Can not send icmp netunreach packet
Subsystems (2)
PATCH 1/1 Can not send icmp netunreach packet
Hi All
There is a bug about icmp netunreach.
If the kernel does not find a route for a packet, it must send a icmp netunreach packet to the source host, and discard the packet. But the kernel does not send a icmp netunreach packet because of the fib_lookup return value of -ESRCH when a route is not found.
core/fib_rules.c 2008-02-25 13:16:01.000000000 +0800 renamed from core_back/fib_rules.c 2008-02-25 13:15:37.000000000 +0800 +1 -1
@@ -188,7 +188,7 @@ jumped: } } - err = -ESRCH; + err = -ENETUNREACH; out: rcu_read_unlock();
Quoted 8 lines · Li Yewang
Hi All
There is a bug about icmp netunreach.
If the kernel does not find a route for a packet, it must send a icmp netunreach packet to the source host, and discard the packet. But the kernel does not send a icmp netunreach packet because of thefib_lookupreturn value of-ESRCHwhen a route is not found.
...or because some function doesn't handle -ESRCH return from fib_lookup? It seems changing this to -ESRCH was needed in some cases. And you don't explain enough why it can't be handled later (like in ipv4/route.c: ip_route_input_slow)?
Regards,
Jarek P.
Quoted 15 lines
Signed-off-by Li Yewang <[email protected]>diff -Nurp
net/core_back/fib_rules.cnet/core/fib_rules.ccore/fib_rules.c 2008-02-25 13:16:01.000000000 +0800 renamed from core_back/fib_rules.c 2008-02-25 13:15:37.000000000 +0800 +1
-1@@ -188,7 +188,7 @@ jumped: } } - err = -ESRCH; + err = -ENETUNREACH; out: rcu_read_unlock();
Jarek Poplawski wrote:
Maybe ip_error() does not handle the ESRCH error. In this place ESRCH eq to ENETUNREACH?
static int ip_error(struct sk_buff *skb)
{
struct rtable *rt = (struct rtable*)skb->dst;
unsigned long now;
int code;
switch (rt->u.dst.error) {
case EINVAL:
default:
goto out;
case EHOSTUNREACH:
code = ICMP_HOST_UNREACH;
break;
case ENETUNREACH:
code = ICMP_NET_UNREACH;
break;
case EACCES:
code = ICMP_PKT_FILTERED;
break;
}...............snip....................
}
Quoted 38 lines · Li Yewang
Quoted 10 lines
Hi All
There is a bug about icmp netunreach.
If the kernel does not find a route for a packet, it must send a icmp netunreach packet to the source host, and discard the packet. But the kernel does not send a icmp netunreach packet because of thefib_lookupreturn value of-ESRCHwhen a route is not found....or because some function doesn't handle
-ESRCHreturn fromfib_lookup? It seems changing this to-ESRCHwas needed in some cases. And you don't explain enough why it can't be handled later (like inipv4/route.c:ip_route_input_slow)?Regards,
Jarek P.Quoted 15 lines
Signed-off-by Li Yewang <[email protected]>diff -Nurp
net/core_back/fib_rules.cnet/core/fib_rules.ccore/fib_rules.c 2008-02-25 13:16:01.000000000 +0800 renamed from core_back/fib_rules.c 2008-02-25 13:15:37.000000000 +0800 +1
-1@@ -188,7 +188,7 @@ jumped: } } - err = -ESRCH; + err = -ENETUNREACH; out: rcu_read_unlock();
Quoted 4 lines · Wei Yongjun
Jarek Poplawski wrote:
Maybe
ip_error()does not handle the ESRCH error. In this place ESRCH eq to ENETUNREACH?
It doesn't handle ESRCH for sure... Current solution seems to expect it is changed earlier to ENETUNREACH. It looks reasonable because otherwise all other places checking for this should be updated too.
But, IMHO, it could be tested if such a change here helps in current problem, and then maybe found where it was skipped? On the other hand, probably checking with grep for all such ENETUNREACH cases, and adding ESRCH where needed could be much simpler and safer...
Jarek P.
Quoted 63 lines
static int
ip_error(struct sk_buff *skb){ struct rtable *rt = (struct rtable*)skb->dst; unsigned long now; int code; switch (rt->u.dst.error) { case EINVAL: default: goto out; case EHOSTUNREACH: code = ICMP_HOST_UNREACH; break; case ENETUNREACH: code = ICMP_NET_UNREACH; break; case EACCES: code = ICMP_PKT_FILTERED; break; }...............snip....................
}Quoted 36 lines · Li Yewang
Quoted 9 lines
Hi All
There is a bug about icmp netunreach.
