Skip to contentSkip to conversation

gre: used time_before for comparing jiffies

PATCH In review Wei Yongjun 3 participants 6 messages Feb 20, 2009 – Feb 24, 2009

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/3 gre: used time_before for comparing jiffies

The functions time_before is more robust for comparing jiffies against other values.

---
 net/ipv4/ip_gre.c |    5 +++--
 1 files changed, 3 insertions(+), 2 deletions(-)
net/ipv4/ip_gre.c +3 -2
index 0101521..811bef7 100644
@@ -432,7 +432,7 @@ static void ipgre_err(struct sk_buff *skb, u32 info)
 	if (t->parms.iph.ttl == 0 && type == ICMP_TIME_EXCEEDED)
 		goto out;
 
-	if (jiffies - t->err_time < IPTUNNEL_ERR_TIMEO)
+	if (time_before(jiffies, t->err_time + IPTUNNEL_ERR_TIMEO))
 		t->err_count++;
 	else
 		t->err_count = 1;
@@ -744,7 +744,8 @@ static int ipgre_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 #endif
 
 	if (tunnel->err_count > 0) {
-		if (jiffies - tunnel->err_time < IPTUNNEL_ERR_TIMEO) {
+		if (time_before(jiffies,
+				tunnel->err_time + IPTUNNEL_ERR_TIMEO)) {
 			tunnel->err_count--;
 
 			dst_link_failure(skb);
Signature · 1 line
1.5.3.8
maintainer
#1 lore raw

From: Wei Yongjun <[email protected]>
Date: Fri, 20 Feb 2009 12:06:44 +0800

Quoted 4 lines

The functions time_before is more robust for comparing jiffies against other values.

There is some history of this in the networking code.

If you inspect carefully, the open-coded versions of time comparisons in the networking have a larger window of acceptance.

So actually, it is a regression fo use time_*() helpers in these cases.

If these things have been like this for so long, it is very likely there is a good reason :-)

#2 lore raw
Quoted 6 lines · David Miller

There is some history of this in the networking code.

If you inspect carefully, the open-coded versions of time comparisons in the networking have a larger window of acceptance.

Hmm, it looks like the macro version is actually better in this case.

- if (jiffies - t->err_time < IPTUNNEL_ERR_TIMEO)

This wraps at jiffies = LONG_MAX + t->err_time + 1

+ if (time_before(jiffies, t->err_time + IPTUNNEL_ERR_TIMEO))

This expands to

	(long)jiffies - (long)(t->err_time + IPTUNNEL_ERR_TIMEO) < 0

which wraps at jiffies = LONG_MAX + t->err_time + IPTUNNEL_ERR_TIMEO + 1

So assuming that IPTUNNEL_ERR_TIMEO > 0, then the macro version wraps around after the open-coded version, which would seem to mean that it's better, no?

Cheers,

Signature · 4 lines
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
maintainer
#3 lore raw

From: Herbert Xu <[email protected]>
Date: Fri, 20 Feb 2009 16:39:35 +0800

Quoted 3 lines

So assuming that IPTUNNEL_ERR_TIMEO > 0, then the macro version wraps around after the open-coded version, which would seem to mean that it's better, no?

Seems that way...

Let me see if I can find Alexey's email where he explains all of this. I really don't want to touch these things without being able to read that again.

If anyone else can find his posting about this stuff from years ago, please post the link.

Thanks.

#4 lore raw
In reply to Wei Yongjun
Quoted 4 lines · Wei Yongjun

The functions time_before is more robust for comparing jiffies against other values.

Please resubmit these patches so that they can be reviewed again.

Thanks,

Signature · 4 lines
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
maintainer
#5 lore raw

From: Herbert Xu <[email protected]>
Date: Tue, 24 Feb 2009 15:13:52 +0800

Quoted 7 lines
Quoted 4 lines · Wei Yongjun

The functions time_before is more robust for comparing jiffies against other values.

Please resubmit these patches so that they can be reviewed again.

Indeed, I think my original objections have been cleared up after some research.