I would suggest the following changes using getaddrinfo() instead of inet_aton() to avoid address-family dependency. Is cap_getaddrinfo() allowed by capdns?

@@ -147,8 +163,9 @@
 const char *typelimit = "ADDR";
 int familylimit;
 const char *ipstr = "127.0.0.1";
-struct in_addr ip;
-struct hostent *hp;
+char hname[NI_MAXHOST];
+struct addrinfo hints, *res;
+int error;

Looks good to me.

This is my submission, so I will leave someone else to do further review of this change. Please do not count my "accept revision" as an immediate approval.

IMHO strvisx(3) in machine.c should have a VIS_OCTAL flag as in my original submission because meta char encoding of strvisx(3), which uses iscntrl(3), does not make sense for most of multibyte encodings. Displaying characters not for the current locale in octal (or hexadecimal) number looks more intuitive to me.

Reverting a change does not need any approval or review. Please document the reason in the commit log instead of just saying "rollback".

Please split this changes into two commits; one is to revert r335836 and another is to add setlocale() and remove printable() function. A single commit including the both is confusing.

Looks good to me. Thank you for taking care of it!

The 64-bit default prefixlen was removed at rev.277241. I did not notice it but I do not think it is a correct change because it has several bad side-effects. I will take a look at route.c by this weekend.

If I understand this correctly, when a 0xffff checksum comes in via wire and the receiving NIC has no RXCSUM offloading, dhclient with this patch does not check the checksum. Is this a correct behavior? I think the conditional (usum != 0xffff) should be added only when the NIC has IFCAP_RXCSUM and it is enabled. Whether this flag is activated or not can be obtained by SIOSGIFCAP ioctl (ifr->ifr_curcap).

Looks good to me.

Looks good to me, but you need to remove the names in contrib.committers.xml.

I think you need to add CFLAGS+=-DLIBWRAP in Makefile. The others look fine to me.