Skip to contentSkip to conversation

netfilter: xtables: stricter ruleset validation

PATCH In review Florian Westphal 4 participants 27 messages Apr 1, 2016 – Apr 13, 2016

Subsystems (2)
lore
Share & tools
lore

Florian Westphal

author maintainer
#0 lore raw

cover netfilter: xtables: stricter ruleset validation

This series adds more checks on xtables (arp, ip, ip6tables) rulesets.

- check all offsets (target, next) of all rules during initial pass after copy from userspace.
- check targets of jumps (-j bla): offset should be start of a rule
- assert that alleged target size is at least as big as minimum target structure
- change CONFIG_COMPAT code path to push ruleset via normal setsockopt path after initial 32->64 bit conversion to avoid duplicating checks
- use a common helper to copy counters from userspace instead of the ip/ip6/arp implementation.

Tested:
- iptables.git iptables-test.py passes
- made a few performance tests w. really silly rulesets to verify that things don't slow down too much, see individual patches for details.

 include/linux/netfilter/x_tables.h |   12 +
 net/ipv4/netfilter/arp_tables.c    |  303 ++++++++++------------------------
 net/ipv4/netfilter/ip_tables.c     |  327 +++++++++----------------------------
 net/ipv6/netfilter/ip6_tables.c    |  320 ++++++++----------------------------
 net/netfilter/x_tables.c           |  244 +++++++++++++++++++++++++++
 5 files changed, 506 insertions(+), 700 deletions(-)
author maintainer
#1 lore raw

PATCH 1/17 netfilter: x_tables: don't move to non-existent next rule

Ben Hawkes says:

In the mark_source_chains function (net/ipv4/netfilter/ip_tables.c) it is possible for a user-supplied ipt_entry structure to have a large next_offset field. This field is not bounds checked prior to writing a counter value at the supplied offset.

Base chains enforce absolute verdict.

User defined chains are supposed to end with an unconditional return, xtables userspace adds them automatically.

But if such return is missing we will move to non-existent next rule.

---
 net/ipv4/netfilter/arp_tables.c | 8 +++++---
 net/ipv4/netfilter/ip_tables.c  | 4 ++++
 net/ipv6/netfilter/ip6_tables.c | 4 ++++
 3 files changed, 13 insertions(+), 3 deletions(-)
net/ipv4/netfilter/arp_tables.c +5 -3
index 4133b0f..82a434b 100644
@@ -439,6 +439,8 @@ static int mark_source_chains(const struct xt_table_info *newinfo,
 				size = e->next_offset;
 				e = (struct arpt_entry *)
 					(entry0 + pos + size);
+				if (pos + size >= newinfo->size)
+					return 0;
 				e->counters.pcnt = pos;
 				pos += size;
 			} else {
@@ -461,6 +463,8 @@ static int mark_source_chains(const struct xt_table_info *newinfo,
 				} else {
 					/* ... this is a fallthru */
 					newpos = pos + e->next_offset;
+					if (newpos >= newinfo->size)
+						return 0;
 				}
 				e = (struct arpt_entry *)
 					(entry0 + newpos);
@@ -691,10 +695,8 @@ static int translate_table(struct xt_table_info *newinfo, void *entry0,
 		}
 	}
 
-	if (!mark_source_chains(newinfo, repl->valid_hooks, entry0)) {
-		duprintf("Looping hook\n");
+	if (!mark_source_chains(newinfo, repl->valid_hooks, entry0))
 		return -ELOOP;
-	}
 
 	/* Finally, each sanity check must pass */
 	i = 0;
net/ipv4/netfilter/ip_tables.c +4 -0
index 631c100..e301a3d 100644
@@ -520,6 +520,8 @@ mark_source_chains(const struct xt_table_info *newinfo,
 				size = e->next_offset;
 				e = (struct ipt_entry *)
 					(entry0 + pos + size);
+				if (pos + size >= newinfo->size)
+					return 0;
 				e->counters.pcnt = pos;
 				pos += size;
 			} else {
@@ -541,6 +543,8 @@ mark_source_chains(const struct xt_table_info *newinfo,
 				} else {
 					/* ... this is a fallthru */
 					newpos = pos + e->next_offset;
+					if (newpos >= newinfo->size)
+						return 0;
 				}
 				e = (struct ipt_entry *)
 					(entry0 + newpos);
net/ipv6/netfilter/ip6_tables.c +4 -0
index 86b67b7..7b3335b 100644
@@ -532,6 +532,8 @@ mark_source_chains(const struct xt_table_info *newinfo,
 				size = e->next_offset;
 				e = (struct ip6t_entry *)
 					(entry0 + pos + size);
+				if (pos + size >= newinfo->size)
+					return 0;
 				e->counters.pcnt = pos;
 				pos += size;
 			} else {
@@ -553,6 +555,8 @@ mark_source_chains(const struct xt_table_info *newinfo,
 				} else {
 					/* ... this is a fallthru */
 					newpos = pos + e->next_offset;
+					if (newpos >= newinfo->size)
+						return 0;
 				}
 				e = (struct ip6t_entry *)
 					(entry0 + newpos);
Signature · 1 line
2.7.3
author maintainer
#2 lore raw

PATCH 2/17 netfilter: x_tables: validate targets of jumps

In reply to Florian Westphal

When we see a jump also check that the offset gets us to beginning of a rule (an ipt_entry).

The extra overhead is negible, even with absurd cases.

300k custom rules, 300k jumps to 'next' user chain:
[ plus one jump from INPUT to first userchain ]:

Before:

real    0m24.874s
user    0m7.532s
sys     0m16.076s

After:

real    0m27.464s
user    0m7.436s
sys     0m18.840s
---
 net/ipv4/netfilter/arp_tables.c | 16 ++++++++++++++++
 net/ipv4/netfilter/ip_tables.c  | 16 ++++++++++++++++
 net/ipv6/netfilter/ip6_tables.c | 16 ++++++++++++++++
 3 files changed, 48 insertions(+)
net/ipv4/netfilter/arp_tables.c +16 -0
index 82a434b..ec37f7c 100644
@@ -367,6 +367,18 @@ static inline bool unconditional(const struct arpt_entry *e)
 	       memcmp(&e->arp, &uncond, sizeof(uncond)) == 0;
 }
 
+static bool find_jump_target(const struct xt_table_info *t,
+			     const struct arpt_entry *target)
+{
+	struct arpt_entry *iter;
+
+	xt_entry_foreach(iter, t->entries, t->size) {
+		 if (iter == target)
+			return true;
+	}
+	return false;
+}
+
 /* Figures out from what hook each rule can be called: returns 0 if
  * there are loops.  Puts hook bitmask in comefrom.
  */
@@ -460,6 +472,10 @@ static int mark_source_chains(const struct xt_table_info *newinfo,
 					/* This a jump; chase it. */
 					duprintf("Jump rule %u -> %u\n",
 						 pos, newpos);
+					e = (struct arpt_entry *)
+						(entry0 + newpos);
+					if (!find_jump_target(newinfo, e))
+						return 0;
 				} else {
 					/* ... this is a fallthru */
 					newpos = pos + e->next_offset;
net/ipv4/netfilter/ip_tables.c +16 -0
index e301a3d..503038e 100644
@@ -443,6 +443,18 @@ ipt_do_table(struct sk_buff *skb,
 #endif
 }
 
+static bool find_jump_target(const struct xt_table_info *t,
+			     const struct ipt_entry *target)
+{
+	struct ipt_entry *iter;
+
+	xt_entry_foreach(iter, t->entries, t->size) {
+		 if (iter == target)
+			return true;
+	}
+	return false;
+}
+
 /* Figures out from what hook each rule can be called: returns 0 if
    there are loops.  Puts hook bitmask in comefrom. */
 static int
@@ -540,6 +552,10 @@ mark_source_chains(const struct xt_table_info *newinfo,
 					/* This a jump; chase it. */
 					duprintf("Jump rule %u -> %u\n",
 						 pos, newpos);
+					e = (struct ipt_entry *)
+						(entry0 + newpos);
+					if (!find_jump_target(newinfo, e))
+						return 0;
 				} else {
 					/* ... this is a fallthru */
 					newpos = pos + e->next_offset;
net/ipv6/netfilter/ip6_tables.c +16 -0
index 7b3335b..126f2a0 100644
@@ -455,6 +455,18 @@ ip6t_do_table(struct sk_buff *skb,
 #endif
 }
 
+static bool find_jump_target(const struct xt_table_info *t,
+			     const struct ip6t_entry *target)
+{
+	struct ip6t_entry *iter;
+
+	xt_entry_foreach(iter, t->entries, t->size) {
+		 if (iter == target)
+			return true;
+	}
+	return false;
+}
+
 /* Figures out from what hook each rule can be called: returns 0 if
    there are loops.  Puts hook bitmask in comefrom. */
 static int
@@ -552,6 +564,10 @@ mark_source_chains(const struct xt_table_info *newinfo,
 					/* This a jump; chase it. */
 					duprintf("Jump rule %u -> %u\n",
 						 pos, newpos);
+					e = (struct ip6t_entry *)
+						(entry0 + newpos);
+					if (!find_jump_target(newinfo, e))
+						return 0;
 				} else {
 					/* ... this is a fallthru */
 					newpos = pos + e->next_offset;
Signature · 1 line
2.7.3
#3 lore raw

Re: [PATCH nf 02/17] netfilter: x_tables: validate targets of jumps

Quoted 7 lines · Florian Westphal

The extra overhead is negible, even with absurd cases.

300k custom rules, 300k jumps to 'next' user chain:
[ plus one jump from INPUT to first userchain ]:

sys 0m16.076s
sys 0m18.840s

I wouldn't call a 17% increase negligible. But, correctness before speed, I'm onboard even with a 200% jump.

author maintainer
#4 lore raw

PATCH 3/17 netfilter: x_tables: add and use xt_check_entry_offsets

In reply to Florian Westphal

Currently arp/ip and ip6tables each implement a short helper to check that the target offset is large enough to hold one xt_entry_target struct and that t->u.target_size fits within the current rule.

Unfortunately these checks are not sufficient.

To avoid adding new tests to all of ip/ip6/arptables move the current checks into a helper, then extend this helper in followup patches.

---
 include/linux/netfilter/x_tables.h |  4 ++++
 net/ipv4/netfilter/arp_tables.c    | 11 +----------
 net/ipv4/netfilter/ip_tables.c     | 12 +-----------
 net/ipv6/netfilter/ip6_tables.c    | 12 +-----------
 net/netfilter/x_tables.c           | 34 ++++++++++++++++++++++++++++++++++
 5 files changed, 41 insertions(+), 32 deletions(-)
include/linux/netfilter/x_tables.h +4 -0
index 80a305b..0de0862 100644
@@ -242,6 +242,10 @@ void xt_unregister_match(struct xt_match *target);
 int xt_register_matches(struct xt_match *match, unsigned int n);
 void xt_unregister_matches(struct xt_match *match, unsigned int n);
 
+int xt_check_entry_offsets(const void *base,
+			   unsigned int target_offset,
+			   unsigned int next_offset);
+
 int xt_check_match(struct xt_mtchk_param *, unsigned int size, u_int8_t proto,
 		   bool inv_proto);
 int xt_check_target(struct xt_tgchk_param *, unsigned int size, u_int8_t proto,
net/ipv4/netfilter/arp_tables.c +1 -10
index ec37f7c..74668c1 100644
@@ -496,19 +496,10 @@ next:
 
 static inline int check_entry(const struct arpt_entry *e)
 {
-	const struct xt_entry_target *t;
-
 	if (!arp_checkentry(&e->arp))
 		return -EINVAL;
 
-	if (e->target_offset + sizeof(struct xt_entry_target) > e->next_offset)
-		return -EINVAL;
-
-	t = arpt_get_target_c(e);
-	if (e->target_offset + t->u.target_size > e->next_offset)
-		return -EINVAL;
-
-	return 0;
+	return xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 }
 
 static inline int check_target(struct arpt_entry *e, const char *name)
net/ipv4/netfilter/ip_tables.c +1 -11
index 503038e..71c204d 100644
@@ -590,20 +590,10 @@ static void cleanup_match(struct xt_entry_match *m, struct net *net)
 static int
 check_entry(const struct ipt_entry *e)
 {
-	const struct xt_entry_target *t;
-
 	if (!ip_checkentry(&e->ip))
 		return -EINVAL;
 
