From: Amery Hung <ameryhung@gmail.com>
To: bpf@vger.kernel.org
Cc: netdev@vger.kernel.org, alexei.starovoitov@gmail.com,
andrii@kernel.org, daniel@iogearbox.net, eddyz87@gmail.com,
memxor@gmail.com, martin.lau@kernel.org,
mykyta.yatsenko5@gmail.com, ameryhung@gmail.com,
kernel-team@meta.com
Subject: [PATCH bpf-next v6 07/13] bpf: Unify referenced object tracking in verifier
Date: Thu, 28 May 2026 18:49:30 -0700 [thread overview]
Message-ID: <20260529014936.2811085-8-ameryhung@gmail.com> (raw)
In-Reply-To: <20260529014936.2811085-1-ameryhung@gmail.com>
Helpers and kfuncs independently tracked referenced object metadata
using standalone id fields in their respective arg_meta structs.
This led to duplicated logic and inconsistent error handling between the
two paths.
Introduce struct ref_obj_desc to consolidate id and parent_id along with
a count of how many arguments carry a reference. Add update_ref_obj() to
populate it from a bpf_reg_state, replacing open-coded assignments in
check_func_arg(), check_kfunc_args(), and process_iter_arg(). Add
validate_ref_obj() to check for ambiguous ref_obj before using it.
For ref_obj releasing helpers and kfuncs, keep checking it before
calling update_ref_obj() for now. A later patch will make these
functions not depending on ref_obj. For other users of ref_obj, move the
checks to the use locations. For helper, this means moving the checks
inside helper_multiple_ref_obj_use() to use locations.
is_acquire_function() is dropped as ref_obj is never used.
Pass ref_obj_desc into process_dynptr_func()/mark_stack_slots_dynptr()
instead of a bare parent_id to make it less confusing.
Drop the selftest introduced in 7ec899ac90a2 ("selftests/bpf: Negative
test case for ref_obj_id in args") since the verifier no longer
complains about ambiguous ref_obj if it is not used.
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Signed-off-by: Amery Hung <ameryhung@gmail.com>
---
include/linux/bpf_verifier.h | 14 +-
kernel/bpf/verifier.c | 122 +++++++++---------
.../bpf/progs/test_ringbuf_map_key.c | 11 +-
tools/testing/selftests/bpf/verifier/calls.c | 24 ----
4 files changed, 78 insertions(+), 93 deletions(-)
diff --git a/include/linux/bpf_verifier.h b/include/linux/bpf_verifier.h
index 75b287d8d92f..b0521ba7787a 100644
--- a/include/linux/bpf_verifier.h
+++ b/include/linux/bpf_verifier.h
@@ -1424,6 +1424,18 @@ struct bpf_dynptr_desc {
u32 parent_id;
};
+/*
+ * The last seen rereferenced object; Updated by update_ref_obj() when a register refers to a
+ * referenced object. Used when the helper or kfunc is releasing a referenced object, casting
+ * a referenced object, returning allocated memory derived from referenced object or creating
+ * a dynptr with a referenced object as parent.
