add TH_AE capabilities to ppp and pf

[rscheff]

Summary

Split out changes from D47063 which are strictly
speaking changing the functionality.

Add support for the AE Flag in the TCP header to pf and ppp

[kp]

pf bits look good.

usr.sbin/ppp/ip.c

563

It looks like we only use this in a loop that ends when mask == 0x40, so we're never going to see ECE, CWR or AE there, right?

[rscheff]

• missed function definition

[cc]

sys/netpfil/ipfilter/netinet/ip_compat.h

698

I found TCPF_ALL is used in printtcpflags() from sbin/ipf/libipf/printtcpflags.c.

Would you please also add these in flagset[] and flags[] from sbin/ipf/libipf/flags.c?

[rscheff]

• decode ACE in libipf too
• expand ipf for AccECN compatibility

[rscheff]

• prepare more ipf tools for AccECN

[rscheff]

• log TCP flags properly in ppp

[rscheff]

In D47106#1073744, @kp wrote:

pf bits look good.

usr.sbin/ppp/ip.c

563

Then let's remove the magic number, and decode all TCP header flags properly for logging instead...

[cc]

My current concern is that the definition and the usage of the super set macro TH_FLAGS or TCPF_ALL are inconsistent. For example, TH_ECE is in TH_FLAGS, but TH_ECN is in TCPF_ALL.

sbin/ipf/ipsend/ipsend.c

368–393 ↗

(On Diff #144885)

Looks the same typo is among these lines, shall use __tcp_get_flags() inside, like this:

	__tcp_set_flags(tcp, __tcp_get_flags(tcp) | TH_SYN);

sbin/ipf/ipsend/iptests.c

927 ↗

(On Diff #144885)

I am thinking if we can just use the super set macro TH_FLAGS from sys/netinet/tcp.h or TCPF_ALL from sys/netpfil/ipfilter/netinet/ip_compat.h. But looks there is duplicated definition as TH_ECE == TH_ECN. I am confused about which one to use. But since you are using TH_ECE, so I take it for granted to use TH_FLAGS from sys/netinet/tcp.h.

sbin/ipf/libipf/printpacket.c

103–105

Shall be W for TH_CWR.

[rscheff]

• fix typo, remove redundant all-flags definitions

[rscheff]

Thanks for the thorough review!

I did build world, but the typo wasn't flagged - is that source code not compiled by GENERIC?

sbin/ipf/ipsend/iptests.c

927 ↗

(On Diff #144885)

Yes, there is no point of having multiple independent definitions to sum up all the TCP header flags. Removing the redundant definitions and using TH_FLAGS is the better path. Thanks!

[rscheff]

@kp can you please review the PF-related changes in this Diff?

[kp]

In D47106#1090312, @rscheff wrote:

@kp can you please review the PF-related changes in this Diff?

I'm happy with the pf bits.