Evan Hunt [Fri, 5 Jun 2020 06:13:54 +0000 (23:13 -0700)]
modify reference counting within netmgr
- isc__nmhandle_get() now attaches to the sock in the nmhandle object.
the caller is responsible for dereferencing the original socket
pointer when necessary.
- tcpdns listener sockets attach sock->outer to the outer tcp listener
socket. tcpdns connected sockets attach sock->outerhandle to the handle
for the tcp connected socket.
- only listener sockets need to be attached/detached directly. connected
sockets should only be accessed and reference-counted via their
associated handles.
Evan Hunt [Thu, 4 Jun 2020 21:54:36 +0000 (14:54 -0700)]
make isc_nmsocket_{attach,detach}{} functions private
there is no need for a caller to reference-count socket objects.
they need tto be able tto close listener sockets (i.e., those
returned by isc_nm_listen{udp,tcp,tcpdns}), and an isc_nmsocket_close()
function has been added for that. other sockets are only accessed via
handles.
Ondřej Surý [Thu, 11 Jun 2020 11:01:26 +0000 (13:01 +0200)]
Add missing acquire memory barrier in isc_nmhandle_unref
The ThreadSanitizer uses system synchronization primitives to check for
data race. The netmgr handle->references was missing acquire memory
barrier before resetting and reusing the memory occupied by isc_nmhandle_t.
Mark Andrews [Wed, 10 Jun 2020 07:07:52 +0000 (17:07 +1000)]
The dsset returned by dns_keynode_dsset needs to be thread safe.
- clone keynode->dsset rather than return a pointer so that thread
use is independent of each other.
- hold a reference to the dsset (keynode) so it can't be deleted
while in use.
- create a new keynode when removing DS records so that dangling
pointers to the deleted records will not occur.
- use a rwlock when accessing the rdatalist to prevent instabilities
when DS records are added.
Witold Kręcicki [Wed, 10 Jun 2020 14:19:16 +0000 (16:19 +0200)]
Fix a race in TCP accepting.
There's a possibility of a race in TCP accepting code:
T1 accepts a connection C1
T2 accepts a connection C2
T1 tries to accept a connection C3, but we hit a quota,
isc_quota_cb_init() sets quota_accept_cb for the socket,
we return from accept_connection
T2 drops C2, but we race in quota_release with accepting C3 so
we don't see quota->waiting is > 0, we don't launch the callback
T1 accepts a connection C4, we are able to get the quota we clear
the quota_accept_cb from sock->quotacb
T1 drops C1, tries to call the callback which is zeroed, sigsegv.
Michał Kępień [Tue, 9 Jun 2020 13:35:54 +0000 (15:35 +0200)]
Disable temporarily unsupported tests on Windows
Due to the changes introduced by the Automake migration, system tests
requiring Python (chain, pipelined, qmin, tcp), dynamic loading of
shared objects (dlzexternal, dyndb, filter-aaaa), or LMDB (nzd2nzf)
currently do not work on Windows. Temporarily disable them on that
platform by moving them from the PARALLEL_COMMON list to the
PARALLEL_UNIX list until the situation is rectified.
Michał Kępień [Tue, 9 Jun 2020 13:35:54 +0000 (15:35 +0200)]
Fix SYSTEMTESTTOP on Windows
Without SYSTEMTESTTOP=.. lines in tests.sh scripts, SYSTEMTESTTOP is
being set to an absolute path. On Windows, this means that an absolute
Cygwin path gets passed as a command line argument to native Windows
binaries, which cannot work and causes system tests to break. Fix by
passing SYSTEMTESTTOP through cygpath on Windows, which causes that
variable to be set to an absolute "mixed mode" path (Windows path with
forward slashes).
Michał Kępień [Tue, 9 Jun 2020 12:47:06 +0000 (14:47 +0200)]
Restore GitLab CI job building release tarballs
With "make dist" producing usable source tarballs and documentation
building working again, restore the script which allows a release
tarball to be built by a GitLab CI job, only making minimal adjustments
required due to the changes in the documentation building process and
due to dropping the "version" file.
Michał Kępień [Tue, 9 Jun 2020 12:47:06 +0000 (14:47 +0200)]
Fix respdiff job in GitLab CI
As the "configure" script is no longer stored in the Git repository, run
"autoreconf -fi" at the beginning of the respdiff job in GitLab CI in
order to enable that job to work properly.
Michał Kępień [Tue, 9 Jun 2020 12:47:06 +0000 (14:47 +0200)]
Fix source tarball creation job in GitLab CI
For the time being, "make all" needs to be run before "make dist" can
succeed as parts of the documentation are generated by programs compiled
during the regular build process.
As only one source tarball is published for each BIND release, make sure
the tarball creation job in GitLab CI only contains one tarball in the
desired format among its artifacts.
Drop the TARBALL_COMPRESSOR .gitlab-ci.yml variable as it is no longer
used in the source tarball creation process.
Michał Kępień [Tue, 9 Jun 2020 12:47:06 +0000 (14:47 +0200)]
Revise "srcid" file handling
The "srcid" file present in each BIND source tarball contains a
shortened hash of the Git commit corresponding to a given BIND release.
This allows a Git reference to be included in an archive that otherwise
lacks any Git information.
Before the move to Automake, if an "srcid" file was present in the root
source directory at the time ./configure was run, its contents were used
as the value of a compile-time constant which was then baked into BIND
binaries; otherwise, "git rev-parse" was used to determine the value of
that constant.
With Automake, a similar approach was attempted that required the
"srcid" file to be present at autoreconf time in order for it to be
used. However, note that this means that even if that file is present
in a source tarball created using "make dist", its contents are not
going to influence the value of the aforementioned compile-time constant
because autoreconf hardcodes the output of "git rev-parse" into the
configure script at autoreconf time.
To make things more clear, always use "git rev-parse" for determining
the value of the PACKAGE_SRCID compile-time constant when running
autoreconf. This causes "srcid" to be an empty string in source
tarballs built from other source tarballs, but that is not deemed to be
much of an issue as "make dist" is expected to be run from Git
repository clones. Remove stderr redirections to /dev/null to ensure
errors caused e.g. by running "make dist" from outside a Git repository
clone are not hidden. Trim the Git commit hash to 7 characters for
consistency between Unix and Windows systems.
Despite the above, ensure the "srcid" file is present in source tarballs
created using "make dist" as that file is used by the build process on
Windows.
Witold Kręcicki [Thu, 21 May 2020 12:31:09 +0000 (14:31 +0200)]
Fix a data access race in resolver
We were passing client address to dns_resolver_createfetch as a pointer
and it was saved as a pointer. The client (with its address) could be
gone before the fetch is finished, and in a very odd scenario
log_formerr would call isc_sockaddr_format() which first checks if the
address family is valid (and at this point it still is), then the
sockaddr is cleared, and then isc_netaddr_fromsockaddr is called which
fails an assertion as the address family is now invalid.
Michal Nowak [Thu, 4 Jun 2020 17:02:27 +0000 (19:02 +0200)]
Fix "make dist"
Make various adjustments necessary to enable "make dist" to build a BIND
source tarball whose contents are complete enough to build binaries, run
unit & system tests, and generate documentation on Unix systems.
Known outstanding issues:
- "make distcheck" does not work yet.
- Tests do not work for out-of-tree source-tarball-based builds.
- Source tarballs are not complete enough for building on Windows.
Michal Nowak [Thu, 4 Jun 2020 16:53:47 +0000 (18:53 +0200)]
Get rid of "helper" *.c files
Merge lib/isc/unix/ifiter_getifaddrs.c into lib/isc/unix/interfaceiter.c
and lib/isc/xoshiro128starstar.c into lib/isc/random.c. This avoids the
need for extra Automake directives required to process the "helper" *.c
files properly and makes the code more localized.
Michał Kępień [Fri, 5 Jun 2020 11:19:49 +0000 (13:19 +0200)]
Refactor lib/isc/fsaccess.c
Turn the static check_bad_bits() function used by both Unix and Windows
systems into a "private" function and extract the "private" parts of
lib/isc/fsaccess.c to lib/isc/fsaccess_common_p.h. Instead of including
lib/isc/fsaccess.c from lib/isc/{unix,win32}/fsaccess.c, make the former
an independent C source file.
Rename lib/isc/fsaccess.c to lib/isc/fsaccess_common.c to prevent build
issues on Windows caused by multiple source files (lib/isc/fsaccess.c,
lib/isc/win32/fsaccess.c) being compiled into the same object file.
These changes improve consistency with the way "private" functions and
macros are treated elsewhere in the source tree.
Mark Andrews [Tue, 17 Mar 2020 04:25:47 +0000 (15:25 +1100)]
Verify the question section when transfering in.
There was a case where an primary server sent a response
on the wrong TCP connection and failure to check the question
section resulted in a truncated zone being served.
