bpf: add plumbing for file-backed dynptr

Add the necessary verifier plumbing for the new file-backed dynptr type.
Introduce two kfuncs for its lifecycle management:
 * bpf_dynptr_from_file() for initialization
 * bpf_dynptr_file_discard() for destruction

Currently there is no mechanism for kfunc to release dynptr, this patch
add one:
 * Dynptr release function sets meta->release_regno
 * Call unmark_stack_slots_dynptr() if meta->release_regno is set and
 dynptr ref_obj_id is set as well.

Signed-off-by: Mykyta Yatsenko <[email protected]>
Acked-by: Eduard Zingerman <[email protected]>

Author: mykyta5

include/linux/bpf.h

@@ -792,12 +792,15 @@ enum bpf_type_flag {
 	/* DYNPTR points to skb_metadata_end()-skb_metadata_len() */
 	DYNPTR_TYPE_SKB_META	= BIT(19 + BPF_BASE_TYPE_BITS),
 
+	/* DYNPTR points to file */
+	DYNPTR_TYPE_FILE	= BIT(20 + BPF_BASE_TYPE_BITS),
+
 	__BPF_TYPE_FLAG_MAX,
 	__BPF_TYPE_LAST_FLAG	= __BPF_TYPE_FLAG_MAX - 1,
 };
 
 #define DYNPTR_TYPE_FLAG_MASK	(DYNPTR_TYPE_LOCAL | DYNPTR_TYPE_RINGBUF | DYNPTR_TYPE_SKB \
-				 | DYNPTR_TYPE_XDP | DYNPTR_TYPE_SKB_META)
+				 | DYNPTR_TYPE_XDP | DYNPTR_TYPE_SKB_META | DYNPTR_TYPE_FILE)
 
 /* Max number of base types. */
 #define BPF_BASE_TYPE_LIMIT	(1UL << BPF_BASE_TYPE_BITS)
@@ -1385,6 +1388,8 @@ enum bpf_dynptr_type {
 	BPF_DYNPTR_TYPE_XDP,
 	/* Points to skb_metadata_end()-skb_metadata_len() */
 	BPF_DYNPTR_TYPE_SKB_META,
+	/* Underlying data is a file */
+	BPF_DYNPTR_TYPE_FILE,
 };
 
 int bpf_dynptr_check_size(u64 size);

kernel/bpf/helpers.c

@@ -4252,6 +4252,16 @@ __bpf_kfunc int bpf_task_work_schedule_resume(struct task_struct *task, struct b
 	return bpf_task_work_schedule(task, tw, map__map, callback, aux__prog, TWA_RESUME);
 }
 
+__bpf_kfunc int bpf_dynptr_from_file(struct file *file, u32 flags, struct bpf_dynptr *ptr__uninit)
+{
+	return 0;
+}
+
+__bpf_kfunc int bpf_dynptr_file_discard(struct bpf_dynptr *dynptr)
+{
+	return 0;
+}
+
 __bpf_kfunc_end_defs();
 
 static void bpf_task_work_cancel_scheduled(struct irq_work *irq_work)
@@ -4429,6 +4439,8 @@ BTF_ID_FLAGS(func, bpf_cgroup_read_xattr, KF_RCU)
 BTF_ID_FLAGS(func, bpf_stream_vprintk, KF_TRUSTED_ARGS)
 BTF_ID_FLAGS(func, bpf_task_work_schedule_signal, KF_TRUSTED_ARGS)
 BTF_ID_FLAGS(func, bpf_task_work_schedule_resume, KF_TRUSTED_ARGS)
+BTF_ID_FLAGS(func, bpf_dynptr_from_file, KF_TRUSTED_ARGS)
+BTF_ID_FLAGS(func, bpf_dynptr_file_discard)
 BTF_KFUNCS_END(common_btf_ids)
 
 static const struct btf_kfunc_id_set common_kfunc_set = {

kernel/bpf/log.c

@@ -500,6 +500,8 @@ const char *dynptr_type_str(enum bpf_dynptr_type type)
 		return "xdp";
 	case BPF_DYNPTR_TYPE_SKB_META:
 		return "skb_meta";
+	case BPF_DYNPTR_TYPE_FILE:
+		return "file";
 	case BPF_DYNPTR_TYPE_INVALID:
 		return "<invalid>";
 	default:

kernel/bpf/verifier.c

@@ -692,6 +692,8 @@ static enum bpf_dynptr_type arg_to_dynptr_type(enum bpf_arg_type arg_type)
 		return BPF_DYNPTR_TYPE_XDP;
 	case DYNPTR_TYPE_SKB_META:
 		return BPF_DYNPTR_TYPE_SKB_META;
+	case DYNPTR_TYPE_FILE:
+		return BPF_DYNPTR_TYPE_FILE;
 	default:
 		return BPF_DYNPTR_TYPE_INVALID;
 	}
@@ -710,14 +712,16 @@ static enum bpf_type_flag get_dynptr_type_flag(enum bpf_dynptr_type type)
 		return DYNPTR_TYPE_XDP;
 	case BPF_DYNPTR_TYPE_SKB_META:
 		return DYNPTR_TYPE_SKB_META;
+	case BPF_DYNPTR_TYPE_FILE:
+		return DYNPTR_TYPE_FILE;
 	default:
 		return 0;
 	}
 }
 
 static bool dynptr_type_refcounted(enum bpf_dynptr_type type)
 {
-	return type == BPF_DYNPTR_TYPE_RINGBUF;
+	return type == BPF_DYNPTR_TYPE_RINGBUF || type == BPF_DYNPTR_TYPE_FILE;
 }
 
 static void __mark_dynptr_reg(struct bpf_reg_state *reg,
@@ -12291,6 +12295,8 @@ enum special_kfunc_type {
 	KF_bpf_res_spin_unlock,
 	KF_bpf_res_spin_lock_irqsave,
 	KF_bpf_res_spin_unlock_irqrestore,
+	KF_bpf_dynptr_from_file,
+	KF_bpf_dynptr_file_discard,
 	KF___bpf_trap,
 	KF_bpf_task_work_schedule_signal,
 	KF_bpf_task_work_schedule_resume,
@@ -12363,6 +12369,8 @@ BTF_ID(func, bpf_res_spin_lock)
 BTF_ID(func, bpf_res_spin_unlock)
 BTF_ID(func, bpf_res_spin_lock_irqsave)
 BTF_ID(func, bpf_res_spin_unlock_irqrestore)
+BTF_ID(func, bpf_dynptr_from_file)
+BTF_ID(func, bpf_dynptr_file_discard)
 BTF_ID(func, __bpf_trap)
 BTF_ID(func, bpf_task_work_schedule_signal)
 BTF_ID(func, bpf_task_work_schedule_resume)
@@ -13326,6 +13334,11 @@ static int check_kfunc_args(struct bpf_verifier_env *env, struct bpf_kfunc_call_
 				dynptr_arg_type |= DYNPTR_TYPE_XDP;
 			} else if (meta->func_id == special_kfunc_list[KF_bpf_dynptr_from_skb_meta]) {
 				dynptr_arg_type |= DYNPTR_TYPE_SKB_META;
+			} else if (meta->func_id == special_kfunc_list[KF_bpf_dynptr_from_file]) {
+				dynptr_arg_type |= DYNPTR_TYPE_FILE;
+			} else if (meta->func_id == special_kfunc_list[KF_bpf_dynptr_file_discard]) {
+				dynptr_arg_type |= DYNPTR_TYPE_FILE;
+				meta->release_regno = regno;
 			} else if (meta->func_id == special_kfunc_list[KF_bpf_dynptr_clone] &&
 				   (dynptr_arg_type & MEM_UNINIT)) {
 				enum bpf_dynptr_type parent_type = meta->initialized_dynptr.type;
@@ -14006,12 +14019,18 @@ static int check_kfunc_call(struct bpf_verifier_env *env, struct bpf_insn *insn,
 	 * PTR_TO_BTF_ID in bpf_kfunc_arg_meta, do the release now.
 	 */
 	if (meta.release_regno) {
-		err = release_reference(env, regs[meta.release_regno].ref_obj_id);
-		if (err) {
-			verbose(env, "kfunc %s#%d reference has not been acquired before\n",
-				func_name, meta.func_id);
-			return err;
+		struct bpf_reg_state *reg = &regs[meta.release_regno];
+
+		if (meta.initialized_dynptr.ref_obj_id) {
+			err = unmark_stack_slots_dynptr(env, reg);
+		} else {
+			err = release_reference(env, reg->ref_obj_id);
+			if (err)
+				verbose(env, "kfunc %s#%d reference has not been acquired before\n",
+					func_name, meta.func_id);
 		}
+		if (err)
+			return err;
 	}
 
 	if (meta.func_id == special_kfunc_list[KF_bpf_list_push_front_impl] ||