+ */
+struct ref_obj_desc {
+ u32 id;
+ u32 parent_id;
+ u8 cnt;
+};
+
struct bpf_kfunc_call_arg_meta {
/* In parameters */
struct btf *btf;
@@ -1432,7 +1444,6 @@ struct bpf_kfunc_call_arg_meta {
const struct btf_type *func_proto;
const char *func_name;
/* Out parameters */
- u32 id;
u8 release_regno;
bool r0_rdonly;
u32 ret_btf_id;
@@ -1470,6 +1481,7 @@ struct bpf_kfunc_call_arg_meta {
} iter;
struct bpf_map_desc map;
struct bpf_dynptr_desc dynptr;
+ struct ref_obj_desc ref_obj;
u64 mem_size;
};
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 3db0f1db475a..1a49e5194bb4 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -231,9 +231,28 @@ static void bpf_map_key_store(struct bpf_insn_aux_data *aux, u64 state)
(poisoned ? BPF_MAP_KEY_POISON : 0ULL);
}
+static void update_ref_obj(struct ref_obj_desc *ref_obj, struct bpf_reg_state *reg)
+{
+ ref_obj->id = reg->id;
+ ref_obj->parent_id = reg->parent_id;
+ ref_obj->cnt++;
+}
+
+static int validate_ref_obj(struct bpf_verifier_env *env, struct ref_obj_desc *ref_obj)
+{
+ if (ref_obj->cnt > 1) {
+ verifier_bug(env, "function expects only one referenced object but got %d\n",
+ ref_obj->cnt);
+ return -EFAULT;
+ }
+
+ return 0;
+}
+
struct bpf_call_arg_meta {
struct bpf_map_desc map;
struct bpf_dynptr_desc dynptr;
+ struct ref_obj_desc ref_obj;
bool raw_mode;
bool pkt_access;
u8 release_regno;
@@ -241,7 +260,6 @@ struct bpf_call_arg_meta {
int access_size;
int mem_size;
u64 msize_max_value;
- u32 id;
int func_id;
struct btf *btf;
u32 btf_id;
@@ -528,20 +546,6 @@ bool bpf_is_may_goto_insn(struct bpf_insn *insn)
return insn->code == (BPF_JMP | BPF_JCOND) && insn->src_reg == BPF_MAY_GOTO;
}
-static bool helper_multiple_ref_obj_use(enum bpf_func_id func_id,
- const struct bpf_map *map)
-{
- int ref_obj_uses = 0;
-
- if (is_ptr_cast_function(func_id))
- ref_obj_uses++;
- if (is_acquire_function(func_id, map))
- ref_obj_uses++;
-
- return ref_obj_uses > 1;
-}
-
-
static bool is_spi_bounds_valid(struct bpf_func_state *state, int spi, int nr_slots)
{
int allocated_slots = state->allocated_stack / BPF_REG_SIZE;
@@ -670,11 +674,11 @@ static int destroy_if_dynptr_stack_slot(struct bpf_verifier_env *env,
struct bpf_func_state *state, int spi);
static int mark_stack_slots_dynptr(struct bpf_verifier_env *env, struct bpf_reg_state *reg,
- enum bpf_arg_type arg_type, int insn_idx, int parent_id,
- struct bpf_dynptr_desc *dynptr)
+ enum bpf_arg_type arg_type, int insn_idx,
+ struct ref_obj_desc *ref_obj, struct bpf_dynptr_desc *dynptr)
{
struct bpf_func_state *state = bpf_func(env, reg);
- int spi, i, err;
+ int spi, i, err, parent_id = 0;
enum bpf_dynptr_type type;
spi = dynptr_get_spi(env, reg);
@@ -707,6 +711,13 @@ static int mark_stack_slots_dynptr(struct bpf_verifier_env *env, struct bpf_reg_
return -EINVAL;
if (dynptr->type == BPF_DYNPTR_TYPE_INVALID) { /* dynptr constructors */
+ err = validate_ref_obj(env, ref_obj);
+ if (err)
+ return err;
+
+ /* Track parent's id if the parent is a referenced object */
+ parent_id = ref_obj->id;
+
if (dynptr_type_referenced(type)) {
int id;
