rbldnsd (0.999~20180516-3) unstable; urgency=medium
* Disabled dh_auto_test for the time being.
-- Marco d'Itri <md@linux.it> Thu, 29 Aug 2019 16:01:06 +0200
rbldnsd (0.999~20180516-2) unstable; urgency=medium
* Depend on pkg-config.
-- Marco d'Itri <md@linux.it> Thu, 29 Aug 2019 14:27:27 +0200
rbldnsd (0.999~20180516-1) unstable; urgency=medium
* New upstream snapshot.
* Added support for systemd and socket activation.
-- Marco d'Itri <md@linux.it> Thu, 29 Aug 2019 12:47:49 +0200
rbldnsd (0.998b~pre1-1) unstable; urgency=medium
* New upstream snapshot.
* New upstream maintainer.
* New Debian maintainer.
-- Marco d'Itri <md@linux.it> Tue, 27 Sep 2016 02:04:33 +0200
rbldnsd (0.998-1) UNRELEASED; urgency=low
* configure.lib-fix-building-with-as-needed.patch (Closes: #641556)
(cherry-picked). Thanks to Matthias Klose <doko@ubuntu.com> for
reporting this and providing the fix.
* add build-{arch,indep} targets to d/rules
* use dpkg-buildflags for LDFLAGS CPPFLAGS CFLAGS
* add Vcs headers to d/control
-- Michael Tokarev <mjt@tls.msk.ru> Sun, 20 Oct 2013 21:00:39 +0400
rbldnsd (0.997a-1) unstable; urgency=low
* change the way how `make dist' behaves
(no need for a temporary subdir anymore)
* fixed sed expression to determine version number, needed for *BSD
-- Michael Tokarev <mjt@tls.msk.ru> Tue, 23 Jul 2013 20:32:28 +0400
rbldnsd (0.997) unstable; urgency=low
[ Jeff Dairiki ]
* configure: add --enable-asserts option to enable compilation of
assertions; assertion checking is disabled by default
* tests/test_acl.py, tests/test_ip4trie.py: new, very basic, tests
for the acl and ip4trie datasets
* rbldnsd_acl.c, rbldnsd_ip4trie.c: allow zero-length "wildcard" IP4
CIDR prefixes
* rbldnsd_acl.c: add support for ip6 addresses in ACLs
* rbldnsd_util.c: delete the (now unused) ip4trie code
* rbldnsd_acl.c: use new LC-TBM trie implementation
* rbldnsd_ip4trie.c: use new LC-TBM trie implementation instead of
the previous ip4trie; this decreased the memory consumption of
this dataset by roughly a factor of three
* rbldnsd_ip6tset.c: use new dump_ip6()
* Makefile.in: add 'check' target to run btrie.c self-tests and new
python-driven tests
* feature: ip6trie - new dataset supports listing of arbitrary length
ip6 CIDRs, along with individual A/TXT values for each prefix
* rbldnsd_util.c(dump_ip6, dump_ip6range): new functions to support
zone dump of ip6 zones
* btrie.c: LC-TBM trie implementation; supports arbitrary length prefixes
in a memory- and cpu- efficient manner
* configure: add test for __SIZEOF_POINTER__ - define if the compiler
does not already do so
* configure: add test for byte sex - define WORDS_BIGENDIAN in config.h
on big-endian architectures
* ip6addr.c: add support for compressed ("::") ip6 address notation
* mempool.c: bug fix for a theoretically possible buffer overrun
[ Michael Tokarev ]
* feature: ip6tset - new dataset supports listing of ip6 /64 subnets and
the exclusion of /128 subnets; only supports a single A/TXT value for
the entire dataset
-- Michael Tokarev <mjt@corpit.ru> Sat, 29 Jun 2013 12:02:54 +0400
rbldnsd (0.996b) unstable; urgency=low
* cleanup: #ifdefed dump_a_txt() properly so it will not be compiled
if --disable-master-dump was specified.
* implement "base template" ($=) feature.
* feature: (experimental) support for dynamically-loaded extensions (DSOs)
(disabled by default, use --enable-dso configure option)
* portability and readability fixes from Victor Duchovni
* added configure test for inline and __inline keywords, and only
use #warning keyword if __GNUC__ is defined (more portability fixes
from Victor Duchovni)
* misc type conversions here and there, and change alignment in
mempool.c to be sizeof(void*) instead of sizeof(int), to help
64bit platforms. Thanks to Mike Quintero for an excellent
bugreport.
