This should re-enable the name lookup - can you test ? Frank On 03/20/12 10:47, Ryo ONODERA wrote:
From: Frank Kardel<kardel%netbsd.org@localhost>, Date: Tue, 20 Mar 2012 10:13:27 +0100On 03/20/12 09:59, Martin Husemann wrote:On Tue, Mar 20, 2012 at 08:39:43AM +0100, Frank Kardel wrote:Could you check this patch. ntpq.c exercises the new AI_ADDRCONFIG option in getaddrinfo(). There seems to be a PNPE (programmed null pointer exception).See also PR pkg/46206 (likely the same problem) MartinLooks like it from the PR. I can't test right now but the patch should be very much the correct fix as the current code is definitely a *0 when AI_ADDRCONFIG is set. Can anyone test ?I have reproduced the problem of firefox. With your patch, firefox runs (without your patch core dumps). But firefox cannot resolve hostname. For example, on firefox, http://www.yahoo.co.jp/ is "Server not found". http://203.216.227.245/ is displayed properly. This will help?
Index: getaddrinfo.c
===================================================================
RCS file: /cvsroot/src/lib/libc/net/getaddrinfo.c,v
retrieving revision 1.99
diff -u -r1.99 getaddrinfo.c
--- getaddrinfo.c       20 Mar 2012 10:34:33 -0000      1.99
+++ getaddrinfo.c       20 Mar 2012 10:55:57 -0000
@@ -503,8 +503,11 @@
        for (ex = explore; ex->e_af >= 0; ex++) {
                *pai = ai0;
 
+
                /* ADDRCONFIG check */
-               if ((((uint64_t)1 << ex->e_af) & mask) == 0)
+               /* PF_UNSPEC entries are prepared for DNS queries only */
+               if (ex->e_af != PF_UNSPEC &&
+                   (((uint64_t)1 << ex->e_af) & mask) == 0)
                        continue;
 
                /* require exact match for family field */