Mark Andrews [Tue, 28 Apr 2020 05:37:19 +0000 (15:37 +1000)]
Reject primary zones with an DS record at the zone apex.
DS records only belong at delegation points and if present
at the zone apex are invariably the result of administrative
errors. Additionally they can't be queried for with modern
resolvers as the parent servers will be queried.
Mark Andrews [Fri, 29 May 2020 06:12:50 +0000 (16:12 +1000)]
Preserve PATH and LD_LIBRARY_PATH when calling make
When ./run.sh <test> is invoked, it acts as a wrapper around
`env - TESTS="<test>" make -e check` to preserve the ability to build
files defined only in the `check` target. Unfortunately, cleaning the
full environment had a side-effect of some tests failing due to missing
binaries and libraries. We now preserve the two most important
variables - PATH and LD_LIBRARY_PATH.
Michał Kępień [Thu, 4 Jun 2020 12:11:46 +0000 (14:11 +0200)]
Embed SoftHSM version in CI job names
To indicate the SoftHSM version used in each CI job while avoiding the
need to add another token to job names, replace "pkcs11" with
"softhsm2.4" and "fedora31:amd64" with "softhsm2.6".
Michał Kępień [Thu, 4 Jun 2020 12:11:46 +0000 (14:11 +0200)]
Make Fedora CI jobs PKCS#11-enabled
Various SoftHSM versions differ in algorithm support. Since Fedora
tends to have the latest SoftHSM version available in its stock package
repositories, enable PKCS#11 support in Fedora jobs to test multiple
SoftHSM versions in GitLab CI.
Michal Nowak [Fri, 22 May 2020 13:44:27 +0000 (15:44 +0200)]
Revise installation locations for BIND binaries
Move BIND binaries which are neither daemons nor administrative programs
to $bindir. This results in only the following binaries being left in
$sbindir:
- ddns-confgen
- named
- rndc
- rndc-confgen
- tsig-confgen
Petr Mensik [Fri, 27 Mar 2020 20:03:40 +0000 (21:03 +0100)]
Delay kserver cleanup until all tasks finishes
It might be possible some pending task would run when kserver is already
cleaned up. Postpone gsstsig structures cleanup after task and timer
managers are destroyed. No pending threads are possible after it.
Make action in maybeshutdown only if doshutdown was not already called.
Might be called from getinput event.
Ondřej Surý [Thu, 4 Jun 2020 09:09:27 +0000 (11:09 +0200)]
Stop building release notes as a separate sphinx-doc document
The release notes were previously built as a separate document
(including the PDF version). It was agreed that this doesn't make much
sense, so the release notes are now included only as an appendix to the
BIND 9 ARM.
This includes reorganization of the lists of RFCs supported by BIND 9.
I included all the RFCs and notes from the list identified by Vicky in
any DNS-related RFCs written by current ISC engineers, on the assumption
that BIND would comply with them.
Witold Kręcicki [Tue, 2 Jun 2020 08:39:05 +0000 (10:39 +0200)]
isc_uv_import must pass UV__IPC_SOCKET_XFER_TCP_CONNECTION, not SERVER.
As a leftover from old TCP accept code isc_uv_import passed TCP_SERVER
flag when importing a socket on Windows.
Since now we're importing/exporting accepted connections it needs to
pass TCP_CONNECTION flag.
Michał Kępień [Wed, 3 Jun 2020 13:45:28 +0000 (15:45 +0200)]
Prevent invalid warnings about missing identifiers
The Danger script inspects differences between the current version of a
given merge request's target branch and the merge request branch. If
the latter falls behind the former, the Danger script will wrongly warn
about missing GitLab/RT identifiers because it incorrectly treats the
"+++" diff marker as an indication of the merge request adding new lines
to a file. Tweak the relevant conditional expression to prevent such
invalid warnings from being raised.
Michał Kępień [Wed, 3 Jun 2020 13:45:28 +0000 (15:45 +0200)]
Make fetching target branch reliable
As GitLab Runner Docker executor caches Git repositories between jobs,
prevent the Danger script from attempting to update local refs to ensure
"git fetch" returns with an exit code of 0. Use the FETCH_HEAD ref for
determining the differences between the merge request branch and its
target branch.
Michał Kępień [Wed, 3 Jun 2020 13:45:28 +0000 (15:45 +0200)]
Tweak condition for missing log message warning
Commits adding CHANGES entries and/or release notes do not need a commit
log message. Do not warn about a missing commit log message for such
commits to make the warning more meaningful.
projects / thirdparty / bind9.git / log
