Fra: |
"Paul Smedley" <ecs-isp@2rosenthals.com> |
Full Headers Undecoded message |
Emne: |
Re: [eCS-ISP] Bind 9.11.37 issue - ticket #784 - update |
Dato: |
Thu, 4 Jul 2024 18:49:09 +0930 |
Til: |
eCS ISP Mailing List <ecs-isp@2rosenthals.com> |
|
---|
Hi Steven,
On 4/7/24 12:58, Steven Levine wrote:
In <list-10146255@2rosenthals.com>, on 07/03/24
at 07:17 PM, "Paul Smedley" <ecs-isp@2rosenthals.com> said:
Hi Paul,
As I suspected the 100% issue is on some other thread - specifically
thread 2.
You might recall that a build a monitorthreads utility back when and it
showed the looping thread immediately.
The looping code seems to include
lib\isc\unix\socket.c:1418
while (cmsgp != NULL) {
socket_log(sock, NULL, TRACE,
isc_msgcat, ISC_MSGSET_SOCKET, ISC_MSG_PROCESSCMSG,
"processing cmsg %p", cmsgp);
and the message queue never empties. Assuming that there really are
messages to be logged, I need to figure out how to turn on the logging for
is class of messages.
Under the debugger this code:
lib\isc\unix\socket.c:1423
#ifdef ISC_PLATFORM_HAVEIN6PKTINFO
if (cmsgp->cmsg_level == IPPROTO_IPV6
&& cmsgp->cmsg_type == IPV6_PKTINFO) {
seems to be enabled. Should ISC_PLATFORM_HAVEIN6PKTINFO really be
defined?
Nope..... https://smedley.id.au/tmp/bind-9.11.37-os2-20240704.zip should fix...
Backstory is I have some stub ipv6 stuff to get some things compiling - seems this includes a definition of IN6PKTINFO - which is used despite named being built with ipv6 support disabled...
I'll look into the creation of a bind-os2 repo in github when I get time. We're about to head away for a week or so. This build should also fix the ascii/binary issue you reported.
Thanks for investigating!
Cheers,
Paul
|