* bugfix: combined dataset - improper return of query() routine in some
cases
* internal code reorg:
- move firstword[_lc]() to _util.c
- use two structs instead of a set of 2-element arrays in dnset
* bugfix: lowercase base zone names given on command line and in `combined'
dataset, or else they wont be recognized in queries
* added an lsb info to Debian initscript (Closes: #468886)
-- Michael Tokarev <mjt@corpit.ru> Sat, 29 Mar 2008 17:38:49 +0300
rbldnsd (0.996a) unstable; urgency=low
* the "34-Birthday Release"
* use setitimer() instead of alarm() on systems which supports it
(configure & rbldnsd.c). This works around some broken linux
kernel behaviour where we sometimes lose a signal (SIGALRM in this
case) and rbldnsd stops checking for updates.
* cosmetic code cleanups:
- use void* instead of char* in mempool.[ch] where appropriate.
- wrong type in check_expires(): unsigned instead of time_t.
- wrong type (signedness) for domain name in ds_dnset_dump()
- use appropriate char/uchar for domain names in ds_generic_dump()
- istream signedness (warning) fixes
* debian:
- Build-Depends on debhelper>>4, and set DH_COMPAT to 4
- Build-Depends on zlib1g-dev, and pass --enable-zlib to ./configure
- bump Standards-Version to 3.7.2
- add Depends: on adduser (Closes: #398560)
* bugfix: fix dataset "un-expiration" time. Previously, once a dataset
has expired, it never "un-expires" again even if new expire time is
in future. Due to missing reset of ds->ds_expire field.
* bugfix: fix configure breakages:
- portability: for f; do => for f in "$@"; do
- fix broken GNU C (mis)detection
* don't treat -a as experimental and mention it will be the default.
Add -A option.
-- Michael Tokarev <mjt@corpit.ru> Thu, 27 Jul 2006 13:36:30 +0400
rbldnsd (0.996) unstable; urgency=low
* 0.996 release
* portability fix in istream.c:
EPROTO=>ENOEXEC if EPROTO isn't defined
-- Michael Tokarev <mjt@corpit.ru> Sun, 19 Feb 2006 16:41:16 +0300
rbldnsd (0.995.99) unstable; urgency=low
* add a 'pass' entry processing into ACL "dataset"
* (internal) reorganize ds_special() (no user-visible changes)
* (internal) reorganized zone reloading code to eliminate
start_loading()+longjump() hack
* data expiration support, in form
$TIMESTAMP created expires
-- Michael Tokarev <mjt@corpit.ru> Sun, 29 Jan 2006 00:43:51 +0000
rbldnsd (0.995.98) unstable; urgency=low
* deal with realloc(smaller_size) returning NULL
* add next-reload-size hint to dnset too
* (cosmetic code) unify several structure member names across
different *sets
* up MAX_NS from 20 to 32, per request from Spamhaus
* fix a nasty typo in vssprintf() (bufsiz => bufsz). DO #undefine
constants which aren't needed anymore!
* rearrange code a bit: move readslines() from rbldnsd_util.c to
rbldnsd_zones.c, make some functions static and remove declarations
from rbldnsd.h. No code changes.
* cosmetic again: NOSTATS, NOSTDINT_H etc => NO_STATS, NO_STDINT_H etc.
Note if you had automated building procedure you have to change
your #defines.
* new configure script, using shell functions in configure.lib - much
more manageable.
* configure script now accepts some --disable- and --enable- options
(--disable-ipv6)
* istream.[ch] - helper module, light buffering input somewhat similar
to stdio, much faster, supports line oriented input, and is stackable
(one istream can read from another or something else). Reading speed
improved significantly.
* support of automatic on-the-fly de-compression of gzip-compressed
data files (with a help of zlib).
-- Michael Tokarev <mjt@corpit.ru> Tue, 20 Dec 2005 00:15:56 +0300
rbldnsd (0.995) unstable; urgency=low
* released 0.995
* ensure we do not return more than 255 answers (which can happen when
EDNS0 size extensions are enabled), as the code isn't prepared to handle
2-byte numanswers field in DNS packet header.
-- Michael Tokarev <mjt@corpit.ru> Thu, 28 Apr 2005 23:57:32 +0400
rbldnsd (0.994.94) unstable; urgency=low
* fixed EDNS0 (it now really works)
* fixed NS+glue records - if there's no room for glue but auth section
can be added, add auth w/o glue
* remove p_..cnt #defines (to be p_..cnt2), as they're confusing
-- Michael Tokarev <mjt@corpit.ru> Thu, 28 Apr 2005 15:24:10 +0400
rbldnsd (0.994.93) unstable; urgency=low
* glue records for NSes, finally.
