Skip to contentSkip to conversation

cls_flower: call nla_ok() before nla_next()

PATCH Superseded Cong Wang 2 participants 6 messages Jan 14, 2021

Subsystems (3)

v3 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Cong Wang

author
#0 lore raw

PATCH v3 1/1 cls_flower: call nla_ok() before nla_next()

From: Cong Wang <[email protected]>

fl_set_enc_opt() simply checks if there are still bytes left to parse, but this is not sufficent as syzbot seems to be able to generate malformatted netlink messages. nla_ok() is more strict so should be used to validate the next nlattr here.

And nla_validate_nested_deprecated() has less strict check too, it is probably too late to switch to the strict version, but we can just call nla_ok() too after it.

Reported-and-tested-by [email protected]
Fixes 0a6e77784f49 ("net/sched: allow flower to match tunnel options")
Fixes 79b1011cb33d ("net: sched: allow flower to match erspan options")
Cc Jamal Hadi Salim <[email protected]>
Cc Xin Long <[email protected]>
Cc Jiri Pirko <[email protected]>
Cc Jakub Kicinski <[email protected]>
---
 net/sched/cls_flower.c | 22 +++++++++++++---------
 1 file changed, 13 insertions(+), 9 deletions(-)
net/sched/cls_flower.c +13 -9
index 1319986693fc..280eab0c833f 100644
@@ -1272,6 +1272,10 @@ static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key,
 
 		nla_opt_msk = nla_data(tb[TCA_FLOWER_KEY_ENC_OPTS_MASK]);
 		msk_depth = nla_len(tb[TCA_FLOWER_KEY_ENC_OPTS_MASK]);
+		if (!nla_ok(nla_opt_msk, msk_depth)) {
+			NL_SET_ERR_MSG(extack, "Invalid attribute for masks");
+			return -EINVAL;
+		}
 	}
 
 	nla_for_each_attr(nla_opt_key, nla_enc_key,
@@ -1307,9 +1311,6 @@ static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key,
 				NL_SET_ERR_MSG(extack, "Key and mask miss aligned");
 				return -EINVAL;
 			}
-
-			if (msk_depth)
-				nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
 			break;
 		case TCA_FLOWER_KEY_ENC_OPTS_VXLAN:
 			if (key->enc_opts.dst_opt_type) {
@@ -1340,9 +1341,6 @@ static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key,
 				NL_SET_ERR_MSG(extack, "Key and mask miss aligned");
 				return -EINVAL;
 			}
-
-			if (msk_depth)
-				nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
 			break;
 		case TCA_FLOWER_KEY_ENC_OPTS_ERSPAN:
 			if (key->enc_opts.dst_opt_type) {
@@ -1373,14 +1371,20 @@ static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key,
 				NL_SET_ERR_MSG(extack, "Key and mask miss aligned");
 				return -EINVAL;
 			}
-
-			if (msk_depth)
-				nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
 			break;
 		default:
 			NL_SET_ERR_MSG(extack, "Unknown tunnel option type");
 			return -EINVAL;
 		}
+
+		if (!nla_opt_msk)
+			continue;
+
+		if (!nla_ok(nla_opt_msk, msk_depth)) {
+			NL_SET_ERR_MSG(extack, "A mask attribute is invalid");
+			return -EINVAL;
+		}
+		nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
 	}
 
 	return 0;
Signature · 1 line
2.25.1
maintainer
#1 lore raw

net/sched/cls_flower.c static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key, · line 1381

Quoted 10 lines · On Thu, 14 Jan 2021 13:07:49 -0800 Cong Wang wrote:
net/sched/cls_flower.c +3 -2
-			if (msk_depth)
-				nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
 			break;
 		default:
 			NL_SET_ERR_MSG(extack, "Unknown tunnel option type");
 			return -EINVAL;
 		}
+
+		if (!nla_opt_msk)
+			continue;

Why the switch from !msk_depth to !nla_opt_msk?

Seems like previously providing masks for only subset of options would have worked.

author
#2 lore raw
Quoted 13 lines · Jakub Kicinski

net/sched/cls_flower.c static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key, · line 1381

Quoted 10 lines · On Thu, 14 Jan 2021 13:07:49 -0800 Cong Wang wrote:
net/sched/cls_flower.c +3 -2
-                     if (msk_depth)
-                             nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
                      break;
              default:
                      NL_SET_ERR_MSG(extack, "Unknown tunnel option type");
                      return -EINVAL;
              }
+
+             if (!nla_opt_msk)
+                     continue;
Quoted 1 line

Why the switch from !msk_depth to !nla_opt_msk?

It is the same, when nla_opt_msk is NULL, msk_depth is 0. Checking nla_opt_msk is NULL is more readable to express that mask is not provided.

Quoted 3 lines

Seems like previously providing masks for only subset of options would have worked.

I don't think so, every type has this check:

if (key->enc_opts.len != mask->enc_opts.len) {
NL_SET_ERR_MSG(extack, "Key and mask

miss aligned");
                                return -EINVAL;
                        }

which guarantees the numbers are aligned.

