rta_addattr_l() copies alen bytes but advances the parent by
RTA_ALIGN(len), so the 1-3 padding bytes are never written. This affects
every caller that passes an unaligned length. Nests built in stack
buffers (lwtunnel encap, RTA_MULTIPATH, RTA_METRICS) then send those
bytes to the kernel. For example, in
ip nexthop add id 53 encap mpls 101 ttl 10 via 192.0.2.1 dev d0
the three bytes after MPLS_IPTUNNEL_TTL are stack contents:
05 00 02 00 0a 7f 00 00 MPLS_IPTUNNEL_TTL, ttl 10, stack bytes
05 00 02 00 0a 00 00 00 with this patch
and valgrind reports "sendmsg(msg.msg_iov[0]) points to uninitialised
byte(s)".
The kernel skips padding when parsing, so nothing fails, but its own
encoder zeroes it (__nla_reserve()), and so does addraw_l() here. Do
the same.
Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Dave Seddon <redacted>
---
v2:
- Drop the Fixes: tag; the gap goes back to the initial import (Stephen)
- Cite addraw_l() as the precedent
- Correct the note on addattr_l()
addattr_l() doesn't zero its padding either (07f9436201eb added the
memset to addraw_l()), but its callers zero their request first, so it
is left alone. I can send that as a separate patch if wanted.
v1: https://lore.kernel.org/netdev/20261104052901.2735000-1-dave.seddon.ca@gmail.com/ (local)
---
lib/libnetlink.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/libnetlink.c b/lib/libnetlink.c
index edf3a8ba..c4a8e694 100644
--- a/lib/libnetlink.c
+++ b/lib/libnetlink.c
@@ -1502,6 +1502,7 @@ int rta_addattr_l(struct rtattr *rta, int maxlen, int type,
subrta->rta_len = len;
if (alen)
memcpy(RTA_DATA(subrta), data, alen);
+ memset((void *)subrta + len, 0, RTA_ALIGN(len) - len);
rta->rta_len = NLMSG_ALIGN(rta->rta_len) + RTA_ALIGN(len);
return 0;
}
--
2.56.0