---
base-commit: 71b500afd2f7336f5b6c6026f2af546fc079be26
change-id: 20260414-subprog-token-fix-96c29ffa7e12

[PATCH bpf-next v3 1/2] bpf: copy BPF token from main program to subprograms

[Eduard Zingerman]

bpf_jit_subprogs() copies various fields from the main program's aux to
each subprogram's aux, but omits the BPF token. This causes
bpf_prog_kallsyms_add() to fail for subprograms loaded via BPF token,
as bpf_token_capable() falls back to capable() in init_user_ns when
token is NULL.

Copy prog->aux->token to func[i]->aux->token so that subprograms
inherit the same capability delegation as the main program.

Fixes: d79a35497547 ("bpf: Consistently use BPF token throughout BPF verifier logic")
Signed-off-by: Tao Chen <ctao@meta.com>
Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
---
 kernel/bpf/fixups.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/kernel/bpf/fixups.c b/kernel/bpf/fixups.c
index 67c9b28767e1..dd00a680e4ea 100644
--- a/kernel/bpf/fixups.c
+++ b/kernel/bpf/fixups.c
@@ -1110,6 +1110,7 @@ int bpf_jit_subprogs(struct bpf_verifier_env *env)
 		func[i]->aux->exception_cb = env->subprog_info[i].is_exception_cb;
 		func[i]->aux->changes_pkt_data = env->subprog_info[i].changes_pkt_data;
 		func[i]->aux->might_sleep = env->subprog_info[i].might_sleep;
+		func[i]->aux->token = prog->aux->token;
 		if (!i)
 			func[i]->aux->exception_boundary = env->seen_exception;
 

-- 
2.53.0

[PATCH bpf-next v3 2/2] selftests/bpf: verify kallsyms entries for token-loaded subprograms

[Eduard Zingerman]

Add a test that loads an XDP program with a global subprogram using a
BPF token from a user namespace, then verifies that both the main
program and the subprogram appear in /proc/kallsyms.

This exercises the bpf_prog_kallsyms_add() path for subprograms and
would have caught the missing aux->token copy in bpf_jit_subprogs().

load_kallsyms_local() filters out kallsyms with zero addresses.
For a process with limited capabilities to read kallsym addresses the
following sysctl variables have to be set to zero:
- /proc/sys/kernel/perf_event_paranoid
- /proc/sys/kernel/kptr_restrict
Set these variables using sysctl_set() utility function extracted from
unpriv_bpf_disabled.c to a separate c/header.
Since the test modifies global system state, mark it as serial.

Signed-off-by: Eduard Zingerman <eddyz87@gmail.com>
---
 tools/testing/selftests/bpf/Makefile               |  1 +
 tools/testing/selftests/bpf/prog_tests/token.c     | 83 +++++++++++++++++++++-
 .../selftests/bpf/prog_tests/unpriv_bpf_disabled.c | 21 +-----
 tools/testing/selftests/bpf/progs/token_kallsyms.c | 19 +++++
 tools/testing/selftests/bpf/sysctl_helpers.c       | 26 +++++++
 tools/testing/selftests/bpf/sysctl_helpers.h       |  7 ++
 6 files changed, 134 insertions(+), 23 deletions(-)

diff --git a/tools/testing/selftests/bpf/Makefile b/tools/testing/selftests/bpf/Makefile
index f75c4f52c028..da81a1853cfd 100644
--- a/tools/testing/selftests/bpf/Makefile
+++ b/tools/testing/selftests/bpf/Makefile
@@ -750,6 +750,7 @@ TRUNNER_EXTRA_SOURCES := test_progs.c		\
 			 btf_helpers.c		\
 			 cap_helpers.c		\
 			 unpriv_helpers.c 	\
+			 sysctl_helpers.c	\
 			 netlink_helpers.c	\
 			 jit_disasm_helpers.c	\
 			 io_helpers.c		\
diff --git a/tools/testing/selftests/bpf/prog_tests/token.c b/tools/testing/selftests/bpf/prog_tests/token.c
index b81dde283052..cf5b94512cff 100644
--- a/tools/testing/selftests/bpf/prog_tests/token.c
+++ b/tools/testing/selftests/bpf/prog_tests/token.c
@@ -1,9 +1,7 @@
 // SPDX-License-Identifier: GPL-2.0
 /* Copyright (c) 2023 Meta Platforms, Inc. and affiliates. */
 #define _GNU_SOURCE
-#include <test_progs.h>
 #include <bpf/btf.h>
-#include "cap_helpers.h"
 #include <fcntl.h>
 #include <sched.h>
 #include <signal.h>
@@ -15,9 +13,17 @@
 #include <sys/stat.h>
 #include <sys/syscall.h>
 #include <sys/un.h>
+
+#include "bpf_util.h"
+#include "cap_helpers.h"
+#include "sysctl_helpers.h"
+#include "test_progs.h"
+#include "trace_helpers.h"
+
 #include "priv_map.skel.h"
 #include "priv_prog.skel.h"
 #include "dummy_st_ops_success.skel.h"
+#include "token_kallsyms.skel.h"
 #include "token_lsm.skel.h"
 #include "priv_freplace_prog.skel.h"
 
