Michal Nowak [Wed, 5 Feb 2020 10:03:09 +0000 (10:03 +0000)]
Windows: Prevent tools from clashing with named in system tests
In system tests on Windows tool's local port can sometimes clash with
'named'. On Unix the system is poked for the minimal local port,
otherwise is set to 32768 as a sane minimum. For Windows we don't
poke but set a hardcoded limit; this change aligns the limit with
Unix and changes it to 32768.
Mark Andrews [Wed, 5 Feb 2020 05:53:43 +0000 (16:53 +1100)]
'dispatch' must be non NULL, remove test.
10067 cleanup:
CID 1452683 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking dispatch suggests that it
may be null, but it has already been dereferenced on all
paths leading to the check.
10068 if (dispatch != NULL)
10069 isc_mem_put(server->mctx, dispatch, sizeof(*dispatch));
Mark Andrews [Wed, 5 Feb 2020 05:51:01 +0000 (16:51 +1100)]
'dctx' must be non NULL, remove test.
1549 cleanup:
1550 if (dctx->dbiter != NULL)
1551 dns_dbiterator_destroy(&dctx->dbiter);
1552 if (dctx->db != NULL)
1553 dns_db_detach(&dctx->db);
CID 1452686 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking dctx suggests that it may
be null, but it has already been dereferenced on all paths
leading to the check.
1554 if (dctx != NULL)
1555 isc_mem_put(mctx, dctx, sizeof(*dctx));
Mark Andrews [Wed, 5 Feb 2020 05:49:09 +0000 (16:49 +1100)]
'dir_list' must be non NULL, remove test.
707 complete_allnds:
CID 1452689 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking dir_list suggests that it
may be null, but it has already been dereferenced on all
paths leading to the check.
708 if (dir_list != NULL) {
709 /* clean up entries from list. */
Mark Andrews [Wed, 5 Feb 2020 05:45:59 +0000 (16:45 +1100)]
'lcfg' must be non NULL, remove test.
389 else
CID 1452695 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking lcfg suggests that it may
be null, but it has already been dereferenced on all paths
leading to the check.
390 if (lcfg != NULL)
391 isc_logconfig_destroy(&lcfg);
Mark Andrews [Wed, 5 Feb 2020 05:43:12 +0000 (16:43 +1100)]
's' must be non NULL, remove test.
122 cleanup:
CID 1452696 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking s suggests that it may be
null, but it has already been dereferenced on all paths
leading to the check.
Mark Andrews [Wed, 5 Feb 2020 05:41:03 +0000 (16:41 +1100)]
'tql' must be non NULL, remove test.
255 flag_fail:
256 /* get rid of what was build of the query list */
CID 1452697 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking tql suggests that it may
be null, but it has already been dereferenced on all paths
leading to the check.
257 if (tql != NULL)
258 destroy_querylist(mctx, &tql);
CID 1452700 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking closest suggests that it
may be null, but it has already been dereferenced on all
paths leading to the check.
6415 if (closest != NULL)
6416 free_noqname(mctx, &closest);
CID 1452701 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking tmpPath suggests that it
may be null, but it has already been dereferenced on all
paths leading to the check.
Mark Andrews [Wed, 5 Feb 2020 05:28:56 +0000 (16:28 +1100)]
'stub' cannot be non NULL, remove test.
13429 cleanup:
13430 cancel_refresh(zone);
CID 1452702 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking stub suggests that it may
be null, but it has already been dereferenced on all paths
leading to the check.
CID 1452704 (#1 of 1): Dereference before null check
(REVERSE_INULL) check_after_deref: Null-checking noqname
suggests that it may be null, but it has already been
dereferenced on all paths leading to the check.
6370 if (noqname != NULL)
6371 free_noqname(mctx, &noqname);
Mark Andrews [Wed, 5 Feb 2020 05:15:35 +0000 (16:15 +1100)]
'dctx' must be non NULL, remove test.
11030 cleanup:
CID 1452705 (#1 of 1): Dereference before null check
(REVERSE_INULL) check_after_deref: Null-checking dctx
suggests that it may be null, but it has already been
dereferenced on all paths leading to the check.
11031 if (dctx != NULL)
11032 dumpcontext_destroy(dctx);
11033 return (result);
Mark Andrews [Wed, 5 Feb 2020 05:11:11 +0000 (16:11 +1100)]
'event' must be non NULL, remove test.
1401 }
CID 1453455 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking event suggests that it may be null,
but it has already been dereferenced on all paths leading to the check.
1402 if (event != NULL)
1403 isc_event_free(ISC_EVENT_PTR(&event));
Mark Andrews [Wed, 5 Feb 2020 04:41:26 +0000 (15:41 +1100)]
'buffer' must be non-NULL as isc_buffer_allocate can no longer fail.
1636 cleanup:
CID 1458130 (#1 of 1): Dereference before null check (REVERSE_INULL)
check_after_deref: Null-checking buffer suggests that it may be
null, but it has already been dereferenced on all paths leading to
the check.
