Skip to contentSkip to conversation

xfrm: Pass template address family to xfrm_state_look_at

PATCH In review Anthony DeRossi 2 participants 5 messages Nov 3, 2020

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

PATCH 1/1 xfrm: Pass template address family to xfrm_state_look_at

This fixes a regression where valid selectors are incorrectly skipped when xfrm_state_find is called with a non-matching address family (e.g. when using IPv6-in-IPv4 ESP in transport mode).

The state's address family is matched against the template's family (encap_family) in xfrm_state_find before checking the selector in xfrm_state_look_at. The template's family should also be used for selector matching, otherwise valid selectors may be skipped.

Fixes e94ee171349d ("xfrm: Use correct address family in xfrm_state_find")
---
 net/xfrm/xfrm_state.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
net/xfrm/xfrm_state.c +2 -2
index ee6ac32bb06d..065f1bd8479a 100644
@@ -1075,7 +1075,7 @@ xfrm_state_find(const xfrm_address_t *daddr, const xfrm_address_t *saddr,
 		    tmpl->mode == x->props.mode &&
 		    tmpl->id.proto == x->id.proto &&
 		    (tmpl->id.spi == x->id.spi || !tmpl->id.spi))
-			xfrm_state_look_at(pol, x, fl, family,
+			xfrm_state_look_at(pol, x, fl, encap_family,
 					   &best, &acquire_in_progress, &error);
 	}
 	if (best || acquire_in_progress)
@@ -1092,7 +1092,7 @@ xfrm_state_find(const xfrm_address_t *daddr, const xfrm_address_t *saddr,
 		    tmpl->mode == x->props.mode &&
 		    tmpl->id.proto == x->id.proto &&
 		    (tmpl->id.spi == x->id.spi || !tmpl->id.spi))
-			xfrm_state_look_at(pol, x, fl, family,
+			xfrm_state_look_at(pol, x, fl, encap_family,
 					   &best, &acquire_in_progress, &error);
 	}
 
Signature · 1 line
2.26.2
maintainer
#1 lore raw
Quoted 9 lines · Anthony DeRossi

This fixes a regression where valid selectors are incorrectly skipped when xfrm_state_find is called with a non-matching address family (e.g. when using IPv6-in-IPv4 ESP in transport mode).

The state's address family is matched against the template's family (encap_family) in xfrm_state_find before checking the selector in xfrm_state_look_at. The template's family should also be used for selector matching, otherwise valid selectors may be skipped.

Quoted 5 lines
Fixes e94ee171349d ("xfrm: Use correct address family in xfrm_state_find")
---
 net/xfrm/xfrm_state.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Your patch reintroduces the same bug that my patch was trying to fix, namely that when you do the comparison on flow you must use the original family and not some other value.

Cheers,

Signature · 3 lines
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
#2 lore raw
Quoted 17 lines · Herbert Xu
Quoted 9 lines · Anthony DeRossi

This fixes a regression where valid selectors are incorrectly skipped when xfrm_state_find is called with a non-matching address family (e.g. when using IPv6-in-IPv4 ESP in transport mode).

The state's address family is matched against the template's family (encap_family) in xfrm_state_find before checking the selector in xfrm_state_look_at. The template's family should also be used for selector matching, otherwise valid selectors may be skipped.

Quoted 5 lines
Fixes e94ee171349d ("xfrm: Use correct address family in xfrm_state_find")
---
 net/xfrm/xfrm_state.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
Quoted 3 lines

Your patch reintroduces the same bug that my patch was trying to fix, namely that when you do the comparison on flow you must use the original family and not some other value.

My mistake, I misunderstood the original bug.

Anthony

maintainer
#3 lore raw
In reply to Anthony DeRossi
Quoted 3 lines · Anthony DeRossi

This fixes a regression where valid selectors are incorrectly skipped when xfrm_state_find is called with a non-matching address family (e.g. when using IPv6-in-IPv4 ESP in transport mode).

Why are we even allowing v6-over-v4 in transport mode? Isn't that the whole point of BEET mode?

Cheers,

Signature · 3 lines
Email: Herbert Xu <[email protected]>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
#4 lore raw
Quoted 8 lines · Herbert Xu
Quoted 3 lines · Anthony DeRossi

This fixes a regression where valid selectors are incorrectly skipped when xfrm_state_find is called with a non-matching address family (e.g. when using IPv6-in-IPv4 ESP in transport mode).

Why are we even allowing v6-over-v4 in transport mode? Isn't that the whole point of BEET mode?

I'm not sure. This is the outgoing policy that strongSwan creates for an IPv6-in-IPv4 tunnel when compression is enabled:

src fd02::/16 dst fd02::2/128

        dir out priority 326271 ptype main
        tmpl src 10.0.0.8 dst 192.168.1.231
                proto comp spi 0x0000d00e reqid 1 mode tunnel
        tmpl src 0.0.0.0 dst 0.0.0.0
                proto esp spi 0xc543e950 reqid 1 mode transport

After your patch, outgoing IPv6 packets fail to match the associated state:

src 10.0.0.8 dst 192.168.1.231

        proto esp spi 0xc543e950 reqid 1 mode transport
        replay-window 0
        auth-trunc hmac(sha256)
0x143b570f59b23eaa560905f19a922451c6dfa5694ba2e45e1b065bb1863421aa 128
        enc cbc(aes) 0x526ed144ca087125ce30e36c8f20d972
        encap type espinudp sport 4501 dport 4500 addr 0.0.0.0
        anti-replay context: seq 0x0, oseq 0x0, bitmap 0x00000000
        sel src 0.0.0.0/0 dst 0.0.0.0/0

Is this an invalid configuration?

Anthony