-	if (e->target_offset + sizeof(struct xt_entry_target) >
-	    e->next_offset)
-		return -EINVAL;
-
-	t = ipt_get_target_c(e);
-	if (e->target_offset + t->u.target_size > e->next_offset)
-		return -EINVAL;
-
-	return 0;
+	return xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 }
 
 static int
net/ipv6/netfilter/ip6_tables.c +1 -11
index 126f2a0..24ae7f4 100644
@@ -602,20 +602,10 @@ static void cleanup_match(struct xt_entry_match *m, struct net *net)
 static int
 check_entry(const struct ip6t_entry *e)
 {
-	const struct xt_entry_target *t;
-
 	if (!ip6_checkentry(&e->ipv6))
 		return -EINVAL;
 
-	if (e->target_offset + sizeof(struct xt_entry_target) >
-	    e->next_offset)
-		return -EINVAL;
-
-	t = ip6t_get_target_c(e);
-	if (e->target_offset + t->u.target_size > e->next_offset)
-		return -EINVAL;
-
-	return 0;
+	return xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 }
 
 static int check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
net/netfilter/x_tables.c +34 -0
index 582c9cf..1f44bfa 100644
@@ -541,6 +541,40 @@ int xt_compat_match_to_user(const struct xt_entry_match *m,
 EXPORT_SYMBOL_GPL(xt_compat_match_to_user);
 #endif /* CONFIG_COMPAT */
 
+/**
+ * xt_check_entry_offsets - validate arp/ip/ip6t_entry
+ *
+ * @base: pointer to arp/ip/ip6t_entry
+ * @target_offset: the arp/ip/ip6_t->target_offset
+ * @next_offset: the arp/ip/ip6_t->next_offset
+ *
+ * validates that target_offset and next_offset are sane.
+ *
+ * The arp/ip/ip6t_entry structure @base must have passed following tests:
+ * - it must point to a valid memory location
+ * - base to base + next_offset must be accessible, i.e. not exceed allocated
+ *   length.
+ *
+ * Return: 0 on success, negative errno on failure.
+ */
+int xt_check_entry_offsets(const void *base,
+			   unsigned int target_offset,
+			   unsigned int next_offset)
+{
+	const struct xt_entry_target *t;
+	const char *e = base;
+
+	if (target_offset + sizeof(*t) > next_offset)
+		return -EINVAL;
+
+	t = (void *)(e + target_offset);
+	if (target_offset + t->u.target_size > next_offset)
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL(xt_check_entry_offsets);
+
 int xt_check_target(struct xt_tgchk_param *par,
 		    unsigned int size, u_int8_t proto, bool inv_proto)
 {
Signature · 1 line
2.7.3
author maintainer
#5 lore raw

PATCH 4/17 netfilter: x_tables: kill check_entry helper

In reply to Florian Westphal

Once we add more sanity testing to xt_check_entry_offsets it becomes relvant if we're expecting a 32bit 'config_compat' blob or a normal one.

Since we already have a lot of similar-named functions (check_entry, compat_check_entry, find_and_check_entry, etc.) and the current incarnation is short just fold its contents into the callers.

---
 net/ipv4/netfilter/arp_tables.c | 19 ++++++++-----------
 net/ipv4/netfilter/ip_tables.c  | 20 ++++++++------------
 net/ipv6/netfilter/ip6_tables.c | 20 ++++++++------------
 3 files changed, 24 insertions(+), 35 deletions(-)
net/ipv4/netfilter/arp_tables.c +8 -11
index 74668c1..24ad92a 100644
@@ -494,14 +494,6 @@ next:
 	return 1;
 }
 
-static inline int check_entry(const struct arpt_entry *e)
-{
-	if (!arp_checkentry(&e->arp))
-		return -EINVAL;
-
-	return xt_check_entry_offsets(e, e->target_offset, e->next_offset);
-}
-
 static inline int check_target(struct arpt_entry *e, const char *name)
 {
 	struct xt_entry_target *t = arpt_get_target(e);
@@ -597,7 +589,10 @@ static inline int check_entry_size_and_hooks(struct arpt_entry *e,
 		return -EINVAL;
 	}
 
-	err = check_entry(e);
+	if (!arp_checkentry(&e->arp))
+		return -EINVAL;
+
+	err = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 	if (err)
 		return err;
 
@@ -1256,8 +1251,10 @@ check_compat_entry_size_and_hooks(struct compat_arpt_entry *e,
 		return -EINVAL;
 	}
 
-	/* For purposes of check_entry casting the compat entry is fine */
-	ret = check_entry((struct arpt_entry *)e);
+	if (!arp_checkentry(&e->arp))
+		return -EINVAL;
+
+	ret = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 	if (ret)
 		return ret;
 
net/ipv4/netfilter/ip_tables.c +8 -12
index 71c204d..cdf1850 100644
@@ -588,15 +588,6 @@ static void cleanup_match(struct xt_entry_match *m, struct net *net)
 }
 
 static int
-check_entry(const struct ipt_entry *e)
-{
-	if (!ip_checkentry(&e->ip))
-		return -EINVAL;
-
-	return xt_check_entry_offsets(e, e->target_offset, e->next_offset);
-}
-
-static int
 check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
 {
 	const struct ipt_ip *ip = par->entryinfo;
@@ -760,7 +751,10 @@ check_entry_size_and_hooks(struct ipt_entry *e,
 		return -EINVAL;
 	}
 
-	err = check_entry(e);
+	if (!ip_checkentry(&e->ip))
+		return -EINVAL;
+
+	err = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 	if (err)
 		return err;
 
@@ -1516,8 +1510,10 @@ check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
 		return -EINVAL;
 	}
 
-	/* For purposes of check_entry casting the compat entry is fine */
-	ret = check_entry((struct ipt_entry *)e);
+	if (!ip_checkentry(&e->ip))
+		return -EINVAL;
+
+	ret = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 	if (ret)
 		return ret;
 
net/ipv6/netfilter/ip6_tables.c +8 -12
index 24ae7f4..e378311 100644
@@ -599,15 +599,6 @@ static void cleanup_match(struct xt_entry_match *m, struct net *net)
 	module_put(par.match->me);
 }
 
-static int
-check_entry(const struct ip6t_entry *e)
-{
-	if (!ip6_checkentry(&e->ipv6))
-		return -EINVAL;
-
-	return xt_check_entry_offsets(e, e->target_offset, e->next_offset);
-}
-
 static int check_match(struct xt_entry_match *m, struct xt_mtchk_param *par)
 {
 	const struct ip6t_ip6 *ipv6 = par->entryinfo;
@@ -772,7 +763,10 @@ check_entry_size_and_hooks(struct ip6t_entry *e,
 		return -EINVAL;
 	}
 
-	err = check_entry(e);
+	if (!ip6_checkentry(&e->ipv6))
+		return -EINVAL;
+
+	err = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 	if (err)
 		return err;
 
@@ -1528,8 +1522,10 @@ check_compat_entry_size_and_hooks(struct compat_ip6t_entry *e,
 		return -EINVAL;
 	}
 
-	/* For purposes of check_entry casting the compat entry is fine */
-	ret = check_entry((struct ip6t_entry *)e);
+	if (!ip6_checkentry(&e->ipv6))
+		return -EINVAL;
+
+	ret = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
 	if (ret)
 		return ret;
 
Signature · 1 line
2.7.3
author maintainer
#6 lore raw

PATCH 5/17 netfilter: x_tables: assert minimum target size

In reply to Florian Westphal

The target size includes the size of the xt_entry_target struct.

---
 net/netfilter/x_tables.c | 3 +++
 1 file changed, 3 insertions(+)
net/netfilter/x_tables.c +3 -0
index 1f44bfa..ec1b718 100644
@@ -568,6 +568,9 @@ int xt_check_entry_offsets(const void *base,
 		return -EINVAL;
 
 	t = (void *)(e + target_offset);
+	if (t->u.target_size < sizeof(*t))
+		return -EINVAL;
+
 	if (target_offset + t->u.target_size > next_offset)
 		return -EINVAL;
 
Signature · 1 line
2.7.3
author maintainer
#7 lore raw

PATCH 6/17 netfilter: x_tables: add compat version of xt_check_entry_offsets

In reply to Florian Westphal

32bit rulesets have different layout and alignment requirements, so once more integrity checks get added to xt_check_entry_offsets it will reject well-formed 32bit rulesets.

---
 include/linux/netfilter/x_tables.h |  3 +++
 net/ipv4/netfilter/arp_tables.c    |  3 ++-
 net/ipv4/netfilter/ip_tables.c     |  3 ++-
 net/ipv6/netfilter/ip6_tables.c    |  3 ++-
 net/netfilter/x_tables.c           | 22 ++++++++++++++++++++++
 5 files changed, 31 insertions(+), 3 deletions(-)
include/linux/netfilter/x_tables.h +3 -0
index 0de0862..08de48b 100644
@@ -494,6 +494,9 @@ void xt_compat_target_from_user(struct xt_entry_target *t, void **dstptr,
 				unsigned int *size);
 int xt_compat_target_to_user(const struct xt_entry_target *t,
 			     void __user **dstptr, unsigned int *size);
+int xt_compat_check_entry_offsets(const void *base,
+				  unsigned int target_offset,
+				  unsigned int next_offset);
 
 #endif /* CONFIG_COMPAT */
 #endif /* _X_TABLES_H */
net/ipv4/netfilter/arp_tables.c +2 -1
index 24ad92a..ab8952a 100644
@@ -1254,7 +1254,8 @@ check_compat_entry_size_and_hooks(struct compat_arpt_entry *e,
 	if (!arp_checkentry(&e->arp))
 		return -EINVAL;
 
-	ret = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
+	ret = xt_compat_check_entry_offsets(e, e->target_offset,
+					    e->next_offset);
 	if (ret)
 		return ret;
 
net/ipv4/netfilter/ip_tables.c +2 -1
index cdf1850..7d24c87 100644
@@ -1513,7 +1513,8 @@ check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
 	if (!ip_checkentry(&e->ip))
 		return -EINVAL;
 
-	ret = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
+	ret = xt_compat_check_entry_offsets(e,
+					    e->target_offset, e->next_offset);
 	if (ret)
 		return ret;
 
net/ipv6/netfilter/ip6_tables.c +2 -1
index e378311..73eee7b 100644
@@ -1525,7 +1525,8 @@ check_compat_entry_size_and_hooks(struct compat_ip6t_entry *e,
 	if (!ip6_checkentry(&e->ipv6))
 		return -EINVAL;
 
-	ret = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
+	ret = xt_compat_check_entry_offsets(e,
+					    e->target_offset, e->next_offset);
 	if (ret)
 		return ret;
 
net/netfilter/x_tables.c +22 -0
index ec1b718..fa206ce 100644
@@ -539,6 +539,27 @@ int xt_compat_match_to_user(const struct xt_entry_match *m,
 	return 0;
 }
 EXPORT_SYMBOL_GPL(xt_compat_match_to_user);
+
+int xt_compat_check_entry_offsets(const void *base,
+				  unsigned int target_offset,
+				  unsigned int next_offset)
+{
+	const struct compat_xt_entry_target *t;
+	const char *e = base;
+
+	if (target_offset + sizeof(*t) > next_offset)
+		return -EINVAL;
+
+	t = (void *)(e + target_offset);
+	if (t->u.target_size < sizeof(*t))
+		return -EINVAL;
+
+	if (target_offset + t->u.target_size > next_offset)
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL(xt_compat_check_entry_offsets);
 #endif /* CONFIG_COMPAT */
 
 /**
@@ -549,6 +570,7 @@ EXPORT_SYMBOL_GPL(xt_compat_match_to_user);
  * @next_offset: the arp/ip/ip6_t->next_offset
  *
  * validates that target_offset and next_offset are sane.
+ * Also see xt_compat_check_entry_offsets for CONFIG_COMPAT version.
  *
  * The arp/ip/ip6t_entry structure @base must have passed following tests:
  * - it must point to a valid memory location
Signature · 1 line
2.7.3
author maintainer
#8 lore raw

PATCH 7/17 netfilter: x_tables: check standard target size too

In reply to Florian Westphal

We have targets and standard targets -- the latter carries a verdict.

