From 4d96f633af0fe6c02f85da24d6b9c563761be99a Mon Sep 17 00:00:00 2001
From: Eric Dumazet <edumazet@google.com>
Date: Wed, 24 Mar 2021 14:53:37 -0700
Subject: [PATCH] inet: use bigger hash table for IP ID generation
Patch-mainline: v5.13-rc1
Git-commit: aa6dd211e4b1dde9d5dc25d699d35f789ae7eeba
References: CVE-2021-45486 bsc#1194087

In commit 73f156a6e8c1 ("inetpeer: get rid of ip_id_count")
I used a very small hash table that could be abused
by patient attackers to reveal sensitive information.

Switch to a dynamic sizing, depending on RAM size.

Typical big hosts will now use 128x more storage (2 MB)
to get a similar increase in security and reduction
of hash collisions.

As a bonus, use of alloc_large_system_hash() spreads
allocated memory among all NUMA nodes.

Fixes: 73f156a6e8c1 ("inetpeer: get rid of ip_id_count")
Reported-by: Amit Klein <aksecurity@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Denis Kirjanov <denis.kirjanov@suse.com>
---
 net/ipv4/route.c | 45 ++++++++++++++++++++++++++++++---------------
 1 file changed, 30 insertions(+), 15 deletions(-)

diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index bb2a2d5ef662..bbe854786934 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -70,6 +70,8 @@
 #include <linux/types.h>
 #include <linux/kernel.h>
 #include <linux/mm.h>
+#include <linux/bootmem.h>
+#include <linux/memblock.h>
 #include <linux/string.h>
 #include <linux/socket.h>
 #include <linux/sockios.h>
@@ -481,8 +483,10 @@ static void ipv4_confirm_neigh(const struct dst_entry *dst, const void *daddr)
 	__ipv4_confirm_neigh(dev, *(__force u32 *)pkey);
 }
 
-#define IP_IDENTS_SZ 2048u
-
+/* Hash tables of size 2048..262144 depending on RAM size.
+ * Each bucket uses 8 bytes.
+ */
+static u32 ip_idents_mask __read_mostly;
 static atomic_t *ip_idents __read_mostly;
 static u32 *ip_tstamps __read_mostly;
 
@@ -494,12 +494,16 @@ static u32 *ip_tstamps __read_mostly;
  */
 u32 ip_idents_reserve(u32 hash, int segs)
 {
-	u32 *p_tstamp = ip_tstamps + hash % IP_IDENTS_SZ;
-	atomic_t *p_id = ip_idents + hash % IP_IDENTS_SZ;
-	u32 old = READ_ONCE(*p_tstamp);
-	u32 now = (u32)jiffies;
+	u32 bucket, old, now = (u32)jiffies;
+	atomic_t *p_id;
+	u32 *p_tstamp;
 	u32 delta = 0;
 
+	bucket = hash & ip_idents_mask;
+	p_tstamp = ip_tstamps + bucket;
+	p_id = ip_idents + bucket;
+	old = READ_ONCE(*p_tstamp);
+
 	if (old != now && cmpxchg(p_tstamp, old, now) == old)
 		delta = prandom_u32_max(now - old);
 
@@ -3055,18 +3063,25 @@ struct ip_rt_acct __percpu *ip_rt_acct __read_mostly;
 
 int __init ip_rt_init(void)
 {
-	int rc = 0;
+	void *idents_hash;
 	int cpu;
 
-	ip_idents = kmalloc(IP_IDENTS_SZ * sizeof(*ip_idents), GFP_KERNEL);
-	if (!ip_idents)
-		panic("IP: failed to allocate ip_idents\n");
+	/* For modern hosts, this will use 2 MB of memory */
+	idents_hash = alloc_large_system_hash("IP idents",
+					      sizeof(*ip_idents) + sizeof(*ip_tstamps),
+					      0,
+					      16, /* one bucket per 64 KB */
+					      HASH_ZERO,
+					      NULL,
+					      &ip_idents_mask,
+					      2048,
+					      256*1024);
 
-	prandom_bytes(ip_idents, IP_IDENTS_SZ * sizeof(*ip_idents));
+	ip_idents = idents_hash;
 
-	ip_tstamps = kcalloc(IP_IDENTS_SZ, sizeof(*ip_tstamps), GFP_KERNEL);
-	if (!ip_tstamps)
-		panic("IP: failed to allocate ip_tstamps\n");
+	prandom_bytes(ip_idents, (ip_idents_mask + 1) * sizeof(*ip_idents));
+
+	ip_tstamps = idents_hash + (ip_idents_mask + 1) * sizeof(*ip_idents);
 
 	for_each_possible_cpu(cpu) {
 		struct uncached_list *ul = &per_cpu(rt_uncached_list, cpu);
@@ -3111,7 +3126,7 @@ int __init ip_rt_init(void)
 #endif
 	register_pernet_subsys(&rt_genid_ops);
 	register_pernet_subsys(&ipv4_inetpeer_ops);
-	return rc;
+	return 0;
 }
 
 #ifdef CONFIG_SYSCTL
-- 
2.16.4