summaryrefslogtreecommitdiffstats
path: root/input/flow
Commit message (Collapse)AuthorAgeFilesLines
* Use `NFPROTO_*` constants for protocol familiesJeremy Sowden2025-06-011-11/+12
| | | | | | | | | | | | | | | | Netfilter has a set of `NFPROTO_*` constants for the protocol families that it supports, in part because it supports protocols and pseudo-protocols that do not have `PF_*` (and `AF_*`) constants. Currently, ulogd uses `AF_*` constants for protocol families, because it does not support any families which do not have `AF_*` constants. Switch to `NFPROTO_*` constants instead, so we can add ARP support later. In the IP2* filters, retain `AF_*` for address family variables. Remove a stray semicolon. Signed-off-by: Jeremy Sowden <jeremy@azazel.net> Signed-off-by: Florian Westphal <fw@strlen.de>
* nfct: add network namespace supportCorubba Smith2025-05-201-13/+69
| | | | | | | | | | | | | | | | | | | Allow the plugin to fetch data from a different network namespace. This is possible by changing the network namespace before opening the netlink socket, and immediately changing back to the original network namespace once the socket is open. The number of nfct_open usages here warranted a dedicated wrapper function. If changing back to the original network namespace fails, ulogd will log an error, but continue to run in a different network namespace than it was started in, which may cause unexpected behaviour. But I don't see a way to properly "escalate" it such that ulogd aborts entirely. Also slightly adjust the error log messages to specify which socket failed to open. Signed-off-by: Corubba Smith <corubba@gmx.de> Signed-off-by: Florian Westphal <fw@strlen.de>
* nfct: add flow end timestamp on hashtable purgeCorubba Smith2025-03-251-0/+1
| | | | | | | | | | | | | | | | | In polling mode during normal operation, as well as in event mode with hashtable when an overrun occurs, the hashtable is fully re-synced against conntrack. When removing flows from the hashtable that are no longer in conntrack, there is no way to get the actual end timestamp of the flow from conntrack because it is already gone. Since the last conntrack data in the hashtable for these flows will never contain an end timestamp in this case, set_timestamp_from_ct() will always fall back to using the current time, aka when the plugin determines that the flow disappeared from conntrack. That is only an approximation, but should be good enough; and certainly more accurate than no end timestamp at all. Signed-off-by: Corubba Smith <corubba@gmx.de> Signed-off-by: Florian Westphal <fw@strlen.de>
* nfct: fix counter-reset without hashtableCorubba Smith2025-03-251-0/+3
| | | | | | | | | | | | | | | In event mode the hashtable is optional, and sending SIGUSR2 to ulogd will call get_ctr_zero(). The dump_reset_handler will try to update the hashtable regardless of whether it is used (and thus initialized), which results in a segfault if it isn't. Instead just short-circuit the handler, and skip any further result processing because it's not used in this case anyway. All flow counters in conntrack are reset regardless of the return value of the handler/callback. Signed-off-by: Corubba Smith <corubba@gmx.de> Signed-off-by: Florian Westphal <fw@strlen.de>
* all: remove trivial configure hooksCorubba Smith2025-03-121-13/+0
| | | | | | | These are now covered by the default implementation. Signed-off-by: Corubba Smith <corubba@gmx.de> Signed-off-by: Florian Westphal <fw@strlen.de>
* all: use config_parse_file function in all pluginsCorubba Smith2025-03-121-1/+1
| | | | | | | | | | Replace all usages of `config_parse_file()` in plugins with the new `ulogd_parse_configfile()` function, adding error handling where it was missing. I used the same codestyle as the surrounding code, which varies between plugins. Signed-off-by: Corubba Smith <corubba@gmx.de> Signed-off-by: Florian Westphal <fw@strlen.de>
* nfct: add icmpv6Corubba Smith2025-03-121-0/+26
| | | | | | | | | | | | | Add two new dedicated fields to provide the ICMPv6 code and type. While libnetfilter_conntrack uses the same attribute for both ICMPv4 and v6, there are no version-agnostic ICMP IEs in IPFIX. The fields are annotated with the appropriate IPFIX metadata, which is currently not actually used anywhere. You may call it consistency, future-proofing or cargo-culting. Signed-off-by: Corubba Smith <corubba@gmx.de> Signed-off-by: Florian Westphal <fw@strlen.de>
* nfct: add newline to reliable log messageCorubba Smith2025-03-111-1/+1
| | | | | | Fixes: 4bc3b22e426d ("NFCT: add `reliable' config option to enable reliable flow-based logging") Signed-off-by: Corubba Smith <corubba@gmx.de> Signed-off-by: Florian Westphal <fw@strlen.de>
* nfct: fix calloc argument orderCorubba Smith2025-03-091-5/+5