* when queried for NS or ANY to the base zone, return NS recs in
both answer and authority sections.
* in acl "dataset", default action is now "ignore", not "always-listed"
-- Michael Tokarev <mjt@corpit.ru> Wed, 27 Apr 2005 19:04:56 +0400
rbldnsd (0.994.92) unstable; urgency=low
* finally: EDNS0 support. Patch was here for a long time, now applied.
* ACL (initial, experimental) support, with quite some changes all over
to make it possible/easier:
- add peer address info pointers to struct dnspacket
(and made both struct packet and peer sockaddr to be static
in rbldnsd.c); remove peer address parameters from
logreply().
- add DSTF_SPECIAL flag, to indicate this dataset type can't be nested;
modify rbldnsd_combined.c accordingly
- remove (unused) DSTF_ZERODN flag
- define two new helper macros, dstype(name) and isdstype(dst,name),
to refer to types of datasets, and use the macros in the code
- add ACL-specific RR info into struct dnspacket
- dataset query types now return bitflags, not true/false:
NSQUERY_FOUND (1) - found a matching record
NSQUERY_ADDPEER (2) - for "always listed" acl, we should add
the "always listed" ACL RR into the reply packet.
return NSQUERY_FOUND instead of generic `1' in all dataset->queryfns.
- extend qi_tflag to also include ACL-specific flags:
NSQUERY_IGNORE, NSQUERY_REFUSE, NSQUERY_EMPTY, NSQUERY_ALWAYS
- add check_query_overwrites() macro to test the above flags and
return NSQUERY_ADDPEER or other bits in dataset->queryfn routines,
and use this macro in all non-metadata datasets
- when constructing reply, collect flags from queryfns, not boolean.
- add g_dsacl and zone->z_dsacl pointers (global and zone-specific
datasets)
- modify rbldnsd_zones.c:addzone() to recognize ACLs and to disallow
empty base zone domain name.
- disallow $NS and $SOA for ACL-type datasets in ds_special()
- new dataset file: rbldnsd_acl.c. Add acl dataset into global
dataset type list.
- actually call ACL-specific routine -- ds_acl_query() -- when
constructing reply to a query
- when at the end of constructing answer section of the reply
we notice "always-listed" ACL has been triggered, add
ACL-specific A+TXT records into the reply
-- Michael Tokarev <mjt@corpit.ru> Sat, 16 Apr 2005 17:43:43 +0400
rbldnsd (0.994b) unstable; urgency=low
* use of uninitialized pointer in ip4set and ip4trie
datasets when input data file (A+TXT template for a given
entry) is invalid, instead of rejecting the line. This can
lead to "random" crashes.
-- Michael Tokarev <mjt@corpit.ru> Thu, 10 Mar 2005 01:57:14 +0300
rbldnsd (0.994a) unstable; urgency=low
* fixed wrong NXDOMAIN return for a query for base subzone in
combined dataset (should be NODATA at least, not NXDOMAIN)
-- Michael Tokarev <mjt@corpit.ru> Thu, 10 Mar 2005 01:43:22 +0300
rbldnsd (0.994) unstable; urgency=low
* released 0.994
-- Michael Tokarev <mjt@corpit.ru> Sat, 18 Dec 2004 17:07:24 +0300
rbldnsd (0.993.9) unstable; urgency=low
* fixed wrong return in ds_ip4trie_dump_octets(), which
prevented some ranges from being expanded properly when
creating master-format dump (-d) file.
* generalized and moved ds_ip4trie_dump_range() (renamed to
dump_ip4range()) to rbldnsd_utils.c to made it generally
useful (to be used in rbldnsd_ip4set.c too), and made it
void (cf the wrong return above)
* made txtsubst() public, and move it and dump_a_txt() into
rbldnsd_utils.c where the stuff belongs to, out from
rbldnsd_packets.c which is too large and has nothing to
do with that stuff.
* new helper dump_ip4(), analogous to dump_ip4range()
* use dump_ip4() in ip4tset
* rewrite ds_ip4set_dump(), to handle that problrmatic case
with entries 127/8 and 127.0.0.2: should also emit
*.0.127 and *.0.0.127
* recognize NO_MASTER_DUMP #define (disable -d option)
* fix memleak when loading $n substitutions (mp_strdup vs estrdup)
-- Michael Tokarev <mjt@corpit.ru> Mon, 13 Dec 2004 04:40:27 +0300
rbldnsd (0.993.1) unstable; urgency=low
* don't remove rbldns user in debian/postrm (Closes: bug#258012).