The ip/ip6tables validation functions will access t->verdict for the standard targets to fetch the jump offset or verdict for chainloop detection, but this happens before the targets get checked/validated.

Thus we also need to check for verdict presence here, else t->verdict can point right after a blob.

Spotted with UBSAN while testing malformed blobs.

---
 net/netfilter/x_tables.c | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
net/netfilter/x_tables.c +15 -0
index fa206ce..1cb7a27 100644
@@ -540,6 +540,13 @@ int xt_compat_match_to_user(const struct xt_entry_match *m,
 }
 EXPORT_SYMBOL_GPL(xt_compat_match_to_user);
 
+/* non-compat version may have padding after verdict */
+struct compat_xt_standard_target {
+	struct compat_xt_entry_target t;
+	compat_uint_t verdict;
+};
+
+/* see xt_check_entry_offsets */
 int xt_compat_check_entry_offsets(const void *base,
 				  unsigned int target_offset,
 				  unsigned int next_offset)
@@ -557,6 +564,10 @@ int xt_compat_check_entry_offsets(const void *base,
 	if (target_offset + t->u.target_size > next_offset)
 		return -EINVAL;
 
+	if (strcmp(t->u.user.name, XT_STANDARD_TARGET) == 0 &&
+	    target_offset + sizeof(struct compat_xt_standard_target) != next_offset)
+		return -EINVAL;
+
 	return 0;
 }
 EXPORT_SYMBOL(xt_compat_check_entry_offsets);
@@ -596,6 +607,10 @@ int xt_check_entry_offsets(const void *base,
 	if (target_offset + t->u.target_size > next_offset)
 		return -EINVAL;
 
+	if (strcmp(t->u.user.name, XT_STANDARD_TARGET) == 0 &&
+	    target_offset + sizeof(struct xt_standard_target) != next_offset)
+		return -EINVAL;
+
 	return 0;
 }
 EXPORT_SYMBOL(xt_check_entry_offsets);
Signature · 1 line
2.7.3
author maintainer
#9 lore raw

PATCH 8/17 netfilter: x_tables: check for bogus target offset

In reply to Florian Westphal

We're currently asserting that targetoff + targetsize <= nextoff.

Extend it to also check that targetoff is >= sizeof(xt_entry). Since this is generic code, add an argument pointing to the start of the match/target, we can then derive the base structure size from the delta.

We also need the e->elems pointer in a followup change to validate matches.

---
 include/linux/netfilter/x_tables.h |  4 ++--
 net/ipv4/netfilter/arp_tables.c    |  5 +++--
 net/ipv4/netfilter/ip_tables.c     |  5 +++--
 net/ipv6/netfilter/ip6_tables.c    |  5 +++--
 net/netfilter/x_tables.c           | 17 +++++++++++++++--
 5 files changed, 26 insertions(+), 10 deletions(-)
include/linux/netfilter/x_tables.h +2 -2
index 08de48b..30cfb1e 100644
@@ -242,7 +242,7 @@ void xt_unregister_match(struct xt_match *target);
 int xt_register_matches(struct xt_match *match, unsigned int n);
 void xt_unregister_matches(struct xt_match *match, unsigned int n);
 
-int xt_check_entry_offsets(const void *base,
+int xt_check_entry_offsets(const void *base, const char *elems,
 			   unsigned int target_offset,
 			   unsigned int next_offset);
 
@@ -494,7 +494,7 @@ void xt_compat_target_from_user(struct xt_entry_target *t, void **dstptr,
 				unsigned int *size);
 int xt_compat_target_to_user(const struct xt_entry_target *t,
 			     void __user **dstptr, unsigned int *size);
-int xt_compat_check_entry_offsets(const void *base,
+int xt_compat_check_entry_offsets(const void *base, const char *elems,
 				  unsigned int target_offset,
 				  unsigned int next_offset);
 
net/ipv4/netfilter/arp_tables.c +3 -2
index ab8952a..95ed4e4 100644
@@ -592,7 +592,8 @@ static inline int check_entry_size_and_hooks(struct arpt_entry *e,
 	if (!arp_checkentry(&e->arp))
 		return -EINVAL;
 
-	err = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
+	err = xt_check_entry_offsets(e, e->elems, e->target_offset,
+				     e->next_offset);
 	if (err)
 		return err;
 
@@ -1254,7 +1255,7 @@ check_compat_entry_size_and_hooks(struct compat_arpt_entry *e,
 	if (!arp_checkentry(&e->arp))
 		return -EINVAL;
 
-	ret = xt_compat_check_entry_offsets(e, e->target_offset,
+	ret = xt_compat_check_entry_offsets(e, e->elems, e->target_offset,
 					    e->next_offset);
 	if (ret)
 		return ret;
net/ipv4/netfilter/ip_tables.c +3 -2
index 7d24c87..baab033d 100644
@@ -754,7 +754,8 @@ check_entry_size_and_hooks(struct ipt_entry *e,
 	if (!ip_checkentry(&e->ip))
 		return -EINVAL;
 
-	err = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
+	err = xt_check_entry_offsets(e, e->elems, e->target_offset,
+				     e->next_offset);
 	if (err)
 		return err;
 
@@ -1513,7 +1514,7 @@ check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
 	if (!ip_checkentry(&e->ip))
 		return -EINVAL;
 
-	ret = xt_compat_check_entry_offsets(e,
+	ret = xt_compat_check_entry_offsets(e, e->elems,
 					    e->target_offset, e->next_offset);
 	if (ret)
 		return ret;
net/ipv6/netfilter/ip6_tables.c +3 -2
index 73eee7b..6957627 100644
@@ -766,7 +766,8 @@ check_entry_size_and_hooks(struct ip6t_entry *e,
 	if (!ip6_checkentry(&e->ipv6))
 		return -EINVAL;
 
-	err = xt_check_entry_offsets(e, e->target_offset, e->next_offset);
+	err = xt_check_entry_offsets(e, e->elems, e->target_offset,
+				     e->next_offset);
 	if (err)
 		return err;
 
@@ -1525,7 +1526,7 @@ check_compat_entry_size_and_hooks(struct compat_ip6t_entry *e,
 	if (!ip6_checkentry(&e->ipv6))
 		return -EINVAL;
 
-	ret = xt_compat_check_entry_offsets(e,
+	ret = xt_compat_check_entry_offsets(e, e->elems,
 					    e->target_offset, e->next_offset);
 	if (ret)
 		return ret;
net/netfilter/x_tables.c +15 -2
index 1cb7a27..e2a6f2a 100644
@@ -546,14 +546,17 @@ struct compat_xt_standard_target {
 	compat_uint_t verdict;
 };
 
-/* see xt_check_entry_offsets */
-int xt_compat_check_entry_offsets(const void *base,
+int xt_compat_check_entry_offsets(const void *base, const char *elems,
 				  unsigned int target_offset,
 				  unsigned int next_offset)
 {
+	long size_of_base_struct = elems - (const char *)base;
 	const struct compat_xt_entry_target *t;
 	const char *e = base;
 
+	if (target_offset < size_of_base_struct)
+		return -EINVAL;
+
 	if (target_offset + sizeof(*t) > next_offset)
 		return -EINVAL;
 
@@ -577,12 +580,16 @@ EXPORT_SYMBOL(xt_compat_check_entry_offsets);
  * xt_check_entry_offsets - validate arp/ip/ip6t_entry
  *
  * @base: pointer to arp/ip/ip6t_entry
+ * @elems: pointer to first xt_entry_match, i.e. ip(6)t_entry->elems
  * @target_offset: the arp/ip/ip6_t->target_offset
  * @next_offset: the arp/ip/ip6_t->next_offset
  *
  * validates that target_offset and next_offset are sane.
  * Also see xt_compat_check_entry_offsets for CONFIG_COMPAT version.
  *
+ * This function does not validate the targets or matches themselves, it
+ * only tests that all the offsets and sizes are correct.
+ *
  * The arp/ip/ip6t_entry structure @base must have passed following tests:
  * - it must point to a valid memory location
  * - base to base + next_offset must be accessible, i.e. not exceed allocated
@@ -591,12 +598,18 @@ EXPORT_SYMBOL(xt_compat_check_entry_offsets);
  * Return: 0 on success, negative errno on failure.
  */
 int xt_check_entry_offsets(const void *base,
+			   const char *elems,
 			   unsigned int target_offset,
 			   unsigned int next_offset)
 {
+	long size_of_base_struct = elems - (const char *)base;
 	const struct xt_entry_target *t;
 	const char *e = base;
 
+	/* target start is within the ip/ip6/arpt_entry struct */
+	if (target_offset < size_of_base_struct)
+		return -EINVAL;
+
 	if (target_offset + sizeof(*t) > next_offset)
 		return -EINVAL;
 
Signature · 1 line
2.7.3
author maintainer
#10 lore raw

PATCH 9/17 netfilter: x_tables: validate all offsets and sizes in a rule

In reply to Florian Westphal

Validate that all matches (if any) add up to the beginning of the target and that each match covers at least the base structure size.

The compat path should be able to safely re-use the function as the structures only differ in alignment; added a BUILD_BUG_ON just in case we have an arch that adds padding as well.

---
 net/netfilter/x_tables.c | 81 +++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 76 insertions(+), 5 deletions(-)
net/netfilter/x_tables.c +76 -5
index e2a6f2a..f9aa971 100644
@@ -416,6 +416,47 @@ int xt_check_match(struct xt_mtchk_param *par,
 }
 EXPORT_SYMBOL_GPL(xt_check_match);
 
+/** xt_check_entry_match - check that matches end before start of target
+ *
+ * @match: beginning of xt_entry_match
+ * @target: beginning of this rules target (alleged end of matches)
+ * @alignment: alignment requirement of match structures
+ *
+ * Validates that all matches add up to the beginning of the target,
+ * and that each match covers at least the base structure size.
+ *
+ * Return: 0 on success, negative errno on failure.
+ */
+static int xt_check_entry_match(const char *match, const char *target,
+				const size_t alignment)
+{
+	const struct xt_entry_match *pos;
+	int length = target - match;
+
+	if (length == 0) /* no matches */
+		return 0;
+
+	pos = (struct xt_entry_match *)match;
+	do {
+		if ((unsigned long)pos % alignment)
+			return -EINVAL;
+
+		if (length < (int)sizeof(struct xt_entry_match))
+			return -EINVAL;
+
+		if (pos->u.match_size < sizeof(struct xt_entry_match))
+			return -EINVAL;
+
+		if (pos->u.match_size > length)
+			return -EINVAL;
+
+		length -= pos->u.match_size;
+		pos = ((void *)((char *)(pos) + (pos)->u.match_size));
+	} while (length > 0);
+
+	return 0;
+}
+
 #ifdef CONFIG_COMPAT
 int xt_compat_add_offset(u_int8_t af, unsigned int offset, int delta)
 {
@@ -571,7 +612,14 @@ int xt_compat_check_entry_offsets(const void *base, const char *elems,
 	    target_offset + sizeof(struct compat_xt_standard_target) != next_offset)
 		return -EINVAL;
 
