Skip to contentSkip to conversation

netfilter : xt_time should not assume CONFIG_KTIME_SCALAR

NET Applied Eric Dumazet 4 participants 8 messages Nov 13, 2007 – Nov 14, 2007

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

It is not correct to assume one can get nsec from a ktime directly by using .tv64 field.

net/netfilter/xt_time.c +1 -1
index ff44f86..f9c55dc 100644
@@ -170,7 +170,7 @@ static bool xt_time_match(const struct sk_buff *skb,
 	if (skb->tstamp.tv64 == 0)
 		__net_timestamp((struct sk_buff *)skb);
 
-	stamp = skb->tstamp.tv64;
+	stamp = ktime_to_ns(skb->tstamp);
 	do_div(stamp, NSEC_PER_SEC);
 
 	if (info->flags & XT_TIME_LOCAL_TZ)
maintainer Applied
#1 lore raw

From: Eric Dumazet <[email protected]>
Date: Tue, 13 Nov 2007 12:30:37 +0100

Quoted 4 lines

It is not correct to assume one can get nsec from a ktime directly by using .tv64 field.

Patrick, this is very clearly a correct bug fix, so I'm going to apply this directly.

Applied, thanks Eric.

#2 lore raw
Quoted 9 lines · David Miller

From: Eric Dumazet <[email protected]>
Date: Tue, 13 Nov 2007 12:30:37 +0100

Quoted 3 lines

It is not correct to assume one can get nsec from a ktime directly by using .tv64 field.

Patrick, this is very clearly a correct bug fix, so I'm going to apply this directly.

Thanks, thats obviously correct. Still, I would prefer if people would CC netfilter-devel and myself, it has become a bad habit lately especially to skip netfilter-devel.

#3 lore raw
Quoted 8 lines · Patrick McHardy
Quoted 7 lines · David Miller

From: Eric Dumazet <[email protected]>
Date: Tue, 13 Nov 2007 12:30:37 +0100

Quoted 3 lines

It is not correct to assume one can get nsec from a ktime directly by using .tv64 field.

Quoted 9 lines
Quoted 2 lines

Patrick, this is very clearly a correct bug fix, so I'm going to apply this directly.

Thanks, thats obviously correct. Still, I would prefer if people would CC netfilter-devel and myself, it has become a bad habit lately especially to skip netfilter-devel.

Sorry for forgeting netfilter-devel, I'll try to remember for next time...

Thank you

#5 lore raw

[NET] random : secure_tcp_sequence_number should not assume CONFIG_KTIME_SCALAR

In reply to David Miller

I discovered one other incorrect use of .tv64 (coming from me, I must confess)

I guess this patch is needed for 2.6.24 and stable (2.6.22 & 2.6.23)

Thank you

[NET] random : secure_tcp_sequence_number should not assume CONFIG_KTIME_SCALAR

All 32 bits machines but i386 dont have CONFIG_KTIME_SCALAR. On these machines, ktime.tv64 is more than 4 times the (correct) result given by ktime_to_ns()

Again on these machines, using ktime_get_real().tv64 >> 6 give a 32bits rollover every 64 seconds, which is not wanted (less than the 120 s MSL)

Using ktime_to_ns() is the portable way to get nsecs from a ktime, and have correct code.

drivers/char/random.c +3 -3
index 1756b1f..5fee056 100644
@@ -1494,7 +1494,7 @@ __u32 secure_tcpv6_sequence_number(__be32 *saddr, __be32 *daddr,
 	seq = twothirdsMD4Transform((const __u32 *)daddr, hash) & HASH_MASK;
 	seq += keyptr->count;
 
-	seq += ktime_get_real().tv64;
+	seq += ktime_to_ns(ktime_get_real());
 
 	return seq;
 }
@@ -1556,7 +1556,7 @@ __u32 secure_tcp_sequence_number(__be32 saddr, __be32 daddr,
 	 *	overlaps less than one time per MSL (2 minutes).
 	 *	Choosing a clock of 64 ns period is OK. (period of 274 s)
 	 */
-	seq += ktime_get_real().tv64 >> 6;
+	seq += ktime_to_ns(ktime_get_real()) >> 6;
 #if 0
 	printk("init_seq(%lx, %lx, %d, %d) = %d\n",
 	       saddr, daddr, sport, dport, seq);
@@ -1616,7 +1616,7 @@ u64 secure_dccp_sequence_number(__be32 saddr, __be32 daddr,
 	seq = half_md4_transform(hash, keyptr->secret);
 	seq |= ((u64)keyptr->count) << (32 - HASH_BITS);
 
-	seq += ktime_get_real().tv64;
+	seq += ktime_to_ns(ktime_get_real());
 	seq &= (1ull << 48) - 1;
 #if 0
 	printk("dccp init_seq(%lx, %lx, %d, %d) = %d\n",
#4 lore raw
In reply to Eric Dumazet
Quoted 7 lines · Eric Dumazet
Quoted 6 lines · Patrick McHardy
Quoted 3 lines · David Miller

Patrick, this is very clearly a correct bug fix, so I'm going to apply this directly.

Quoted 6 lines
Quoted 5 lines

Thanks, thats obviously correct. Still, I would prefer if people would CC netfilter-devel and myself, it has become a bad habit lately especially to skip netfilter-devel.

Sorry for forgeting netfilter-devel, I'll try to remember for next time...

Thanks Eric. I missed that I actually was CCed because greylisting took a bit longer to let it through :)

maintainer
#6 lore raw

Re: [NET] random : secure_tcp_sequence_number should not assume CONFIG_KTIME_SCALAR

In reply to Eric Dumazet

From: Eric Dumazet <[email protected]>
Date: Tue, 13 Nov 2007 14:41:19 +0100

Quoted 4 lines

I discovered one other incorrect use of .tv64 (coming from me, I must confess)

I guess this patch is needed for 2.6.24 and stable (2.6.22 & 2.6.23)

...

Quoted 2 lines

[NET] random : secure_tcp_sequence_number should not assume CONFIG_KTIME_SCALAR

Applied thanks Eric, I'll queue it up for -stable too.

Perhaps ktime_t->tv64 should be renamed to "->__tv64" or similar to prevent future mistakes like this. Only the ktime implementation should be touching that thing.

#7 lore raw

Re: [stable] [NET] random : secure_tcp_sequence_number should not assume CONFIG_KTIME_SCALAR

In reply to Eric Dumazet
Quoted 6 lines · Eric Dumazet

I discovered one other incorrect use of .tv64 (coming from me, I must confess)

I guess this patch is needed for 2.6.24 and stable (2.6.22 & 2.6.23)

Thank you

Thanks for sending this to stable@, but David does a great job of forwarding these to us, so I'll wait for him to relay it to us before taking it into our tree.

thanks,

greg k-h