[PATCH v2] bpf: reject direct access to nullable PTR_TO_BUF pointers

[PATCH v2] bpf: reject direct access to nullable PTR_TO_BUF pointers

[Qi Tang]

check_mem_access() matches PTR_TO_BUF via base_type() which strips
PTR_MAYBE_NULL, allowing direct dereference without a null check.

Map iterator ctx->key and ctx->value are PTR_TO_BUF | PTR_MAYBE_NULL.
On stop callbacks these are NULL, causing a kernel NULL dereference.

Add a type_may_be_null() guard to the PTR_TO_BUF branch, matching the
existing PTR_TO_BTF_ID pattern.

  BUG: kernel NULL pointer dereference, address: 0000000000000000
  Oops: Oops: 0000 [#1] SMP KASAN NOPTI
  RIP: 0010:bpf_prog_5f0e2be830ac3243_null_deref_iter+0x10/0x25
  Call Trace:
   bpf_iter_run_prog+0x1c2/0x2e0
   __bpf_hash_map_seq_show+0x120/0x180
   bpf_seq_read+0x29c/0x530
   vfs_read+0x179/0x930
   ksys_read+0xef/0x1c0
   do_syscall_64+0xe0/0x1290
   entry_SYSCALL_64_after_hwframe+0x76/0x7e
  Kernel panic - not syncing: Fatal exception

Fixes: 20b2aff4bc15 ("bpf: Introduce MEM_RDONLY flag")
Signed-off-by: Qi Tang <tpluszz77@gmail.com>
---
Changes in v2:
  - Add selftest

Link: https://lore.kernel.org/all/20260331172720.29938-1-tpluszz77@gmail.com/
---
 kernel/bpf/verifier.c                         |  3 +-
 .../selftests/bpf/progs/iter_buf_null_fail.c  | 49 +++++++++++++++++++
 2 files changed, 51 insertions(+), 1 deletion(-)
 create mode 100644 tools/testing/selftests/bpf/progs/iter_buf_null_fail.c

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 937c9adf1b3d..17850836943b 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -7905,7 +7905,8 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, u32 regn
 	} else if (reg->type == CONST_PTR_TO_MAP) {
 		err = check_ptr_to_map_access(env, regs, regno, off, size, t,
 					      value_regno);
-	} else if (base_type(reg->type) == PTR_TO_BUF) {
+	} else if (base_type(reg->type) == PTR_TO_BUF &&
+		   !type_may_be_null(reg->type)) {
 		bool rdonly_mem = type_is_rdonly_mem(reg->type);
 		u32 *max_access;
 
diff --git a/tools/testing/selftests/bpf/progs/iter_buf_null_fail.c b/tools/testing/selftests/bpf/progs/iter_buf_null_fail.c
new file mode 100644
index 000000000000..997389f01ed3
--- /dev/null
+++ b/tools/testing/selftests/bpf/progs/iter_buf_null_fail.c
@@ -0,0 +1,49 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2026 Qi Tang */
+
+#include <vmlinux.h>
+#include <bpf/bpf_helpers.h>
+#include "bpf_misc.h"
+
+char _license[] SEC("license") = "GPL";
+
+struct {
+	__uint(type, BPF_MAP_TYPE_HASH);
+	__uint(max_entries, 1);
+	__type(key, __u32);
+	__type(value, __u64);
+} hashmap SEC(".maps");
+
+/*
+ * Verify that the verifier rejects direct access to nullable PTR_TO_BUF
+ * (ctx->key) without a null check.  On the iterator stop callback,
+ * ctx->key is NULL, so unconditional access would be a NULL deref.
+ */
+SEC("iter/bpf_map_elem")
+__failure __msg("invalid mem access")
+int iter_buf_null_deref(struct bpf_iter__bpf_map_elem *ctx)
+{
+	/* ctx->key is PTR_TO_BUF | PTR_MAYBE_NULL | MEM_RDONLY.
+	 * Direct access without null check must be rejected.
+	 */
+	volatile __u32 v = *(__u32 *)ctx->key;
+
+	return 0;
+}
+
+/*
+ * Verify that access after a null check is still accepted.
+ */
+SEC("iter/bpf_map_elem")
+__success
+int iter_buf_null_check_ok(struct bpf_iter__bpf_map_elem *ctx)
+{
+	__u32 *key = ctx->key;
+
+	if (!key)
+		return 0;
+
+	volatile __u32 v = *key;
+
+	return 0;
+}
-- 
2.43.0

Re: [PATCH v2] bpf: reject direct access to nullable PTR_TO_BUF pointers

[Kumar Kartikeya Dwivedi]

On Wed, 1 Apr 2026 at 16:38, Qi Tang <tpluszz77@gmail.com> wrote:
>
> check_mem_access() matches PTR_TO_BUF via base_type() which strips
> PTR_MAYBE_NULL, allowing direct dereference without a null check.
>
> Map iterator ctx->key and ctx->value are PTR_TO_BUF | PTR_MAYBE_NULL.
> On stop callbacks these are NULL, causing a kernel NULL dereference.
>
> Add a type_may_be_null() guard to the PTR_TO_BUF branch, matching the
> existing PTR_TO_BTF_ID pattern.

Just this much info is enough, please drop the stack trace from the commit log.

>
>   BUG: kernel NULL pointer dereference, address: 0000000000000000
>   Oops: Oops: 0000 [#1] SMP KASAN NOPTI
>   RIP: 0010:bpf_prog_5f0e2be830ac3243_null_deref_iter+0x10/0x25
>   Call Trace:
>    bpf_iter_run_prog+0x1c2/0x2e0
>    __bpf_hash_map_seq_show+0x120/0x180
>    bpf_seq_read+0x29c/0x530
>    vfs_read+0x179/0x930
>    ksys_read+0xef/0x1c0
>    do_syscall_64+0xe0/0x1290
>    entry_SYSCALL_64_after_hwframe+0x76/0x7e
>   Kernel panic - not syncing: Fatal exception
>
> Fixes: 20b2aff4bc15 ("bpf: Introduce MEM_RDONLY flag")
> Signed-off-by: Qi Tang <tpluszz77@gmail.com>
> ---

The fix looks fine, but please split the kernel fix and selftests into
separate patches.
E.g. look at this series:
https://lore.kernel.org/bpf/20260401-bpf_map_torn_writes-v1-0-782d071c55e7@meta.com
Also, use the subject prefix [PATCH bpf ...] for v3.

pw-bot: cr

> [...]