Hello, I am experiencing a problem similar to the one reported in Bug# 5402 - which his still open - however, in this case I see it the other way around - busybox's tftp fails to connect to V6-and-V4 targets from my V6-only box. The use of AI_ADDRCONFIG, as proposed during the investigation of 5402, addressed that problem, and at least in my configuration (busybox 1.23, glibc 2.22, linux ARM 4.1.12) I could not identify any of the side effects reported at that time. Is there any plan to revisit the patch below? https://bugs.busybox.net/attachment.cgi?id=4730 Thanks. ============================== Steps to recreate the problem: ============================== 1 - Configure a Linux box with only IPv6 addressing. 2 - Make sure you have a TFTP client running on a dual IPv4-and-IPv6 host Ex: [root@localhost ~]# dig silvaws.smd.devops.dell.com. any silvaws.smd.devops.dell.com. 86400 IN A 10.36.1.2 silvaws.smd.devops.dell.com. 86400 IN AAAA fd76:6c:61:6e36:a0b2:e2a7:c09:3d59 3 - Attempt to get a file from this host's TFTP server. tftp -g -r hello.txt silvaws.smd.devops.dell.com Actual Results ============== tftp: sendto: Network is unreachable Analysis ======== The call to getaddrinfo (routine str2sockaddr() in libbb/xconnect.c) returns both IPv6 and IPv4 address in that oerder. However, due to the configuration option ENABLE_FEATURE_PREFER_IPV4_ADDRESS, the IPv4 address is chosen. When the AI_ADDRCONFIG is passed, no IPv4 is returned, so the IPv6 address is used.
(In reply to luisapsilva from comment #0) > 1 - Configure a Linux box with only IPv6 addressing. > ... > The call to getaddrinfo (routine str2sockaddr() in libbb/xconnect.c) returns both IPv6 and IPv4 address in that oerder. The resolver returning IPv4 addresses for a pure IPv6 box would cause problems all over the place, not only in bbox. Configure resolver not to do that. Perhaps /etc/resolv.conf "options inet6". If resolver can't be configured this way, try running DNS cache service locally, configure it to only return IPv6, and point resolver to 127.0.0.1.