Evan Hunt [Sat, 22 Feb 2020 08:37:05 +0000 (00:37 -0800)]
improve calculation of database transfer size
- change name of 'bytes' to 'xfrsize' in dns_db_getsize() parameter list
and related variables; this is a more accurate representation of what
the function is doing
- change the size calculations in dns_db_getsize() to more accurately
represent the space needed for a *XFR message or journal file to contain
the data in the database. previously we returned the sizes of all
rdataslabs, including header overhead and offset tables, which
resulted in the database size being reported as much larger than the
equivalent *XFR or journal.
- map files caused a particular problem here: the fullname can't be
determined from the node while a file is being deserialized, because
the uppernode pointers aren't set yet. so we store "full name length"
in the dns_rbtnode structure while serializing, and clear it after
deserialization is complete.
Evan Hunt [Sat, 22 Feb 2020 01:05:04 +0000 (17:05 -0800)]
dns_journal_iter_init() can now return the size of the delta
the call initailizing a journal iterator can now optionally return
to the caller the size in bytes of an IXFR message (not including
DNS header overhead, signatures etc) containing the differences from
the beginning to the ending serial number.
this is calculated by scanning the journal transaction headers to
calculate the transfer size. since journal file records contain a length
field that is not included in IXFR messages, we subtract out the length
of those fields from the overall transaction length.
this necessitated adding an "RR count" field to the journal transaction
header, so we know how many length fields to subract. NOTE: this will
make existing journal files stop working!
Witold Kręcicki [Tue, 3 Mar 2020 09:09:17 +0000 (10:09 +0100)]
Destroy query in killoldestquery under a lock.
Fixes a race between ns_client_killoldestquery and ns_client_endrequest -
killoldestquery takes a client from `recursing` list while endrequest
destroys client object, then killoldestquery works on a destroyed client
object. Prevent it by holding reclist lock while cancelling query.
Michal Nowak [Wed, 4 Mar 2020 16:06:31 +0000 (16:06 +0000)]
Fix "pkcs11" system test
- Define the SLOT environment variable before starting the test. This
variable defaults to 0 and that does not work with SoftHSM 2.
- The system test expects the PIN environment variable to be set to
"1234" while bin/tests/prepare-softhsm2.sh sets it to "0000".
Update bin/tests/prepare-softhsm2.sh so that it sets the PIN to
"1234".
- Move contents of bin/tests/system/pkcs11/prereq.sh to
bin/tests/system/pkcs11/setup.sh as the former was creating a file
called "supported" that was getting removed by the latter before
bin/tests/system/pkcs11/tests.sh could access it.
Evan Hunt [Wed, 4 Mar 2020 06:46:58 +0000 (22:46 -0800)]
list "validate-except" entries in "rndc nta -d" and "rndc secroots"
- no longer exclude these entries when dumping the NTA table
- indicate "validate-except" entries with the keyword "permanent" in
place of an expiry date
- add a test for this feature, and update other tests to account for
the presence of extra lines in some rndc outputs
- incidentally removed the unused function dns_ntatable_dump()
- CHANGES, release note
Witold Kręcicki [Wed, 5 Feb 2020 11:35:54 +0000 (12:35 +0100)]
Fix a race in isc_socket destruction.
There was a very slim chance of a race between isc_socket_detach and
process_fd: isc_socket_detach decrements references to 0, and before it
calls destroy gets preempted. Second thread calls process_fd, increments
socket references temporarily to 1, and then gets preempted, first thread
then hits assertion in destroy() as the reference counter is now 1 and
not 0.
Mark Andrews [Thu, 27 Feb 2020 06:35:18 +0000 (17:35 +1100)]
Call set_resigntime() in receive_secure_serial()
With RRSIG records no longer being signed with the full
sig-validity-interval we need to ensure the zone->resigntime
as it may need to be set to a earlier time.
Witold Kręcicki [Mon, 17 Feb 2020 09:37:39 +0000 (10:37 +0100)]
Badcache with multiple locks.
Previously badcache used one single mutex for everything, which
was causing performance issues. Use one global rwlock for the whole
hashtable and per-bucket mutexes.
Michał Kępień [Fri, 28 Feb 2020 14:22:29 +0000 (15:22 +0100)]
Add ZLIB_LIBS to ISCLIBS
When --with-zlib is passed to ./configure (or when the latter
autodetects zlib's presence), libisc uses certain zlib functions and
thus libisc's users should be linked against zlib in that case. Adjust
Makefile variables appropriately to prevent shared build failures caused
by underlinking.
Evan Hunt [Thu, 20 Feb 2020 02:26:48 +0000 (18:26 -0800)]
send bursts of queries in parallel to reduce test runtime
sending each group of queries simultaneously, and then checking the
output after the last one finishes, reduces the runtime of the
serve-stale test by about six minutes.
Evan Hunt [Wed, 19 Feb 2020 23:46:26 +0000 (15:46 -0800)]
stop testing 'serve-stale yes' and 'serve-fail no' alternate syntax
"yes" and "no" are permissible synonyms for "on" and "off", which
use exactly the same code paths. making sure they work isn't a good
use of 80 seconds of test time.
Witold Kręcicki [Mon, 17 Feb 2020 14:52:09 +0000 (15:52 +0100)]
Don't issue ns_client_endrequest on a NS_CLIENTSTATE_READY client.
Fix a potential assertion failure on shutdown in ns__client_endrequest.
Scenario:
1. We are shutting down, interface->clientmgr is gone.
2. We receive a packet, it gets through ns__client_request
3. mgr == NULL, return
4. isc_nmhandle_detach calls ns_client_reset_cb
5. ns_client_reset_cb calls ns_client_endrequest
6. INSIST(client->state == NS_CLIENTSTATE_WORKING ||
client->state == NS_CLIENTSTATE_RECURSING) is not met
- we haven't started processing this packet so
client->state == NS_CLIENTSTATE_READY.
As a solution - don't do anything in ns_client_reset_cb if the client
is still in READY state.
Evan Hunt [Wed, 19 Feb 2020 08:17:03 +0000 (00:17 -0800)]
minor cosmetic fixes
- the configuration summary reported zlib compression was not
supported even when it was.
- when bind.keys.h was regenerated it violated clang-format style.
projects / thirdparty / bind9.git / log
