Since the BIND 9 QA repository has been made public, adjust the relevant
URLs and paths used in .gitlab-ci.yml so that they work with the public
version of that repository.
(cherry picked from commit 25ec1d79e4)
Since pre-release testing is usually carried out for branches in which
CHANGES entries are intentionally malformed to prevent entry numbering
conflicts down the road, do not run the "changes" GitLab CI job in
pipelines that are triggered by a parent pipeline (which can currently
only be a pre-release testing pipeline) to prevent triggering job
failures that would be meaningless anyway.
(cherry picked from commit 8c2503947f)
Checking the contents of the CHANGES file currently requires invoking
multiple shell scripts. These invocations are conflated with those for
other test scripts in the "misc" GitLab CI job. Extract the commands
checking the contents of the CHANGES file to a separate GitLab CI job,
"changes", to improve readability. Remove similar checks for the
CHANGES.SE file altogether as they are only relevant for BIND -S and
therefore should not be present in an open source branch.
(cherry picked from commit 1335e139f2)
To enable GitLab CI jobs in other projects to trigger pipelines in the
BIND 9 project using their CI_JOB_TOKEN, add "pipelines" to the relevant
GitLab CI job trigger lists.
The dnstap system test fails permanently. Given that we don't have the
resources to investigate and fix the issue in this EoL branch, turn off
dnstap globally.
I:dnstap:checking that named-checkconf detects no error in good-fstrm-set-reopen-interval.conf
I:dnstap:wait for servers to finish loading
I:dnstap:checking initial message counts
I:dnstap:checking UDP message counts
I:dnstap:ns3 3 expected 4
I:dnstap:failed
I:dnstap:checking TCP message counts
I:dnstap:checking AUTH_QUERY message counts
I:dnstap:checking AUTH_RESPONSE message counts
I:dnstap:checking CLIENT_QUERY message counts
I:dnstap:checking CLIENT_RESPONSE message counts
I:dnstap:ns3 1 expected 2
I:dnstap:failed
But enable dnstap in Debian "stretch" CI jobs to ensure it builds and
gets at least minimal testing.
IO::Socket::INET6 is no longer being maintained and its functionality
has been replaced by IO::Socket::IP.
(cherry picked from commit d1983adc22)
(cherry picked from commit fff8de8430)
Correct some errors in previous non-threaded changes. With these changes
unit tests in non-threaded versions pass again.
Related to [GL #4424] and [GL #4459], fixup of !8786.
dns__cacherbt_expireheader can unlink / free header_prev underneath
it. Use ISC_LIST_TAIL after calling dns__cacherbt_expireheader
instead to get the next pointer to be processed.
(cherry picked from commit 7ce2e86024)
Improve the selecting of the new signing key by remembering where
we stopped the iteration and just continue from that place instead
of iterating from the start over and over again each time.
Split the task manager queues into normal and slow task queues, so we
can move the tasks that blocks processing for a long time (like DNSSEC
validation) into the slow queue which doesn't block fast
operations (like responding from the cache). This mitigates the whole
class of KeyTrap-like issues.
Be more strict when encountering DNSSEC validation failures - fail on
the first failure. This will break domains that have DNSSEC signing
keys with duplicate key ids, but this is something that's much easier
to fix on the authoritative side, so we are just going to be strict
on the resolver side where it is causing performance problems.
We were missing a test where a single owner name would have multiple
types with a different case. The generated RRSIGs and NSEC records will
then have different case than the signed records and message parser have
to cope with that and treat everything as the same owner.
(cherry picked from commit c8b623d87f)
The case insensitive matching in isc_ht was basically completely broken
as only the hashvalue computation was case insensitive, but the key
comparison was always case sensitive.
(cherry picked from commit c462d65b2f)
The fix for CVE-2023-4408 introduced a regression in the message
parser, which could cause a crash if an rdata type that can only
occur in the question was found in another section.
(cherry picked from commit 510f1de8a6)
the fix for CVE-2023-4408 introduced a regression in the message
parser, which could cause a crash if duplicate rdatasets were found
in the question section. this commit ensures that rdatasets are
correctly disassociated and freed when this occurs.
(cherry picked from commit 4c19d35614)
The functions dns_message_find and dns_message_movename have been
removed. Remove the symbols from libdns.def.in to fix the windows
build.
(cherry picked from commit c44965af33)
To prevent allocating large hashtable in dns_message, we need to
backport the improvements to isc_ht API from BIND 9.18+ that includes
support for case insensitive keys and incremental rehashing of the
hashtables.
(cherry picked from commit a4baf32415)
When parsing messages use a hashtable instead of a linear search to reduce
the amount of work done in findname when there's more than one name in
the section.
There are two hashtables:
1) hashtable for owner names - that's constructed for each section when we
hit the second name in the section and destroyed right after parsing
that section;
2) per-name hashtable - for each name in the section, we construct a new
hashtable for that name if there are more than one rdataset for that
particular name.
(cherry picked from commit b8a9631754)
This job won't pass without a BIND 9 code change. Something we must
avoid to prevent delivering security patches that might not apply
cleanly for someone without this purely CI-related code change.
The tests have been executed with four jobs for a long time because the
TEST_PARALLEL_JOBS variable was set to 4 in the project's CI/CD
Variables Settings. The variable was recently dropped in favor of
.gitlab-ci.yaml one and now needs to be streamlined with the rest of
BIND 9 maintained branches.
(cherry picked from commit 03d7b45d81)
(cherry picked from commit 6a975344a2)
The "dnstap" system test started to fail very frequently for BIND 9.16
builds running in Oracle Linux 7 Docker containers spawned by GitLab CI.
The test used to be perfectly stable on that platform for the exact same
source tree, so the problem could not have been triggered by a bad code
change. Temporarily disable dnstap for Oracle Linux 7 builds prepared
in GitLab CI until the issue is fixed as it virtually prevents the
relevant GitLab CI jobs from being able to succeed.
See isc-projects/bind9#4240
(cherry picked from commit 8078e7de05)