Thanks.

maintainer
#3 lore raw
Quoted 28 lines · On Thu, 14 Jan 2021 13:57:13 -0800 Cong Wang wrote:
Quoted 13 lines · Jakub Kicinski

net/sched/cls_flower.c static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key, · line 1381

Quoted 10 lines · On Thu, 14 Jan 2021 13:07:49 -0800 Cong Wang wrote:
net/sched/cls_flower.c +3 -2
-                     if (msk_depth)
-                             nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
                      break;
              default:
                      NL_SET_ERR_MSG(extack, "Unknown tunnel option type");
                      return -EINVAL;
              }
+
+             if (!nla_opt_msk)
+                     continue;  
Quoted 1 line

Why the switch from !msk_depth to !nla_opt_msk?

It is the same, when nla_opt_msk is NULL, msk_depth is 0. Checking nla_opt_msk is NULL is more readable to express that mask is not provided.

Quoted 3 lines

Seems like previously providing masks for only subset of options would have worked.

I don't think so, every type has this check:

if (key->enc_opts.len != mask->enc_opts.len) {
NL_SET_ERR_MSG(extack, "Key and mask

Quoted 7 lines
miss aligned");
                                return -EINVAL;
                        }

which guarantees the numbers are aligned.

Thanks.

static int fl_set_vxlan_opt(const struct nlattr *nla, struct fl_flow_key *key,
int depth, int option_len,
struct netlink_ext_ack *extack)

{
	struct nlattr *tb[TCA_FLOWER_KEY_ENC_OPT_VXLAN_MAX + 1];
	struct vxlan_metadata *md;
	int err;

	md = (struct vxlan_metadata *)&key->enc_opts.data[key->enc_opts.len];
	memset(md, 0xff, sizeof(*md));

	if (!depth)
		return sizeof(*md);
		^^^^^^^^^^^^^^^^^^^

The mask is filled with all 1s if attribute is not provided.

author
#4 lore raw
Quoted 53 lines · Jakub Kicinski
Quoted 28 lines · On Thu, 14 Jan 2021 13:57:13 -0800 Cong Wang wrote:
Quoted 13 lines · Jakub Kicinski

net/sched/cls_flower.c static int fl_set_enc_opt(struct nlattr **tb, struct fl_flow_key *key, · line 1381

Quoted 10 lines · On Thu, 14 Jan 2021 13:07:49 -0800 Cong Wang wrote:
net/sched/cls_flower.c +3 -2
-                     if (msk_depth)
-                             nla_opt_msk = nla_next(nla_opt_msk, &msk_depth);
                      break;
              default:
                      NL_SET_ERR_MSG(extack, "Unknown tunnel option type");
                      return -EINVAL;
              }
+
+             if (!nla_opt_msk)
+                     continue;
Quoted 1 line

Why the switch from !msk_depth to !nla_opt_msk?

It is the same, when nla_opt_msk is NULL, msk_depth is 0. Checking nla_opt_msk is NULL is more readable to express that mask is not provided.

Quoted 3 lines

Seems like previously providing masks for only subset of options would have worked.

I don't think so, every type has this check:

if (key->enc_opts.len != mask->enc_opts.len) {
NL_SET_ERR_MSG(extack, "Key and mask

Quoted 7 lines
miss aligned");
                                return -EINVAL;
                        }

which guarantees the numbers are aligned.

Thanks.

static int fl_set_vxlan_opt(const struct nlattr *nla, struct fl_flow_key *key,
int depth, int option_len,
struct netlink_ext_ack *extack)

{
        struct nlattr *tb[TCA_FLOWER_KEY_ENC_OPT_VXLAN_MAX + 1];
        struct vxlan_metadata *md;
        int err;

        md = (struct vxlan_metadata *)&key->enc_opts.data[key->enc_opts.len];
        memset(md, 0xff, sizeof(*md));

        if (!depth)
                return sizeof(*md);
                ^^^^^^^^^^^^^^^^^^^
Quoted 1 line

The mask is filled with all 1s if attribute is not provided.

Hmm, then what is the length comparison check for?

fl_set_vxlan_opt() either turns negative or sizeof(*md), and negitve is already checked when it returns, so when we hit the length comparison it is always equal. So it must be redundant.

(Note, I am only talking about the vxlan case you pick here, because the geneve case is different, as it returns different sizes.)

Thanks.

maintainer
#5 lore raw
Quoted 17 lines · On Thu, 14 Jan 2021 14:50:04 -0800 Cong Wang wrote:
Quoted 15 lines

static int fl_set_vxlan_opt(const struct nlattr *nla, struct fl_flow_key *key,
int depth, int option_len,
struct netlink_ext_ack *extack)

{
        struct nlattr *tb[TCA_FLOWER_KEY_ENC_OPT_VXLAN_MAX + 1];
        struct vxlan_metadata *md;
        int err;

        md = (struct vxlan_metadata *)&key->enc_opts.data[key->enc_opts.len];
        memset(md, 0xff, sizeof(*md));

        if (!depth)
                return sizeof(*md);
                ^^^^^^^^^^^^^^^^^^^
Quoted 1 line

The mask is filled with all 1s if attribute is not provided.

Quoted 8 lines

Hmm, then what is the length comparison check for?

fl_set_vxlan_opt() either turns negative or sizeof(*md), and negitve is already checked when it returns, so when we hit the length comparison it is always equal. So it must be redundant.

(Note, I am only talking about the vxlan case you pick here, because the geneve case is different, as it returns different sizes.)

Good question 🤷🏻‍♂️🤷🏻‍♂️ I'd guess GENEVE came first and and then people copy / pasted the code structure other tunnel types?