-	return 0;
+	/* compat_xt_entry match has less strict aligment requirements,
+	 * otherwise they are identical.  In case of padding differences
+	 * we need to add compat version of xt_check_entry_match.
+	 */
+	BUILD_BUG_ON(sizeof(struct compat_xt_entry_match) != sizeof(struct xt_entry_match));
+
+	return xt_check_entry_match(elems, base + target_offset,
+				    __alignof__(struct compat_xt_entry_match));
 }
 EXPORT_SYMBOL(xt_compat_check_entry_offsets);
 #endif /* CONFIG_COMPAT */
@@ -584,17 +632,39 @@ EXPORT_SYMBOL(xt_compat_check_entry_offsets);
  * @target_offset: the arp/ip/ip6_t->target_offset
  * @next_offset: the arp/ip/ip6_t->next_offset
  *
- * validates that target_offset and next_offset are sane.
- * Also see xt_compat_check_entry_offsets for CONFIG_COMPAT version.
+ * validates that target_offset and next_offset are sane and that all
+ * match sizes (if any) align with the target offset.
  *
  * This function does not validate the targets or matches themselves, it
- * only tests that all the offsets and sizes are correct.
+ * only tests that all the offsets and sizes are correct, that all
+ * match structures are aligned, and that the last structure ends where
+ * the target structure begins.
+ *
+ * Also see xt_compat_check_entry_offsets for CONFIG_COMPAT version.
  *
  * The arp/ip/ip6t_entry structure @base must have passed following tests:
  * - it must point to a valid memory location
  * - base to base + next_offset must be accessible, i.e. not exceed allocated
  *   length.
  *
+ * A well-formed entry looks like this:
+ *
+ * ip(6)t_entry   match [mtdata]  match [mtdata] target [tgdata] ip(6)t_entry
+ * e->elems[]-----'                              |               |
+ *                matchsize                      |               |
+ *                                matchsize      |               |
+ *                                               |               |
+ * target_offset---------------------------------'               |
+ * next_offset---------------------------------------------------'
+ *
+ * elems[]: flexible array member at end of ip(6)/arpt_entry struct.
+ *          This is where matches (if any) and the target reside.
+ * target_offset: beginning of target.
+ * next_offset: start of the next rule; also: size of this rule.
+ * Since targets have a minimum size, target_offset + minlen <= next_offset.
+ *
+ * Every match stores its size, sum of sizes must not exceed target_offset.
+ *
  * Return: 0 on success, negative errno on failure.
  */
 int xt_check_entry_offsets(const void *base,
@@ -624,7 +694,8 @@ int xt_check_entry_offsets(const void *base,
 	    target_offset + sizeof(struct xt_standard_target) != next_offset)
 		return -EINVAL;
 
-	return 0;
+	return xt_check_entry_match(elems, base + target_offset,
+				    __alignof__(struct xt_entry_match));
 }
 EXPORT_SYMBOL(xt_check_entry_offsets);
 
Signature · 1 line
2.7.3
author maintainer
#11 lore raw

PATCH 10/17 netfilter: ip_tables: simplify translate_compat_table args

In reply to Florian Westphal
---
 net/ipv4/netfilter/ip_tables.c | 59 +++++++++++++++++-------------------------
 1 file changed, 24 insertions(+), 35 deletions(-)
net/ipv4/netfilter/ip_tables.c +24 -35
index baab033d..d7041860 100644
@@ -1449,7 +1449,6 @@ compat_copy_entry_to_user(struct ipt_entry *e, void __user **dstptr,
 
 static int
 compat_find_calc_match(struct xt_entry_match *m,
-		       const char *name,
 		       const struct ipt_ip *ip,
 		       int *size)
 {
@@ -1486,8 +1485,7 @@ check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
 				  const unsigned char *base,
 				  const unsigned char *limit,
 				  const unsigned int *hook_entries,
-				  const unsigned int *underflows,
-				  const char *name)
+				  const unsigned int *underflows)
 {
 	struct xt_entry_match *ematch;
 	struct xt_entry_target *t;
@@ -1523,7 +1521,7 @@ check_compat_entry_size_and_hooks(struct compat_ipt_entry *e,
 	entry_offset = (void *)e - (void *)base;
 	j = 0;
 	xt_ematch_foreach(ematch, e) {
-		ret = compat_find_calc_match(ematch, name, &e->ip, &off);
+		ret = compat_find_calc_match(ematch, &e->ip, &off);
 		if (ret != 0)
 			goto release_matches;
 		++j;
@@ -1572,7 +1570,7 @@ release_matches:
 
 static int
 compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
-			    unsigned int *size, const char *name,
+			    unsigned int *size,
 			    struct xt_table_info *newinfo, unsigned char *base)
 {
 	struct xt_entry_target *t;
@@ -1655,14 +1653,9 @@ compat_check_entry(struct ipt_entry *e, struct net *net, const char *name)
 
 static int
 translate_compat_table(struct net *net,
-		       const char *name,
-		       unsigned int valid_hooks,
 		       struct xt_table_info **pinfo,
 		       void **pentry0,
-		       unsigned int total_size,
-		       unsigned int number,
-		       unsigned int *hook_entries,
-		       unsigned int *underflows)
+		       const struct compat_ipt_replace *compatr)
 {
 	unsigned int i, j;
 	struct xt_table_info *newinfo, *info;
@@ -1674,8 +1667,8 @@ translate_compat_table(struct net *net,
 
 	info = *pinfo;
 	entry0 = *pentry0;
-	size = total_size;
-	info->number = number;
+	size = compatr->size;
+	info->number = compatr->num_entries;
 
 	/* Init all hooks to impossible value. */
 	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
@@ -1686,40 +1679,39 @@ translate_compat_table(struct net *net,
 	duprintf("translate_compat_table: size %u\n", info->size);
 	j = 0;
 	xt_compat_lock(AF_INET);
-	xt_compat_init_offsets(AF_INET, number);
+	xt_compat_init_offsets(AF_INET, compatr->num_entries);
 	/* Walk through entries, checking offsets. */
-	xt_entry_foreach(iter0, entry0, total_size) {
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		ret = check_compat_entry_size_and_hooks(iter0, info, &size,
 							entry0,
-							entry0 + total_size,
-							hook_entries,
-							underflows,
-							name);
+							entry0 + compatr->size,
+							compatr->hook_entry,
+							compatr->underflow);
 		if (ret != 0)
 			goto out_unlock;
 		++j;
 	}
 
 	ret = -EINVAL;
-	if (j != number) {
+	if (j != compatr->num_entries) {
 		duprintf("translate_compat_table: %u not %u entries\n",
-			 j, number);
+			 j, compatr->num_entries);
 		goto out_unlock;
 	}
 
 	/* Check hooks all assigned */
 	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
 		/* Only hooks which are valid */
-		if (!(valid_hooks & (1 << i)))
+		if (!(compatr->valid_hooks & (1 << i)))
 			continue;
 		if (info->hook_entry[i] == 0xFFFFFFFF) {
 			duprintf("Invalid hook entry %u %u\n",
-				 i, hook_entries[i]);
+				 i, info->hook_entry[i]);
 			goto out_unlock;
 		}
 		if (info->underflow[i] == 0xFFFFFFFF) {
 			duprintf("Invalid underflow %u %u\n",
-				 i, underflows[i]);
+				 i, info->underflow[i]);
 			goto out_unlock;
 		}
 	}
@@ -1729,17 +1721,17 @@ translate_compat_table(struct net *net,
 	if (!newinfo)
 		goto out_unlock;
 
-	newinfo->number = number;
+	newinfo->number = compatr->num_entries;
 	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
 		newinfo->hook_entry[i] = info->hook_entry[i];
 		newinfo->underflow[i] = info->underflow[i];
 	}
 	entry1 = newinfo->entries;
 	pos = entry1;
-	size = total_size;
-	xt_entry_foreach(iter0, entry0, total_size) {
+	size = compatr->size;
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		ret = compat_copy_entry_from_user(iter0, &pos, &size,
-						  name, newinfo, entry1);
+						  newinfo, entry1);
 		if (ret != 0)
 			break;
 	}
@@ -1749,12 +1741,12 @@ translate_compat_table(struct net *net,
 		goto free_newinfo;
 
 	ret = -ELOOP;
-	if (!mark_source_chains(newinfo, valid_hooks, entry1))
+	if (!mark_source_chains(newinfo, compatr->valid_hooks, entry1))
 		goto free_newinfo;
 
 	i = 0;
 	xt_entry_foreach(iter1, entry1, newinfo->size) {
-		ret = compat_check_entry(iter1, net, name);
+		ret = compat_check_entry(iter1, net, compatr->name);
 		if (ret != 0)
 			break;
 		++i;
@@ -1794,7 +1786,7 @@ translate_compat_table(struct net *net,
 free_newinfo:
 	xt_free_table_info(newinfo);
 out:
-	xt_entry_foreach(iter0, entry0, total_size) {
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		if (j-- == 0)
 			break;
 		compat_release_entry(iter0);
@@ -1839,10 +1831,7 @@ compat_do_replace(struct net *net, void __user *user, unsigned int len)
 		goto free_newinfo;
 	}
 
-	ret = translate_compat_table(net, tmp.name, tmp.valid_hooks,
-				     &newinfo, &loc_cpu_entry, tmp.size,
-				     tmp.num_entries, tmp.hook_entry,
-				     tmp.underflow);
+	ret = translate_compat_table(net, &newinfo, &loc_cpu_entry, &tmp);
 	if (ret != 0)
 		goto free_newinfo;
 
Signature · 1 line
2.7.3
author maintainer
#12 lore raw

PATCH 11/17 netfilter: ip6_tables: simplify translate_compat_table args

In reply to Florian Westphal
---
 net/ipv6/netfilter/ip6_tables.c | 59 +++++++++++++++++------------------------
 1 file changed, 24 insertions(+), 35 deletions(-)
net/ipv6/netfilter/ip6_tables.c +24 -35
index 6957627..8d082c55 100644
@@ -1461,7 +1461,6 @@ compat_copy_entry_to_user(struct ip6t_entry *e, void __user **dstptr,
 
 static int
 compat_find_calc_match(struct xt_entry_match *m,
-		       const char *name,
 		       const struct ip6t_ip6 *ipv6,
 		       int *size)
 {
@@ -1498,8 +1497,7 @@ check_compat_entry_size_and_hooks(struct compat_ip6t_entry *e,
 				  const unsigned char *base,
 				  const unsigned char *limit,
 				  const unsigned int *hook_entries,
-				  const unsigned int *underflows,
-				  const char *name)
+				  const unsigned int *underflows)
 {
 	struct xt_entry_match *ematch;
 	struct xt_entry_target *t;
@@ -1535,7 +1533,7 @@ check_compat_entry_size_and_hooks(struct compat_ip6t_entry *e,
 	entry_offset = (void *)e - (void *)base;
 	j = 0;
 	xt_ematch_foreach(ematch, e) {
-		ret = compat_find_calc_match(ematch, name, &e->ipv6, &off);
+		ret = compat_find_calc_match(ematch, &e->ipv6, &off);
 		if (ret != 0)
 			goto release_matches;
 		++j;
@@ -1584,7 +1582,7 @@ release_matches:
 
 static int
 compat_copy_entry_from_user(struct compat_ip6t_entry *e, void **dstptr,
-			    unsigned int *size, const char *name,
+			    unsigned int *size,
 			    struct xt_table_info *newinfo, unsigned char *base)
 {
 	struct xt_entry_target *t;
@@ -1664,14 +1662,9 @@ static int compat_check_entry(struct ip6t_entry *e, struct net *net,
 
 static int
 translate_compat_table(struct net *net,
-		       const char *name,
-		       unsigned int valid_hooks,
 		       struct xt_table_info **pinfo,
 		       void **pentry0,
-		       unsigned int total_size,
-		       unsigned int number,
-		       unsigned int *hook_entries,
-		       unsigned int *underflows)
+		       const struct compat_ip6t_replace *compatr)
 {
 	unsigned int i, j;
 	struct xt_table_info *newinfo, *info;
@@ -1683,8 +1676,8 @@ translate_compat_table(struct net *net,
 
 	info = *pinfo;
 	entry0 = *pentry0;
-	size = total_size;
-	info->number = number;
+	size = compatr->size;
+	info->number = compatr->num_entries;
 
 	/* Init all hooks to impossible value. */
 	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
@@ -1695,40 +1688,39 @@ translate_compat_table(struct net *net,
 	duprintf("translate_compat_table: size %u\n", info->size);
 	j = 0;
 	xt_compat_lock(AF_INET6);
-	xt_compat_init_offsets(AF_INET6, number);
+	xt_compat_init_offsets(AF_INET6, compatr->num_entries);
 	/* Walk through entries, checking offsets. */
-	xt_entry_foreach(iter0, entry0, total_size) {
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		ret = check_compat_entry_size_and_hooks(iter0, info, &size,
 							entry0,
-							entry0 + total_size,
-							hook_entries,
-							underflows,
-							name);
+							entry0 + compatr->size,
+							compatr->hook_entry,
+							compatr->underflow);
 		if (ret != 0)
 			goto out_unlock;
 		++j;
 	}
 
 	ret = -EINVAL;
-	if (j != number) {
+	if (j != compatr->num_entries) {
 		duprintf("translate_compat_table: %u not %u entries\n",
-			 j, number);
+			 j, compatr->num_entries);
 		goto out_unlock;
 	}
 
 	/* Check hooks all assigned */
 	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
 		/* Only hooks which are valid */
-		if (!(valid_hooks & (1 << i)))
+		if (!(compatr->valid_hooks & (1 << i)))
 			continue;
 		if (info->hook_entry[i] == 0xFFFFFFFF) {
 			duprintf("Invalid hook entry %u %u\n",
-				 i, hook_entries[i]);
+				 i, info->hook_entry[i]);
 			goto out_unlock;
 		}
 		if (info->underflow[i] == 0xFFFFFFFF) {
 			duprintf("Invalid underflow %u %u\n",
-				 i, underflows[i]);
+				 i, info->underflow[i]);
 			goto out_unlock;
 		}
 	}
@@ -1738,17 +1730,17 @@ translate_compat_table(struct net *net,
 	if (!newinfo)
 		goto out_unlock;
 
-	newinfo->number = number;
+	newinfo->number = compatr->num_entries;
 	for (i = 0; i < NF_INET_NUMHOOKS; i++) {
 		newinfo->hook_entry[i] = info->hook_entry[i];
 		newinfo->underflow[i] = info->underflow[i];
 	}
 	entry1 = newinfo->entries;
 	pos = entry1;
-	size = total_size;
-	xt_entry_foreach(iter0, entry0, total_size) {
+	size = compatr->size;
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		ret = compat_copy_entry_from_user(iter0, &pos, &size,
-						  name, newinfo, entry1);
+						  newinfo, entry1);
 		if (ret != 0)
 			break;
 	}
@@ -1758,12 +1750,12 @@ translate_compat_table(struct net *net,
 		goto free_newinfo;
 
 	ret = -ELOOP;
-	if (!mark_source_chains(newinfo, valid_hooks, entry1))
+	if (!mark_source_chains(newinfo, compatr->valid_hooks, entry1))
 		goto free_newinfo;
 
 	i = 0;
 	xt_entry_foreach(iter1, entry1, newinfo->size) {
-		ret = compat_check_entry(iter1, net, name);
+		ret = compat_check_entry(iter1, net, compatr->name);
 		if (ret != 0)
 			break;
 		++i;
@@ -1803,7 +1795,7 @@ translate_compat_table(struct net *net,
 free_newinfo:
 	xt_free_table_info(newinfo);
 out:
-	xt_entry_foreach(iter0, entry0, total_size) {
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		if (j-- == 0)
 			break;
 		compat_release_entry(iter0);
@@ -1848,10 +1840,7 @@ compat_do_replace(struct net *net, void __user *user, unsigned int len)
 		goto free_newinfo;
 	}
 
-	ret = translate_compat_table(net, tmp.name, tmp.valid_hooks,
-				     &newinfo, &loc_cpu_entry, tmp.size,
-				     tmp.num_entries, tmp.hook_entry,
-				     tmp.underflow);
+	ret = translate_compat_table(net, &newinfo, &loc_cpu_entry, &tmp);
 	if (ret != 0)
 		goto free_newinfo;
 
Signature · 1 line
2.7.3
author maintainer
#13 lore raw

PATCH 12/17 netfilter: arp_tables: simplify translate_compat_table args

In reply to Florian Westphal
---
 net/ipv4/netfilter/arp_tables.c | 82 ++++++++++++++++++-----------------------
 1 file changed, 36 insertions(+), 46 deletions(-)
net/ipv4/netfilter/arp_tables.c +36 -46
index 95ed4e4..1d1386d 100644
@@ -1214,6 +1214,18 @@ static int do_add_counters(struct net *net, const void __user *user,
 }
 
 #ifdef CONFIG_COMPAT
+struct compat_arpt_replace {
+	char				name[XT_TABLE_MAXNAMELEN];
+	u32				valid_hooks;
+	u32				num_entries;
+	u32				size;
+	u32				hook_entry[NF_ARP_NUMHOOKS];
+	u32				underflow[NF_ARP_NUMHOOKS];
+	u32				num_counters;
+	compat_uptr_t			counters;
+	struct compat_arpt_entry	entries[0];
+};
+
 static inline void compat_release_entry(struct compat_arpt_entry *e)
 {
 	struct xt_entry_target *t;
@@ -1229,8 +1241,7 @@ check_compat_entry_size_and_hooks(struct compat_arpt_entry *e,
 				  const unsigned char *base,
 				  const unsigned char *limit,
 				  const unsigned int *hook_entries,
-				  const unsigned int *underflows,
-				  const char *name)
+				  const unsigned int *underflows)
 {
 	struct xt_entry_target *t;
 	struct xt_target *target;
@@ -1301,7 +1312,7 @@ out:
 
 static int
 compat_copy_entry_from_user(struct compat_arpt_entry *e, void **dstptr,
-			    unsigned int *size, const char *name,
+			    unsigned int *size,
 			    struct xt_table_info *newinfo, unsigned char *base)
 {
 	struct xt_entry_target *t;
@@ -1334,14 +1345,9 @@ compat_copy_entry_from_user(struct compat_arpt_entry *e, void **dstptr,
 	return ret;
 }
 
-static int translate_compat_table(const char *name,
-				  unsigned int valid_hooks,
-				  struct xt_table_info **pinfo,
+static int translate_compat_table(struct xt_table_info **pinfo,
 				  void **pentry0,
-				  unsigned int total_size,
-				  unsigned int number,
-				  unsigned int *hook_entries,
-				  unsigned int *underflows)
+				  const struct compat_arpt_replace *compatr)
 {
 	unsigned int i, j;
 	struct xt_table_info *newinfo, *info;
@@ -1353,8 +1359,8 @@ static int translate_compat_table(const char *name,
 
 	info = *pinfo;
 	entry0 = *pentry0;
-	size = total_size;
-	info->number = number;
+	size = compatr->size;
+	info->number = compatr->num_entries;
 
 	/* Init all hooks to impossible value. */
 	for (i = 0; i < NF_ARP_NUMHOOKS; i++) {
@@ -1365,40 +1371,39 @@ static int translate_compat_table(const char *name,
 	duprintf("translate_compat_table: size %u\n", info->size);
 	j = 0;
 	xt_compat_lock(NFPROTO_ARP);
-	xt_compat_init_offsets(NFPROTO_ARP, number);
+	xt_compat_init_offsets(NFPROTO_ARP, compatr->num_entries);
 	/* Walk through entries, checking offsets. */
-	xt_entry_foreach(iter0, entry0, total_size) {
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		ret = check_compat_entry_size_and_hooks(iter0, info, &size,
 							entry0,
-							entry0 + total_size,
-							hook_entries,
-							underflows,
-							name);
+							entry0 + compatr->size,
+							compatr->hook_entry,
+							compatr->underflow);
 		if (ret != 0)
 			goto out_unlock;
 		++j;
 	}
 
 	ret = -EINVAL;
-	if (j != number) {
+	if (j != compatr->num_entries) {
 		duprintf("translate_compat_table: %u not %u entries\n",
-			 j, number);
+			 j, compatr->num_entries);
 		goto out_unlock;
 	}
 
 	/* Check hooks all assigned */
 	for (i = 0; i < NF_ARP_NUMHOOKS; i++) {
 		/* Only hooks which are valid */
-		if (!(valid_hooks & (1 << i)))
+		if (!(compatr->valid_hooks & (1 << i)))
 			continue;
 		if (info->hook_entry[i] == 0xFFFFFFFF) {
 			duprintf("Invalid hook entry %u %u\n",
-				 i, hook_entries[i]);
+				 i, info->hook_entry[i]);
 			goto out_unlock;
 		}
 		if (info->underflow[i] == 0xFFFFFFFF) {
 			duprintf("Invalid underflow %u %u\n",
-				 i, underflows[i]);
+				 i, info->underflow[i]);
 			goto out_unlock;
 		}
 	}
@@ -1408,17 +1413,17 @@ static int translate_compat_table(const char *name,
 	if (!newinfo)
 		goto out_unlock;
 
-	newinfo->number = number;
+	newinfo->number = compatr->num_entries;
 	for (i = 0; i < NF_ARP_NUMHOOKS; i++) {
 		newinfo->hook_entry[i] = info->hook_entry[i];
 		newinfo->underflow[i] = info->underflow[i];
 	}
 	entry1 = newinfo->entries;
 	pos = entry1;
-	size = total_size;
-	xt_entry_foreach(iter0, entry0, total_size) {
+	size = compatr->size;
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		ret = compat_copy_entry_from_user(iter0, &pos, &size,
-						  name, newinfo, entry1);
+						  newinfo, entry1);
 		if (ret != 0)
 			break;
 	}
@@ -1428,7 +1433,7 @@ static int translate_compat_table(const char *name,
 		goto free_newinfo;
 
 	ret = -ELOOP;
-	if (!mark_source_chains(newinfo, valid_hooks, entry1))
+	if (!mark_source_chains(newinfo, compatr->valid_hooks, entry1))
 		goto free_newinfo;
 
 	i = 0;
@@ -1439,7 +1444,7 @@ static int translate_compat_table(const char *name,
 			break;
 		}
 
-		ret = check_target(iter1, name);
+		ret = check_target(iter1, compatr->name);
 		if (ret != 0) {
 			xt_percpu_counter_free(iter1->counters.pcnt);
 			break;
@@ -1481,7 +1486,7 @@ static int translate_compat_table(const char *name,
 free_newinfo:
 	xt_free_table_info(newinfo);
 out:
-	xt_entry_foreach(iter0, entry0, total_size) {
+	xt_entry_foreach(iter0, entry0, compatr->size) {
 		if (j-- == 0)
 			break;
 		compat_release_entry(iter0);
@@ -1493,18 +1498,6 @@ out_unlock:
 	goto out;
 }
 
-struct compat_arpt_replace {
-	char				name[XT_TABLE_MAXNAMELEN];
-	u32				valid_hooks;
-	u32				num_entries;
-	u32				size;
-	u32				hook_entry[NF_ARP_NUMHOOKS];
-	u32				underflow[NF_ARP_NUMHOOKS];
-	u32				num_counters;
-	compat_uptr_t			counters;
-	struct compat_arpt_entry	entries[0];
-};
-
 static int compat_do_replace(struct net *net, void __user *user,
 			     unsigned int len)
 {
@@ -1537,10 +1530,7 @@ static int compat_do_replace(struct net *net, void __user *user,
 		goto free_newinfo;
 	}
 
-	ret = translate_compat_table(tmp.name, tmp.valid_hooks,
-				     &newinfo, &loc_cpu_entry, tmp.size,
-				     tmp.num_entries, tmp.hook_entry,
-				     tmp.underflow);
+	ret = translate_compat_table(&newinfo, &loc_cpu_entry, &tmp);
 	if (ret != 0)
 		goto free_newinfo;
 
Signature · 1 line
2.7.3
author maintainer
#14 lore raw

PATCH 13/17 netfilter: x_tables: xt_compat_match_from_user doesn't need a retval

In reply to Florian Westphal

Always returned 0.

---
 include/linux/netfilter/x_tables.h |  2 +-
 net/ipv4/netfilter/arp_tables.c    | 17 +++++------------
 net/ipv4/netfilter/ip_tables.c     | 26 +++++++++-----------------
 net/ipv6/netfilter/ip6_tables.c    | 27 +++++++++------------------
 net/netfilter/x_tables.c           |  5 ++---
 5 files changed, 26 insertions(+), 51 deletions(-)
include/linux/netfilter/x_tables.h +1 -1
index 30cfb1e..e2da9b9 100644
@@ -484,7 +484,7 @@ void xt_compat_init_offsets(u_int8_t af, unsigned int number);
 int xt_compat_calc_jump(u_int8_t af, unsigned int offset);
 
 int xt_compat_match_offset(const struct xt_match *match);
-int xt_compat_match_from_user(struct xt_entry_match *m, void **dstptr,
+void xt_compat_match_from_user(struct xt_entry_match *m, void **dstptr,
 			      unsigned int *size);
 int xt_compat_match_to_user(const struct xt_entry_match *m,
 			    void __user **dstptr, unsigned int *size);
net/ipv4/netfilter/arp_tables.c +5 -12
index 1d1386d..be514c6 100644
@@ -1310,7 +1310,7 @@ out:
 	return ret;
 }
 
-static int
+static void
 compat_copy_entry_from_user(struct compat_arpt_entry *e, void **dstptr,
 			    unsigned int *size,
 			    struct xt_table_info *newinfo, unsigned char *base)
@@ -1319,9 +1319,8 @@ compat_copy_entry_from_user(struct compat_arpt_entry *e, void **dstptr,
 	struct xt_target *target;
 	struct arpt_entry *de;
 	unsigned int origsize;
-	int ret, h;
+	int h;
 
-	ret = 0;
 	origsize = *size;
 	de = (struct arpt_entry *)*dstptr;
 	memcpy(de, e, sizeof(struct arpt_entry));
@@ -1342,7 +1341,6 @@ compat_copy_entry_from_user(struct compat_arpt_entry *e, void **dstptr,
 		if ((unsigned char *)de - base < newinfo->underflow[h])
 			newinfo->underflow[h] -= origsize - *size;
 	}
-	return ret;
 }
 
 static int translate_compat_table(struct xt_table_info **pinfo,
@@ -1421,16 +1419,11 @@ static int translate_compat_table(struct xt_table_info **pinfo,
 	entry1 = newinfo->entries;
 	pos = entry1;
 	size = compatr->size;
-	xt_entry_foreach(iter0, entry0, compatr->size) {
-		ret = compat_copy_entry_from_user(iter0, &pos, &size,
-						  newinfo, entry1);
-		if (ret != 0)
-			break;
-	}
+	xt_entry_foreach(iter0, entry0, compatr->size)
+		compat_copy_entry_from_user(iter0, &pos, &size,
+					    newinfo, entry1);
 	xt_compat_flush_offsets(NFPROTO_ARP);
 	xt_compat_unlock(NFPROTO_ARP);
-	if (ret)
-		goto free_newinfo;
 
 	ret = -ELOOP;
 	if (!mark_source_chains(newinfo, compatr->valid_hooks, entry1))
net/ipv4/netfilter/ip_tables.c +9 -17
index d7041860..5c20eef 100644
@@ -1568,7 +1568,7 @@ release_matches:
 	return ret;
 }
 
-static int
+static void
 compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
 			    unsigned int *size,
 			    struct xt_table_info *newinfo, unsigned char *base)
@@ -1577,10 +1577,9 @@ compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
 	struct xt_target *target;
 	struct ipt_entry *de;
 	unsigned int origsize;
-	int ret, h;
+	int h;
 	struct xt_entry_match *ematch;
 
-	ret = 0;
 	origsize = *size;
 	de = (struct ipt_entry *)*dstptr;
 	memcpy(de, e, sizeof(struct ipt_entry));
@@ -1589,11 +1588,9 @@ compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
 	*dstptr += sizeof(struct ipt_entry);
 	*size += sizeof(struct ipt_entry) - sizeof(struct compat_ipt_entry);
 
-	xt_ematch_foreach(ematch, e) {
-		ret = xt_compat_match_from_user(ematch, dstptr, size);
-		if (ret != 0)
-			return ret;
-	}
+	xt_ematch_foreach(ematch, e)
+		xt_compat_match_from_user(ematch, dstptr, size);
+
 	de->target_offset = e->target_offset - (origsize - *size);
 	t = compat_ipt_get_target(e);
 	target = t->u.kernel.target;
@@ -1606,7 +1603,6 @@ compat_copy_entry_from_user(struct compat_ipt_entry *e, void **dstptr,
 		if ((unsigned char *)de - base < newinfo->underflow[h])
 			newinfo->underflow[h] -= origsize - *size;
 	}
-	return ret;
 }
 
 static int
@@ -1729,16 +1725,12 @@ translate_compat_table(struct net *net,
 	entry1 = newinfo->entries;
 	pos = entry1;
 	size = compatr->size;
-	xt_entry_foreach(iter0, entry0, compatr->size) {
-		ret = compat_copy_entry_from_user(iter0, &pos, &size,
-						  newinfo, entry1);
-		if (ret != 0)
-			break;
-	}
+	xt_entry_foreach(iter0, entry0, compatr->size)
+		compat_copy_entry_from_user(iter0, &pos, &size,
+					    newinfo, entry1);
+
 	xt_compat_flush_offsets(AF_INET);
 	xt_compat_unlock(AF_INET);
-	if (ret)
-		goto free_newinfo;
 
 	ret = -ELOOP;
 	if (!mark_source_chains(newinfo, compatr->valid_hooks, entry1))
net/ipv6/netfilter/ip6_tables.c +9 -18
index 8d082c55..620d54c 100644
@@ -1580,7 +1580,7 @@ release_matches:
 	return ret;
 }
 
-static int
+static void
 compat_copy_entry_from_user(struct compat_ip6t_entry *e, void **dstptr,
 			    unsigned int *size,
 			    struct xt_table_info *newinfo, unsigned char *base)
@@ -1588,10 +1588,9 @@ compat_copy_entry_from_user(struct compat_ip6t_entry *e, void **dstptr,
 	struct xt_entry_target *t;
 	struct ip6t_entry *de;
 	unsigned int origsize;
-	int ret, h;
+	int h;
 	struct xt_entry_match *ematch;
 
-	ret = 0;
 	origsize = *size;
 	de = (struct ip6t_entry *)*dstptr;
 	memcpy(de, e, sizeof(struct ip6t_entry));
@@ -1600,11 +1599,9 @@ compat_copy_entry_from_user(struct compat_ip6t_entry *e, void **dstptr,
 	*dstptr += sizeof(struct ip6t_entry);
 	*size += sizeof(struct ip6t_entry) - sizeof(struct compat_ip6t_entry);
 
-	xt_ematch_foreach(ematch, e) {
-		ret = xt_compat_match_from_user(ematch, dstptr, size);
-		if (ret != 0)
-			return ret;
-	}
+	xt_ematch_foreach(ematch, e)
+		xt_compat_match_from_user(ematch, dstptr, size);
+
 	de->target_offset = e->target_offset - (origsize - *size);
 	t = compat_ip6t_get_target(e);
 	xt_compat_target_from_user(t, dstptr, size);
@@ -1616,7 +1613,6 @@ compat_copy_entry_from_user(struct compat_ip6t_entry *e, void **dstptr,
 		if ((unsigned char *)de - base < newinfo->underflow[h])
 			newinfo->underflow[h] -= origsize - *size;
 	}
-	return ret;
 }
 
 static int compat_check_entry(struct ip6t_entry *e, struct net *net,
@@ -1737,17 +1733,12 @@ translate_compat_table(struct net *net,
 	}
 	entry1 = newinfo->entries;
 	pos = entry1;
-	size = compatr->size;
-	xt_entry_foreach(iter0, entry0, compatr->size) {
-		ret = compat_copy_entry_from_user(iter0, &pos, &size,
-						  newinfo, entry1);
-		if (ret != 0)
-			break;
-	}
+	xt_entry_foreach(iter0, entry0, compatr->size)
+		compat_copy_entry_from_user(iter0, &pos, &size,
+					    newinfo, entry1);
+
 	xt_compat_flush_offsets(AF_INET6);
 	xt_compat_unlock(AF_INET6);
-	if (ret)
-		goto free_newinfo;
 
 	ret = -ELOOP;
 	if (!mark_source_chains(newinfo, compatr->valid_hooks, entry1))
net/netfilter/x_tables.c +2 -3
index f9aa971..7e7173b 100644
@@ -526,8 +526,8 @@ int xt_compat_match_offset(const struct xt_match *match)
 }
 EXPORT_SYMBOL_GPL(xt_compat_match_offset);
 
-int xt_compat_match_from_user(struct xt_entry_match *m, void **dstptr,
-			      unsigned int *size)
+void xt_compat_match_from_user(struct xt_entry_match *m, void **dstptr,
+			       unsigned int *size)
 {
 	const struct xt_match *match = m->u.kernel.match;
 	struct compat_xt_entry_match *cm = (struct compat_xt_entry_match *)m;
@@ -549,7 +549,6 @@ int xt_compat_match_from_user(struct xt_entry_match *m, void **dstptr,
 
 	*size += off;
 	*dstptr += msize;
-	return 0;
 }
 EXPORT_SYMBOL_GPL(xt_compat_match_from_user);
 
Signature · 1 line
2.7.3
author maintainer
#15 lore raw

PATCH 14/17 netfilter: x_tables: do compat validation via translate_table

In reply to Florian Westphal

This looks like refactoring, but its also a bug fix.

Problem is that the compat path (32bit iptables, 64bit kernel) lacks a few sanity tests that are done in the normal path.

For example, we do not check for underflows and the base chain policies.

While its possible to also add such checks to the compat path, its more copy&pastry, for instance we cannot reuse check_underflow() helper as e->target_offset differs in the compat case.

Other problem is that it makes auditing for validation errors harder; two places need to be checked and kept in sync.

At a high level 32 bit compat works like this:
1- initial pass over blob:
validate match/entry offsets, bounds checking lookup all matches and targets
do bookkeeping wrt. size delta of 32/64bit structures assign match/target.u.kernel pointer (points at kernel implementation, needed to access ->compatsize etc.)

2- allocate memory according to the total bookkeeping size to
contain the translated ruleset

3- second pass over original blob:
for each entry, copy the 32bit representation to the newly allocated memory. This also does any special match translations (e.g. adjust 32bit to 64bit longs, etc).

4- check if ruleset is free of loops (chase all jumps)

5-first pass over translated blob:
call the checkentry function of all matches and targets.

The alternative implemented by this patch is to drop steps 3&4 from the compat process, the translation is changed into an intermediate step rather than a full 1:1 translate_table replacement.

In the 2nd pass (step #3), change the 64bit ruleset back to a kernel representation, i.e. put() the kernel pointer and restore ->u.user.name .

This gets us a 64bit ruleset that is in the format generated by a 64bit iptables userspace -- we can then use translate_table() to get the 'native' sanity checks.

This has two drawbacks:

1. we re-validate all the match and target entry structure sizes even though compat translation is supposed to never generate bogus offsets.
2. we put and then re-lookup each match and target.

THe upside is that we get all sanity tests and ruleset validations provided by the normal path and can remove some duplicated compat code.

iptables-restore time of autogenerated ruleset with 300k chains of form -A CHAIN0001 -m limit --limit 1/s -j CHAIN0002
-A CHAIN0002 -m limit --limit 1/s -j CHAIN0003

shows no noticeable differences in restore times:

old:   0m30.796s
new:   0m31.521s

64bit: 0m25.674s

---
 net/ipv4/netfilter/arp_tables.c | 114 ++++++-----------------------
 net/ipv4/netfilter/ip_tables.c  | 155 ++++++++--------------------------------
 net/ipv6/netfilter/ip6_tables.c | 148 ++++++--------------------------------
 net/netfilter/x_tables.c        |   8 +++
 4 files changed, 83 insertions(+), 342 deletions(-)
net/ipv4/netfilter/arp_tables.c +23 -91
index be514c6..705179b 100644
@@ -1234,19 +1234,17 @@ static inline void compat_release_entry(struct compat_arpt_entry *e)
 	module_put(t->u.kernel.target->me);
 }
 
-static inline int
+static int
 check_compat_entry_size_and_hooks(struct compat_arpt_entry *e,
 				  struct xt_table_info *newinfo,
 				  unsigned int *size,
 				  const unsigned char *base,
-				  const unsigned char *limit,
-				  const unsigned int *hook_entries,
-				  const unsigned int *underflows)
+				  const unsigned char *limit)
 {
 	struct xt_entry_target *t;
 	struct xt_target *target;
 	unsigned int entry_offset;
-	int ret, off, h;
+	int ret, off;
 
 	duprintf("check_compat_entry_size_and_hooks %p\n", e);
 	if ((unsigned long)e % __alignof__(struct compat_arpt_entry) != 0 ||
@@ -1291,17 +1289,6 @@ check_compat_entry_size_and_hooks(struct compat_arpt_entry *e,
 	if (ret)
 		goto release_target;
 
-	/* Check hooks & underflows */
-	for (h = 0; h < NF_ARP_NUMHOOKS; h++) {
-		if ((unsigned char *)e - base == hook_entries[h])
-			newinfo->hook_entry[h] = hook_entries[h];
-		if ((unsigned char *)e - base == underflows[h])
-			newinfo->underflow[h] = underflows[h];
-	}
-
-	/* Clear counters and comefrom */
-	memset(&e->counters, 0, sizeof(e->counters));
-	e->comefrom = 0;
 	return 0;
 
 release_target:
@@ -1351,7 +1338,7 @@ static int translate_compat_table(struct xt_table_info **pinfo,
 	struct xt_table_info *newinfo, *info;
 	void *pos, *entry0, *entry1;
 	struct compat_arpt_entry *iter0;
-	struct arpt_entry *iter1;
+	struct arpt_replace repl;
 	unsigned int size;
 	int ret = 0;
 
@@ -1360,12 +1347,6 @@ static int translate_compat_table(struct xt_table_info **pinfo,
 	size = compatr->size;
 	info->number = compatr->num_entries;
 
-	/* Init all hooks to impossible value. */
-	for (i = 0; i < NF_ARP_NUMHOOKS; i++) {
-		info->hook_entry[i] = 0xFFFFFFFF;
-		info->underflow[i] = 0xFFFFFFFF;
-	}
-
 	duprintf("translate_compat_table: size %u\n", info->size);
 	j = 0;
 	xt_compat_lock(NFPROTO_ARP);
@@ -1374,9 +1355,7 @@ static int translate_compat_table(struct xt_table_info **pinfo,
 	xt_entry_foreach(iter0, entry0, compatr->size) {
 		ret = check_compat_entry_size_and_hooks(iter0, info, &size,
 							entry0,
-							entry0 + compatr->size,
-							compatr->hook_entry,
-							compatr->underflow);
+							entry0 + compatr->size);
 		if (ret != 0)
 			goto out_unlock;
 		++j;
@@ -1389,23 +1368,6 @@ static int translate_compat_table(struct xt_table_info **pinfo,
 		goto out_unlock;
 	}
 
-	/* Check hooks all assigned */
-	for (i = 0; i < NF_ARP_NUMHOOKS; i++) {
-		/* Only hooks which are valid */
-		if (!(compatr->valid_hooks & (1 << i)))
-			continue;
-		if (info->hook_entry[i] == 0xFFFFFFFF) {
-			duprintf("Invalid hook entry %u %u\n",
-				 i, info->hook_entry[i]);
-			goto out_unlock;
-		}
-		if (info->underflow[i] == 0xFFFFFFFF) {
-			duprintf("Invalid underflow %u %u\n",
-				 i, info->underflow[i]);
-			goto out_unlock;
-		}
-	}
-
 	ret = -ENOMEM;
 	newinfo = xt_alloc_table_info(size);
 	if (!newinfo)
@@ -1422,55 +1384,26 @@ static int translate_compat_table(struct xt_table_info **pinfo,
 	xt_entry_foreach(iter0, entry0, compatr->size)
 		compat_copy_entry_from_user(iter0, &pos, &size,
 					    newinfo, entry1);
+
+	/* all module references in entry0 are now gone */
+
 	xt_compat_flush_offsets(NFPROTO_ARP);
 	xt_compat_unlock(NFPROTO_ARP);
 
-	ret = -ELOOP;
-	if (!mark_source_chains(newinfo, compatr->valid_hooks, entry1))
-		goto free_newinfo;
-
-	i = 0;
-	xt_entry_foreach(iter1, entry1, newinfo->size) {
-		iter1->counters.pcnt = xt_percpu_counter_alloc();
-		if (IS_ERR_VALUE(iter1->counters.pcnt)) {
-			ret = -ENOMEM;
-			break;
-		}
+	memcpy(&repl, compatr, sizeof(*compatr));
 
-		ret = check_target(iter1, compatr->name);
-		if (ret != 0) {
-			xt_percpu_counter_free(iter1->counters.pcnt);
-			break;
-		}
-		++i;
-		if (strcmp(arpt_get_target(iter1)->u.user.name,
-		    XT_ERROR_TARGET) == 0)
-			++newinfo->stacksize;
-	}
-	if (ret) {
-		/*
-		 * The first i matches need cleanup_entry (calls ->destroy)
-		 * because they had called ->check already. The other j-i
-		 * entries need only release.
-		 */
-		int skip = i;
-		j -= i;
-		xt_entry_foreach(iter0, entry0, newinfo->size) {
-			if (skip-- > 0)
-				continue;
-			if (j-- == 0)
-				break;
-			compat_release_entry(iter0);
-		}
-		xt_entry_foreach(iter1, entry1, newinfo->size) {
-			if (i-- == 0)
-				break;
-			cleanup_entry(iter1);
-		}
-		xt_free_table_info(newinfo);
-		return ret;
+	for (i = 0; i < NF_ARP_NUMHOOKS; i++) {
+		repl.hook_entry[i] = newinfo->hook_entry[i];
+		repl.underflow[i] = newinfo->underflow[i];
 	}
 
+	repl.num_counters = 0;
+	repl.counters = NULL;
+	repl.size = newinfo->size;
+	ret = translate_table(newinfo, entry1, &repl);
+	if (ret)
+		goto free_newinfo;
+
 	*pinfo = newinfo;
 	*pentry0 = entry1;
 	xt_free_table_info(info);
@@ -1478,17 +1411,16 @@ static int translate_compat_table(struct xt_table_info **pinfo,
 
 free_newinfo:
 	xt_free_table_info(newinfo);
-out:
+	return ret;
+out_unlock:
+	xt_compat_flush_offsets(NFPROTO_ARP);
+	xt_compat_unlock(NFPROTO_ARP);
 	xt_entry_foreach(iter0, entry0, compatr->size) {
 		if (j-- == 0)
 			break;
 		compat_release_entry(iter0);
 	}
 	return ret;
-out_unlock:
-	xt_compat_flush_offsets(NFPROTO_ARP);
-	xt_compat_unlock(NFPROTO_ARP);
-	goto out;
 }
 
 static int compat_do_replace(struct net *net, void __user *user,
net/ipv4/netfilter/ip_tables.c +29 -126

239 lines not on this page · raw message

net/ipv6/netfilter/ip6_tables.c +23 -125

227 lines not on this page · raw message

net/netfilter/x_tables.c +8 -0
index 7e7173b..9ec23ffa 100644
@@ -533,6 +533,7 @@ void xt_compat_match_from_user(struct xt_entry_match *m, void **dstptr,
 	struct compat_xt_entry_match *cm = (struct compat_xt_entry_match *)m;
 	int pad, off = xt_compat_match_offset(match);
 	u_int16_t msize = cm->u.user.match_size;
+	char name[sizeof(m->u.user.name)];
 
 	m = *dstptr;
 	memcpy(m, cm, sizeof(*cm));
@@ -546,6 +547,9 @@ void xt_compat_match_from_user(struct xt_entry_match *m, void **dstptr,
 
 	msize += off;
 	m->u.user.match_size = msize;
+	strlcpy(name, match->name, sizeof(name));
+	module_put(match->me);
+	strncpy(m->u.user.name, name, sizeof(m->u.user.name));
 
 	*size += off;
 	*dstptr += msize;
@@ -763,6 +767,7 @@ void xt_compat_target_from_user(struct xt_entry_target *t, void **dstptr,
 	struct compat_xt_entry_target *ct = (struct compat_xt_entry_target *)t;
 	int pad, off = xt_compat_target_offset(target);
 	u_int16_t tsize = ct->u.user.target_size;
+	char name[sizeof(t->u.user.name)];
 
 	t = *dstptr;
 	memcpy(t, ct, sizeof(*ct));
@@ -776,6 +781,9 @@ void xt_compat_target_from_user(struct xt_entry_target *t, void **dstptr,
 
 	tsize += off;
 	t->u.user.target_size = tsize;
+	strlcpy(name, target->name, sizeof(name));
+	module_put(target->me);
+	strncpy(t->u.user.name, name, sizeof(t->u.user.name));
 
 	*size += off;
 	*dstptr += tsize;
Signature · 1 line
2.7.3
author maintainer
#16 lore raw

PATCH 15/17 netfilter: x_tables: remove obsolete overflow check for compat case too

In reply to Florian Westphal

commit 9e67d5a739327c44885adebb4f3a538050be73e4 ("[NETFILTER]: x_tables: remove obsolete overflow check") left the compat parts alone, but we can kill it there as well.

---
 net/ipv4/netfilter/arp_tables.c | 2 --
 net/ipv4/netfilter/ip_tables.c  | 2 --
 net/ipv6/netfilter/ip6_tables.c | 2 --
 3 files changed, 6 deletions(-)
net/ipv4/netfilter/arp_tables.c +0 -2
index 705179b..668c5dc 100644
@@ -1436,8 +1436,6 @@ static int compat_do_replace(struct net *net, void __user *user,
 		return -EFAULT;
 
 	/* overflow check */
-	if (tmp.size >= INT_MAX / num_possible_cpus())
-		return -ENOMEM;
 	if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
 		return -ENOMEM;
 	if (tmp.num_counters == 0)
net/ipv4/netfilter/ip_tables.c +0 -2
index c26ccd8..4585aa7 100644
@@ -1706,8 +1706,6 @@ compat_do_replace(struct net *net, void __user *user, unsigned int len)
 		return -EFAULT;
 
 	/* overflow check */
-	if (tmp.size >= INT_MAX / num_possible_cpus())
-		return -ENOMEM;
 	if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
 		return -ENOMEM;
 	if (tmp.num_counters == 0)
net/ipv6/netfilter/ip6_tables.c +0 -2
index f5a4eb2..fd06251 100644
@@ -1709,8 +1709,6 @@ compat_do_replace(struct net *net, void __user *user, unsigned int len)
 		return -EFAULT;
 
 	/* overflow check */
-	if (tmp.size >= INT_MAX / num_possible_cpus())
-		return -ENOMEM;
 	if (tmp.num_counters >= INT_MAX / sizeof(struct xt_counters))
 		return -ENOMEM;
 	if (tmp.num_counters == 0)
Signature · 1 line
2.7.3
author maintainer
#17 lore raw

PATCH 16/17 netfilter: x_tables: remove obsolete check

In reply to Florian Westphal

Since 'netfilter: x_tables: validate targets of jumps' change we validate that the target aligns exactly with beginning of a rule, so offset test is now redundant.

---
 net/ipv4/netfilter/arp_tables.c | 8 --------
 net/ipv4/netfilter/ip_tables.c  | 7 -------
 net/ipv6/netfilter/ip6_tables.c | 7 -------
 3 files changed, 22 deletions(-)
net/ipv4/netfilter/arp_tables.c +0 -8
index 668c5dc..8cefb7a 100644
@@ -461,14 +461,6 @@ static int mark_source_chains(const struct xt_table_info *newinfo,
 				if (strcmp(t->target.u.user.name,
 					   XT_STANDARD_TARGET) == 0 &&
 				    newpos >= 0) {
-					if (newpos > newinfo->size -
-						sizeof(struct arpt_entry)) {
-						duprintf("mark_source_chains: "
-							"bad verdict (%i)\n",
-								newpos);
-						return 0;
-					}
-
 					/* This a jump; chase it. */
 					duprintf("Jump rule %u -> %u\n",
 						 pos, newpos);
net/ipv4/netfilter/ip_tables.c +0 -7
index 4585aa7..9340ce0 100644
@@ -542,13 +542,6 @@ mark_source_chains(const struct xt_table_info *newinfo,
 				if (strcmp(t->target.u.user.name,
 					   XT_STANDARD_TARGET) == 0 &&
 				    newpos >= 0) {
-					if (newpos > newinfo->size -
-						sizeof(struct ipt_entry)) {
-						duprintf("mark_source_chains: "
-							"bad verdict (%i)\n",
-								newpos);
-						return 0;
-					}
 					/* This a jump; chase it. */
 					duprintf("Jump rule %u -> %u\n",
 						 pos, newpos);
net/ipv6/netfilter/ip6_tables.c +0 -7
index fd06251..aa01085 100644
@@ -554,13 +554,6 @@ mark_source_chains(const struct xt_table_info *newinfo,
 				if (strcmp(t->target.u.user.name,
 					   XT_STANDARD_TARGET) == 0 &&
 				    newpos >= 0) {
-					if (newpos > newinfo->size -
-						sizeof(struct ip6t_entry)) {
-						duprintf("mark_source_chains: "
-							"bad verdict (%i)\n",
-								newpos);
-						return 0;
-					}
 					/* This a jump; chase it. */
 					duprintf("Jump rule %u -> %u\n",
 						 pos, newpos);
Signature · 1 line
2.7.3
author maintainer
#18 lore raw

PATCH 17/17 netfilter: x_tables: introduce and use xt_copy_counters_from_user

In reply to Florian Westphal

The three variants use same copy&pasted code, condense this into a helper and use that.

Make sure info.name is 0-terminated.

---
 include/linux/netfilter/x_tables.h |  3 ++
 net/ipv4/netfilter/arp_tables.c    | 48 +++----------------------
 net/ipv4/netfilter/ip_tables.c     | 48 +++----------------------
 net/ipv6/netfilter/ip6_tables.c    | 49 +++----------------------
 net/netfilter/x_tables.c           | 73 ++++++++++++++++++++++++++++++++++++++
 5 files changed, 91 insertions(+), 130 deletions(-)
include/linux/netfilter/x_tables.h +3 -0
index e2da9b9..4dd9306 100644
@@ -251,6 +251,9 @@ int xt_check_match(struct xt_mtchk_param *, unsigned int size, u_int8_t proto,
 int xt_check_target(struct xt_tgchk_param *, unsigned int size, u_int8_t proto,
 		    bool inv_proto);
 
+void *xt_copy_counters_from_user(const void __user *user, unsigned int len,
+				 struct xt_counters_info *info, bool compat);
+
 struct xt_table *xt_register_table(struct net *net,
 				   const struct xt_table *table,
 				   struct xt_table_info *bootstrap,
net/ipv4/netfilter/arp_tables.c +5 -43

67 lines not on this page · raw message

net/ipv4/netfilter/ip_tables.c +5 -43

67 lines not on this page · raw message

net/ipv6/netfilter/ip6_tables.c +5 -44

67 lines not on this page · raw message

net/netfilter/x_tables.c +73 -0

79 lines not on this page · raw message

Signature · 1 line
2.7.3
#19 lore raw

Re: [PATCH nf 17/17] netfilter: x_tables: introduce and use xt_copy_counters_from_user

reproduce: wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross # save the attached .config to linux build tree…

Hi Florian,

[auto build test ERROR on nf/master]

url:    https://github.com/0day-ci/linux/commits/Florian-Westphal/netfilter-x_tables-don-t-move-to-non-existent-next-rule/20160401-202022
base:   https://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf master
config: avr32-atngw100_defconfig (attached as .config)

reproduce:
wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
make.cross ARCH=avr32

All errors (new ones prefixed by >>):

ERROR: "xt_copy_counters_from_user" [net/ipv6/netfilter/ip6_tables.ko] undefined!

>> ERROR: "xt_copy_counters_from_user" [net/ipv4/netfilter/ip_tables.ko] undefined!

WARNING: modpost: Found 5 section mismatch(es). To see full details build your kernel with:
'make CONFIG_DEBUG_SECTION_MISMATCH=y'

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

#20 lore raw

Re: [PATCH nf 17/17] netfilter: x_tables: introduce and use xt_copy_counters_from_user

In reply to Florian Westphal
reproduce: wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross # save the attached .config to linux build tree…

Hi Florian,

[auto build test ERROR on nf/master]

url:    https://github.com/0day-ci/linux/commits/Florian-Westphal/netfilter-x_tables-don-t-move-to-non-existent-next-rule/20160401-202022
base:   https://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf master
config: sh-titan_defconfig (attached as .config)

reproduce:
wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# save the attached .config to linux build tree
make.cross ARCH=sh

All errors (new ones prefixed by >>):

ERROR: "xt_copy_counters_from_user" [net/ipv6/netfilter/ip6_tables.ko] undefined!
ERROR: "xt_copy_counters_from_user" [net/ipv4/netfilter/ip_tables.ko] undefined!

>> ERROR: "xt_copy_counters_from_user" [net/ipv4/netfilter/arp_tables.ko] undefined!

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

#21 lore raw

Re: [PATCH nf 17/17] netfilter: x_tables: introduce and use xt_copy_counters_from_user

In reply to Florian Westphal
reproduce:

Hi Florian,

[auto build test ERROR on nf/master]

url:    https://github.com/0day-ci/linux/commits/Florian-Westphal/netfilter-x_tables-don-t-move-to-non-existent-next-rule/20160401-202022
base:   https://git.kernel.org/pub/scm/linux/kernel/git/pablo/nf master
config: i386-randconfig-s0-201613 (attached as .config)
reproduce:
        # save the attached .config to linux build tree
        make ARCH=i386 

All errors (new ones prefixed by >>):

>> ERROR: "xt_copy_counters_from_user" [net/ipv6/netfilter/ip6_tables.ko] undefined!

---
0-DAY kernel test infrastructure Open Source Technology Center
https://lists.01.org/pipermail/kbuild-all Intel Corporation

author maintainer
#22 lore raw

PATCH v2 17/17 netfilter: x_tables: introduce and use xt_copy_counters_from_user

In reply to Florian Westphal

The three variants use same copy&pasted code, condense this into a helper and use that.

Make sure info.name is 0-terminated.

---
V2: add EXPORT_SYMBOL(), else iptables=m builds fail.

 include/linux/netfilter/x_tables.h |  3 ++
 net/ipv4/netfilter/arp_tables.c    | 48 +++----------------------
 net/ipv4/netfilter/ip_tables.c     | 48 +++----------------------
 net/ipv6/netfilter/ip6_tables.c    | 49 +++----------------------
 net/netfilter/x_tables.c           | 74 ++++++++++++++++++++++++++++++++++++++
 5 files changed, 92 insertions(+), 130 deletions(-)
include/linux/netfilter/x_tables.h +3 -0
index e2da9b9..4dd9306 100644
@@ -251,6 +251,9 @@ int xt_check_match(struct xt_mtchk_param *, unsigned int size, u_int8_t proto,
 int xt_check_target(struct xt_tgchk_param *, unsigned int size, u_int8_t proto,
 		    bool inv_proto);
 
+void *xt_copy_counters_from_user(const void __user *user, unsigned int len,
+				 struct xt_counters_info *info, bool compat);
+
 struct xt_table *xt_register_table(struct net *net,
 				   const struct xt_table *table,
 				   struct xt_table_info *bootstrap,
net/ipv4/netfilter/arp_tables.c +5 -43

67 lines not on this page · raw message

net/ipv4/netfilter/ip_tables.c +5 -43

67 lines not on this page · raw message

net/ipv6/netfilter/ip6_tables.c +5 -44

67 lines not on this page · raw message

net/netfilter/x_tables.c +74 -0

80 lines not on this page · raw message

Signature · 1 line
2.7.3
#23 lore raw
In reply to Florian Westphal

Hi Florian,

Quoted 17 lines · Florian Westphal

This series adds more checks on xtables (arp, ip, ip6tables) rulesets.

- check all offsets (target, next) of all rules during initial pass after copy from userspace.
- check targets of jumps (-j bla): offset should be start of a rule
- assert that alleged target size is at least as big as minimum target structure
- change CONFIG_COMPAT code path to push ruleset via normal setsockopt path after initial 32->64 bit conversion to avoid duplicating checks
- use a common helper to copy counters from userspace instead of the ip/ip6/arp implementation.

Tested:
- iptables.git iptables-test.py passes
- made a few performance tests w. really silly rulesets to verify that things don't slow down too much, see individual patches for details.

Quoted 6 lines
 include/linux/netfilter/x_tables.h |   12 +
 net/ipv4/netfilter/arp_tables.c    |  303 ++++++++++------------------------
 net/ipv4/netfilter/ip_tables.c     |  327 +++++++++----------------------------
 net/ipv6/netfilter/ip6_tables.c    |  320 ++++++++----------------------------
 net/netfilter/x_tables.c           |  244 +++++++++++++++++++++++++++
 5 files changed, 506 insertions(+), 700 deletions(-)

Nice work, and we got less code to maintain, good :)

I'm starting to consider that, given that this has been broken since day 1, we pass this through nf-next and then later on we request inclusion for -stable.

We'll have more time in case of fallout (I know you have done a great effort to intensively test this) but this batch looks large that why I'm thinking about this route change.

Let me know, thanks!

author maintainer
#24 lore raw
Quoted 25 lines · Pablo Neira Ayuso
Quoted 17 lines · Florian Westphal

This series adds more checks on xtables (arp, ip, ip6tables) rulesets.

- check all offsets (target, next) of all rules during initial pass after copy from userspace.
- check targets of jumps (-j bla): offset should be start of a rule
- assert that alleged target size is at least as big as minimum target structure
- change CONFIG_COMPAT code path to push ruleset via normal setsockopt path after initial 32->64 bit conversion to avoid duplicating checks
- use a common helper to copy counters from userspace instead of the ip/ip6/arp implementation.

Tested:
- iptables.git iptables-test.py passes
- made a few performance tests w. really silly rulesets to verify that things don't slow down too much, see individual patches for details.

Quoted 6 lines
 include/linux/netfilter/x_tables.h |   12 +
 net/ipv4/netfilter/arp_tables.c    |  303 ++++++++++------------------------
 net/ipv4/netfilter/ip_tables.c     |  327 +++++++++----------------------------
 net/ipv6/netfilter/ip6_tables.c    |  320 ++++++++----------------------------
 net/netfilter/x_tables.c           |  244 +++++++++++++++++++++++++++
 5 files changed, 506 insertions(+), 700 deletions(-)
Quoted 5 lines

Nice work, and we got less code to maintain, good :)

I'm starting to consider that, given that this has been broken since day 1, we pass this through nf-next and then later on we request inclusion for -stable.

Fine with me.

#25 lore raw
Quoted 5 lines · Florian Westphal
Quoted 3 lines

I'm starting to consider that, given that this has been broken since day 1, we pass this through nf-next and then later on we request inclusion for -stable.

Fine with me.

Florian, I'm going to send a pull request to David, then resync and place this into nf-next. But first I'm going to send a pull request with the little stuff I have in that tree.

#26 lore raw
In reply to Florian Westphal
Quoted 12 lines · Florian Westphal

This series adds more checks on xtables (arp, ip, ip6tables) rulesets.

- check all offsets (target, next) of all rules during initial pass after copy from userspace.
- check targets of jumps (-j bla): offset should be start of a rule
- assert that alleged target size is at least as big as minimum target structure
- change CONFIG_COMPAT code path to push ruleset via normal setsockopt path after initial 32->64 bit conversion to avoid duplicating checks
- use a common helper to copy counters from userspace instead of the ip/ip6/arp implementation.

Quoted 4 lines

Tested:
- iptables.git iptables-test.py passes
- made a few performance tests w. really silly rulesets to verify that things don't slow down too much, see individual patches for details.

Series applied to nf-next, thanks for your patience.