The cache for unreachable primaries was added to BIND 9 in 2006 via
1372e172d0. It features a 10-slot LRU
array with 600 seconds (10 minutes) fixed delay. During this time, any
primary with a hiccup would be blocked for the whole block duration
(unless overwritten by a different entry).
As this design is not very flexible (i.e. the fixed delay and the fixed
amount of the slots), redesign it based on the badcache.c module, which
was implemented earlier for a similar mechanism.
The differences between the new code and the badcache module were large
enough to create a new module instead of trying to make the badcache
module universal, which could complicate the implementation.
The new design implements an exponential backoff for entries which are
added again soon after expiring, i.e. the next expiration happens in
double the amount of time of the previous expiration, but in no more
time than the defined maximum value.
The initial and the maximum expiration values are hard-coded, but, if
required, it should be trivial to implement configurable knobs.
A "template" statement can contain the same configuration clauses
as a "zone" statement. A "zone" statement can now reference a
template, and all the clauses in that template will be used as
default values for the zone. For example:
template primary {
type primary;
file "$name.db";
initial-file "primary.db";
};
zone example.com {
template primary;
file "different-name.db"; // overrides the template
};
Special tokens can now be specified in a zone "file" option
in order to generate the filename parametrically. The first
instead of "$name" in the "file" option is replaced with the
zone origin, the first instance of "$type" is replaced with the
zone type (i.e., primary, secondary, etc), and the first instance
of "$view" is replaced with the view name..
This simplifies the creation of zones using initial-file templates.
For example:
$ rndc addzone <zonename> \
{ type primary; file "$name.db"; initial-file "template.db"
When loading a primary zone for the first time, if the zonefile
does not exist but an "initial-file" option has been set, then a
new file will be copied into place from the path specified by
"initial-file".
This can be used to simplify the process of adding new zones. For
instance, a template zonefile could be used by running:
$ rndc addzone example.com \
'{ type primary; file "example.db"; initial-file "template.db"; };'
there were some duplicated syntax checks in named_zone_configure()
that are no longer needed, now that we perform those same checks
using isccfg_check_zoneconf().
there were also some syntax checks that were *only* in
named_zone_configure(), which have now been moved to
isccfg_check_zoneconf(). test cases for them have been
added to the checkconf system test.
the function that checks zone syntax in libisccfg was previously
only called when loading named.conf, not when parsing an an
"rndc addzone" or "rndc modzone" command. this has been corrected.
note that some checks are still skipped: those that check for
duplication of filenames, key directories, etc. to fix this, we'd need
to export the symbol tables that are set up when loading named.conf and
preserve them so they could be reused later.
Since 70b1777d8a was commited the OSS-Fuzz
build was broken because the `chunk_get_raw()` was not updated in the
`FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION`-enabled area. Add the `size`
argument to the fuzzing version of the `chunk_get_raw()` function.
commandline.c failed to compile on Solaris because NAME_MAX was
undefined. Include 'isc/dir.h' which defines NAME_MAX for platforms
that don't define it.
In file included from commandline.c:54:
./include/isc/commandline.h:31:38: error: 'NAME_MAX' undeclared here (not in a function)
31 | extern char isc_commandline_progname[NAME_MAX];
| ^~~~~~~~
Using C23 attributes for `counted_by` is broken with clang.
`__has_attribute` is used since `__has_c_attribute` only works with C23
attributes, (`gnu::counted_by`/`clang::counted_by`)
Coverity detected that 'optlen' was not being checked in 'process_opt'.
This is actually already done when the OPT record was initially
parsed. Add an INSIST to silence Coverity as is done in message.c.
There were several methods how we used 'argv[0]'. Some programs had a
static value, some programs did use isc_file_progname(), some programs
stripped 'lt-' from the beginning of the name. And some used argv[0]
directly.
Unify the handling and all the variables into isc_commandline_progname
that gets populated by the new isc_commandline_init(argc, argv) call.
Instead of giving the memory pools names with an explicit call to
isc_mempool_setname(), add the name to isc_mempool_create() call to have
all the memory pools an unconditional name.
Instead of giving the memory context names with an explicit call to
isc_mem_setname(), add the name to isc_mem_create() call to have all the
memory contexts an unconditional name.
The memory context for isc_managers and dst_api units had no name and
that was causing trouble with the statistics channel output. Set the
name for the two memory context that were missing a proper name.
After b171cacf4f, a zone object can
remain in the memory for a while, until garbage collection is run.
Setting the DNS_ZONEFLG_EXITING flag should prevent the zone
maintenance function from running while it's in that state.
Otherwise, a secondary zone could initiate a zone transfer after
it had been deleted.
When request manager shuts down, it also shuts down all its ongoing
requests. Currently it calls their callback functions with a
ISC_R_SHUTTINGDOWN result code for the request. Since a request
manager can shutdown not only during named shutdown but also during
named reconfiguration, instead of sending ISC_R_SHUTTINGDOWN result
code send a ISC_R_CANCELED code to avoid confusion and errors with
the expectation that a ISC_R_SHUTTINGDOWN result code can only be
received during actual shutdown of named.
All the callback functions which are passed to either the
dns_request_create() or the dns_request_createraw() functions have
been analyzed to confirm that they can process both the
ISC_R_SHUTTINGDOWN and ISC_R_CANCELED result codes. Changes were
made where it was necessary.
When the zone.c:refresh_callback() callback function is called during
a SOA request before a zone transfer, it can receive a
ISC_R_SHUTTINGDOWN result for the sent request when named is shutting
down, and in that case it just destroys the request and finishes the
ongoing transfer, without clearing the DNS_ZONEFLG_REFRESH flag of the
zone. This is alright when named is going to shutdown, but currently
the callback can get a ISC_R_SHUTTINGDOWN result also when named is
reconfigured during the ongoibg SOA request. In that case, leaving the
DNS_ZONEFLG_REFRESH flag set results in the zone never being able
to refresh again, because any new attempts will be caneled while
the flag is set. Clear the DNS_ZONEFLG_REFRESH flag on the 'exiting'
error path of the callback function.
replace the pattern `for (result = dns_rdataset_first(x); result ==
ISC_R_SUCCES; result = dns_rdataset_next(x)` with a new
`DNS_RDATASET_FOREACH` macro throughout BIND.
the import_rdataset() function can't return any value other
than ISC_R_SUCCESS, so it's been changed to void and its callers
don't rely on its return value any longer.
the comments for some calls in the dns_message API specified
requirements which were not actually enforced in the functions.
in most cases, this has now been corrected by adding the missing
REQUIREs. in one case, the comment was incorrect and has been
revised.
previously, ISC_LIST_FOREACH and ISC_LIST_FOREACH_SAFE were
two separate macros, with the _SAFE version allowing entries
to be unlinked during the loop. ISC_LIST_FOREACH is now also
safe, and the separate _SAFE macro has been removed.
similarly, the ISC_LIST_FOREACH_REV macro is now safe, and
ISC_LIST_FOREACH_REV_SAFE has also been removed.
Before implementing adaptive chunk sizing, it was necessary to ensure
that a chunk could hold up to 48 twigs, but the new logic will size-up
new chunks to ensure that the current allocation can succeed.
We exploit the new logic in two ways:
- We make the minimum chunk size smaller than the old limit of 2^6,
reducing memory consumption.
- We make the maximum chunk size larger, as it has been observed that
it improves resolver performance.
qp-tries allocate their nodes (twigs) in chunks to reduce allocator
pressure and improve memory locality. The choice of chunk size presents
a tradeoff: larger chunks benefit qp-tries with many values (as seen
in large zones and resolvers) but waste memory in smaller use cases.
Previously, our fixed chunk size of 2^10 twigs meant that even an
empty qp-trie would consume 12KB of memory, while reducing this size
would negatively impact resolver performance.
This commit implements an adaptive chunking strategy that:
- Tracks the size of the most recently allocated chunk.
- Doubles the chunk size for each new allocation until reaching a
predefined maximum.
This approach effectively balances memory efficiency for small tries
while maintaining the performance benefits of larger chunk sizes for
bigger data structures.
This commit also splits the callback freeing qpmultis into two
phases, one that frees the underlying qptree, and one that reclaims
the qpmulti memory. In order to prevent races between the qpmulti
destructor and chunk garbage collection jobs, the second phase is
protected by reference counting.
The `max-rsa-exponent-size` could limit the exponents of the RSA
public keys during the DNSSEC verification. Instead of providing
a cryptic (not cryptographic) knob, hardcode the max exponent to
be 4096 (the theoretical maximum for DNSSEC).
The DST API has been cleaned up, duplicate functions has been squashed
into single call (verify and verify2 functions), and couple of unused
functions have been completely removed (createctx2, computesecret,
paramcompare, and cleanup).
This new option sets the delay, in seconds, to wait before sending
a set of NOTIFY messages for a zone. Whenever a NOTIFY message is
ready to be sent, sending will be deferred for this duration.
In a previous change, the "algorithm" value passed to
dns_tsigkey_create() was changed from a DNS name to an integer;
the name was then chosen from a table of known algorithms. A
side effect of this change was that a query using an unknown TSIG
algorithm was no longer handled correctly, and could trigger an
assertion failure. This has been corrected.
The dns_tsigkey struct now stores the signing algorithm
as dst_algorithm_t value 'alg' instead of as a dns_name,
but retains an 'algname' field, which is used only when the
algorithm is DST_ALG_UNKNOWN. This allows the name of the
unrecognized algorithm name to be returned in a BADKEY
response.
When we optimised the closest encloser NSEC3 discovery the maxlabels
variable was used in the binary search. The updated value was later
used to add the NO QNAME NSEC3 but that block of code needed the
original value. This resulted in the wrong NSEC3 sometimes being
chosen to perform this role.
If a call_rcu thread is running, there is a possible race condition
where the destructors run before all call_rcu callbacks have finished
running. This can happen, for example, if the call_rcu callback tries to
log something after the logging context has been torn down.
In !10394, we tried to counter this by explicitely creating a call_rcu
thread an shutting it down before running the destructors, but it is
possible for things to "slip" and end up on the default call_rcu thread.
As a quickfix, this commit moves an rcu_barrier() that was in the mem
context destructor earlier, so that it "protects" all libisc
destructors.
Previously all kinds of TCP timeouts had a single getter and setter
functions. Separate each timeout to its own getter/setter functions,
because in majority of cases only one is required at a time, and it's
not optimal expanding those functions every time a new timeout value
is implemented.
The checkds_send_toaddr() function uses hardcoded timeout values
for both UDP and TCP, however, with TCP named has configurable
timeout values. Slightly refactor the timeouts calculation part
and use the configured 'tcp-initial-timeout' value as the connect
timeout.
The notify_send_toaddr() function uses hardcoded timeout values
for both UDP and TCP, however, with TCP named has configurable
timeout values. Slightly refactor the timeouts calculation part
and use the configured 'tcp-initial-timeout' value as the connect
timeout.
The new 'tcp-primaries-timeout' configuration option works the same way
as the existing 'tcp-initial-timeout' option, but applies only to the
TCP connections made to the primary servers, so that the timeout value
can be set separately for them. The default is 15 seconds.
Also, while accommodating zone.c's code to support the new option, make
a light refactoring with the way UDP timeouts are calculated by using
definitions instead of hardcoded values.
The 'qpnode->nsec' structure member isn't protected by a lock and
there's a data race between the reading and writing parts in the
qpcache_addrdataset() function. Use a node read lock for accessing
'qpnode->nsec' in qpcache_addrdataset(). Add an additional
'qpnode->nsec != DNS_DB_NSEC_HAS_NSEC' check under a write lock
to be sure that no other competing thread changed it in the time
when the read lock is unlocked and a write lock is not acquired
yet.
When 'stale-answer-client-timeout' is 0, named is allowed to return
a stale answer immediately, while also initiating a new query to get
the real answer. This mode is activated in ns__query_start() by setting
the 'qctx->options.stalefirst' optoin to 'true' before calling the
query_lookup() function, but not when the zone is known to be
authoritative to the server. When the zone is authoritative, and
query_looup() finds out that the requested name is a delegation,
then before proceeding with the query, named tries to look it up
in the cache first. Here comes the issue that it doesn't consider
enabling 'qctx->options.stalefirst' in this case, and so the
'stale-answer-client-timeout 0' setting doesn't work for those
delegated zones - instead of immediately returning the stale answer
(if it exists), named tries to resolve it.
Fix this issue by enabling 'qctx->options.stalefirst' in the
query_zone_delegation() function just before named looks up the name
in the cache using a new query_lookup() call. Also, if nothing was
found in the cache, don't initiate another query_lookup() from inside
query_notfound(), and let query_notfound() do its work, i.e. it will
call query_delegation() for further processing.
Split the YAML display of the EDNS COOKIE option into CLIENT and SERVER
parts. The STATUS of the EDNS COOKIE in the reply is now a YAML element
rather than a comment.
The offical EDNS option name for "UL" is "UPDATE-LEASE". We now
emit "UPDATE-LEASE" instead of "UL", when printing messages, but
"UL" has been retained as an alias on the command line.
Update leases consist of 1 or 2 values, LEASE and KEY-LEASE. These
components are now emitted separately so they can be easily extracted
from YAML output. Tests have been added to check YAML correctness.
When rendering text, such as domain names or the EXTRA-TEXT
field of the EDE option, backslashes and quotation marks must
be escaped to ensure that the emitted message is valid YAML.
The CHAIN and REPORT-CHANNEL EDNS options are both domain names, so they
can be combined. THE CLIENT-TAG and SERVER-TAG EDNS options are both 16
bit integers, so they can be combined.
The custom allocation API for libxml2 is deprecated starting in macOS
Sequoia 15.4, iOS 18.4, tvOS 18.4, visionOS 2.4, and tvOS 18.4.
Disable the memory function override for libxml2 when
LIBXML_HAS_DEPRECATED_MEMORY_ALLOCATION_FUNCTIONS is defined as Apple
broke the system-wide libxml2 starting with macOS Sequoia 15.4.
When isc__thread_initialize() is called from a library constructor, it
could be called before we fork the main process. This happens with
named, and then we have the call_rcu_thread attached to the pre-fork
process and not the post-fork process, which means that the initial
process will never shutdown, because there's noone to tell it so.
Move the isc__thread_initialize() and isc__thread_shutdown() to the
isc_loop unit where we call it before creating the extra thread and
after joining all the extra threads respectively.
The *DB_VIRTUAL value was introduced to allow the clients (presumably
ns_clients) that has been running for some time to access the cached
data that was valid at the time of its inception. The default value
of 5 minutes is way longer than longevity of the ns_client object as
the resolver will give up after 2 minutes.
Reduce the value to 10 seconds to accomodate to honour the original
more closely, but still allow some leeway for clients that started some
time in the past.
Our measurements show that even setting this value to 0 has no
statistically significant effect, thus the value of 10 seconds should be
on the safe side.
This will help identify the broken server if we happen to break
EDNS version negotiation. It will also help protect the client
from spoofed BADVERSION responses.
We were failing to account for the length byte before the OID.
See RFC 4034.
Algorithm number 254 is reserved for private use and will never be
assigned to a specific algorithm. The public key area in the DNSKEY
RR and the signature area in the RRSIG RR begin with an unsigned
length byte followed by a BER encoded Object Identifier (ISO OID) of
that length. The OID indicates the private algorithm in use, and the
remainder of the area is whatever is required by that algorithm.
Entities should only use OIDs they control to designate their private
algorithms.
If the nested DNS validator ends up in the same fetch because of the
loops, the code could be copying the EDE codes from the same source EDE
context as the destination EDE context. Skip copying the EDE codes if
the source and the destination is the same.
Instead of passing the edectx from the fetchctx into all subvalidators,
make the ede context ownership explict for dns_resolver_createfetch()
callers, and copy the ede result codes from the children validators to
the parent when finishing the validation process.
Profiles show that an high amount of CPU time spent in memset.
By removing zero initalization of certain large buffers we improve
performance in certain authoritative workloads.
use the ISC_LIST_FOREACH pattern in places where lists had
been iterated using a different pattern from the typical
`for` loop: for example, `while (!ISC_LIST_EMPTY(...))` or
`while ((e = ISC_LIST_HEAD(...)) != NULL)`.
the pattern `for (x = ISC_LIST_HEAD(...); x != NULL; ISC_LIST_NEXT(...)`
has been changed to `ISC_LIST_FOREACH` throughout BIND, except in a few
cases where the change would be excessively complex.
in most cases this was a straightforward change. in some places,
however, the list element variable was referenced after the loop
ended, and the code was refactored to avoid this necessity.
also, because `ISC_LIST_FOREACH` uses typeof(list.head) to declare
the list elements, compilation failures can occur if the list object
has a `const` qualifier. some `const` qualifiers have been removed
from function parameters to avoid this problem, and where that was not
possible, `UNCONST` was used.
ISC_LIST_FOREACH and related macros now use 'typeof(list.head)' to
declare the list elements automatically; the caller no longer needs
to do so.
ISC_LIST_FOREACH_SAFE also now implicitly declares its own 'next'
pointer, so it only needs three parameters instead of four.
Add a function that checks if a 'hostname' is not a valid IPv4 or IPv6
address. Returns 'true' if the hostname is likely a domain name, and
'false' if it represents an IP address.
When allocating memory under -m trace|record, the __FILE__ pointer is
stored, so it can be printed out later in order to figure out in which
file an allocation leaked. (among others, like the line number).
However named crashes when called with -m record and using a plugin
leaking memory. The reason is that plugins are unloaded earlier than
when the leaked allocations are dumped (obviously, as it's done as late
as possible). In such circumstances, __FILE__ is dangling because the
dynamically loaded library (the plugin) is not in memory anymore.
Fix the crash by systematically copying the __FILE__ string
instead of copying the pointer. Of course, this make each allocation to
consume a bit more memory (and longer, as it needs to calculate the
length of __FILE__) but this occurs only under -m trace|record debugging
flags.
In term of unit test, because grepping in C is not fun, and because the
whole "syntax" of the dump output is tested in other tests, this simply
search for a substring in the whole buffer to make sure the expected
allocations are found.
In the code base it is very common to iterate over all names in a message
section and all rdatasets for each name, but various idioms are used for
iteration.
This commit standardizes them as much as possible to a single idiom,
using the macro MSG_SECTION_FOREACH, similar to the existing
ISC_LIST_FOREACH.
the code in query_dns64() that applies the dns64 prefixes to
an A rdataset has been moved into the dns_dns64 module, and
dns_dns64_destroy() now unlinks the dns64 object from its
containing list. with these changes, we no longer need the
list-manipulation API calls dns_dns64_next() and
dns_dns64_unlink().
This is the core implementation of the SIEVE algorithm described in the
following paper:
Zhang, Yazhuo, Juncheng Yang, Yao Yue, Ymir Vigfusson, and K V
Rashmi. “SIEVE Is Simpler than LRU: An Efficient Turn-Key Eviction
Algorithm for Web Caches,” n.d.. available online from
https://junchengyang.com/publication/nsdi24-SIEVE.pdf
In QPcache, there were two places that tried to upgrade the lock. In
clean_stale_header(), the code would try to upgrade the lock and cleanup
the header, and in qpzonode_release(), the tree lock would be optionally
upgraded, so we can cleanup the node directly if empty. These
optimizations are not needed and they have no effect on the performance.
All DNSKEY keys are able to authenticate. The DNS_KEYTYPE_NOAUTH
(and DNS_KEYTYPE_NOCONF) flags were defined for the KEY rdata type,
and are not applicable to DNSKEY.
Previously, because the DNSKEY implementation was built on top of
KEY, the NOAUTH flag prevented authentication in DNSKEYs as well.
This has been corrected.
Use enums for DNS_KEYFLAG_, DNS_KEYTYPE_, DNS_KEYOWNER_, DNS_KEYALG_,
and DNS_KEYPROTO_ values.
Remove values that are never used.
Eliminate the obsolete DNS_KEYFLAG_SIGNATORYMASK. Instead, add three
more RESERVED bits for the key flag values that it covered but which
were never used.
when sending a query to a forwarder for a name within a secure domain,
the first query is now sent with CD=0. when the forwarder itself
is validating, this will give it a chance to detect bogus data and
replace it with valid data before answering. this reduces our chances
of being stuck with data that can't be validated.
if the forwarder returns SERVFAIL to the initial query, the query
will be repeated with CD=1, to allow for the possibility that the
forwarder's validator is faulty or that the bogus answer is covered
by an NTA.
note: previously, CD=1 was only sent when the query name was in a
secure domain. today, validating servers have a trust anchor at the
root by default, so virtually all queries are in a secure domain.
therefore, the code has been simplified. as long as validation is
enabled, any forward query that receives a SERVFAIL response will be
retried with CD=1.
This can be set at the option, view and server levels and causes
named to add an EDNS ZONEVERSION option to requests. Replies are
logged to the 'zoneversion' category.
If there was an EDNS ZONEVERSION option in the DNS request and the
answer was from a zone, return the zone's serial and number of
labels excluding the root label with the type set to 0 (ZONE-SERIAL).
when searching a DNSKEY or KEY rrset for the key that matches
a particular algorithm and ID, it's a waste of time to convert
every key into a dst_key object; it's faster to compute the key
ID by checksumming the region, and then only do the full key
conversion once we know we've found the correct key.
this optimization was already in use in the validator, but it's
been refactored for code clarity, and is now also used in query.c
and message.c.
dns_zonekey_iszonekey() was the only function defined in the
dns_zonekey module, and was only called from one place. it
makes more sense to group this with dns_dnssec functions.
This merge request resolves some performance regressions introduced
with the change from isc_symtab_t to isc_hashmap_t.
The key improvements are:
1. Using a faster hash function than both isc_hashmap_t and
isc_symtab_t. The previous implementation used SipHash, but the
hashflood resistance properties of SipHash are unneeded for config
parsing.
2. Shrinking the initial size of the isc_hashmap_t used inside
isc_symtab_t. Symtab is mainly used for config parsing, and the
when used that way it will have between 1 and ~50 keys, but the
previous implementation initialized a map with 128 slots.
By initializing a smaller map, we speed up mallocs and optimize for
the typical case of few config keys.
3. Slight optimization of the string matching in the hashmap, so that
the tail is handled in a single load + comparison, instead of byte
by byte.
Of the three improvements, this is the least important.
Only set the next time the keymgr should run if the value is non zero.
Otherwise we default back to one hour. This may happen if there is one
or more key with an unlimited lifetime.
The keymgr never set the expected timing metadata when CDS/CDNSKEY
records for the corresponding key will be removed from the zone. This
is not troublesome, as key states dictate when this happens, but with
the new pytest we use the timing metadata to determine if the CDS and/or
CDNSKEY for the given key needs to be published.
There are a couple of cases where the safety intervals are added
inappropriately:
1. When setting the PublishCDS/SyncPublish timing metadata, we don't
need to add the publish-safety value if we are calculating the time
when the zone is completely signed for the first time. This value
is for when the DNSKEY has been published and we add a safety
interval before considering the DNSKEY omnipresent.
2. The retire-safety value should only be added to ZSK rollovers if
there is an actual rollover happening, similar to adding the sign
delay.
3. The retire-safety value should only be added to KSK rollovers if
there is an actual rollover happening. We consider the new DS
omnipresent a bit later, so that we are forced to keep the old DS
a bit longer.
While converting the kasp system test to pytest, I encountered a small
bug in the keymgr code. We retire keys when there is more than one
key matching a 'keys' line from the dnssec-policy. But if there are
multiple identical 'keys' lines, as is the case for the test zone
'checkds-doubleksk.kasp', we retire one of the two keys that have the
same properties.
Fix this by checking if there are double matches. This is not fool proof
because there may be many keys for a few identical 'keys' lines, but it
is good enough for now. In practice it makes no sense to have a policy
that dictates multiple keys with identical properties.
The fetch context that held these values could be freed while there
were still active pointers to the memory. Using a reference counted
pointer avoids this.
When a response times out the fctx_cancelquery() function
incorrectly calculates it in the 'dns_resstatscounter_queryrtt5'
counter (i.e. >=1600 ms). To avoid this, the rctx_timedout()
function should make sure that 'rctx->finish' is NULL. And in order
to adjust the RTT values for the timed out server, 'rctx->no_response'
should be true. Update the rctx_timedout() function to make those
changes.
The resquery_response() function increases the response counter without
checking if the response was successful. Increase the counter only when
the result indicates success.
There were kludges to help process responses from authoritative servers
giving RRs in wrong sections (mentioning BIND 8). These should just go
away and such responses should not be processed.
Add missing locks in dns_zone_getxfrsource4 et al. Addresses CID
468706, 468708, 468741, 468742, 468785 and 468778.
Cleanup dns_zone_setxfrsource4 et al to now return void.
Remove double copies with dns_zone_getprimaryaddr and dns_zone_getsourceaddr.
- dns_rdatatype_ismulti() returns true if a given type can have
multiple answers: ANY, RRSIG, or SIG.
- dns_rdatatype_issig() returns true for a signature: RRSIG or SIG.
- dns_rdatatype_isaddr() returns true for an address: A or AAAA.
- dns_rdatatype_isalias() returns true for an alias: CNAME or DNAME.
the step() function (used for stepping to the prececessor or
successor of a database node) could overlook a node because
there was an rdataset marked IGNORE because it had been rolled
back, covering an active rdataset under it.
when a key is revoked its key ID changes, due to the inclusion
of the "revoke" flag. a collision between this changed key ID and
that of an unrelated public-only key could cause a crash in
dnssec-signzone.
When performing QNAME minimization, named now sends an NS
query for the original QNAME, to prevent the parent zone from
receiving the QTYPE.
For example, when looking up example.com/A, we now send NS queries
for both com and example.com before sending the A query to the
servers for example.com. Previously, an A query for example.com
would have been sent to the servers for com.
Several system tests needed to be adjusted for the new query pattern:
- Some queries in the serve-stale test were sent to the wrong server.
- The synthfromdnssec test could fail due to timing issues; this
has been addressed by adding a 1-second delay.
- The cookie test could fail due to the a change in the count of
TSIG records received in the "check that missing COOKIE with a
valid TSIG signed response does not trigger TCP fallback" test case.
- The GL #4652 regression test case in the chain system test depends
on a particular query order, which no longer occurs when QNAME
minimization is active. We now disable qname-minimization
for that test.
If a timeout occurs when sending a QMIN query, QNAME
minimization should be disabled. This now causes a hard
failure in strict mode, or a fallback to non-minimized queries
in relaxed mode.
The calling fetch has already called fcount_incr() for this zone;
calling it again for a QMIN query results in double counting.
When resuming after a QMIN query is answered, however, we do now
ensure before continuing that the fetches-per-zone limit has not
been exceeded.
Extended DNS Error message EDE 20 (Not Authoritative) is now sent when
client request recursion (RD) but the server has recursion disabled.
RFC 8914 mention EDE 20 should also be returned if the client doesn't
have the RD bit set (and recursion is needed) but it doesn't apply for
BIND as BIND would try to resolve from the "deepest" referral in
AUTHORITY section. For example, if the client asks for "www.isc.org/A"
but the server only knows the root domain, it will returns NOERROR but
no answer for "www.isc.og/A", just the list of other servers to ask.
Extended DNS Error messages EDE 7 (expired key) and EDE 8 (validity
period of the key not yet started) are now sent in case of such DNSSEC
validation failures.
Refactor the existing validator extended error APIs in order to make it
easy to have a consisdent extra info (with domain/type) in the various
use case (i.e. when the EDE depends on validator state,
validate_extendederror or when the EDE doesn't depend of any state but
can be called directly in a specific flow).
The isc_mem API is one of the most commonly used APIs that didn't
used ISC_REFCOUNT_DECL and ISC_REFCOUNT_IMPL macros. Replace the
implementation of isc_mem_attach(), isc_mem_detach() and
isc_mem_destroy() with the respective macros.
This also removes the legacy isc_mem_destroy() functionality that would
check whether all references had been detached from the memory context
as it doesn't work reliably when using the call_rcu() API. Instead of
doing this individually, call isc_mem_checkdestroyed(stderr) from the
isc_mem_destroy() macro to keep the extra check that all contexts were
freed when the program is exiting.
ZONEMD needs to be able to digest SIG and RRSIG records. The signer
field can be compressed in SIG so we need to call dns_name_digest().
While for RRSIG the records the signer field is not compressed the
canonical form has the signer field downcased (RFC 4034, 6.2). This
also implies that compare_rrsig needs to downcase the signer field
during comparison.
The qpcache_findzonecut() accepts two "foundnames": 'foundname' and
'dcname' could be NULL. Originally, when 'dcname' would be NULL, the
'dcname' would be set to 'foundname'. Then code like this was present:
result = find_deepest_zonecut(&search, node, nodep, foundname,
rdataset,
sigrdataset DNS__DB_FLARG_PASS);
dns_name_copy(foundname, dcname);
Which basically means that we are copying the .ndata over itself for no
apparent reason. Cleanup the dcname vs foundname usage.
Co-authored-by: Evan Hunt <each@isc.org>
Co-authored-by: Ondřej Surý <ondrej@isc.org>
This commit bumps the total number of active streams (= the opened
streams for which a request is received, but response is not ready) to
60% of the total streams limit.
The previous limit turned out to be too tight as revealed by
longer (≥1h) runs of "stress:long:rpz:doh+udp:linux:*" tests.
Previously, the code would try to avoid sending any data regardless of
what it is unless:
a) The flush limit is reached;
b) There are no sends in flight.
This strategy is used to avoid too numerous send requests with little
amount of data. However, it has been proven to be too aggressive and,
in fact, harms performance in some cases (e.g., on longer (≥1h) runs
of "stress:long:rpz:doh+udp:linux:*").
Now, additionally to the listed cases, we also:
c) Flush the buffer and perform a send operation when there is an
outgoing DNS message passed to the code (which is indicated by the
presence of a send callback).
That helps improve performance for "stress:long:rpz:doh+udp:linux:*"
tests.
Previously, a function for continuing IO processing on the next UV
tick was introduced (http_do_bio_async()). The intention behind this
function was to ensure that http_do_bio() is eventually called at
least once in the future. However, the current implementation allows
queueing multiple such delayed requests needlessly. There is currently
no need for these excessive requests as http_do_bio() can requeue them
if needed. At the same time, each such request can lead to a memory
allocation, particularly in BIND 9.18.
This commit ensures that the number of enqueued delayed IO processing
requests never exceeds one in order to avoid potentially bombarding IO
threads with the delayed requests needlessly.
This commit significantly simplifies the code flow in the
http_do_bio() function, which is responsible for processing incoming
and outgoing HTTP/2 data. It seems that the way it was structured
before was indirectly caused by the presence of the missing callback
calls bug, fixed in 8b8f4d500d.
The change introduced by this commit is known to remove a bottleneck
and allows reproducible and measurable performance improvement for
long runs (>= 1h) of "stress:long:rpz:doh+udp:linux:*" tests.
Additionally, it fixes a similar issue with potentially missing send
callback calls processing and hardens the code against use-after-free
errors related to the session object (they can potentially occur).
Previously, a gcc < 4.6 shim for _Static_assert() was included. Such an
old compiler is not supported now anyway, so the macro variant has been
removed in favor of a single definition using _Static_assert().
Previously, the LOCK()/UNLOCK() and friends macros were defined in the
isc/util.h header. Those macros were moved to their respective headers
as those would have to be included anyway if that particular lock was in
use.
Formerly, isc/util.h would pull a few extra headers (isc/list.h,
isc/attributes.h, isc/result.h and errno.h). These includes were
removed in favor of including them directly when used.
The short convenience list macros were used very sparingly and
inconsistenly in the code base. As the consistency is prefered over
the convenience, all shortened list macro were removed in favor of
their ISC_LIST API targets.
When all the addresses were already iterated over, the
dns_remote_curraddr() function asserts. So before calling it,
dns_zone_getprimaryaddr() now checks the address list using the
dns_remote_done() function. This also means that instead of
returning 'isc_sockaddr_t' it now returns 'isc_result_t' and
writes the primary's address into the provided pointer only when
returning success.
Since algorithm fetching is handled purely in libisc, FIPS mode toggling
can be purely done in within the library instead of provider fetching in
the binary for OpenSSL >=3.0.
Disabling FIPS mode isn't a realistic requirement and isn't done
anywhere in the codebase. Make the FIPS mode toggle enable-only to
reflect the situation.
Answers to an "ANY" query which are processed by the RPZ "passthru"
policy have the response-policy's 'max-policy-ttl' value unexpectedly
applied. Do not change the records' TTL when RPZ uses a policy which
does not alter the answer.
when the caching of a negative record failed because of the
presence of a positive one, ncache_adderesult() could override
this to ISC_R_SUCCESS. this could cause CNAME and DNAME responses
to be handled incorrectly. ncache_adderesult() now sets the result
code correctly in such cases.
the target name parameter to dns_adb_createfind() was always passed as
NULL, so we can safely remove it.
relatedly, the 'target' field in the dns_adbname structure was never
referenced after being set. the 'expire_target' field was used, but
only as a way to check whether an ADB name represents a CNAME or DNAME,
and that information can be stored as a single flag.
Named was stopping nameserver address resolution attempts too soon
when dual stack servers are configured. Dual stack servers are
used when there are *not* addresses for the server in a particular
address family so find->status == DNS_ADB_NOMOREADDRESSES is not a
sufficient stopping condition when dual stack servers are available.
Call fctx_try to see if the alternate servers can be used.
DNSKEY, KEY, RRSIG and SIG constraints have been relaxed to allow
empty key and signature material after the algorithm identifier for
PRIVATEOID and PRIVATEDNS. It is arguable whether this falls within
the expected use of these types as no key material is shared and
the signatures are ineffective but these are private algorithms and
they can be totally insecure.
if the NS_QUERY_DONE_BEGIN or NS_QUERY_DONE_SEND hook is
used in a plugin and returns NS_HOOK_RETURN, some of the
cleanup in ns_query_done() can be skipped over, leading
to reference leaks that can cause named to hang on shut
down.
this has been addressed by adding more housekeeping
code after the cleanup: tag in ns_query_done().
previously, dns_name_fromtext() took both a target name and an
optional target buffer parameter, which could override the name's
dedicated buffer. this interface is unnecessarily complex.
we now have two functions, dns_name_fromtext() to convert text
into a dns_name that has a dedicated buffer, and dns_name_wirefromtext()
to convert text into uncompressed DNS wire format and append it to a
target buffer.
in cases where it really is necessary to have both, we can use
dns_name_fromtext() to load the dns_name, then dns_name_towire()
to append the wire format to the target buffer.
the target buffer passed to dns_name_concatenate() was never
used (except for one place in dig, where it wasn't actually
needed, and has already been removed in a prior commit).
we can safely remove the parameter.
this parameter was added as a (minor) optimization for
cases where dns_name_towire() is run repeatedly with the
same compression context, as when rendering all of the rdatas
in an rdataset. it is currently only used in one place.
we now simplify the interface by removing the extra parameter.
the compression offset value is now part of the compression
context, and can be activated when needed by calling
dns_compress_setmultiuse(). multiuse mode is automatically
deactivated by any subsequent call to dns_compress_permitted().
the dns_rdataslab_fromrdataset() function creates a slab
from an rdataset. if the source rdataset already uses a slab,
then no processing is necessary; we can just copy the existing
slab to a new location.
Previously, the hashmap iterator for fetches-per-zone was destroy
outside the rwlock. This could lead to an assertion failure due to a
timing race with the internal rehashing of the hashmap table as the
rehashing process requires no iterators to be running when rehashing the
hashmap table. This has been fixed by moving the destruction of the
iterator inside the read locked section.
The third argument to set_offsets() was only used in
dns_name_fromregion() and not really needed. We can remove the third
argument and then manually check whether the last label is root label.
There was just a single use of passing an extra buffer to
dns_name_downcase() which have been replaced by simple call to
isc_ascii_lowercase() and the 'target' argument from dns_name_downcase()
function has been removed.
The MAKE_EMPTY() macro was clearing up the output variable in case of
the failure. However, this was breaking the usual design pattern that
the output variables are left in indeterminate state or we don't touch
them at all when a failure occurs. Remove the macro and change the
dns_name_downcase() to not touch the name contents until success.
There was a back-and-forth between static arrays and the pointers to the
offsets. Since we are now only using the static arrays, we can cleanup
the usage of the pointers that would previously point either to the
static array or name->offsets if available.
The offsets were meant to speed-up the repeated dns_name operations, but
it was experimentally proven that there's actually no real-world
benefit. Remove the offsets and labels fields from the dns_name and the
static offsets fields to save 128 bytes from the fixedname in favor of
calculating labels and offsets only when needed.
The DNS header shows if a message has multiple questions or invalid
NOTIFY sections. We can drop these messages early, right after parsing
the question. This matches RFC 9619 for multi-question messages and
Unbound's handling of NOTIFY.
To further add further robustness, we include an additional check for
unknown opcodes, and also drop those messages early.
Add early_sanity_check() function to check for these conditions:
- Messages with more than one question, as required by RFC 9619
- NOTIFY query messages containing answer sections (like Unbound)
- NOTIFY messages containing authority sections (like Unbound)
- Unknown opcodes.
A change in 6aba56ae8 (checking whether a rejected RRset was identical
to the data it would have replaced, so that we could still cache a
signature) inadvertently introduced cases where processing of a
response would continue when previously it would have been skipped.
Acquire the database refernce in the detachnode() to prevent the last
reference to be release while the NODE_LOCK being locked. The NODE_LOCK
is locked/unlocked inside the RCU critical section, thus it is most
probably this should not pose a problem as the database uses call_rcu
memory reclamation, but this it is still safer to acquire the reference
before releasing the node.
As the default_call_rcu_thread can't be forced to flush all the work
during the executable shutdown, create one call_rcu_thread explicitly
and assign it to the all created threads.
This allows this explicit call_rcu_thread to be unassociated from the
main thread and freed before the executable destructor exits.
Instead of relying on unreliable order of execution of the library
constructors and destructors, move them to individual binaries. The
advantage is that the execution time and order will remain constant and
will not depend on the dynamic load dependency solver.
This requires more work, but that was mitigated by a simple requirement,
any executable using libisc and libdns, must include <isc/lib.h> and
<dns/lib.h> respectively (in this particular order). In turn, these two
headers must not be included from within any library as they contain
inlined functions marked with constructor/destructor attributes.
Previously, the dns_resolver_dumpfetches() would go over the fetch
counters. Alas, because of the earlier optimization, the fetch counters
would be increased only when fetches-per-zone was not 0, otherwise the
whole counting was skipped for performance reasons.
Instead of using the auxiliary fetch counters hash table, use the real
hash table that stores the fetch contexts to dump the ongoing fetches to
the recursing file.
Additionally print more information about the fetch context like start
and expiry times, number of fetch responses, number of queries and count
of allowed and dropped fetches.
The order of the fetch context hash table rwlock and the individual
fetch context was reversed when calling the release_fctx() function.
This was causing a problem when iterating the hash table, and thus the
ordering has been corrected in a way that the hash table rwlock is now
always locked on the outside and the fctx lock is the interior lock.
In the next commit, we need to know whether the timer has been started
or stopped. Add isc_timer_running() function that returns true if the
timer has been started.
After a reconfiguration the old view can be left without a valid
'rpzs' member, because when the RPZ is not changed during the named
reconfiguration 'rpzs' "migrate" from the old view into the new
view, so when a query resumes it can find that 'qctx->view->rpzs'
is NULL which query_resume() currently doesn't expect to happen if
it's recursing and 'qctx->rpz_st' is not NULL.
Fix the issue by adding a NULL-check. In order to not split the log
message to two different log messages depending on whether
'qctx->view->rpzs' is NULL or not, change the message to not log
the RPZ policy's "version" which is just a runtime counter and is
most likely not very useful for the users.
The isc_counter_create() doesn't need the return value (it was always
ISC_R_SUCCESS), use the macros to implement the reference counting,
little style cleanup, and expand the unit test.
Checking whether the authority section is properly signed should
be left to the validator. Checking in getsection (dns_message_parse)
was way too early and resulted in resolution failures of lookups
that should have otherwise succeeded.
Previously a hard-coded limitation of maximum two key or message
verification checks were introduced when checking the message's
SIG(0) signature. It was done in order to protect against possible
DoS attacks. The logic behind choosing the number two was that more
than one key should only be required only during key rotations, and
in that case two keys are enough. But later it became apparent that
there are other use cases too where even more keys are required, see
issue number #5050 in GitLab.
This change introduces two new configuration options for the views,
sig0key-checks-limit and sig0message-checks-limit, which define how
many keys are allowed to be checked to find a matching key, and how
many message verifications are allowed to take place once a matching
key has been found. The latter protects against expensive cryptographic
operations when there are keys with colliding tags and algorithm
numbers, with default being 2, and the former protects against a bit
less expensive key parsing operations and defaults to 16.
Running jobs which were entered into the isc_quota queue is the
responsibility of the isc_quota_release() function, which, when
releasing a previously acquired quota, checks whether the queue
is empty, and if it's not, it runs a job from the queue without touching
the 'quota->used' counter. This mechanism is susceptible to a possible
hangup of a newly queued job in case when between the time a decision
has been made to queue it (because used >= max) and the time it was
actually queued, the last quota was released. Since there is no more
quotas to be released (unless arriving in the future), the newly
entered job will be stuck in the queue.
Fix the wrong memory ordering for 'quota->used', as the relaxed
ordering doesn't ensure that data modifications made by one thread
are visible in other threads.
Add checks in both isc_quota_release() and isc_quota_acquire_cb()
to make sure that the described hangup does not happen. Also see
code comments.
Expose the average transfer rate (in bytes-per-second) during the
last full 'min-transfer-rate-in <bytes> <minutes>' minutes interval.
If no such interval has passed yet, then the overall average rate is
reported instead.
This new option sets a minimum amount of transfer rate for
an incoming zone transfer that will abort a transfer, which
for some network related reasons run very slowly.
Add a new dns_rdataset_equals() function to check whether two
rdatasets are equal in DNSSEC terms.
When an rdataset being cached is rejected because its trust
level is lower than the existing rdataset, we now check to see
whether the rejected data was identical to the existing data.
This allows us to cache a potentially useful RRSIG when handling
CD=1 queries, while still rejecting RRSIGs that would definitely
have resulted in a validation failure.
The "raw" version of the header was used for the noqname and the closest
proofs to save around 152 bytes of the dns_slabheader_t while bringing
an additional complexity. Remove the raw version of the dns_slabheader
API at the slight expense of having unused dns_slabheader_t data sitting
in front of the proofs.
reduce the number of rdata comparisons needed by walking
through the original slab once to determine whether the rdata
in it is duplicated in the slab to be subtracted, and then
write out the rdatas that aren't. previously, this was
done twice: once when determining the size of the target buffer
and then again when copying data into it.
when merging two rdata slabs, we now check once to see
whether an item in the new slab has a duplicate in the
old. previously this was done twice; once to determine the
size of the target buffer required, and then again when
copying the data into it.
we also minimize the number of rdata comparisons necessary,
by remembering which items in the old slab have already been
found to be duplicates.
The function name dns_slabheader_fromrdataset() was too similar
to dns_rdataslab_fromrdataset(). Instead, we now have an rdataset
method 'getheader' which is implemented for slab-type rdatasets.
A new NOHEADER rdataset attribute is set for rdatasets using
raw slabs (i.e., noqname and closest encloser proofs); when
called on rdatasets with that flag set, dns_rdataset_getheader()
returns NULL.
when dns_rdataslab_fromrdataset() is run, in addition to
allocating space for a slab header, it also partially
initializes it, setting the type match rdataset->type and
rdataset->covers, the trust to rdataset->trust, and the TTL to
rdataset->ttl.
there are now two functions for creating an rdataslab from an
rdataset: dns_rdataslab_fromrdataset() creates a full slab (including
space for a slab header), and dns_rdataslab_raw_fromrdataset() creates
a raw slab.
- there are now two functions for getting rdataslab size:
dns_rdataslab_size() is for full slabs and dns_rdataslab_sizeraw()
for raw slabs. there is no longer a need for a reservelen parameter.
- dns_rdataslab_count() also no longer takes a reservelen parameter.
(currently it's never used for raw slabs, so there is no _countraw()
function.)
- dns_rdataslab_rdatasize() has been removed, because
dns_rdataslab_sizeraw() can do the same thing.
- dns_rdataslab_merge() and dns_rdataslab_subtract() both take
slabheader parameters instead of character buffers, and the
reservelen parameter has been removed.
if both rdataslabs being compared have zero length, return true.
also, since these functions are only ever called on slabheaders
with sizeof(dns_slabheader_t) as the reserve length, we can
simplify the API: remove the reservelen argument, and pass the
slabs as type dns_slabheader_t * instead of unsigned char *.
The dns_slabheader object uses the 'next' pointer for two purposes.
In the first header for any given type, 'next' points to the first
header for the next type. But 'down' points to the next header of
the same type, and in that record, 'next' points back up.
This design made the code confusing to read. We now use a union
so that the 'next' pointer can also be called 'up'.
The value returned by http_send_outgoing() is not used anywhere, so we
make it not return anything (void). Probably it is an omission from
older times.
When handling outgoing data, there were a couple of rarely executed
code paths that would not take into account that the callback MUST be
called.
It could lead to potential memory leaks and consequent shutdown hangs.
This commit changes the way how the number of active HTTP streams is
calculated and allows it to scale with the values of the maximum
amount of streams per connection, instead of effectively capping at
STREAM_CLIENTS_PER_CONN.
The original limit, which is intended to define the pipelining limit
for TCP/DoT. However, it appeared to be too restrictive for DoH, as it
works quite differently and implements pipelining at protocol level by
the means of multiplexing multiple streams. That renders each stream
to be effectively a separate connection from the point of view of the
rest of the codebase.
Previously we would limit the amount of incoming data to process based
solely on the presence of not completed send requests. That worked,
however, it was found to severely degrade performance in certain
cases, as was revealed during extended testing.
Now we switch to keeping track of how much data is in flight (or ready
to be in flight) and limit the amount of processed incoming data when
the amount of in flight data surpasses the given threshold, similarly
to like we do in other transports.
Database versions are not used in cache databases. Some places in
qpcache.c required the version argument to be NULL; others marked it
as UNUSED. Unify all cases to require version to be NULL.
Add new related_headers() function that simplifies the code
flow in qpcache_findrdataset(). Also use check_stale_header() function
to remove code duplication.
Add a helper function both_headers() that unifies the slabheader
matching for simple type: it returns true when both the type and
the matching RRSIG have been found.
The new maybe_update_headers() function unifies the LRU updates to the
slabheaders that was scattered all over the place. More calls to update
headers after bindrdatasets() were also added for completeness.
This removes code duplication between the dual bindrdataset() calls. It
also unifies the handling as there were small differences between the
calls: one variant was checking for !NEGATIVE(found) condition and one
wasn't, and it is technically ok to do the check for all variants.