1637 if (buffer != NULL)
1638 isc_buffer_free(&buffer);
Ondřej Surý [Sat, 1 Feb 2020 16:13:45 +0000 (17:13 +0100)]
Fix comparison between type uint16_t and wider type size_t in a loop
Found by LGTM.com (see below for description), and while it should not
happen as EDNS OPT RDLEN is uint16_t, the fix is easy. A little bit
of cleanup is included too.
> In a loop condition, comparison of a value of a narrow type with a value
> of a wide type may result in unexpected behavior if the wider value is
> sufficiently large (or small). This is because the narrower value may
> overflow. This can lead to an infinite loop.
Matthijs Mekking [Thu, 23 Jan 2020 13:34:43 +0000 (14:34 +0100)]
Increase TTL in serve-stale test
Increase the short lived record TTL and negative SOA TTL to make
this test less vulnerable to timing issues. The drawback is that we
also have to sleep longer in this test.
Matthijs Mekking [Wed, 15 Jan 2020 15:11:43 +0000 (16:11 +0100)]
Add test for "Others" rrtype stat counter
Add queries and checks for CAA RRtype in the serve-stale test.
Ensure that the "Others" rrtype stat counter is incremented and
decremented properly if the RRset becomes stale/ancient.
The low max-stale-ttl config option needs to be increased in order
to match the timing when things expire (aka become ancient).
Matthijs Mekking [Fri, 17 Jan 2020 07:41:06 +0000 (08:41 +0100)]
Simplify cachedb rrset statistic counters
This commit simplifies the cachedb rrset statistics in two ways:
- Introduce new rdtypecounter arithmetics, allowing bitwise
operations.
- Remove the special DLV statistic counter.
New rdtypecounter arithmetics
-----------------------------
"The rdtypecounter arithmetics is a brain twister". Replace the
enum counters with some defines. A rdtypecounter is now 8 bits for
RRtypes and 3 bits for flags:
Matthijs Mekking [Fri, 17 Jan 2020 07:35:12 +0000 (08:35 +0100)]
Remove the DLV statistics counter
This also removes counting the DLV RRtype separately. Since we have
deprecated the lookaside validation it makes no sense to keep this
special statistic counter.
Michał Kępień [Tue, 4 Feb 2020 10:39:27 +0000 (11:39 +0100)]
Update GitLab CI to OpenBSD 6.6
Since OpenBSD 6.6 is the current OpenBSD release, replace OpenBSD 6.5
GitLab CI jobs with their up-to-date counterparts.
As CI jobs for OpenBSD 6.6 will be run by a generalized libvirt executor
rather than an OpenBSD-specific one, make the necessary tag and variable
adjustments as well.
Evan Hunt [Thu, 30 Jan 2020 23:30:32 +0000 (15:30 -0800)]
Correctly handle catalog zone entries containing slashes
- Add quotes before and after zone name when generating "addzone"
input so avoid "unexpected token" errors.
- Use a hex digest for zone filenames when the zone or view name
contains a slash.
- Test with a domain name containing a slash.
- Incidentally added 'catzhash.py' to contrib/scripts to generate
hash labels for catalog zones, as it was needed to write the test.
Ondřej Surý [Sun, 2 Feb 2020 07:35:46 +0000 (08:35 +0100)]
Refactor the isc_buffer_allocate() usage using the semantic patch
The isc_buffer_allocate() function now cannot fail with ISC_R_MEMORY.
This commit removes all the checks on the return code using the semantic
patch from previous commit, as isc_buffer_allocate() now returns void.
Ondřej Surý [Sun, 2 Feb 2020 07:50:41 +0000 (08:50 +0100)]
Refactor the isc_mempool_create() usage using the semantic patch
The isc_mempool_create() function now cannot fail with ISC_R_MEMORY.
This commit removes all the checks on the return code using the semantic
patch from previous commit, as isc_mempool_create() now returns void.
Michał Kępień [Wed, 29 Jan 2020 13:50:26 +0000 (14:50 +0100)]
List atypical failures in system test summary
Each system test can be marked as failed not only due to some tested
component(s) not behaving as expected, but also because of core dumps,
assertion failures, and/or ThreadSanitizer reports being found among its
artifacts. Make the system test summary list the tests which exhibit
such atypical symptoms to more clearly present the nature of problems
found.
Tony Finch [Thu, 16 Jan 2020 15:46:04 +0000 (15:46 +0000)]
Send NOFITY messages after deleting private-type records.
The `rndc signing -clear` command cleans up the private-type records
that keep track of zone signing activity, but before this change it
did not tell the secondary servers that the zone has changed.
Diego Fronza [Wed, 15 Jan 2020 18:22:06 +0000 (15:22 -0300)]
Added test for the proposed fix
Added test to ensure that NXDOMAIN is returned when BIND is queried for a
non existing domain in CH class (if a view of CHAOS class is configured)
and that it also doesn't crash anymore in those cases.
Diego Fronza [Wed, 15 Jan 2020 17:39:38 +0000 (14:39 -0300)]
Fixed crash when querying for non existing domain in chaos class
Function dns_view_findzonecut in view.c wasn't correctly handling
classes other than IN (chaos, hesiod, etc) whenever the name being
looked up wasn't in cache or in any of the configured zone views' database.
That resulted in a NULL fname being used in resolver.c:4900, which
in turn was triggering abort.