Bind 9 DNS Server chroot cannot work on Solaris 10 u6 and u7

Question

My Old verion "Solaris 10 x86 Generic_Patch_118844-30" can be run "Bind 9 DNS Server" in chroot mode.
And this is no any problem.
I have been try to use "Bind 9 DNS Server" in Solairs 10 x86 u6 or u7.
The result is not any problem.
But when I turn it run in the chroot mode. the "Bind 9 DNS Server" cannot run.
I have been manuelly run the named:-
/usr/sbin/named -c /etc/named.conf -t /chroot/dns -u named -f -g
The respond is :-
13-May-2009 02:17:46.623 starting BIND 9.3.6-P1 -c /etc/named.conf -t /chroot/dns -u named -f -g
13-May-2009 02:17:46.624 found 1 CPU, using 1 worker thread
13-May-2009 02:17:46.627 socket.c:3259: unexpected error:
13-May-2009 02:17:46.627 open(/dev/poll) failed: No such file or directory
13-May-2009 02:17:46.628 ./main.c:495: unexpected error:
13-May-2009 02:17:46.628 isc_socketmgr_create() failed: file not found
13-May-2009 02:17:46.629 create_managers() failed: unexpected error
13-May-2009 02:17:46.629 exiting (due to early fatal error)
Look like the bind 9 runing in the chroot mode after that cannot find /dev/poll
Even I use Bind 9 version 9.6.0, the result is same.
So, I don't sure the problem are the Bind 9 or Solaris 10 u6/u7
I try to continous install New verion Bind 9 in my old version Solaris 10.
THE Result is NO ANY PROBLEM in old version Solaris 10.
And I already bypass the SMF problem.
Anyone can tell me what the problem in solaris 10 u6/u7?

Answer 1

Looks like something reported similar bug:
[BIND fails to start|http://bugs.opensolaris.org/view_bug.do%3Bjsessionid=376e1152f0ddc75829ed1725542e?bug_id=6799867]
but I am somewhat puzzled why there is no follow up on the bug fixing.
Ok, I found the source (may be?):
From named:
http://src.opensolaris.org/source/xref/sfw/usr/src/cmd/bind/bind-9.3.6-P1/bin/named/main.c
    462 static isc_result_t
    463 create_managers(void) {
    464      isc_result_t result;
    465      unsigned int socks;
    466
    467 #ifdef ISC_PLATFORM_USETHREADS
    468      if (ns_g_cpus == 0)
    469           ns_g_cpus = ns_g_cpus_detected;
    470      isc_log_write(ns_g_lctx, NS_LOGCATEGORY_GENERAL, NS_LOGMODULE_SERVER,
    471                 ISC_LOG_INFO, "found %u CPU%s, using %u worker thread%s",
    472                 ns_g_cpus_detected, ns_g_cpus_detected == 1 ? "" : "s",
    473                 ns_g_cpus, ns_g_cpus == 1 ? "" : "s");
    474 #else
    475      ns_g_cpus = 1;
    476 #endif
    477      result = isc_taskmgr_create(ns_g_mctx, ns_g_cpus, 0, &ns_g_taskmgr);
    478      if (result != ISC_R_SUCCESS) {
    479           UNEXPECTED_ERROR(__FILE__, __LINE__,
    480                      "isc_taskmgr_create() failed: %s",
    481                      isc_result_totext(result));
    482           return (ISC_R_UNEXPECTED);
    483      }
    484
    485      result = isc_timermgr_create(ns_g_mctx, &ns_g_timermgr);
    486      if (result != ISC_R_SUCCESS) {
    487           UNEXPECTED_ERROR(__FILE__, __LINE__,
    488                      "isc_timermgr_create() failed: %s",
    489                      isc_result_totext(result));
    490           return (ISC_R_UNEXPECTED);
    491      }
    492
    493      result = isc_socketmgr_create2(ns_g_mctx, &ns_g_socketmgr, maxsocks);===========================> here. (notice the error message and the actual function called are not the same).
    494      if (result != ISC_R_SUCCESS) {
    495           UNEXPECTED_ERROR(__FILE__, __LINE__,
    496                      "isc_socketmgr_create() failed: %s",
    497                      isc_result_totext(result));
    498           return (ISC_R_UNEXPECTED);
    499      }
    500      result = isc_socketmgr_getmaxsockets(ns_g_socketmgr, &socks);
    501      if (result == ISC_R_SUCCESS) {
    502           isc_log_write(ns_g_lctx, NS_LOGCATEGORY_GENERAL,
    503                      NS_LOGMODULE_SERVER,
    504                      ISC_LOG_INFO, "using up to %u sockets", socks);
    505      }
    506
    507      result = isc_entropy_create(ns_g_mctx, &ns_g_entropy);
    508      if (result != ISC_R_SUCCESS) {
    509           UNEXPECTED_ERROR(__FILE__, __LINE__,
    510                      "isc_entropy_create() failed: %s",
    511                      isc_result_totext(result));
    512           return (ISC_R_UNEXPECTED);
    513      }
    514
    515      result = isc_hash_create(ns_g_mctx, ns_g_entropy, DNS_NAME_MAXWIRE);
    516      if (result != ISC_R_SUCCESS) {
    517           UNEXPECTED_ERROR(__FILE__, __LINE__,
    518                      "isc_hash_create() failed: %s",
    519                      isc_result_totext(result));
    520           return (ISC_R_UNEXPECTED);
    521      }
    522
    523      return (ISC_R_SUCCESS);
    524 }And in isc_socketmgr_create2():
   3384 isc_result_t
   3385 isc_socketmgr_create2(isc_mem_t *mctx, isc_socketmgr_t **managerp,
   3386                 unsigned int maxsocks)
   3387 {
<SNIP>
   3488
   3489      /*
   3490       * Set up initial state for the select loop
   3491       */
   3492      result = setup_watcher(mctx, manager);===============================>will call open() on /dev/poll.
   3493      if (result != ISC_R_SUCCESS)
   3494           goto cleanup;
   3495      memset(manager->fdstate, 0, manager->maxsocks * sizeof(int));
Other the above, I cannot really understand or help further. But I know it should fail in somewhere in setup_watcher() in socket.c.