| | | | | | | | | | | | The first argument to calloc() is the number of elements, the second is the size of a single element. Having the arguments switched shouldn't make any difference during runtime, but GCC warns about it when using -Wcalloc-transposed-args [0]. [0] https://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#index-Wcalloc-transposed-args Signed-off-by: Corubba Smith <corubba@gmx.de> Signed-off-by: Florian Westphal <fw@strlen.de>
* build: use correct automake variable for library dependenciesJeremy Sowden2021-11-151-1/+2
| | | | | | | | A couple of library dependencies are specified in `_LDFLAGS` variables. They are supposed to be specified in `_LIBADD` variables. Move them. Signed-off-by: Jeremy Sowden <jeremy@azazel.net> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* build: delete commented-out codeJeremy Sowden2021-11-151-4/+1
| | | | | | | | There are a few of commented-out variable definitions left over from the introduction of Automake. Remove them. Signed-off-by: Jeremy Sowden <jeremy@azazel.net> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* build: skip sub-directories containing disabled pluginsJeremy Sowden2021-11-151-2/+0
| | | | | | | | | | | | | | | Currently, make enters all sub-directories containing source-code, even if they only contain optional targets which are not configured to be built. Instead, change the Makefiles so that the sub-directories are optional, rather than the targets. Group sub-directory definitions consistently at the top of the Makefiles that contain them. Trim a few leading and trailing blank lines. Signed-off-by: Jeremy Sowden <jeremy@azazel.net> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* build: add Make_global.am for common flagsJeremy Sowden2021-11-151-2/+2
| | | | | | | | | | Move `${regular_CFLAGS}` from configure.ac to Make_global.am, renaming it to `AM_CFLAGS`. Add `AM_CPPFGLAGS` to include `$(top_srcdir)/include`. Include the new file in the Makefiles that require it. Signed-off-by: Jeremy Sowden <jeremy@azazel.net> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* IPFIX: Add IPFIX output pluginAnder Juaristi2019-04-301-2/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This patch adds an IPFIX output plugin to ulogd2. It generates NetFlow/IPFIX traces and sends them to a remote server (collector) via TCP or UDP. Based on original work by Holger Eitzenberger <holger@eitzenberger.org>. How to test this ---------------- I am currently testing this with the NFCT input and Wireshark. Place the following in ulogd.conf: # this will print all flows on screen loglevel=1 # load NFCT and IPFIX plugins plugin="/lib/ulogd/ulogd_inpflow_NFCT.so" plugin="/lib/ulogd/ulogd_output_IPFIX.so" stack=ct1:NFCT,ipfix1:IPFIX [ct1] netlink_socket_buffer_size=217088 netlink_socket_buffer_maxsize=1085440 accept_proto_filter=tcp,sctp [ipfix1] oid=1 host="127.0.0.1" #port=4739 #send_template="once" I am currently testing it by launching a plain NetCat listener on port 4739 (the default for IPFIX) and then running Wireshark and see that it dissects the IPFIX/NetFlow traffic correctly (obviously this relies on the Wireshark NetFlow dissector being correct). First: nc -vvvv -l 127.0.0.1 4739 Then: sudo ulogd -vc ulogd.conf Signed-off-by: Ander Juaristi <a@juaristi.eus> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* nfct: use start timestamp if providedEric Leblond2014-02-221-4/+14
| | | | | | | | When hash table is not used, the start timestamp was not used even if the kernel is sending it. This patch modifies the code to use it when available. This allows to log connection with start and end with a single message per connection and without the cost of maintaining the hash table.
* nfct: make NFCT packet counter/length 64 bitUlrich Weber2013-10-101-12/+12
| | | | | | | | | | | | | | Kernel and libnetfilter_conntrack counters are 64bit, so use 64bit too in ulogd instead of 32bit. Worked fine on little endian systems but big endian systems had zero counter... Didn't test ipfix output, but RFC allows template with either 32 or 64 counters, so should be safe. Signed-off-by: Ulrich Weber <uw@xyne.com> Signed-off-by: Eric Leblond <eric@regit.org>
* nfct: use timestamp of conntrack object.Eric Leblond2013-03-051-10/+32
| | | | | | | If conntrack object sent by connection tracking system is containing a timestamp we use it instead of a gettimeofday() based counter. Signed-off-by: Eric Leblond <eric@regit.org>
* nfct: add protocol filterEric Leblond2013-01-051-2/+58
| | | | | | | This patch adds a new configuration variable which is used to limit conntrack event to connection of these protocols: For example: accept_proto_filter=tcp,sctp
* nfct: implement src and dst filterEric Leblond2013-01-051-2/+214
| | | | | | | This patch implements two filtering options in NFCT input plugin. If 'accept_src_filter' is set to a network it will only catch the event where the source is that specific network. 'accept_dst_filter' does the same for the destination.