If the kernel does not find a route for a packet, it must send
a icmp netunreach packet to the source host, and discard the
packet. But the kernel does not send a icmp netunreach packet
because of thefib_lookup
return value of-ESRCHwhen a route is not found....or because some function doesn't handle
-ESRCHreturn fromfib_lookup? It seems changing this to-ESRCHwas needed in some cases. And you don't explain enough why it can't be handled later (like inipv4/route.c:ip_route_input_slow)?Regards,
Jarek P.Quoted 14 lines
Signed-off-by Li Yewang <[email protected]>diff -Nurp
net/core_back/fib_rules.cnet/core/fib_rules.ccore/fib_rules.c 2008-02-25 13:16:01.000000000 +0800 renamed from core_back/fib_rules.c 2008-02-25 13:15:37.000000000 +0800 +1
-0@@ -188,7 +188,7 @@ jumped: } } - err = -ESRCH; + err = -ENETUNREACH; out: rcu_read_unlock();
Quoted 62 lines · On Tue, 26 Feb 2008 18:59:08 +0900 Wei Yongjun <[email protected]> wrote:
Jarek Poplawski wrote:
Maybe
ip_error()does not handle the ESRCH error. In this place ESRCH eq to ENETUNREACH?static int
ip_error(struct sk_buff *skb){ struct rtable *rt = (struct rtable*)skb->dst; unsigned long now; int code; switch (rt->u.dst.error) { case EINVAL: default: goto out; case EHOSTUNREACH: code = ICMP_HOST_UNREACH; break; case ENETUNREACH: code = ICMP_NET_UNREACH; break; case EACCES: code = ICMP_PKT_FILTERED; break; }...............snip....................
}Quoted 32 lines · Li Yewang
Quoted 10 lines
Hi All
There is a bug about icmp netunreach.
If the kernel does not find a route for a packet, it must send a icmp netunreach packet to the source host, and discard the packet. But the kernel does not send a icmp netunreach packet because of thefib_lookupreturn value of-ESRCHwhen a route is not found....or because some function doesn't handle
-ESRCHreturn fromfib_lookup? It seems changing this to-ESRCHwas needed in some cases. And you don't explain enough why it can't be handled later (like inipv4/route.c:ip_route_input_slow)?Regards,
Jarek P.Quoted 9 lines
Signed-off-by Li Yewang <[email protected]>diff -Nurp
net/core_back/fib_rules.cnet/core/fib_rules.ccore/fib_rules.c 2008-02-25 13:16:01.000000000 +0800 renamed from core_back/fib_rules.c 2008-02-25 13:15:37.000000000 +0800 +0
-0@@ -188,7 +188,7 @@ jumped: } }
Quoted 6 lines
Quoted 6 lines
core/fib_rules.c 2008-02-25 13:16:01.000000000 +0800 jumped: · line 194
Quoted 6 lines
core/fib_rules.c 2008-02-25 13:16:01.000000000 +0800 renamed from core_back/fib_rules.c 2008-02-25 13:15:37.000000000 +0800 +1
-1- err = -ESRCH; + err = -ENETUNREACH; out: rcu_read_unlock();
The switch shouldn't see a problem because ENETUNREACH is already substituted for ESRCH in:
static int ip_route_input_slow(struct sk_buff *skb, __be32 daddr, __be32 saddr,
u8 tos, struct net_device *dev)
{
...
/*
* Now we are ready to route packet.
*/
if ((err = fib_lookup(net, &fl, &res)) != 0) {
if (!IN_DEV_FORWARD(in_dev))
goto e_hostunreach;
goto no_route;...no_route:
RT_CACHE_STAT_INC(in_no_route); spec_dst = inet_select_addr(dev, 0, RT_SCOPE_UNIVERSE); res.type = RTN_UNREACHABLE; if (err == -ESRCH) err = -ENETUNREACH; goto local_input;
David Miller
maintainerFrom: Jarek Poplawski <[email protected]>
Date: Tue, 26 Feb 2008 07:35:38 +0000
Quoted 10 lines · Li Yewang
Quoted 8 lines
Hi All
There is a bug about icmp netunreach.
If the kernel does not find a route for a packet, it must send a icmp netunreach packet to the source host, and discard the packet. But the kernel does not send a icmp netunreach packet because of thefib_lookupreturn value of-ESRCHwhen a route is not found.
Quoted 4 lines
...or because some function doesn't handle
-ESRCHreturn fromfib_lookup? It seems changing this to-ESRCHwas needed in some cases. And you don't explain enough why it can't be handled later (like inipv4/route.c:ip_route_input_slow)?
This was changed to -ESRCH so that the proper statistics would be bumped. So if we change it back, the statistics will be broken again.
Quoted 13 lines · David Miller
From: Jarek Poplawski <[email protected]>
Date: Tue, 26 Feb 2008 07:35:38 +0000Quoted 10 lines · Li Yewang
Quoted 8 lines
Hi All
There is a bug about icmp netunreach.
If the kernel does not find a route for a packet, it must send a icmp netunreach packet to the source host, and discard the packet. But the kernel does not send a icmp netunreach packet because of thefib_lookupreturn value of-ESRCHwhen a route is not found.
Quoted 8 lines
Quoted 4 lines
...or because some function doesn't handle
-ESRCHreturn fromfib_lookup? It seems changing this to-ESRCHwas needed in some cases. And you don't explain enough why it can't be handled later (like inipv4/route.c:ip_route_input_slow)?This was changed to
-ESRCHso that the proper statistics would be bumped. So if we change it back, the statistics will be broken again.
Actually, I liked more this sophisticated explanation from the changelog:
BTW, it's really hard to find any place which could still behave as described by Li. Wasn't it with some older kernel?
Jarek P.