@@ -7182,7 +7193,7 @@ static int process_kptr_func(struct bpf_verifier_env *env, int regno,
*/
static int process_dynptr_func(struct bpf_verifier_env *env, struct bpf_reg_state *reg,
argno_t argno, int insn_idx, enum bpf_arg_type arg_type,
- int parent_id, struct bpf_dynptr_desc *dynptr)
+ struct ref_obj_desc *ref_obj, struct bpf_dynptr_desc *dynptr)
{
int spi, err = 0;
@@ -7223,7 +7234,7 @@ static int process_dynptr_func(struct bpf_verifier_env *env, struct bpf_reg_stat
return err;
}
- err = mark_stack_slots_dynptr(env, reg, arg_type, insn_idx, parent_id, dynptr);
+ err = mark_stack_slots_dynptr(env, reg, arg_type, insn_idx, ref_obj, dynptr);
} else /* OBJ_RELEASE and None case from above */ {
/* For the reg->type == PTR_TO_STACK case, bpf_dynptr is never const */
if (reg->type == CONST_PTR_TO_DYNPTR && (arg_type & OBJ_RELEASE)) {
@@ -7271,13 +7282,6 @@ static int process_dynptr_func(struct bpf_verifier_env *env, struct bpf_reg_stat
return err;
}
-static u32 iter_ref_id(struct bpf_verifier_env *env, struct bpf_reg_state *reg, int spi)
-{
- struct bpf_func_state *state = bpf_func(env, reg);
-
- return state->stack[spi].spilled_ptr.id;
-}
-
static bool is_iter_kfunc(struct bpf_kfunc_call_arg_meta *meta)
{
return meta->kfunc_flags & (KF_ITER_NEW | KF_ITER_NEXT | KF_ITER_DESTROY);
@@ -7310,6 +7314,7 @@ static bool is_kfunc_arg_iter(struct bpf_kfunc_call_arg_meta *meta, int arg_idx,
static int process_iter_arg(struct bpf_verifier_env *env, struct bpf_reg_state *reg, argno_t argno, int insn_idx,
struct bpf_kfunc_call_arg_meta *meta)
{
+ struct bpf_func_state *state = bpf_func(env, reg);
const struct btf_type *t;
u32 arg_idx = arg_idx_from_argno(argno);
int spi, err, i, nr_slots, btf_id;
@@ -7381,7 +7386,7 @@ static int process_iter_arg(struct bpf_verifier_env *env, struct bpf_reg_state *
/* remember meta->iter info for process_iter_next_call() */
meta->iter.spi = spi;
meta->iter.frameno = reg->frameno;
- meta->id = iter_ref_id(env, reg, spi);
+ update_ref_obj(&meta->ref_obj, &state->stack[spi].spilled_ptr);
if (is_iter_destroy_kfunc(meta)) {
err = unmark_stack_slots_iter(env, reg, nr_slots);
@@ -8160,6 +8165,7 @@ static int check_func_arg(struct bpf_verifier_env *env, u32 arg,
u32 regno = BPF_REG_1 + arg;
struct bpf_reg_state *reg = reg_state(env, regno);
enum bpf_arg_type arg_type = fn->arg_type[arg];
+ argno_t argno = argno_from_arg(arg + 1);
enum bpf_reg_type type = reg->type;
u32 *arg_btf_id = NULL;
u32 key_size;
@@ -8226,15 +8232,8 @@ static int check_func_arg(struct bpf_verifier_env *env, u32 arg,
meta->release_regno = regno;
}
- if (reg_is_referenced(env, reg) && base_type(arg_type) != ARG_KPTR_XCHG_DEST) {
- if (meta->id) {
- verbose(env, "more than one arg with referenced id R%d %u %u",
- regno, reg->id,
- meta->id);
- return -EACCES;
- }
- meta->id = reg->id;
- }
+ if (reg_is_referenced(env, reg))
+ update_ref_obj(&meta->ref_obj, reg);
switch (base_type(arg_type)) {
case ARG_CONST_MAP_PTR:
@@ -8373,7 +8372,7 @@ static int check_func_arg(struct bpf_verifier_env *env, u32 arg,
true, meta);
break;
case ARG_PTR_TO_DYNPTR:
- err = process_dynptr_func(env, reg, argno_from_reg(regno), insn_idx, arg_type, 0,
+ err = process_dynptr_func(env, reg, argno_from_reg(regno), insn_idx, arg_type, &meta->ref_obj,
&meta->dynptr);
if (err)
return err;
@@ -9036,6 +9035,7 @@ static int btf_check_func_arg_match(struct bpf_verifier_env *env, int subprog,
struct bpf_subprog_info *sub = subprog_info(env, subprog);
struct bpf_func_state *caller = cur_func(env);
struct bpf_verifier_log *log = &env->log;
+ struct ref_obj_desc ref_obj = {};
u32 i;
int ret, err;
@@ -9113,7 +9113,7 @@ static int btf_check_func_arg_match(struct bpf_verifier_env *env, int subprog,
if (ret)
return ret;
- ret = process_dynptr_func(env, reg, argno, -1, arg->arg_type, 0, NULL);
+ ret = process_dynptr_func(env, reg, argno, -1, arg->arg_type, &ref_obj, NULL);
if (ret)
return ret;
} else if (base_type(arg->arg_type) == ARG_PTR_TO_BTF_ID) {
@@ -10119,8 +10119,8 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
err = -EINVAL;
if (arg_type_is_dynptr(fn->arg_type[meta.release_regno - BPF_REG_1])) {
err = unmark_stack_slots_dynptr(env, ®s[meta.release_regno]);
- } else if (func_id == BPF_FUNC_kptr_xchg && meta.id) {
- u32 id = meta.id;
+ } else if (func_id == BPF_FUNC_kptr_xchg && meta.ref_obj.id) {
+ u32 id = meta.ref_obj.id;
bool in_rcu = in_rcu_cs(env);
struct bpf_func_state *state;
struct bpf_reg_state *reg;
@@ -10139,10 +10139,10 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
}
}));
}
- } else if (meta.id) {
- err = release_reference(env, meta.id);
+ } else if (meta.ref_obj.id) {
+ err = release_reference(env, meta.ref_obj.id);
} else if (bpf_register_is_null(®s[meta.release_regno])) {
- /* meta.id can only be 0 if register that is meant to be
+ /* meta.ref_obj.id can only be 0 if register that is meant to be
* released is NULL, which must be > R0.
*/
err = 0;
@@ -10407,17 +10407,15 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
if (type_may_be_null(regs[BPF_REG_0].type))
regs[BPF_REG_0].id = ++env->id_gen;
- if (helper_multiple_ref_obj_use(func_id, meta.map.ptr)) {
- verifier_bug(env, "func %s#%d sets ref_obj_id more than once",
- func_id_name(func_id), func_id);
- return -EFAULT;
- }
-
if (is_ptr_cast_function(func_id) &&
- find_reference_state(env->cur_state, meta.id)) {
+ find_reference_state(env->cur_state, meta.ref_obj.id)) {
struct bpf_verifier_state *branch;
struct bpf_reg_state *r0;
+ err = validate_ref_obj(env, &meta.ref_obj);
+ if (err)
+ return err;
+
/*
* In order for a release of any of the original or cast pointers
* to invalidate all other pointers, reuse the same reference id for
@@ -10435,7 +10433,7 @@ static int check_helper_call(struct bpf_verifier_env *env, struct bpf_insn *insn
r0->type = SCALAR_VALUE;
regs[BPF_REG_0].type &= ~PTR_MAYBE_NULL;
- regs[BPF_REG_0].id = meta.id;
+ regs[BPF_REG_0].id = meta.ref_obj.id;
} else if (is_acquire_function(func_id, meta.map.ptr)) {
int id = acquire_reference(env, insn_idx, 0);
@@ -11909,13 +11907,13 @@ static int check_kfunc_args(struct bpf_verifier_env *env, struct bpf_kfunc_call_
}
if (reg_is_referenced(env, reg)) {
- if (is_kfunc_release(meta) && meta->id) {
- verifier_bug(env, "more than one arg with referenced id %s %u %u",
- reg_arg_name(env, argno), reg->id,
- meta->id);
+ if (is_kfunc_release(meta) && meta->ref_obj.cnt) {
+ verbose(env, "more than one arg with referenced id %s %u %u",
+ reg_arg_name(env, argno), reg->id,
+ meta->ref_obj.id);
return -EFAULT;
}
- meta->id = reg->id;
+ update_ref_obj(&meta->ref_obj, reg);
if (is_kfunc_release(meta)) {
if (regno < 0) {
verbose(env, "%s release arg cannot be a stack argument\n",
@@ -12098,7 +12096,7 @@ static int check_kfunc_args(struct bpf_verifier_env *env, struct bpf_kfunc_call_
}
ret = process_dynptr_func(env, reg, argno, insn_idx, dynptr_arg_type,
- meta->id, &meta->dynptr);
+ &meta->ref_obj, &meta->dynptr);
if (ret < 0)
return ret;
break;
@@ -13042,8 +13040,12 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
regs[BPF_REG_0].type |= MEM_RDONLY;
/* Ensures we don't access the memory after a release_reference() */
- if (meta.id)
- regs[BPF_REG_0].parent_id = meta.id;
+ if (meta.ref_obj.id) {
+ err = validate_ref_obj(env, &meta.ref_obj);
+ if (err)
+ return err;
+ regs[BPF_REG_0].parent_id = meta.ref_obj.id;
+ }
if (is_kfunc_rcu_protected(&meta))
regs[BPF_REG_0].type |= MEM_RCU;
diff --git a/tools/testing/selftests/bpf/progs/test_ringbuf_map_key.c b/tools/testing/selftests/bpf/progs/test_ringbuf_map_key.c
index 21bb7da90ea5..0efafa927a3d 100644
--- a/tools/testing/selftests/bpf/progs/test_ringbuf_map_key.c
+++ b/tools/testing/selftests/bpf/progs/test_ringbuf_map_key.c
@@ -35,7 +35,7 @@ SEC("fentry/" SYS_PREFIX "sys_getpgid")
int test_ringbuf_mem_map_key(void *ctx)
{
int cur_pid = bpf_get_current_pid_tgid() >> 32;
- struct sample *sample, sample_copy;
+ struct sample *sample;
int *lookup_val;
if (cur_pid != pid)
@@ -55,16 +55,11 @@ int test_ringbuf_mem_map_key(void *ctx)
lookup_val = (int *)bpf_map_lookup_elem(&hash_map, sample);
__sink(lookup_val);
- /* workaround - memcpy is necessary so that verifier doesn't
- * complain with:
- * verifier internal error: more than one arg with ref_obj_id R3
- * when trying to do bpf_map_update_elem(&hash_map, sample, &sample->seq, BPF_ANY);
- *
+ /*
* Since bpf_map_lookup_elem above uses 'sample' as key, test using
* sample field as value below
*/
- __builtin_memcpy(&sample_copy, sample, sizeof(struct sample));
- bpf_map_update_elem(&hash_map, &sample_copy, &sample->seq, BPF_ANY);
+ bpf_map_update_elem(&hash_map, sample, &sample->seq, BPF_ANY);
bpf_ringbuf_submit(sample, 0);
return 0;
diff --git a/tools/testing/selftests/bpf/verifier/calls.c b/tools/testing/selftests/bpf/verifier/calls.c
index 0bb4337552c8..42d523a21a43 100644
--- a/tools/testing/selftests/bpf/verifier/calls.c
+++ b/tools/testing/selftests/bpf/verifier/calls.c
@@ -2410,27 +2410,3 @@
.errstr_unpriv = "",
.prog_type = BPF_PROG_TYPE_CGROUP_SKB,
},
-{
- "calls: several args with ref_obj_id",
- .insns = {
- /* Reserve at least sizeof(struct iphdr) bytes in the ring buffer.
- * With a smaller size, the verifier would reject the call to
- * bpf_tcp_raw_gen_syncookie_ipv4 before we can reach the
- * ref_obj_id error.
- */
- BPF_MOV64_IMM(BPF_REG_2, 20),
- BPF_MOV64_IMM(BPF_REG_3, 0),
- BPF_LD_MAP_FD(BPF_REG_1, 0),
- BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0, BPF_FUNC_ringbuf_reserve),
- /* if r0 == 0 goto <exit> */
- BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 3),
- BPF_MOV64_REG(BPF_REG_1, BPF_REG_0),
- BPF_MOV64_REG(BPF_REG_2, BPF_REG_0),
- BPF_RAW_INSN(BPF_JMP | BPF_CALL, 0, 0, 0, BPF_FUNC_tcp_raw_gen_syncookie_ipv4),
- BPF_EXIT_INSN(),
- },
- .fixup_map_ringbuf = { 2 },
- .result = REJECT,
- .errstr = "more than one arg with ref_obj_id",
- .prog_type = BPF_PROG_TYPE_SCHED_CLS,
-},
--
2.53.0-Meta
next prev parent reply other threads:[~2026-05-29 1:49 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-05-29 1:49 [PATCH bpf-next v6 00/13] Refactor verifier object relationship tracking Amery Hung
2026-05-29 1:49 ` [PATCH bpf-next v6 01/13] bpf: Simplify mark_stack_slot_obj_read() and callers Amery Hung
2026-05-29 1:49 ` [PATCH bpf-next v6 02/13] bpf: Unify dynptr handling in the verifier Amery Hung
2026-05-29 2:39 ` bot+bpf-ci
2026-05-29 1:49 ` [PATCH bpf-next v6 03/13] bpf: Assign reg->id when getting referenced kptr from ctx Amery Hung
2026-05-29 1:49 ` [PATCH bpf-next v6 04/13] bpf: Preserve reg->id of pointer objects after null-check Amery Hung
2026-05-29 1:49 ` [PATCH bpf-next v6 05/13] bpf: Refactor object relationship tracking and fix dynptr UAF bug Amery Hung
2026-05-29 2:54 ` bot+bpf-ci
2026-05-29 1:49 ` [PATCH bpf-next v6 06/13] bpf: Remove redundant dynptr arg check for helper Amery Hung
2026-05-29 1:49 ` Amery Hung [this message]
2026-05-29 2:40 ` [PATCH bpf-next v6 07/13] bpf: Unify referenced object tracking in verifier bot+bpf-ci
2026-05-29 1:49 ` [PATCH bpf-next v6 08/13] bpf: Unify release handling for helpers and kfuncs Amery Hung
2026-05-29 1:49 ` [PATCH bpf-next v6 09/13] bpf: Fix dynptr ref counting to scan all call frames Amery Hung
2026-05-29 1:49 ` [PATCH bpf-next v6 10/13] selftests/bpf: Test creating dynptr from dynptr data and slice Amery Hung
2026-05-29 1:49 ` [PATCH bpf-next v6 11/13] selftests/bpf: Test using slice after invalidating dynptr clone Amery Hung
2026-05-29 2:39 ` bot+bpf-ci
2026-05-29 1:49 ` [PATCH bpf-next v6 12/13] selftests/bpf: Test using file dynptr after the reference on file is dropped Amery Hung
2026-05-29 1:49 ` [PATCH bpf-next v6 13/13] selftests/bpf: Test using dynptr after freeing the underlying object Amery Hung
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20260529014936.2811085-8-ameryhung@gmail.com \
--to=ameryhung@gmail.com \
--cc=alexei.starovoitov@gmail.com \
--cc=andrii@kernel.org \
--cc=bpf@vger.kernel.org \
--cc=daniel@iogearbox.net \
--cc=eddyz87@gmail.com \
--cc=kernel-team@meta.com \
--cc=martin.lau@kernel.org \
--cc=memxor@gmail.com \
--cc=mykyta.yatsenko5@gmail.com \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox