Netdev List
 help / color / mirror / Atom feed
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 v5 10/14] bpf: Fix dynptr ref counting to scan all call frames
Date: Tue, 19 May 2026 11:13:08 -0700	[thread overview]
Message-ID: <20260519181314.2731658-11-ameryhung@gmail.com> (raw)
In-Reply-To: <20260519181314.2731658-1-ameryhung@gmail.com>

When checking whether a referenced dynptr can be overwritten,
destroy_if_dynptr_stack_slot only counted sibling dynptrs in the
current call frame. If a clone sharing the same virtual ref parent
existed in a different frame (e.g., passed to a subprog), it would
not be counted, causing the verifier to incorrectly reject the
overwrite with "cannot overwrite referenced dynptr".

Fix by extracting the counting into dynptr_ref_cnt() which uses
bpf_for_each_reg_in_vstate_mask() to scan dynptr stack slots across
all call frames.

Fixes: 017f5c4ef73c ("bpf: Allow overwriting referenced dynptr when refcnt > 1")
Reported-by: Eduard Zingerman <eddyz87@gmail.com>
Signed-off-by: Amery Hung <ameryhung@gmail.com>
---
 kernel/bpf/verifier.c                         | 52 +++++++++++--------
 .../selftests/bpf/progs/wakeup_source_fail.c  |  2 +-
 2 files changed, 30 insertions(+), 24 deletions(-)

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index b7984c42ca14..97eb8033d732 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -786,10 +786,29 @@ static void mark_reg_invalid(const struct bpf_verifier_env *env, struct bpf_reg_
 		__mark_reg_unknown(env, reg);
 }
 
+static int dynptr_ref_cnt(struct bpf_verifier_env *env, int v_parent_id)
+{
+	struct bpf_stack_state *stack;
+	struct bpf_func_state *state;
+	struct bpf_reg_state *reg;
+	int ref_cnt = 0;
+
+	bpf_for_each_reg_in_vstate_mask(env->cur_state, state, reg, stack, 1 << STACK_DYNPTR, ({
+		if (!stack || stack->slot_type[0] != STACK_DYNPTR)
+			continue;
+		if (!stack->spilled_ptr.dynptr.first_slot)
+			continue;
+		if (stack->spilled_ptr.parent_id == v_parent_id)
+			ref_cnt++;
+	}));
+
+	return ref_cnt;
+}
+
 static int destroy_if_dynptr_stack_slot(struct bpf_verifier_env *env,
 				        struct bpf_func_state *state, int spi)
 {
-	int i, err = 0;
+	int err = 0;
 
 	/* We always ensure that STACK_DYNPTR is never set partially,
 	 * hence just checking for slot_type[0] is enough. This is
@@ -803,28 +822,15 @@ static int destroy_if_dynptr_stack_slot(struct bpf_verifier_env *env,
 	if (!state->stack[spi].spilled_ptr.dynptr.first_slot)
 		spi = spi + 1;
 
-	if (dynptr_type_referenced(state->stack[spi].spilled_ptr.dynptr.type)) {
-		int v_parent_id = state->stack[spi].spilled_ptr.parent_id;
-		int ref_cnt = 0;
-
-		/*
-		 * A referenced dynptr can be overwritten only if there is at
-		 * least one other dynptr sharing the same virtual ref parent,
-		 * ensuring the reference can still be properly released.
-		 */
-		for (i = 0; i < state->allocated_stack / BPF_REG_SIZE; i++) {
-			if (state->stack[i].slot_type[0] != STACK_DYNPTR)
-				continue;
-			if (!state->stack[i].spilled_ptr.dynptr.first_slot)
-				continue;
-			if (state->stack[i].spilled_ptr.parent_id == v_parent_id)
-				ref_cnt++;
-		}
-
-		if (ref_cnt <= 1) {
-			verbose(env, "cannot overwrite referenced dynptr\n");
-			return -EINVAL;
-		}
+	/*
+	 * A referenced dynptr can be overwritten only if there is at
+	 * least one other dynptr sharing the same virtual ref parent,
+	 * ensuring the reference can still be properly released.
+	 */
+	if (dynptr_type_referenced(state->stack[spi].spilled_ptr.dynptr.type) &&
+	    dynptr_ref_cnt(env, state->stack[spi].spilled_ptr.parent_id) <= 1) {
+		verbose(env, "cannot overwrite referenced dynptr\n");
+		return -EINVAL;
 	}
 
 	/* Invalidate the dynptr and any derived slices */
diff --git a/tools/testing/selftests/bpf/progs/wakeup_source_fail.c b/tools/testing/selftests/bpf/progs/wakeup_source_fail.c
index b8bbb61d4d4e..d4d0f1610853 100644
--- a/tools/testing/selftests/bpf/progs/wakeup_source_fail.c
+++ b/tools/testing/selftests/bpf/progs/wakeup_source_fail.c
@@ -42,7 +42,7 @@ int wakeup_source_access_lock_fields(void *ctx)
 }
 
 SEC("syscall")
-__failure __msg("type=scalar expected=fp")
+__failure __msg("release kfunc bpf_wakeup_sources_read_unlock expects referenced PTR_TO_BTF_ID passed to R1")
 int wakeup_source_unlock_no_lock(void *ctx)
 {
 	struct bpf_ws_lock *lock = (void *)0x1;
-- 
2.53.0-Meta


  parent reply	other threads:[~2026-05-19 18:13 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-05-19 18:12 [PATCH bpf-next v5 00/14] Refactor verifier object relationship tracking Amery Hung
2026-05-19 18:12 ` [PATCH bpf-next v5 01/14] bpf: Simplify mark_stack_slot_obj_read() and callers Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 02/14] bpf: Unify dynptr handling in the verifier Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 03/14] bpf: Assign reg->id when getting referenced kptr from ctx Amery Hung
2026-05-19 18:56   ` bot+bpf-ci
2026-05-19 20:17     ` Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 04/14] bpf: Preserve reg->id of pointer objects after null-check Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 05/14] bpf: Refactor object relationship tracking and fix dynptr UAF bug Amery Hung
2026-05-20 21:47   ` Eduard Zingerman
2026-05-21  7:18     ` Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 06/14] bpf: Remove redundant dynptr arg check for helper Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 07/14] bpf: Unify referenced object tracking in verifier Amery Hung
2026-05-20 22:28   ` Eduard Zingerman
2026-05-19 18:13 ` [PATCH bpf-next v5 08/14] bpf: Unify release handling for helpers and kfuncs Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 09/14] bpf: Fold ref_obj_id into id and introduce virtual references Amery Hung
2026-05-19 18:13 ` Amery Hung [this message]
2026-05-20 19:59   ` [PATCH bpf-next v5 10/14] bpf: Fix dynptr ref counting to scan all call frames Eduard Zingerman
2026-05-20 22:41     ` Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 11/14] selftests/bpf: Test creating dynptr from dynptr data and slice Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 12/14] selftests/bpf: Test using dynptr after freeing the underlying object Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 13/14] selftests/bpf: Test using slice after invalidating dynptr clone Amery Hung
2026-05-19 18:13 ` [PATCH bpf-next v5 14/14] selftests/bpf: Test using file dynptr after the reference on file is dropped 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=20260519181314.2731658-11-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