Skip to contentSkip to conversation

ip route: Set rtm_dst_len to 32 for all ip route get requests

PATCH In review David Ahern 3 participants 4 messages May 17, 2019 – May 23, 2019

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ip route: Set rtm_dst_len to 32 for all ip route get requests

From: David Ahern <[email protected]>

Jason reported that ip route get with a prefix length is now failing:

    $ 192.168.50.0/24
    RTNETLINK answers: Invalid argument

iproute2 now uses strict mode and strict mode in the kernel requires rtm_dst_len to be 32. Non-strict mode ignores the prefix length, so this allows ip to work without affecting existing users who add a prefix length to the request.

Fixes aea41afcfd6d6 ("ip bridge: Set NETLINK_GET_STRICT_CHK on socket")
---
 ip/iproute.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
ip/iproute.c +5 -1
index 2b3dcc5dbd53..d980b86ffd42 100644
@@ -2035,7 +2035,11 @@ static int iproute_get(int argc, char **argv)
 			if (addr.bytelen)
 				addattr_l(&req.n, sizeof(req),
 					  RTA_DST, &addr.data, addr.bytelen);
-			req.r.rtm_dst_len = addr.bitlen;
+			/* kernel ignores prefix length on 'route get'
+			 * requests; to allow ip to work with strict mode
+			 * but not break existing users, just set to 32
+			 */
+			req.r.rtm_dst_len = 32;
 			address_found = true;
 		}
 		argc--; argv++;
Signature · 1 line
2.11.0
#1 lore raw
Quoted 26 lines · On Fri, 17 May 2019 10:59:13 -0700 David Ahern <[email protected]> wrote:

From: David Ahern <[email protected]>

Jason reported that ip route get with a prefix length is now failing:

    $ 192.168.50.0/24
    RTNETLINK answers: Invalid argument

iproute2 now uses strict mode and strict mode in the kernel requires rtm_dst_len to be 32. Non-strict mode ignores the prefix length, so this allows ip to work without affecting existing users who add a prefix length to the request.

Fixes aea41afcfd6d6 ("ip bridge: Set NETLINK_GET_STRICT_CHK on socket")
---
 ip/iproute.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
ip/iproute.c +0 -0
index 2b3dcc5dbd53..d980b86ffd42 100644
@@ -2035,7 +2035,11 @@ static int iproute_get(int argc, char **argv)
 			if (addr.bytelen)
 				addattr_l(&req.n, sizeof(req),

ip/iproute.c static int iproute_get(int argc, char **argv) · line 2045

Quoted 10 lines
ip/iproute.c +5 -1
 					  RTA_DST, &addr.data, addr.bytelen);
-			req.r.rtm_dst_len = addr.bitlen;
+			/* kernel ignores prefix length on 'route get'
+			 * requests; to allow ip to work with strict mode
+			 * but not break existing users, just set to 32
+			 */
+			req.r.rtm_dst_len = 32;
 			address_found = true;
 		}
 		argc--; argv++;

I don't like silently ignoring things. It was wrong before and it is trapped now.

Probably better to error out in iproute2 if any prefix is given.

#2 lore raw
In reply to David Ahern
Quoted 6 lines · On Fri, 17 May 2019 10:59:13 -0700 David Ahern <[email protected]> wrote:

index 2b3dcc5dbd53..d980b86ffd42 100644

ip/iproute.c +0 -0
@@ -2035,7 +2035,11 @@ static int iproute_get(int argc, char **argv)
 			if (addr.bytelen)
 				addattr_l(&req.n, sizeof(req),

ip/iproute.c static int iproute_get(int argc, char **argv) · line 2045

Quoted 10 lines
ip/iproute.c +5 -1
 					  RTA_DST, &addr.data, addr.bytelen);
-			req.r.rtm_dst_len = addr.bitlen;
+			/* kernel ignores prefix length on 'route get'
+			 * requests; to allow ip to work with strict mode
+			 * but not break existing users, just set to 32
+			 */
+			req.r.rtm_dst_len = 32;
 			address_found = true;
 		}
 		argc--; argv++;

Why not warn user that any prefix length (ie not 32) is ignored, then do what you propose.

#3 lore raw
Quoted 14 lines · Stephen Hemminger
Quoted 11 lines · On Fri, 17 May 2019 10:59:13 -0700 David Ahern <[email protected]> wrote:

index 2b3dcc5dbd53..d980b86ffd42 100644

ip/iproute.c +3 -1
@@ -2035,7 +2035,11 @@ static int iproute_get(int argc, char **argv)
 			if (addr.bytelen)
 				addattr_l(&req.n, sizeof(req),
 					  RTA_DST, &addr.data, addr.bytelen);
-			req.r.rtm_dst_len = addr.bitlen;
+			/* kernel ignores prefix length on 'route get'
+			 * requests; to allow ip to work with strict mode
+			 * but not break existing users, just set to 32
Quoted 9 lines

ip/iproute.c static int iproute_get(int argc, char **argv) · line 2045

Quoted 5 lines
ip/iproute.c +2 -0
+			 */
+			req.r.rtm_dst_len = 32;
 			address_found = true;
 		}
 		argc--; argv++;

Why not warn user that any prefix length (ie not 32) is ignored, then do what you propose.

My first version did that. I thought people would complain about a stderr message. At least with failing the 'route get' scripts can be fixed.

After more thought even changing the prefix length for users presents a limitation on future changes.