Skip to contentSkip to conversation

net: fix divide by zero in tcp algorithm illinois

PATCH Applied Jesper Dangaard Brouer 4 participants 5 messages Oct 31, 2012 – Nov 1, 2012

Subsystems (3)
Networking Network drivers TCP
lore
Share & tools
lore

Jesper Dangaard Brouer

author
#0 lore raw

PATCH v2 1/1 net: fix divide by zero in tcp algorithm illinois

Reading TCP stats when using TCP Illinois congestion control algorithm can cause a divide by zero kernel oops.

The division by zero occur in tcp_illinois_info() at:
do_div(t, ca->cnt_rtt);
where ca->cnt_rtt can become zero (when rtt_reset is called)

Steps to Reproduce:
1. Register tcp_illinois:

     # sysctl -w net.ipv4.tcp_congestion_control=illinois
 2. Monitor internal TCP information via command "ss -i"
     # watch -d ss -i
 3. Establish new TCP conn to machine

Either it fails at the initial conn, or else it needs to wait for a loss or a reset.

This is only related to reading stats. The function avg_delay() also performs the same divide, but is guarded with a (ca->cnt_rtt > 0) at its calling point in update_params(). Thus, simply fix tcp_illinois_info().

Function tcp_illinois_info() / get_info() is called without socket lock. Thus, eliminate any race condition on ca->cnt_rtt by using a local stack variable. Simply reuse info.tcpv_rttcnt, as its already set to ca->cnt_rtt.
Function avg_delay() is not affected by this race condition, as its called with the socket lock.

Cc Petr Matousek <[email protected]>
Signed-off-by Jesper Dangaard Brouer <[email protected]>

---
V2:
Address Eric Dumazets input:
- Save 2 bytes of stack, by using info.tcpv_rttcnt.
- Help compiler, and define "u64 t" inside if() lexical scope.

 net/ipv4/tcp_illinois.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)
net/ipv4/tcp_illinois.c +5 -3
index 813b43a..834857f 100644
@@ -313,11 +313,13 @@ static void tcp_illinois_info(struct sock *sk, u32 ext,
 			.tcpv_rttcnt = ca->cnt_rtt,
 			.tcpv_minrtt = ca->base_rtt,
 		};
-		u64 t = ca->sum_rtt;
 
-		do_div(t, ca->cnt_rtt);
-		info.tcpv_rtt = t;
+		if (info.tcpv_rttcnt > 0) {
+			u64 t = ca->sum_rtt;
 
+			do_div(t, info.tcpv_rttcnt);
+			info.tcpv_rtt = t;
+		}
 		nla_put(skb, INET_DIAG_VEGASINFO, sizeof(info), &info);
 	}
 }
Acked-by on v2 1/1
#1 lore raw
Quoted 8 lines · Jesper Dangaard Brouer

Reading TCP stats when using TCP Illinois congestion control algorithm can cause a divide by zero kernel oops.

The division by zero occur in tcp_illinois_info() at:
do_div(t, ca->cnt_rtt);
where ca->cnt_rtt can become zero (when rtt_reset is called)

Quoted 4 lines
Cc Petr Matousek <[email protected]>
Signed-off-by Jesper Dangaard Brouer <[email protected]>

---

Acked-by Eric Dumazet <[email protected]>
maintainer Applied
#2 lore raw

From: Eric Dumazet <[email protected]>
Date: Wed, 31 Oct 2012 18:14:38 +0100

Quoted 9 lines · Jesper Dangaard Brouer
Quoted 7 lines

Reading TCP stats when using TCP Illinois congestion control algorithm can cause a divide by zero kernel oops.

The division by zero occur in tcp_illinois_info() at:
do_div(t, ca->cnt_rtt);
where ca->cnt_rtt can become zero (when rtt_reset is called)

Quoted 6 lines
Quoted 4 lines
Cc Petr Matousek <[email protected]>
Signed-off-by Jesper Dangaard Brouer <[email protected]>

---

Acked-by Eric Dumazet <[email protected]>

Applied, thanks everyone.

Acked-by on v2 1/1
#3 lore raw
In reply to Jesper Dangaard Brouer
Quoted 53 lines · On Wed, 31 Oct 2012 13:45:32 +0100 Jesper Dangaard Brouer <[email protected]>…

Reading TCP stats when using TCP Illinois congestion control algorithm can cause a divide by zero kernel oops.

The division by zero occur in tcp_illinois_info() at:
do_div(t, ca->cnt_rtt);
where ca->cnt_rtt can become zero (when rtt_reset is called)

Steps to Reproduce:
1. Register tcp_illinois:

     # sysctl -w net.ipv4.tcp_congestion_control=illinois
 2. Monitor internal TCP information via command "ss -i"
     # watch -d ss -i
 3. Establish new TCP conn to machine

Either it fails at the initial conn, or else it needs to wait for a loss or a reset.

This is only related to reading stats. The function avg_delay() also performs the same divide, but is guarded with a (ca->cnt_rtt > 0) at its calling point in update_params(). Thus, simply fix tcp_illinois_info().

Function tcp_illinois_info() / get_info() is called without socket lock. Thus, eliminate any race condition on ca->cnt_rtt by using a local stack variable. Simply reuse info.tcpv_rttcnt, as its already set to ca->cnt_rtt.
Function avg_delay() is not affected by this race condition, as its called with the socket lock.

Cc Petr Matousek <[email protected]>
Signed-off-by Jesper Dangaard Brouer <[email protected]>

---
V2:
Address Eric Dumazets input:
- Save 2 bytes of stack, by using info.tcpv_rttcnt.
- Help compiler, and define "u64 t" inside if() lexical scope.

 net/ipv4/tcp_illinois.c |    8 +++++---
 1 files changed, 5 insertions(+), 3 deletions(-)
net/ipv4/tcp_illinois.c +0 -2
index 813b43a..834857f 100644
@@ -313,11 +313,13 @@ static void tcp_illinois_info(struct sock *sk, u32 ext,
 			.tcpv_rttcnt = ca->cnt_rtt,
 			.tcpv_minrtt = ca->base_rtt,
 		};
-		u64 t = ca->sum_rtt;
 
-		do_div(t, ca->cnt_rtt);

net/ipv4/tcp_illinois.c static void tcp_illinois_info(struct sock *sk, u32 ext, · line 325

Quoted 10 lines
net/ipv4/tcp_illinois.c +5 -1
-		info.tcpv_rtt = t;
+		if (info.tcpv_rttcnt > 0) {
+			u64 t = ca->sum_rtt;
 
+			do_div(t, info.tcpv_rttcnt);
+			info.tcpv_rtt = t;
+		}
 		nla_put(skb, INET_DIAG_VEGASINFO, sizeof(info), &info);
 	}
 }
Acked-by Stephen Hemminger <[email protected]>
author
#4 lore raw
Quoted 5 lines · Stephen Hemminger
Quoted 2 lines · On Wed, 31 Oct 2012 13:45:32 +0100 Jesper Dangaard Brouer <[email protected]>…

Reading TCP stats when using TCP Illinois congestion control algorithm can cause a divide by zero kernel oops.

[...]

Quoted 2 lines
Acked-by Stephen Hemminger <[email protected]>

When DaveM is without power, people even have/get time to review and ack my patches ;-). Thanks Stephen and Eric.

--Jesper