* create /var/lib/rbldns root-owned, to stop encouraging usage of
rbldns userid to store datafiles.
* add README.user explaining rbldns userid and /var/lib/rbldns dir usage
* #include <sys/time.h> in rbldnsd.c to get struct timeval definition
(needed e.g. on older glibc)
-- Michael Tokarev <mjt@corpit.ru> Thu, 29 Jul 2004 17:46:51 +0400
rbldnsd (0.993) unstable; urgency=low
* 0.993 release, finally
* cleaned up debian/rules a bit per recommendations by Santiago Vila
* fixed a bug when rbldnsd was using 0.0.0.0 as A value instead
of a real IP address (when construct like ":127.0.0.2" is specified),
introduced in 0.992
* manpage: described various :A:TXT cases; reformatted to use
.SS instead of .IP in some cases.
* remove a note about ip4trie and -d option: ip4trie is dumpable
now.
-- Michael Tokarev <mjt@corpit.ru> Thu, 1 Jul 2004 20:23:01 +0400
rbldnsd (0.993p2) unstable; urgency=low
* warn when $DATASET specifies no subzones
* recognize # in $DATASET line only after a space,
so that one may have names with embedded #s.
* allow names for subdatasets in combined dataset,
for better logging. Specify :name after type
in $DATASET line, like
$DATASET ip4set:http proxies @
$DATASET ip4set:relays relays @
-- Michael Tokarev <mjt@corpit.ru> Thu, 1 Jul 2004 15:27:30 +0400
rbldnsd (0.993p1) unstable; urgency=low
* implement and enforce $MAXRANGE4 special
* refine logging a bit, make it less verbose
* (internal) reorganize loading process a bit to
have dsctx parameter (load context) to be passed
as an argument, instead of global ds_loading.
(internal) moved ip4parse_*() into the only places
where that stuff was used.
(internal) minor cleanups in combined dataset
* ignore incomplete last lines (lines w/o end-of-line
terminator) in data files
* check for data file changes during reloads (while
reading data), and abort loading if a change is
detected
* another parsing error in ip4parse.c (ip4addr() and
ip4prefix(), unlike claimed, was able to return 0)
* more cleanups and clarifications in ip4parse.c.
Do not treat bare numbers as /8s anymore.
Do not allow ranges like 1.2-2 or 1.2-1.2.3 -
number of octets on both sides should match
(with the exception of 1.2.3.4-5 - last part
may have only one octet)
* really require complete IPv4 addresses (not
prefixes) in ip4tset.
* fixed query logging (-l) with background reloading
(missing fflush() before fork() and exit())
* use writev/readv to transfer stats in 2-process reload
(may be a bit faster)
* cleanup NOSTATS #defines/usage a bit
* bugfix: wrong MX pref in dump (-d) in generic dataset.
Fix by Andreas Pommer, apommer at cosy.sbg.ac.at
* bugfix: wrong subzone in $ORIGIN when dumping combined dataset.
Fix by Andreas Pommer, apommer at cosy.sbg.ac.at
* bugfix: incorect (opposite) evaluation of maxttl.
Fix by Bradley Baetz bradley.baetz at optusnet.com.au
* some hooks, to build custom versions (rbldnsd_hooks.[ch])
-- Michael Tokarev <mjt@corpit.ru> Thu, 10 Jun 2004 02:46:57 +0400
rbldnsd (0.992) unstable; urgency=low
* add an ability to specify A value but inherit default
TXT value:
entry :addr: - specific A, no TXT
entry :addr - specific A, default TXT
* remove redundrant CNAMEs from master-file dump in ip4set
* fix syslog statistics logging mess introduced in pre4
-- Michael Tokarev <mjt@corpit.ru> Sun, 7 Mar 2004 01:35:46 +0300
rbldnsd (0.992pre4) unstable; urgency=low
* reviewed statistics again:
- use different structure (smaller and simpler anyway)
- moved all stats updating to rbldnsd_packet.c
- different syslogging (simpler)
- different statslogging (more details)
* extended -t (TTL) option: -t defttl:minttl:maxttl.
* remove some more mess from memstats logging
* changed rbldnsd.init to not rely on readlink and be less strict
* rework 2-process reload, add workarounds for lost signals
-- Michael Tokarev <mjt@corpit.ru> Wed, 3 Mar 2004 17:43:38 +0300
rbldnsd (0.992pre3) unstable; urgency=low
* added ip4tset ("trivial" set of IPv4 addresses) -
faster and smaller but limited version of ip4set
* continue processing queries during reloads (-f option)
-- Michael Tokarev <mjt@corpit.ru> Mon, 1 Mar 2004 19:36:15 +0300
rbldnsd (0.992pre2) unstable; urgency=low
* cleaned up statistics printing
* removed -s option (log memory usage and (re)load times) -
turned (reworked) equivalent always on
* reworked memusage/times logging
* added -s back, to mean something different: log short
stats line on every check (-c) invocation, to help
collecting data for RRD-like applications
* remove some bash-isms from rbldnsd.init
* lazy mode is still experimental...
-- Michael Tokarev <mjt@corpit.ru> Sat, 28 Feb 2004 04:57:36 +0300
rbldnsd (0.992pre1) unstable; urgency=low
* add `lazy' mode (-a, experimental) to stop returing
AUTH section by default, return it only when asked
explicitly
* recognize `-' as single-NS-comment char in $NS line
* use O_LARGEFILE and #define _LARGEFILE64_SOURCE in rbldnsd.c
to be able to write larger logfiles
* removed usage of NI_WITHSCOPEID (was used for unknwon reason
anyway and it have problems with new Solaris)
* fix ds_generic_dump() that was broken for quite some time
* fix example dataset in manpage (missing preference for MX)
-- Michael Tokarev <mjt@corpit.ru> Wed, 4 Feb 2004 15:55:08 +0300
rbldnsd (0.991) unstable; urgency=low
* implemented master format dump for ip4trie dataset
* removed all references to osirusoft.com
* switch to integer TTL everywhere
* PACK32S() and PACK16S() aka stpcpy() (incrementing destination)
* $NS records change: accept all NS in single line or in multiple
lines
* increase dataset reading line length from 512 bytes to 8k to
allow large list of nameservers in one $NS record
* use ns records from FIRST dataset (with non-empty $NS) only
* add new dataset to the end of a list, not to top (cosmetic change,
specifies loading order and thus logging order only)
* disallow zero TTL in command line
* reworked duplicate RR detection and choosing of minTTL.
Fix TTL even when we have exactly the same RR in packet already.
* simplified zone NS records handling
* new routine: zlog(loglevel, zone, fmt, ...)
* log number of zones at startup
-- Michael Tokarev <mjt@corpit.ru> Sun, 30 Nov 2003 18:34:04 +0300
rbldnsd (0.990pre0) unstable; urgency=low
* ensure we return all RRs of the same type (if there
are several of them) with the same (smallest) TTL.
Problem spotted by Victor Duchovny, MorganStanley.
* "randomize" order of RRs of the same type in `generic'
dataset. The randomization is dumb, but it is better
than nothing anyway. Suggested by Furio Ercolessi.
-- Michael Tokarev <mjt@corpit.ru> Sun, 30 Nov 2003 14:28:52 +0300
rbldnsd (0.99) unstable; urgency=medium
* autoconf-style configuration. Run ./configure before make.
- #include "config.h"
- uint32_t
- STATS_LL => uint64_t/long long/long, PRI_DNSCNT: PRIu64/llu/lu
Thanks to Christian Krackowizer for testing and patience.
* change ip4addr parsing routines to return -1 in case of error,
instead of 0, and treat 0 to be valid return value too (0/0)
* fix utimes printing (time rbldnsd spent when loading zones)
* move VERSION[_DATE] extraction from Makefile to configure,
made Makefile depend on debian/changelog.
* debian/rules: do not attempt to make distclean if there's no
Makefile
* another config test: <sys/select.h> (NOSELECT_H)
* some mods to ip4trie
* cast [ug]id_t to int for printf
* config test for vsnprintf(), fatal if not found for now (OSF lacks it)
* remove EasynetDynablock2rbldnsd.pl since easynet.nl provides dynablock
in rbldnsd format now.
* remove osirusoft2rbldns.pl since relays.osirusoft.com is dead.
