Kirill A. Korinsky <[email protected]> wrote:
> Seems that new chromium plays with TOS at UDP for HTTP/3 and our pledge
> prevents it, and kills the process. It dumps ~100Mb core and after that it
> fallback to HTTP/1.1.
We are encountering this too many places, so I'd like everyone to consider
this 2-step approach, becuase Linux dual-stack has utterly poisoned the
ecosystem and noone looks at setsockopt return values accurately.
1. netinet6 accepts IPPROTO_IP/IP_TOS and converts it to
IPPROTO_IPV6/IPV6_TCLASS
2. For programs which are pledged, allow that through
Index: netinet6/ip6_output.c
===================================================================
RCS file: /cvs/src/sys/netinet6/ip6_output.c,v
diff -u -p -u -r1.308 ip6_output.c
--- netinet6/ip6_output.c 5 Aug 2026 09:43:19 -0000 1.308
+++ netinet6/ip6_output.c 8 Sep 2026 23:53:46 -0000
@@ -1032,6 +1032,12 @@ ip6_ctloutput(int op, struct socket *so,
privileged = (inp->inp_socket->so_state & SS_PRIV);
uproto = (int)so->so_proto->pr_protocol;
+ /* Linux poisoned the ecosystem */
+ if (level == IPPROTO_IP && optname == IP_TOS) {
+ level = IPPROTO_IPV6;
+ optname = IPV6_TCLASS;
+ }
+
if (level != IPPROTO_IPV6)
return (EINVAL);
Index: kern/kern_pledge.c
===================================================================
RCS file: /cvs/src/sys/kern/kern_pledge.c,v
diff -u -p -u -r1.365 kern_pledge.c
--- kern/kern_pledge.c 4 Sep 2026 02:38:28 -0000 1.365
+++ kern/kern_pledge.c 8 Sep 2026 23:37:52 -0000
@@ -1451,9 +1464,17 @@ pledge_sockopt(struct proc *p, int set,
}
break;
case AF_INET6:
- if (level == IPPROTO_IPV6) {
+ switch (level) {
+ case IPPROTO_IPV6:
switch (optname) {
case IPV6_TCLASS:
+ return (0);
+ }
+ break;
+ /* Because Linux makes the universe stink */
+ case IPPROTO_IP:
+ switch (optname) {
+ case IP_TOS:
return (0);
}
}
