diff options
| author | Tom Gundersen <teg@jklm.no> | 2015-11-26 03:58:08 +0100 |
|---|---|---|
| committer | Tom Gundersen <teg@jklm.no> | 2015-11-26 04:03:08 +0100 |
| commit | 95d46fcaa4f27bc5e675e8de39ab3acc4732e39b (patch) | |
| tree | df45102f6735f2e23a0addc14ee3244ee499ed39 /src/resolve/resolved-bus.c | |
| parent | 09eaf68ce4e19a653ff25312e75c6e226b127fd6 (diff) | |
resolved: bus - follow CNAME chains when resolving addresses
It may be unexpected to find a CNAME record when doing a reverse lookup, as we
expect to find a PTR record directly. However, it is explicitly supported
according to <https://tools.ietf.org/html/rfc2181#section-10.2>, and there
seems to be no benefit to not supporting it.
Diffstat (limited to 'src/resolve/resolved-bus.c')
| -rw-r--r-- | src/resolve/resolved-bus.c | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/src/resolve/resolved-bus.c b/src/resolve/resolved-bus.c index a43aa59cc0..da1b5014bf 100644 --- a/src/resolve/resolved-bus.c +++ b/src/resolve/resolved-bus.c @@ -298,7 +298,15 @@ static void bus_method_resolve_address_complete(DnsQuery *q) { goto finish; } - /* We don't process CNAME for PTR lookups. */ + r = dns_query_process_cname(q); + if (r == -ELOOP) { + r = sd_bus_reply_method_errorf(q->request, BUS_ERROR_CNAME_LOOP, "CNAME loop detected, or CNAME resolving disabled on '%s'", dns_question_name(q->question)); + goto finish; + } + if (r < 0) + goto finish; + if (r > 0) /* This was a cname, and the query was restarted. */ + return; r = sd_bus_message_new_method_return(q->request, &reply); if (r < 0) |