* nfct: call register callback on opened handlerEric Leblond2013-01-051-1/+1
| | | | | nfctp_callback_register was called on the regular handler instead of begin called on the newly opened handler dedicated to the dump.
* build: avoid per-target CFLAGSJan Engelhardt2012-11-271-2/+1
| | | | | | | | There is no harm in putting the includes in the global AM_CPPFLAGS; this can generally save number of compilations of a file when a source file is used multiple times. Signed-off-by: Jan Engelhardt <jengelh@inai.de>
* configure: add flag to disable NFCT buildEric Leblond2012-11-041-0/+2
|
* src: fix version that -V displaysPablo Neira Ayuso2012-08-031-1/+1
| | | | | | | It was wrong, use VERSION constant which uses the version information available in configure.ac. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* input: NFCT: fix compilation warning with gcc-4.7Pablo Neira Ayuso2012-08-031-2/+0
| | | | | | | | CC ulogd_inpflow_NFCT_la-ulogd_inpflow_NFCT.lo ulogd_inpflow_NFCT.c: In function 'configure_nfct': ulogd_inpflow_NFCT.c:977:28: warning: unused variable 'cpi' [-Wunused-variable] Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: fix crash in polling mode if used by two stacksEric Leblond2012-07-201-5/+5
| | | | | | | | | | | | The polling timer initialisation code was put in the configurator code. It was then created for all instances. But only first one has a valid NFCT handle. This was resulting in a crash. This patch moves the timer initialisation in the constructor which is called only once on the main NFCT instance. Signed-off-by: Eric Leblond <eric@regit.org> Reported-by: Gomathivinayagam Muthuvinayagam <sankarmail@gmail.com> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* build: use pkglibdir instead of pkglibexecdir for automakeBjörn Lässig2012-05-181-1/+1
| | | | | | | | | | | This fixes the following problem while running `autoreconf -fi` `pkglibexecdir' is not a legitimate directory for `LTLIBRARIES' variable `ulogd_filter_PRINTPKT_la_SOURCES' is defined but no program or library has `ulogd_filter_PRINTPKT_la' as canonical name (possible typo) Signed-off-by: Björn Lässig <laessig@bitformer.net> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: add `reliable' config option to enable reliable flow-based loggingPablo Neira Ayuso2012-03-051-1/+18
| | | | | | | | | | | | Reliability comes at the cost of dropping new flows if the destroy event that ctnetlink delivers to us is lost. Under heavy stress this may imply dropping packets, you've been warned. If you do want not to lose one single flow-logging information, enable this. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: use nfct_copy() instead of deprecated nfct_sizeof() functionPablo Neira Ayuso2011-02-251-1/+1
| | | | | | | | | | nfct_sizeof() allows memcpy() of ct objects, which violates the library design (that aims to hide binary layout). We use nfct_copy() with the override flag instead as added by libnetfilter_conntrack 0.9.1. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* build: use appropriate location for program modulesJan Engelhardt2011-02-011-1/+1
| | | | | | | Modules - since they are dependent on the executable - generally go to libexec/. Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
* NFCT: fix bug in polling modePablo Neira Ayuso2010-12-091-0/+6
| | | | | | | | | This closes the following bug: http://bugzilla.netfilter.org/show_bug.cgi?id=684 This problem was introduced with the XML output plugin. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* build: propagate global CFLAGSJan Engelhardt2010-11-051-1/+1
| | | | | | | | | We must not override CFLAGS, because that will break when the user overrides CFLAGS again at make time (which he is entitled to). So, name our CFLAGS regular_CFLAGS, and also include that across all Makefiles so that they are actually uesd for all the code. Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
* build: remove unused $(all_includes)Jan Engelhardt2010-11-051-1/+1
| | | | Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
* build: remove -fPIC flagJan Engelhardt2010-11-051-1/+1
| | | | | | libtool automatically adds PIC flags as needed. Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
* NFCT: copy the conntrack object to the plugin oncePablo Neira Ayuso2010-06-171-1/+8
| | | | | | | With this patch, we copy the conntrack object that we propagate to the output plugin instances *only once*. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: fix plugin re-use in different stacksPablo Neira Ayuso2010-06-171-4/+6
| | | | | | | | | | This patch fixes a problem in configurations that use the NFCT plugin as input in several stacks. The first plugin loaded contains the hashtable and other important NFCT private data. Other plugin instances of NFCT are dummies that are only used to store the output keys. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* output: add new plugin XML to output logs in XMLPablo Neira Ayuso2010-06-171-0/+18
| | | | | | | This patch adds XML that allows to log information in XML for ulogd2. It supports packet and flow-based accounting. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: fix reset counters via SIGUSR2 signalPablo Neira Ayuso2010-01-171-3/+52
| | | | | | | This patch fixes a feature that allows to force the logging of the existing entries and reset the counters. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: fix number of options (missing one)Pablo Neira Ayuso2010-01-171-1/+1
| | | | | | | This patch fixes the number of options in NFCT that is actually 8, not 7. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: split event handler if hashtable is used or notPablo Neira Ayuso2010-01-171-28/+39
| | | | | | | | | This patch splits event_handler into two functions: event_handler_hashtable and event_handler_no_hashtable. Thus, we register the appropriate handler during the initialization time. This patch is a cleanup. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: use new hashtable implementation for better performancePablo Neira Ayuso2010-01-171-40/+79
| | | | | | | | This patch replaces the existing hashtable implementation with a newer that provide better performance since it reduces the number of hash computations. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: change `pollinterval' behaviourPablo Neira Ayuso2010-01-171-8/+120
| | | | | | | | | | | | | | | This patch adds support for poll-based logging. Basically, ulogd polls from the kernel periodically to log entries. You can use the `pollinterval' option in the configuration file to set the polling period. This patch changes the current behaviour of `pollinterval' that allowed to mix both the event-driven logging with polling periodically from the kernel. I have tried to look for anyone in google (and asking Eric Leblond) using this feature but I found noone. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: cleanup constructor and destructor functionsPablo Neira Ayuso2010-01-111-41/+67
| | | | | | | | This patch cleans up the destructor and the destructor functions in the NFCT plugin. I know, this patch isn't easy to review because it includes too many changes in one. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: avoid spamming report about netlink overrunsPablo Neira Ayuso2009-07-221-1/+9
| | | | | | | This patch reduces the verbosity of the log messages that report netlink overruns. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: add configurable option to set the value of the resynchronization timerPablo Neira Ayuso2009-07-221-8/+15
| | | | | | | | | This patch adds `netlink_resync_timeout' that allows you to set the number of seconds that we wait to perform a resynchronization due to a netlink overrun. This patch changes the default timeout from 2 to 60 seconds (less agressive). Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: improve netlink overrun handlingPablo Neira Ayuso2009-07-221-12/+12
| | | | | | | | | | With this patch, we schedule one resynchronization against the kernel conntrack table that will occur in two seconds (still we need a patch to make this configurable). Before this, we scheduled a resynchronization for every overrun, that is very bad in a scenario in which overruns occurs very frequently. Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* NFCT: fix NULL dereference when hashtable is fullPablo Neira Ayuso2009-06-231-0/+9
| | | | | | | | This patch fixes a NULL dereference to the timestamp structure when hashtable_add() fails, for example, because the hashtable is full. Reported-by: Bernhard Schmidt <berni@birkenwald.de> Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
* Delete timer in destructor function.Eric Leblond2009-02-081-0/+2
| | | | This patch deletes the overrun timer in the destructor function.
* Replace INCLUDES by AM_CPPFLAGS in Makefile.am.Eric Leblond2009-01-221-1/+1
| | | | | This patch fixes autotools warning about deprecated usage of INCLUDES in Makefile.am.
* build: use -avoid-version for modulesJan Engelhardt2009-01-201-2/+2
| | | | | | | The modules are pretty much bound to ulogd, and it does not seem to make sense to specially version these. Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
* build: compile fixJan Engelhardt2009-01-181-0/+1
| | | | | | | | | ulogd_inpflow_NFCT.c: In function 'propagate_ct': ulogd_inpflow_NFCT.c:483: error: 'IPPROTO_UDPLITE' undeclared (first use in this function) (and more for IPPROTO_SCTP) Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
generated by cgit v1.2.3 (git 2.25.0) at 2026-02-23 23:46:52 +0000