Ondřej Surý [Mon, 28 Oct 2019 20:04:38 +0000 (15:04 -0500)]
Disable NSEC Aggressive Cache (synth-from-dnssec) by default
It was found that NSEC Aggressive Caching has a significant performance impact
on BIND 9 when used as recursor. This commit disables the synth-from-dnssec
configuration option by default to provide immediate remedy for people running
BIND 9.12+. The NSEC Aggressive Cache will be enabled again after a proper fix
will be prepared.
Michał Kępień [Wed, 16 Oct 2019 20:06:00 +0000 (22:06 +0200)]
Fix cppcheck 1.89 warnings
cppcheck 1.89 enabled certain value flow analysis mechanisms [1] which
trigger null pointer dereference false positives in lib/dns/rpz.c:
lib/dns/rpz.c:584:7: warning: Possible null pointer dereference: tgt_ip [nullPointer]
if (KEY_IS_IPV4(tgt_prefix, tgt_ip)) {
^
lib/dns/rpz.c:1425:44: note: Calling function 'adj_trigger_cnt', 4th argument '(void*)0' value is 0
adj_trigger_cnt(rpzs, rpz_num, rpz_type, NULL, 0, true);
^
lib/dns/rpz.c:584:7: note: Null pointer dereference
if (KEY_IS_IPV4(tgt_prefix, tgt_ip)) {
^
lib/dns/rpz.c:598:7: warning: Possible null pointer dereference: tgt_ip [nullPointer]
if (KEY_IS_IPV4(tgt_prefix, tgt_ip)) {
^
lib/dns/rpz.c:1425:44: note: Calling function 'adj_trigger_cnt', 4th argument '(void*)0' value is 0
adj_trigger_cnt(rpzs, rpz_num, rpz_type, NULL, 0, true);
^
lib/dns/rpz.c:598:7: note: Null pointer dereference
if (KEY_IS_IPV4(tgt_prefix, tgt_ip)) {
^
lib/dns/rpz.c:612:7: warning: Possible null pointer dereference: tgt_ip [nullPointer]
if (KEY_IS_IPV4(tgt_prefix, tgt_ip)) {
^
lib/dns/rpz.c:1425:44: note: Calling function 'adj_trigger_cnt', 4th argument '(void*)0' value is 0
adj_trigger_cnt(rpzs, rpz_num, rpz_type, NULL, 0, true);
^
lib/dns/rpz.c:612:7: note: Null pointer dereference
if (KEY_IS_IPV4(tgt_prefix, tgt_ip)) {
^
It seems that cppcheck no longer treats at least some REQUIRE()
assertion failures as fatal, so add extra assertion macro definitions to
lib/isc/include/isc/util.h that are only used when the CPPCHECK
preprocessor macro is defined; these definitions make cppcheck 1.89
behave as expected.
There is an important requirement for these custom definitions to work:
cppcheck must properly treat abort() as a function which does not
return. In order for that to happen, the __GNUC__ macro must be set to
a high enough number (because system include directories are used and
system headers compile attributes away if __GNUC__ is not high enough).
__GNUC__ is thus set to the major version number of the GCC compiler
used, which is what that latter does itself during compilation.
Michał Kępień [Tue, 15 Oct 2019 19:57:58 +0000 (21:57 +0200)]
Remove remnants of the --with-cc-alg option
Commit afa81ee4e4e863fa646177947c55e8c6b1475f47 omitted some spots in
the source tree which are still referencing the removed --with-cc-alg
"configure" option. Make sure the latter is removed completely.
Michał Kępień [Tue, 15 Oct 2019 19:35:18 +0000 (21:35 +0200)]
Limit triggers for OpenBSD system test jobs
When a GitLab CI runner is not under load, a single OpenBSD system test
job completes in about 12 minutes, which is considered decent. However,
such jobs are usually multiplexed with other system test jobs on the
same host, which causes each of them to take even 40 minutes to
complete. Taking retries into account, this is completely unacceptable
for everyday use, so only start OpenBSD system test jobs for pipelines
created through GitLab's web interface and for pipelines created for Git
tags.
Michał Kępień [Tue, 15 Oct 2019 18:49:08 +0000 (20:49 +0200)]
Tweak dependencies for the Windows build job
Since the Windows build job does not use the files created as a result
of running "autoreconf -fi" in the "autoreconf:sid:amd64" job, set its
dependencies to an empty list.
Since it is currently not possible to use "needs: []" for jobs which do
not belong to the first stage of a pipeline, set the "needs" key for the
Windows build job to the "autoreconf:sid:amd64" job so that all build
jobs are started at the same time (without this change, the Windows
build job does not start until all jobs in the "precheck" stage are
finished).
As a side note, these changes also attempt to eliminate intermittent,
bogus GitLab error messages ("There has been a missing dependency
failure").
Michał Kępień [Tue, 15 Oct 2019 18:49:08 +0000 (20:49 +0200)]
Fix artifacts created by the "autoreconf" CI job
The intended purpose of the "autoreconf:sid:amd64" GitLab CI job is to
run "autoreconf -fi" and then pass the updated files on to subsequent
non-Windows build jobs. However, the artifacts currently created by
that job only include files which are not tracked by Git. Since we
currently do track e.g. "configure" with Git, the aforementioned job is
essentially a no-op. Fix by manually specifying the files generated by
the "autoreconf:sid:amd64" job that should be passed on to subsequent
build jobs.
Michał Kępień [Tue, 15 Oct 2019 14:38:04 +0000 (16:38 +0200)]
Add OpenBSD to GitLab CI
Ensure BIND can be tested on OpenBSD in GitLab CI to more quickly catch
build and test errors on that operating system.
Some notes:
- While GCC is packaged for OpenBSD, only old versions (4.2.1, 4.9.4)
are readily available and none of them is the default system
compiler, so we are only doing Clang builds in GitLab CI.
- Unit tests are currently not run on OpenBSD because it ships with an
old version of kyua which does not handle skipped tests properly.
These jobs will be added when we move away from using kyua in the
future as the test code itself works fine.
- All OpenBSD jobs are run inside QEMU virtual machines, using GitLab
Runner Custom executor.
However, if the .NOTPARALLEL pseudo-target is added to this Makefile,
"make -k -j6 foo" will return 0 as well.
Since bin/tests/Makefile contains the .NOTPARALLEL pseudo-target,
running "make -k -j6 test" from bin/tests/ on OpenBSD prevents any
errors from being reported through that command's exit code.
Work around the issue by running "make -k -j6 test" in the
bin/tests/system/ directory instead as bin/tests/system/Makefile does
not contain the .NOTPARALLEL pseudo-target and thus things work as
expected there.
Mark Andrews [Sun, 13 Oct 2019 14:59:37 +0000 (10:59 -0400)]
Merge branch '1143-a-minor-documentation-issue-consideration-of-parsing-inconsistencies-in-ipv4s-in-address-match-lists-and-in-a-controls-inet-statement-v9_14' into 'v9_14'
Resolve "A minor documentation issue & consideration of parsing inconsistencies in IPv4s in address match lists and in a controls/inet statement"
Split dns_name_copy() into dns_name_copy() and dns_name_copynf()
The dns_name_copy() function followed two different semanitcs that was driven
whether the last argument was or wasn't NULL. This commit splits the function
in two where now third argument to dns_name_copy() can't be NULL and
dns_name_copynf() doesn't have third argument.
The final round of adding RUNTIME_CHECK() around dns_name_copy() calls
This commit was done by hand to add the RUNTIME_CHECK() around stray
dns_name_copy() calls with NULL as third argument. This covers the edge cases
that doesn't make sense to write a semantic patch since the usage pattern was
unique or almost unique.
Add RUNTIME_CHECK() around result = dns_name_copy(..., NULL) calls
This second commit uses second semantic patch to replace the calls to
dns_name_copy() with NULL as third argument where the result was stored in a
isc_result_t variable. As the dns_name_copy(..., NULL) cannot fail gracefully
when the third argument is NULL, it was just a bunch of dead code.
Couple of manual tweaks (removing dead labels and unused variables) were
manually applied on top of the semantic patch.
Add RUNTIME_CHECK() around plain dns_name_copy(..., NULL) calls using spatch
This commit add RUNTIME_CHECK() around all simple dns_name_copy() calls where
the third argument is NULL using the semantic patch from the previous commit.
Add semantic patches to correctly check dns_name_copy(..., NULL) return code
The dns_name_copy() function cannot fail gracefully when the last argument
(target) is NULL. Add RUNTIME_CHECK()s around such calls.
The first semantic patch adds RUNTIME_CHECK() around any call that ignores the
return value and is very safe to apply.
The second semantic patch attempts to properly add RUNTIME_CHECK() to places
where the return value from `dns_name_copy()` is recorded into `result`
variable. The result of this semantic patch needs to be reviewed by hand.
Both patches misses couple places where the code surrounding the
`dns_name_copy(..., NULL)` usage is more complicated and is better suited to be
fixed by a human being that understands the surrounding code.
Mark Andrews [Tue, 1 Oct 2019 04:06:53 +0000 (14:06 +1000)]
silence clang warning by using local variable.
'isc_commandline_index' is a global variable so it can theoretically
change result between if expressions. Save 'argv[isc_commandline_index]'
to local variable 'arg1' and use 'arg1 == NULL' in if expressions
instead of 'argc < isc_commandline_index + 1'. This allows clang
to correctly determine what code is reachable.
Fix passing NULL after the last typed argument to a variadic function leads to undefined behaviour.
From Cppcheck:
Passing NULL after the last typed argument to a variadic function leads to
undefined behaviour. The C99 standard, in section 7.15.1.1, states that if the
type used by va_arg() is not compatible with the type of the actual next
argument (as promoted according to the default argument promotions), the
behavior is undefined. The value of the NULL macro is an implementation-defined
null pointer constant (7.17), which can be any integer constant expression with
the value 0, or such an expression casted to (void*) (6.3.2.3). This includes
values like 0, 0L, or even 0LL.In practice on common architectures, this will
cause real crashes if sizeof(int) != sizeof(void*), and NULL is defined to 0 or
any other null pointer constant that promotes to int. To reproduce you might be
able to use this little code example on 64bit platforms. If the output includes
"ERROR", the sentinel had only 4 out of 8 bytes initialized to zero and was not
detected as the final argument to stop argument processing via
va_arg(). Changing the 0 to (void*)0 or 0L will make the "ERROR" output go away.
// changing 0 to 0L for the 7th argument (which is intended to act as
// sentinel) makes the error go away on x86_64
f("first", s2, s2, s2, s2, s2, 0, s3, (char*)0);
}
void h() {
int i;
volatile unsigned char a[1000];
for (i = 0; i<sizeof(a); i++)
a[i] = -1;
}
This MR changes the default Debian sid build to wrap make with bear
that creates compilation database and use the compilation database
to run Cppcheck on the source files systematically.
The job is currently set to be allowed to fail as it will take some
time to fix all the Cppcheck detected issues.
- compare key data when checking for a trust anchor match.
- allow for the possibility of multiple trust anchors with the same key ID
so we don't overlook possible matches.
projects / thirdparty / bind9.git / log