@@ -1045,6 +1051,58 @@ static int userns_obj_priv_implicit_token_envvar(int mnt_fd, struct token_lsm *l
 	return -EINVAL;
 }
 
+static bool kallsyms_has_bpf_func(struct ksyms *ksyms, const char *func_name)
+{
+	char name[256];
+	int i;
+
+	for (i = 0; i < ksyms->sym_cnt; i++) {
+		if (sscanf(ksyms->syms[i].name, "bpf_prog_%*[^_]_%255s", name) == 1 &&
+		    strcmp(name, func_name) == 0)
+			return true;
+	}
+	return false;
+}
+
+static int userns_obj_priv_prog_kallsyms(int mnt_fd, struct token_lsm *lsm_skel)
+{
+	const char *func_names[] = { "xdp_main", "token_ksym_subprog" };
+	LIBBPF_OPTS(bpf_object_open_opts, opts);
+	struct token_kallsyms *skel;
+	struct ksyms *ksyms = NULL;
+	char buf[256];
+	int i, err;
+
+	snprintf(buf, sizeof(buf), "/proc/self/fd/%d", mnt_fd);
+	opts.bpf_token_path = buf;
+	skel = token_kallsyms__open_opts(&opts);
+	if (!ASSERT_OK_PTR(skel, "token_kallsyms__open_opts"))
+		return -EINVAL;
+
+	err = token_kallsyms__load(skel);
+	if (!ASSERT_OK(err, "token_kallsyms__load"))
+		goto cleanup;
+
+	ksyms = load_kallsyms_local();
+	if (!ASSERT_OK_PTR(ksyms, "load_kallsyms_local")) {
+		err = -EINVAL;
+		goto cleanup;
+	}
+
+	for (i = 0; i < ARRAY_SIZE(func_names); i++) {
+		if (!ASSERT_TRUE(kallsyms_has_bpf_func(ksyms, func_names[i]),
+				 func_names[i])) {
+			err = -EINVAL;
+			break;
+		}
+	}
+
+cleanup:
+	free_kallsyms_local(ksyms);
+	token_kallsyms__destroy(skel);
+	return err;
+}
+
 #define bit(n) (1ULL << (n))
 
 static int userns_bpf_token_info(int mnt_fd, struct token_lsm *lsm_skel)
@@ -1082,7 +1140,7 @@ static int userns_bpf_token_info(int mnt_fd, struct token_lsm *lsm_skel)
 	return err;
 }
 
-void test_token(void)
+void serial_test_token(void)
 {
 	if (test__start_subtest("map_token")) {
 		struct bpffs_opts opts = {
@@ -1194,4 +1252,23 @@ void test_token(void)
 
 		subtest_userns(&opts, userns_bpf_token_info);
 	}
+	if (test__start_subtest("obj_priv_prog_kallsyms")) {
+		char perf_paranoid_orig[32] = {};
+		char kptr_restrict_orig[32] = {};
+		struct bpffs_opts opts = {
+			.cmds = bit(BPF_BTF_LOAD) | bit(BPF_PROG_LOAD),
+			.progs = bit(BPF_PROG_TYPE_XDP),
+			.attachs = ~0ULL,
+		};
+
+		sysctl_set("/proc/sys/kernel/perf_event_paranoid", perf_paranoid_orig, "0");
+		sysctl_set("/proc/sys/kernel/kptr_restrict", kptr_restrict_orig, "0");
+
+		subtest_userns(&opts, userns_obj_priv_prog_kallsyms);
+
+		if (perf_paranoid_orig[0])
+			sysctl_set("/proc/sys/kernel/perf_event_paranoid", NULL, perf_paranoid_orig);
+		if (kptr_restrict_orig[0])
+			sysctl_set("/proc/sys/kernel/kptr_restrict", NULL, kptr_restrict_orig);
+	}
 }
diff --git a/tools/testing/selftests/bpf/prog_tests/unpriv_bpf_disabled.c b/tools/testing/selftests/bpf/prog_tests/unpriv_bpf_disabled.c
index 472f4f9fa95f..64404602b9ab 100644
--- a/tools/testing/selftests/bpf/prog_tests/unpriv_bpf_disabled.c
+++ b/tools/testing/selftests/bpf/prog_tests/unpriv_bpf_disabled.c
@@ -8,6 +8,7 @@
 
 #include "cap_helpers.h"
 #include "bpf_util.h"
+#include "sysctl_helpers.h"
 
 /* Using CAP_LAST_CAP is risky here, since it can get pulled in from
  * an old /usr/include/linux/capability.h and be < CAP_BPF; as a result
@@ -36,26 +37,6 @@ static void process_perfbuf(void *ctx, int cpu, void *data, __u32 len)
 		got_perfbuf_val = *(__u32 *)data;
 }
 
-static int sysctl_set(const char *sysctl_path, char *old_val, const char *new_val)
-{
-	int ret = 0;
-	FILE *fp;
-
-	fp = fopen(sysctl_path, "r+");
-	if (!fp)
-		return -errno;
-	if (old_val && fscanf(fp, "%s", old_val) <= 0) {
-		ret = -ENOENT;
-	} else if (!old_val || strcmp(old_val, new_val) != 0) {
-		fseek(fp, 0, SEEK_SET);
-		if (fprintf(fp, "%s", new_val) < 0)
-			ret = -errno;
-	}
-	fclose(fp);
-
-	return ret;
-}
-
 static void test_unpriv_bpf_disabled_positive(struct test_unpriv_bpf_disabled *skel,
 					      __u32 prog_id, int prog_fd, int perf_fd,
 					      char **map_paths, int *map_fds)
diff --git a/tools/testing/selftests/bpf/progs/token_kallsyms.c b/tools/testing/selftests/bpf/progs/token_kallsyms.c
new file mode 100644
index 000000000000..c9f9344f3eb2
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/token_kallsyms.c
@@ -0,0 +1,19 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2026 Meta Platforms, Inc. and affiliates. */
+
+#include "vmlinux.h"
+#include <bpf/bpf_helpers.h>
+
+char _license[] SEC("license") = "GPL";
+
+__weak
+int token_ksym_subprog(void)
+{
+	return 0;
+}
+
+SEC("xdp")
+int xdp_main(struct xdp_md *xdp)
+{
+	return token_ksym_subprog();
+}
diff --git a/tools/testing/selftests/bpf/sysctl_helpers.c b/tools/testing/selftests/bpf/sysctl_helpers.c
new file mode 100644
index 000000000000..e58c23adaa3e
--- /dev/null
+++ b/tools/testing/selftests/bpf/sysctl_helpers.c
@@ -0,0 +1,26 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <stdio.h>
+#include <errno.h>
+#include <string.h>
+
+#include "sysctl_helpers.h"
+
+int sysctl_set(const char *sysctl_path, char *old_val, const char *new_val)
+{
+	int ret = 0;
+	FILE *fp;
+
+	fp = fopen(sysctl_path, "r+");
+	if (!fp)
+		return -errno;
+	if (old_val && fscanf(fp, "%s", old_val) <= 0) {
+		ret = -ENOENT;
+	} else if (!old_val || strcmp(old_val, new_val) != 0) {
+		fseek(fp, 0, SEEK_SET);
+		if (fprintf(fp, "%s", new_val) < 0)
+			ret = -errno;
+	}
+	fclose(fp);
+
+	return ret;
+}
diff --git a/tools/testing/selftests/bpf/sysctl_helpers.h b/tools/testing/selftests/bpf/sysctl_helpers.h
new file mode 100644
index 000000000000..28d7e209df72
--- /dev/null
+++ b/tools/testing/selftests/bpf/sysctl_helpers.h
@@ -0,0 +1,7 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+#ifndef __SYSCTL_HELPERS_H
+#define __SYSCTL_HELPERS_H
+
+int sysctl_set(const char *sysctl_path, char *old_val, const char *new_val);
+
+#endif

-- 
2.53.0

Re: [PATCH bpf-next v3 2/2] selftests/bpf: verify kallsyms entries for token-loaded subprograms

[sashiko-bot]

Sashiko AI review found 1 potential issue(s):
- [Medium] Unchecked return values of `sysctl_set()` in `serial_test_token()` can lead to confusing test failures.
--

commit e7dc629b9f37dc08742ff9bbddeaaf65c1e34956
Author: Eduard Zingerman <eddyz87@gmail.com>

selftests/bpf: verify kallsyms entries for token-loaded subprograms

This commit adds a test to verify that both the main XDP program and its
global subprogram appear in /proc/kallsyms when loaded using a user
namespace BPF token. It also introduces a sysctl_set utility to configure
system parameters needed to read kallsyms.

> diff --git a/tools/testing/selftests/bpf/prog_tests/token.c b/tools/testing/selftests/bpf/prog_tests/token.c
> index b81dde283052e..cf5b94512cff4 100644
> --- a/tools/testing/selftests/bpf/prog_tests/token.c
> +++ b/tools/testing/selftests/bpf/prog_tests/token.c

[ ... ]

> @@ -1194,4 +1252,23 @@ void test_token(void)
>  
>  		subtest_userns(&opts, userns_bpf_token_info);
>  	}
> +	if (test__start_subtest("obj_priv_prog_kallsyms")) {
> +		char perf_paranoid_orig[32] = {};
> +		char kptr_restrict_orig[32] = {};
> +		struct bpffs_opts opts = {
> +			.cmds = bit(BPF_BTF_LOAD) | bit(BPF_PROG_LOAD),
> +			.progs = bit(BPF_PROG_TYPE_XDP),
> +			.attachs = ~0ULL,
> +		};
> +
> +		sysctl_set("/proc/sys/kernel/perf_event_paranoid", perf_paranoid_orig, "0");
> +		sysctl_set("/proc/sys/kernel/kptr_restrict", kptr_restrict_orig, "0");

Should the return values of these sysctl_set calls be checked?

If the test is executed in an environment where these sysctls are read-only
or inaccessible (such as inside a container), sysctl_set will silently fail.

This would cause load_kallsyms_local to read zeroed addresses because
kptr_restrict wasn't successfully disabled, resulting in confusing test
failures later at ASSERT_TRUE instead of catching the setup failure early.

Would it be better to wrap these with ASSERT_OK?

> +
> +		subtest_userns(&opts, userns_obj_priv_prog_kallsyms);
> +
> +		if (perf_paranoid_orig[0])
> +			sysctl_set("/proc/sys/kernel/perf_event_paranoid", NULL, perf_paranoid_orig);
> +		if (kptr_restrict_orig[0])
> +			sysctl_set("/proc/sys/kernel/kptr_restrict", NULL, kptr_restrict_orig);
> +	}
>  }

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260415-subprog-token-fix-v3-0-6fefe1d51646@gmail.com?part=2

Re: [PATCH bpf-next v3 2/2] selftests/bpf: verify kallsyms entries for token-loaded subprograms

[Eduard Zingerman]

On Wed, 2026-04-15 at 19:17 +0000, sashiko-bot@kernel.org wrote:
> Sashiko AI review found 1 potential issue(s):
> - [Medium] Unchecked return values of `sysctl_set()` in `serial_test_token()` can lead to confusing test failures.
> --
> 
> commit e7dc629b9f37dc08742ff9bbddeaaf65c1e34956
> Author: Eduard Zingerman <eddyz87@gmail.com>
> 
> selftests/bpf: verify kallsyms entries for token-loaded subprograms
> 
> This commit adds a test to verify that both the main XDP program and its
> global subprogram appear in /proc/kallsyms when loaded using a user
> namespace BPF token. It also introduces a sysctl_set utility to configure
> system parameters needed to read kallsyms.
> 
> > diff --git a/tools/testing/selftests/bpf/prog_tests/token.c b/tools/testing/selftests/bpf/prog_tests/token.c
> > index b81dde283052e..cf5b94512cff4 100644
> > --- a/tools/testing/selftests/bpf/prog_tests/token.c
> > +++ b/tools/testing/selftests/bpf/prog_tests/token.c
> 
> [ ... ]
> 
> > @@ -1194,4 +1252,23 @@ void test_token(void)
> >  
> >  		subtest_userns(&opts, userns_bpf_token_info);
> >  	}
> > +	if (test__start_subtest("obj_priv_prog_kallsyms")) {
> > +		char perf_paranoid_orig[32] = {};
> > +		char kptr_restrict_orig[32] = {};
> > +		struct bpffs_opts opts = {
> > +			.cmds = bit(BPF_BTF_LOAD) | bit(BPF_PROG_LOAD),
> > +			.progs = bit(BPF_PROG_TYPE_XDP),
> > +			.attachs = ~0ULL,
> > +		};
> > +
> > +		sysctl_set("/proc/sys/kernel/perf_event_paranoid", perf_paranoid_orig, "0");
> > +		sysctl_set("/proc/sys/kernel/kptr_restrict", kptr_restrict_orig, "0");
> 
> Should the return values of these sysctl_set calls be checked?
> 
> If the test is executed in an environment where these sysctls are read-only
> or inaccessible (such as inside a container), sysctl_set will silently fail.
> 
> This would cause load_kallsyms_local to read zeroed addresses because
> kptr_restrict wasn't successfully disabled, resulting in confusing test
> failures later at ASSERT_TRUE instead of catching the setup failure early.
> 
> Would it be better to wrap these with ASSERT_OK?

Makes sense.

> > +
> > +		subtest_userns(&opts, userns_obj_priv_prog_kallsyms);
> > +
> > +		if (perf_paranoid_orig[0])
> > +			sysctl_set("/proc/sys/kernel/perf_event_paranoid", NULL, perf_paranoid_orig);
> > +		if (kptr_restrict_orig[0])
> > +			sysctl_set("/proc/sys/kernel/kptr_restrict", NULL, kptr_restrict_orig);
> > +	}
> >  }