Add a test ensuring that the amount of work fctx_getaddresses() performs
for any encountered delegation is limited: delegate example.net to a set
of 1,000 name servers in the redirect.com zone, the names of which all
resolve to IP addresses that nothing listens on, and query for a name in
the example.net domain, checking the number of times the findname()
function gets executed in the process; fail if that count is excessively
large.
Since the size of the referral response sent by ans3 is about 20 kB, it
cannot be sent back over UDP (EMSGSIZE) on some operating systems in
their default configuration (e.g. FreeBSD - see the
net.inet.udp.maxdgram sysctl). To enable reliable reproduction of
CVE-2022-2795 (retry patterns vary across BIND 9 versions) and avoid
false positives at the same time (thread scheduling - and therefore the
number of fetch context restarts - vary across operating systems and
across test runs), extend bin/tests/system/resolver/ans3/ans.pl so that
it also listens on TCP and make "ns1" in the "resolver" system test
always use TCP when communicating with "ans3".
Also add a test (foo.bar.sub.tld1/TXT) that ensures the new limitations
imposed on the resolution process by the mitigation for CVE-2022-2795 do
not prevent valid, glueless delegation chains from working properly.
Limit the amount of database lookups that can be triggered in
fctx_getaddresses() (i.e. when determining the name server addresses to
query next) by setting a hard limit on the number of NS RRs processed
for any delegation encountered. Without any limit in place, named can
be forced to perform large amounts of database lookups per each query
received, which severely impacts resolver performance.
The limit used (20) is an arbitrary value that is considered to be big
enough for any sane DNS delegation.
A full backport must have all the commit from the original MR and the
original commit IDs must be referenced in the backport commit messages.
If the criteria above is not met, the MR should be marked as a partial
backport. In that case, any discrepencies are only logged as informative
messages rather than failures.
(cherry picked from commit c617f97784)
When checking a backport MR, ensure that the original MR has been merged
already. This is vital for followup checks that verify commit IDs from
original commits are present in backport commit messages.
(cherry picked from commit 89530f1a1c)
When doing archeology, it is much easier to find stuff if it's properly
linked. This check ensures that backport MR are linked to their original
MR via a "Backport of !XXXX" message.
The regular expression is fairly broad and has been tested to accept the
following variants of the message:
Backport of MR !XXXX
Backport of: !XXXX
backport of mr !XXXX
Backport of !XXXX
Backport of https://gitlab.isc.org/isc-projects/bind9/-/merge_requests/XXXX
(cherry picked from commit 12e0b05738)
Having the MR title clearly marked in its title can be very useful when
looking through older issues/MRs.
This check also ensures that the version from the version label matches
the proper version branch (i.e. v9.16 must be marked with [v9_16]).
(cherry picked from commit 14b027cf83)
- Check that an NS in an authority section returned from a forwarder
which is above the name in a configured "forward first" or "forward
only" zone (i.e., net/NS in a response from a forwarder configured for
local.net) is not cached.
- Test that a DNAME for a parent domain will not be cached when sent
in a response from a forwarder configured to answer for a child.
- Check that glue is rejected if its name falls below that of zone
configured locally.
- Check that an extra out-of-bailiwick data in the answer section is
not cached (this was already working correctly, but was not explicitly
tested before).
- v9_11 backport: Revert primary/secondary to master/slave,
backport rndc helper, backport ns8 config.
(cherry picked from commit bf3fffff67)
When the parse tsan files have text extension they can be viewed
directly in the GitLab web UI without downloading them locally.
(cherry picked from commit 80582073a5)
Compound literals are not used in BIND 9.11, in order to ensure backward
compatibility with ancient compilers. Rework the relevant parts of the
BIND 9.11 backport of the CVE-2021-25220 fix so that compound literals
are not used.
When caching glue, we need to ensure that there is no closer
source of truth for the name. If the owner name for the glue
record would be answered by a locally configured zone, do not
cache.
When caching additional and glue data *not* from a forwarder, we must
check that there is no "forward only" clause covering the owner name
that would take precedence. Such names would normally be allowed by
baliwick rules, but a "forward only" zone introduces a new baliwick
scope.
If we are using a fowarder, in addition to checking that names to
be cached are subdomains of the forwarded namespace, we must also
check that there are no subsidiary forwarded namespaces which would
take precedence. To be safe, we don't cache any responses if the
forwarding configuration has changed since the query was sent.
Also make the script more verbose to identify which patch is being
processed and check for failures in spatch standard error output.
(cherry picked from commit 48c44fe6d4)
This prevents resolver timeouts for the reference (BIND 9.11) servers
used in respdiff tests run on Debian 11 "bullseye".
--with-randomdev=/dev/urandom is part of the "configure" template.
(cherry picked from commit 4d7e343813)
"buster" jobs are now only going to be run in scheduled pipelines.
"--without-gssapi" ./configure option of "bullseye" before it became the
base image is dropped from "bullseye"-the-base-image because it reduces
gcov coverage by 0.38 % (651 lines) and is used in Debian 9 "stretch".
"--enable-openssl-hash" is on purpose not being tested because it fails
linking when either of --with-ecdsa, --with-gost, --with-eddsa, or
--with-aes is used as well because it can't find f.e. HMAC_CTX_new() as
"-lcrypto" is missing:
/usr/bin/ld: ../../lib/isc/libisc.a(hmacmd5.o): in function `isc_hmacmd5_init':
/root/bind9/lib/isc/hmacmd5.c:49: undefined reference to `HMAC_CTX_new'
/usr/bin/ld: /root/bind9/lib/isc/hmacmd5.c:51: undefined reference to `EVP_md5'
/usr/bin/ld: /root/bind9/lib/isc/hmacmd5.c:51: undefined reference to `HMAC_Init_ex'
BIND 9.11 is in security-fixes-only-mode and configure.ac code should
not be fixed to include "-lcrypto" in this corner case; better eliminate
a pairwise hint.
(cherry picked from commit 910d595fbc)
Commit 9aa1c580 dropped dependency of "respdiff" and
"respdiff-third-party" jobs on "tarball-create" job because these jobs
don't need to depend on in (e.g., for its artifacts). This, however,
caused that respdiff jobs weren't started out-of-order and artifacts
from all the "Build" stage jobs plus "unit:gcc:buster:amd64" job were
downloaded to project directory and caused problems with compilation:
Originally, the dependency on "tarball-create" has been added in
45d59c50 to indicate that respdiff "is meant to operate on two different
BIND versions". It seems that the intent didn't work out, and we better
make it obvious that respdiff jobs don't depend on any other job and
should be run out-of-order.
(cherry picked from commit 87578efc71)
The order of directories with reference and test BIND 9 are now reversed
for respdiff.sh.
The data.mdb file has more than 10 GB and makes artifact download take
an unnecessarily long time.
(cherry picked from commit 2ececf2c02)
Alpine Linux image has OpenSSL 3.0.0 from the "edge" repository to test
OpenSSL 3.0.0 support in the BIND 9 "main" branch. However, this breaks
compilation of branches without OpenSSL 3.0.0 support and therefore
OpenSSL deprecated declarations need to be suppressed with
-DOPENSSL_SUPPRESS_DEPRECATED.