* made zonelist really global in rbldnsd.c, reorder some globals to
be in one place, add some more comments
* Fixed range parsing. E.g., 24.217.64-191 did not work (and any range
like this where last two bits where xored into 255)
* allow logging to standard output (-l - or -l +-)
* move logfile handling a bit earlier in the init process, and
do not send initialization messages to stdout in case we're
logging queries there (-l -)
-- Michael Tokarev <mjt@corpit.ru> Tue, 16 Sep 2003 21:33:35 +0400
rbldnsd (0.98) unstable; urgency=low
* use memmove() instead of memcpy() for overlapping regions,
even if dstaddr is less than srcaddr (current memcpy() can
deal with that, but this is out of spec and future versions
may do some more optimizations that'll break things)
* made init_zone_caches() to work for a list of zones, not just
a single zone, and move a call to this routine from newzone()
to init(). Fixes a memory leak in combined dataset (caches
was allocated but never used/freed).
* properly (but still ugly) fix stats counting
* some "ugliness": use more const's and explicitly declare routines
wich does not accept any arguments as such to be foo(void)
(Marco D'Itri md at linux.it)
* handle version.bind CH TXT (and version.server for that matter)
* return REFUSED instead of FORMERR in case unrecognized query
class (IN,ANY,etc) requested
* defttl => def_ttl to be consistent with def_rr
* set ttl for version.bind to be 0
* warn about truncating TXT records. Change interface of parse_a_txt()
to include a line number
* TXT RR is 255 bytes max, not 254 (generic dataset)
* wildcards in dnset: *.example.com and example.com works as before,
but .example.com is now interpreted as BOTH *.example.com and example.com.
* clarifications in manpage: dataset vs zone
* fix logging to various destinations (LOGTO_* constants, vdslog())
* allow to specify multiple listening addresses.
* recognize -b host:port and [ip]:port syntax (only host previously)
* -b option is now mandatory
* do fork() early in the initialization process so that logging will
have proper pid tag. This also slightly simplified init routine.
* made array of listening sockets to be global
* added ip4trie (experimental) dataset: set of IP4 CIDR ranges.
* yet again, fix CIDR parsing (mask vs ~mask)
* fix ip4trie common length calculation in case diff == 0
* fix dslog() to use vprintf, not printf
-- Michael Tokarev <mjt@corpit.ru> Sun, 17 Aug 2003 21:48:54 +0400
rbldnsd (0.97b) unstable; urgency=low
* fixed security bug in per-zone statistic counting.
Thanks Marco D'Itri (md at linux.it) for pointing this out to me.
-- Michael Tokarev <mjt@corpit.ru> Wed, 6 Aug 2003 03:20:36 +0400
rbldnsd (0.97a) unstable; urgency=medium
* recognize and ignore "in" in `generic' dataset, as in:
example.com 1D IN A 127.0.0.1
* do not truncate addresses in CIDR like 1.2.3.4/24 to 1.2.3.0/24
in ip4range()
* fix a bug in ip4set_parse(): propagate error return properly
-- Michael Tokarev <mjt@corpit.ru> Fri, 1 Aug 2003 13:43:28 +0400
rbldnsd (0.97) unstable; urgency=low
* even better mempool align arith
* remove some dnlen usage (only tiny slowdown of zone reloads)
* use switch(firstchar) in ds_special()
* return SOA first to ANY queries
* Also return NS in AUTH in ANY queries inside base zone (cosmetic)
* add SOA/NS even before all other recs when queried base DN
* combined dataset optimization: try to reuse datasets from previous
load instead of reallocating them on every reload to preserve
statistics collected by subdatasets
* in some rare cases, dnset missed one RR for a DN with multiple RRs
* -DRECOGNIZE_IP4IN6 - treat reverse IPv6-mapped IPv4 queries as
plain IPv4 qieries
* Wirehub Dynablock => Easynet Dynablock
* -DSTATS_LL to use long longs for statistic counters
* per-basezone statistic counters
* corrections of various spelling errors in manpage, great thanks to
Bert Driehuis (driehuis{at}playbeing{dot}org) for this.
-- Michael Tokarev <mjt@corpit.ru> Sun, 13 Jul 2003 03:19:56 +0400
rbldnsd (0.96) unstable; urgency=high
* renamed almost all structures and variables, to be more accurate:
struct dataset => struct dsdata (dsd)
struct zonedataset => struct dataset (zds => ds)
struct zonedatalist => struct dslist (zdl => dsl)
struct zonefile => struct dsfile (zf => dsf)
struct dataset_type => struct dstype (dst)
dataset_types[] => ds_types[]
zds_loading => ds_loading
struct dnsqueryinfo => struct dnsqinfo (qi)
connectzonedataset() => connectdataset()
zds_special() => ds_special()
and so on.
* reworked NS and SOA internals, to use more natural data structures
(no more ugly packing of several fields into one char[])
* always assume that SOA and NS refers to zone's base DN -
less work for DN compression routines
* precompute SOA and NS records (pack various fields, pre-compress
domain names) for faster response time
* return NS records if available in AUTHORITY section of positive
replies
* restore MX functionality that was broken for quite some time.
Note that currently, MX records aren't compressed
* do not lowercase domain names used in NS, SOA and MX records.
Add dns_dnequ() routine, remove dns_dntol() call from parse_dn().
* pre-initialize zone caches when creating zone structure (before
startup) - both to avoid memory fragmentation and to ensure all
the required memory is allocated.
* Only add NS records if reply contains some data
* Add NS AUTH record for queries to base zone too (when not asked for NS)
* simplify and clarify warning about too long NS or SOA RRs
* remove zlog()
* Moves:
dslog() &Co from rbldnsd_zones.c to rbldnsd_utils.c
dntoip4addr() from rbldnsd_utils.c to rbldnsd_packet.c
ds_loading from rbldnsd_zones.c to rbldnsd.c
* fixed alignment bug in mempool.c that caused allocation slip
* simplify dssoa allocation
-- Michael Tokarev <mjt@corpit.ru> Thu, 29 May 2003 12:46:37 +0400
rbldnsd (0.95) unstable; urgency=low
* Portability: #include <netinet/in.h> for sockaddr_in definition in
rbldnsd_packet.c
* Allocate dataset to be part of zonedataset -> remove dataset's allocfn
* Added `combined' dummy dataset
* Change loading routines: ds_loadfn become ds_startfn and ds_linefn,
ds_linefn is now passed via zonedataset to readdslines. Preparations
for combined dataset.
* Implemented `combined' dataset, with many changes all other the places
* Notes, TODO items updated
* Fixed problem with DN compression introduced in recent changes.
* cleanup of #include's
* change names of zsoa, zns to reflect content more accurately
* align argument for mp_alloc(), to align object(s) to sizeof(int).
This allows to eliminate one mempool in combined dataset.
* use varlen zns_ttldn[]
* use one memory pool for all nested datasets in combined dataset
* add additional argument to ds_resetfn_t - freeall boolean, to
free all the memory instead of just zeroing it - needed for
combined dataset, but not currently used anyway
* in ip4set, remember size of previously allocated array to
avoid unnecessary realloc()s
* made ip4mask() to be an indexed array access to ip4addr_cidr_netmasks[]
-- Michael Tokarev <mjt@corpit.ru> Tue, 27 May 2003 22:18:20 +0400
rbldnsd (0.94) unstable; urgency=low
* implemented -d option (dump zone data in BIND format to stdout)
* data loading warnings goes to stderr instead of stdout
* Makefile portability tweaks for Solaris
* recognize ';' as comment char in addition to '#'; also, officially
recognize comments after an entry (IP address or domain name) in
ip4set and dnset
-- Michael Tokarev <mjt@corpit.ru> Mon, 26 May 2003 17:49:27 +0400
rbldnsd (0.93) unstable; urgency=low
* reverse change made in 0.91: SOA TTL, when SOA is in AUTHORIRY
section, should be set to MINTTL field of SOA itself.
-- Michael Tokarev <mjt@corpit.ru> Sun, 18 May 2003 19:05:38 +0400
rbldnsd (0.92) unstable; urgency=medium
* bugfix: fixed SOA screwup introduced in 0.91
-- Michael Tokarev <mjt@corpit.ru> Sat, 17 May 2003 05:24:39 +0400
rbldnsd (0.91) unstable; urgency=low
* print version info at build time (Makefile)
* made SIGNALLED_* names action-, not signal-dependant
* allow compilation without IPv6 (-DNOIPv6)
* free addrinfo structure for bind address (not a real leak: it used once)
* rotate nameserver records (simple cyclic rotation)
* do not reallocate memory for dataset headers on every reload
* fixed glitch in dnset - min DN labels was always 0 (tiny speedup only)
* fixed default A RR to be 127.0.0.2, not 2.0.0.127
* added a note about absolute vs relative domain names into manpage
* allow to specify a unit (s, m, h, d, w) for all time values
* changed default ttl to be 35m
* use ISSPACE/SKIPSPACE everywhere
* use PACK32/PACK16 everywhere
* some more variants of parse_{time,uint32}
* use parse_time() for -c option too
* always use SOA's TTL when SOA is included in answer
* added RPM .spec file
-- Michael Tokarev <mjt@corpit.ru> Thu, 15 May 2003 02:35:35 +0400
rbldnsd (0.90) unstable; urgency=low
* 0.90 final. Some little mods for IPv6.
-- Michael Tokarev <mjt@corpit.ru> Sat, 10 May 2003 01:54:52 +0400
rbldnsd (0.89p4.ip6) unstable; urgency=low
* IPv6 transport support. Default is to try any,
specify -6 or -4 to select IPv6 or IPv4.
* -a (ACL) and -L (log ACL) options are gone for now,
because there is no IPv6 versions.
-- Michael Tokarev <mjt@corpit.ru> Thu, 8 May 2003 21:24:02 +0400
rbldnsd (0.89p4) unstable; urgency=low
* NXDOMAIN elimination is gone. Bind9 does that too, so I assume
it is correct behaviour. Many code removed.
* Another prerelease.
-- Michael Tokarev <mjt@corpit.ru> Thu, 8 May 2003 21:24:02 +0400
rbldnsd (0.89p3) unstable; urgency=low
* ip4vset and ip4set merged, ditto for dnvset and dnset.
* $NS in datasets
* $TTL in datasets generic records; now required for $SOA and $NS
* $n substitutions
* more details in NEWS
-- Michael Tokarev <mjt@corpit.ru> Thu, 8 May 2003 16:59:33 +0400
rbldnsd (0.89p2) unstable; urgency=low
* again, many changes. DN-based lists are now ok too from
BIND point of view. Logging improvements. See NEWS file.
-- Michael Tokarev <mjt@corpit.ru> Tue, 6 May 2003 03:33:37 +0400
rbldnsd (0.89p1) unstable; urgency=low
* big number of changes, to match BIND's runtime behaviour.
See NEWS for details
-- Michael Tokarev <mjt@corpit.ru> Sun, 4 May 2003 18:27:42 +0400
rbldnsd (0.84p2) unstable; urgency=low
* rewrote query parsing routine to be much more accurate (see
version 0.83 change entry) and a bit faster.
* cleanups, cleanups.
* return definitive answer to AAAA, PTR and CNAME queries. A
hack for now, seeking for a better way...
-- Michael Tokarev <mjt@corpit.ru> Tue, 29 Apr 2003 04:29:02 +0400
rbldnsd (0.83) unstable; urgency=high
* critical buffer overflow fix in dns query parsing code.
Initial code (0.1 version) was right, but it was rewritten
in 0.2 - the bug was here since 0.2!.. Ughhh!..
-- Michael Tokarev <mjt@corpit.ru> Sat, 19 Apr 2003 05:24:09 +0400
rbldnsd (0.83p1) unstable; urgency=low
* -DNOSTDINT_H to use <sys/types.h> instead of <stdint.h> for uint32_t.
* changed slightly "zone loaded" message
* added minimal access control (-a netlist option)
* added query log filter (-L netlist option), and added logging of
timestamps.
* moved packet receiving/replying stuff into rbldnsd.c
* one-by-off bug - retrying interrupted replies was wrong
* use inline version of qsort in order to be able to use inline
comparision routine. Speed up loading significantly.
* allow to run as non-root; little cleanups
* removed -DNOREMOVEDUPS, -DNOIP4RANGES, -DPRINT_TIMES => -DNOTIMES
* added statistic counters (SIGUSR1/USR2/exit) - disable with -DNOSTATS
* added meminfo logging via mallinfo (-m option) - disable with -DNOMEMINFO
* fixed invocation w/o -b (bind address): rbldnsd wasn't work w/o -b at all
-- Michael Tokarev <mjt@corpit.ru> Sat, 19 Apr 2003 02:25:18 +0400
rbldnsd (0.82) unstable; urgency=low
* recognize another variation of IP address range, for easy use:
127.0.0.1-2 is now treated as 127.0.0.1-127.0.0.2
127.0-200 is now treated as 127.0.0.0-127.200.255.255
* rbldnsd w/o arguments will exit with 1 after printing usage info
-h cause exit with 0
(I forgot to add exit() call after usage())
* example WirehubDynablock2rbldnsd.pl script
* debianized, added startup script for Debian
(rbldnsd is now a native Debian package)
-- Michael Tokarev <mjt@corpit.ru> Sat, 5 Apr 2003 11:40:31 +0400
rbldnsd (0.81-0) unstable; urgency=low
* Initial Debian Release.
Previous CHANGES file is now in CHANGES-0.81
-- Michael Tokarev <mjt@corpit.ru> Sat, 5 Apr 2003 00:30:39 +0400
Local variables:
mode: debian-changelog
End: