]> git.ipfire.org Git - thirdparty/lldpd.git/log
thirdparty/lldpd.git
6 years agobuild: only enable safe-stack if linker supports it feature/safestack 246/head
Vincent Bernat [Thu, 28 Sep 2017 11:11:32 +0000 (13:11 +0200)] 
build: only enable safe-stack if linker supports it

6 years agobuild: enable clang safe-stack feature if available
Vincent Bernat [Thu, 28 Sep 2017 08:33:02 +0000 (10:33 +0200)] 
build: enable clang safe-stack feature if available

6 years agonetlink: be more precise in how we handle link deletion of bridged interface
Vincent Bernat [Wed, 20 Sep 2017 22:25:01 +0000 (00:25 +0200)] 
netlink: be more precise in how we handle link deletion of bridged interface

If link family is AF_BRIDGE, transform deletion into update of
interface. Otherwise, skip the update.

6 years agotests: fix previous testcase to fail without previous commit
Vincent Bernat [Fri, 15 Sep 2017 21:20:28 +0000 (23:20 +0200)] 
tests: fix previous testcase to fail without previous commit

Only the bridge ioctl can trigger this bug.

6 years agonetlink: don't remove interfaces when they are released from a bridge
Vincent Bernat [Fri, 15 Sep 2017 20:38:52 +0000 (22:38 +0200)] 
netlink: don't remove interfaces when they are released from a bridge

When released from a bridge, a RTM_DELLINK is sent. We detect there is
an upper device and turn thay into a RTM_NEWLINK. If the interface was
really removed, we would get two RTM_DELLINK (one with master, one
without).

On most setups, another RTM_NEWLINK was sent due to change of promisc
status.

6 years agotravis: fix broken build due to incomplete python install
Vincent Bernat [Wed, 13 Sep 2017 20:50:57 +0000 (22:50 +0200)] 
travis: fix broken build due to incomplete python install

Unsure of the root cause. See this build log:
 https://travis-ci.org/vincentbernat/lldpd/jobs/275129851

6 years agopriv: add getpid() to seccomp filter
Vincent Bernat [Wed, 13 Sep 2017 17:07:43 +0000 (19:07 +0200)] 
priv: add getpid() to seccomp filter

6 years agosrc: #include <limits.h> when using strtonum()
Vincent Bernat [Thu, 7 Sep 2017 17:52:01 +0000 (19:52 +0200)] 
src: #include <limits.h> when using strtonum()

`strtonum()` seems to work without it, but there is a hint in issue #242
that there is something fishy around that.

6 years agoosx: sync homebrew formula
Vincent Bernat [Mon, 4 Sep 2017 05:03:39 +0000 (07:03 +0200)] 
osx: sync homebrew formula

6 years agoclient: add NEWS entry for before-last commit
Vincent Bernat [Wed, 30 Aug 2017 05:47:34 +0000 (07:47 +0200)] 
client: add NEWS entry for before-last commit

6 years agoclient/tests: remove standalone "return" at the end of functions
Vincent Bernat [Wed, 30 Aug 2017 05:46:03 +0000 (07:46 +0200)] 
client/tests: remove standalone "return" at the end of functions

6 years agoclient: add show interfaces command (#240)
Steven Webster [Wed, 30 Aug 2017 05:43:54 +0000 (01:43 -0400)] 
client: add show interfaces command (#240)

There are several lldpcli configure commands that take port name(s)
as an optional parameter.

Currently, it is not possible to verify interface specific configuration
for lldpd without access to a neighbour switch/host or packet capture.

It would be useful to have a show interfaces command, similar to the
show neighbors command that allows a user/agent to query the local
TLV information.

This commit leverages the existing port display infrastructure to
enable display of local interfaces.  There are a few differences:

- Do not display the chassis RID, as the local index is always 0
- Show TTL as a seperate TLV based on tx_hold, tx_interval config,
  as internally the port TTL atom info is 0, and the chassis TTL
  atom info has been depreciated.

6 years agoclient: fix typos in manual page (#239)
Steven Webster [Tue, 29 Aug 2017 14:12:05 +0000 (10:12 -0400)] 
client: fix typos in manual page (#239)

6 years agolldpd: add option to override default lldpcli config file locations (#238)
Thomas Tannhäuser [Sun, 27 Aug 2017 19:56:30 +0000 (21:56 +0200)] 
lldpd: add option to override default lldpcli config file locations (#238)

6 years agoosx: update Homebrew recipe
Vincent Bernat [Sun, 20 Aug 2017 18:37:24 +0000 (20:37 +0200)] 
osx: update Homebrew recipe

6 years agorelease: prepare 0.9.8 0.9.8
Vincent Bernat [Sun, 20 Aug 2017 18:25:57 +0000 (20:25 +0200)] 
release: prepare 0.9.8

6 years agoclient: fix JSON output when a special character is in string
Vincent Bernat [Tue, 15 Aug 2017 18:27:52 +0000 (20:27 +0200)] 
client: fix JSON output when a special character is in string

6 years agoclient: add "json0" to the list of available formats
Vincent Bernat [Tue, 15 Aug 2017 17:22:00 +0000 (19:22 +0200)] 
client: add "json0" to the list of available formats

This enables a more verbose JSON output whose structure is kept
the same with one or several neighbors. It's easier to parse than the
current structure.

Fix #236

6 years agotests: upgrade pyroute2 version
Vincent Bernat [Mon, 14 Aug 2017 12:37:05 +0000 (14:37 +0200)] 
tests: upgrade pyroute2 version

6 years agotests: replace depcreated pyroute2 functions
Vincent Bernat [Mon, 14 Aug 2017 11:42:50 +0000 (13:42 +0200)] 
tests: replace depcreated pyroute2 functions

6 years agotests: enable test for team device
Vincent Bernat [Mon, 14 Aug 2017 11:37:40 +0000 (13:37 +0200)] 
tests: enable test for team device

6 years agolldpcli: add "configure system hostname ." to use node name
Vincent Bernat [Sun, 13 Aug 2017 13:13:07 +0000 (15:13 +0200)] 
lldpcli: add "configure system hostname ." to use node name

This is a shortcut to "configure system hostname nodename". By default,
lldpd will use the FQDN (resolving the node name using the local
resolver). We let lldpcli interpret "." as a special request to override
this value with the node name (output of "uname -n").

Fix #154

6 years agoMerge pull request #235 from vincentbernat/feature/ethtool-no-root
Vincent Bernat [Sun, 13 Aug 2017 12:46:03 +0000 (14:46 +0200)] 
Merge pull request #235 from vincentbernat/feature/ethtool-no-root

priv: don't run ethtool as root

6 years agointerfaces: show a warning if we don't have permissions for ethtool ioctls feature/ethtool-no-root 235/head
Vincent Bernat [Sun, 13 Aug 2017 04:52:24 +0000 (06:52 +0200)] 
interfaces: show a warning if we don't have permissions for ethtool ioctls

6 years agointerfaces: fix compilation without Dot3 feature
Vincent Bernat [Sun, 13 Aug 2017 04:19:37 +0000 (06:19 +0200)] 
interfaces: fix compilation without Dot3 feature

6 years agointerfaces: don't retry GLINKSETTINGS probing if it failed
Vincent Bernat [Sat, 12 Aug 2017 21:22:23 +0000 (23:22 +0200)] 
interfaces: don't retry GLINKSETTINGS probing if it failed

6 years agopriv: don't run ethtool as root
Vincent Bernat [Sat, 12 Aug 2017 21:15:56 +0000 (23:15 +0200)] 
priv: don't run ethtool as root

Kernels older than 2.6.19 won't get any link information. Kernels 4.6,
4.7 and 4.8 won't get information from GLINKSETTINGS (higher speeds)
but they are not LTS kernels so we should be fine. This removes a
bunch of code.

6 years agointerfaces: use ethtool to get permanent MAC for bonds/teams
Vincent Bernat [Sat, 12 Aug 2017 19:30:17 +0000 (21:30 +0200)] 
interfaces: use ethtool to get permanent MAC for bonds/teams

As this doesn't work for all devices (like dummy and veth), we still
fallback to old code using /proc/net/bonding for bonds. No fallback
available for team devices.

6 years agointerfaces: enable team interfaces like a bond interface
Vincent Bernat [Sat, 12 Aug 2017 18:09:22 +0000 (20:09 +0200)] 
interfaces: enable team interfaces like a bond interface

Unfortunately, we are not able to retrieve the real MAC address as the
kernel doesn't know it. Only the teamd daemon knows it. We could ask for
it through ethtool but it would be the permenant MAC address which may
be incorrect.

Fix #234

6 years agodaemon: enable recent systemd directive by default
Vincent Bernat [Thu, 13 Jul 2017 06:51:13 +0000 (08:51 +0200)] 
daemon: enable recent systemd directive by default

This will generate a warning for some users, but that should be
sustainable.

6 years agodaemon: don't try to restrict address families
Vincent Bernat [Thu, 13 Jul 2017 06:50:15 +0000 (08:50 +0200)] 
daemon: don't try to restrict address families

lldpd needs access to the most invasive address family. It's pointless
to restrict something like that.

6 years agodaemon: systemd unit cannot use ProtectSystem=strict
Vincent Bernat [Thu, 13 Jul 2017 06:47:57 +0000 (08:47 +0200)] 
daemon: systemd unit cannot use ProtectSystem=strict

lldpd needs to create its socket in `/run`. It's put outside the chroot
on purpose but it's not on a directory on its own.

6 years agodoc: add a NEWS entry for the last change
Vincent Bernat [Tue, 11 Jul 2017 17:17:53 +0000 (19:17 +0200)] 
doc: add a NEWS entry for the last change

6 years agolldp: clear station bit if any other capability is enabled
Steven Webster [Tue, 20 Jun 2017 20:04:38 +0000 (16:04 -0400)] 
lldp: clear station bit if any other capability is enabled

The 801.1AB spec states that the station only capability should
not be set in conjunction with any other (capability) bits.

It is possible for the lldpd interface update routine to set the
station bit, while the lldpd chassis update routine may, at a later
time, set the router capability bit. In this case the station bit
is not cleared.

This commit clears the station bit if any other capability is set.

6 years agoclient: install ZSH completions in $(datadir)/zsh/site-functions
Vincent Bernat [Sat, 24 Jun 2017 13:36:04 +0000 (15:36 +0200)] 
client: install ZSH completions in $(datadir)/zsh/site-functions

Instead of Debian-specific location.

Fix #232.

6 years agointerface-linux: fix this statement may fall through [-Wimplicit-fallthrough=] with...
Alexis La Goutte [Fri, 19 May 2017 15:32:27 +0000 (17:32 +0200)] 
interface-linux: fix this statement may fall through [-Wimplicit-fallthrough=] with gcc7

6 years agointerface: fix this statement may fall through [-Wimplicit-fallthrough=] with gcc7
Alexis La Goutte [Fri, 19 May 2017 15:32:06 +0000 (17:32 +0200)] 
interface: fix this statement may fall through [-Wimplicit-fallthrough=] with gcc7

6 years agoprivsep_io: fix this statement may fall through [-Wimplicit-fallthrough=] with gcc7
Alexis La Goutte [Fri, 19 May 2017 15:26:42 +0000 (17:26 +0200)] 
privsep_io: fix this statement may fall through [-Wimplicit-fallthrough=] with gcc7

6 years agodoc: update link to Homebrew
Vincent Bernat [Mon, 8 May 2017 13:24:03 +0000 (15:24 +0200)] 
doc: update link to Homebrew

7 years agobuild: fix get-version script when no git and no dist version
Vincent Bernat [Wed, 5 Apr 2017 15:59:07 +0000 (17:59 +0200)] 
build: fix get-version script when no git and no dist version

7 years agoosx: update Homebrew recipe
Vincent Bernat [Sun, 19 Mar 2017 14:34:41 +0000 (15:34 +0100)] 
osx: update Homebrew recipe

7 years agorelease: prepare 0.9.7 0.9.7
Vincent Bernat [Sun, 19 Mar 2017 14:23:04 +0000 (15:23 +0100)] 
release: prepare 0.9.7

7 years agoci: don't install jansson
Vincent Bernat [Sun, 19 Mar 2017 14:20:13 +0000 (15:20 +0100)] 
ci: don't install jansson

7 years agodoc: document previous commit in NEWS
Vincent Bernat [Fri, 17 Mar 2017 14:27:04 +0000 (15:27 +0100)] 
doc: document previous commit in NEWS

7 years agoIncrease number of lldp neighbors to 32.
Sander Vermin [Fri, 17 Mar 2017 09:03:37 +0000 (10:03 +0100)] 
Increase number of lldp neighbors to 32.

7 years agoredhat: ensure /etc/sysconfig/lldpd is correctly populated
Vincent Bernat [Tue, 7 Mar 2017 10:19:08 +0000 (11:19 +0100)] 
redhat: ensure /etc/sysconfig/lldpd is correctly populated

Patch courtesy of Martin Hauke. See: https://build.opensuse.org/request/show/477422

7 years agodaemon: don't log when going to background
Vincent Bernat [Mon, 20 Feb 2017 14:14:37 +0000 (15:14 +0100)] 
daemon: don't log when going to background

Most daemons do not log such information.

7 years agointerfaces: vxlan interfaces should be ignored
Vincent Bernat [Mon, 20 Feb 2017 14:09:16 +0000 (15:09 +0100)] 
interfaces: vxlan interfaces should be ignored

7 years agointerfaces: include <errno.h> to use errno
Vincent Bernat [Sun, 19 Feb 2017 13:40:24 +0000 (14:40 +0100)] 
interfaces: include <errno.h> to use errno

This fix previous commit. See:
 https://travis-ci.org/vincentbernat/lldpd/jobs/203148213

7 years agointerfaces: don't log an hard error when interface is down
Vincent Bernat [Sun, 19 Feb 2017 13:03:18 +0000 (14:03 +0100)] 
interfaces: don't log an hard error when interface is down

When an interface transitions to the down state, we may be woken up. We
don't want to pollute the logs with errors in this case.

7 years agolldp: don't use chassis TTL anymore
Vincent Bernat [Tue, 14 Feb 2017 15:50:54 +0000 (16:50 +0100)] 
lldp: don't use chassis TTL anymore

Move it to global configuration instead. This is less confusing than
having it at two different places.

7 years agolldp: attach remote TTL to port instead of chassis
Vincent Bernat [Mon, 13 Feb 2017 08:15:13 +0000 (09:15 +0100)] 
lldp: attach remote TTL to port instead of chassis

When receiving a 0 TTL, we should trigger expiry of the remote port, not
the remote chassis. Therefore, TTL is moved to port, except for the
local chassis (we need it even when we don't have a port). So, for
remote ports, TTL is on the port. For local ports, it is on the
chassis. This fixes two cases:

 - a shutdown LLDPU should only expire one local port
 - when a port is refreshed, another port of the same chassis shouldn't
   be refreshed

7 years agolldpd: correctly increase discarded count
Vincent Bernat [Sat, 18 Feb 2017 19:11:47 +0000 (20:11 +0100)] 
lldpd: correctly increase discarded count

When a frame cannot be decoded but has been guessed, increase the
discarded count.

Fix #223.

7 years agotravis: don't install libevent and net-snmp on OSX fix/osx-ci
Vincent Bernat [Wed, 15 Feb 2017 07:12:18 +0000 (08:12 +0100)] 
travis: don't install libevent and net-snmp on OSX

They seem to be already present in Travis environment.

7 years agoclient: built-in JSON support
Vincent Bernat [Mon, 23 Jan 2017 16:28:33 +0000 (17:28 +0100)] 
client: built-in JSON support

Instead of relying on two different libs to serialize JSON, we provide
our own built-in serializer. This reduces the amount of build-related
code (detection of the correct library), as well as the amount of C code
with the exception of UTF-8 handling which was stolen from CCAN.

This will also make the life of some users easier.

The JSON format stays the same: quite verbose with --enable-json0
flag. And inconsistent without it.

Fix #188.
Fix #220.

7 years agoosx: update Homebrew recipe
Vincent Bernat [Sat, 21 Jan 2017 18:41:17 +0000 (19:41 +0100)] 
osx: update Homebrew recipe

7 years agodoc: keyword harmonization for NEWS file
Vincent Bernat [Sat, 21 Jan 2017 18:40:03 +0000 (19:40 +0100)] 
doc: keyword harmonization for NEWS file

7 years agodaemon: when daemonizing, close stdout/stderr
Vincent Bernat [Sat, 21 Jan 2017 12:48:09 +0000 (13:48 +0100)] 
daemon: when daemonizing, close stdout/stderr

Otherwise, when used with systemd/upstart, we write logs in two
different places.

7 years agorelease: prepare 0.9.6 0.9.6
Vincent Bernat [Sat, 21 Jan 2017 12:01:32 +0000 (13:01 +0100)] 
release: prepare 0.9.6

7 years agoredhat: add support for JSON in CentOS >= 6
Vincent Bernat [Sat, 21 Jan 2017 11:59:09 +0000 (12:59 +0100)] 
redhat: add support for JSON in CentOS >= 6

This seems supported in RHEL now, but doesn't work on OSC:
 https://rhn.redhat.com/errata/RHEA-2014-1530.html

Fix #220

7 years agonetlink: clear IFF_SLAVE when no information on upper interface
Vincent Bernat [Sat, 21 Jan 2017 11:40:51 +0000 (12:40 +0100)] 
netlink: clear IFF_SLAVE when no information on upper interface

This means that we have just been enslaved, but don't know the
master. On Linux, enslaved device are handled differently of regular
device. When we are half-enslaved (no upper interface yet), we are still
a regular interface. When we are fully enslaved, we may do nothing
because there is no local change (since flags are the same). Therefore,
clear the flag to avoid this issue.

7 years agoMerge pull request #218 from vincentbernat/feature/netlink-autoadapt
Vincent Bernat [Sat, 21 Jan 2017 11:34:36 +0000 (12:34 +0100)] 
Merge pull request #218 from vincentbernat/feature/netlink-autoadapt

Auto-adapt Netlink socket size

7 years agoMerge pull request #216 from vincentbernat/fix/epollerrr
Vincent Bernat [Sat, 21 Jan 2017 11:32:30 +0000 (12:32 +0100)] 
Merge pull request #216 from vincentbernat/fix/epollerrr

interfaces: on Linux, clear socket error queue on error

7 years agoclient: fix typo in manual page
Vincent Bernat [Fri, 20 Jan 2017 06:55:49 +0000 (07:55 +0100)] 
client: fix typo in manual page

7 years agoMerge pull request #219 from alagoutte/fix
Vincent Bernat [Mon, 16 Jan 2017 07:23:28 +0000 (08:23 +0100)] 
Merge pull request #219 from alagoutte/fix

Fix Expression '!iface' is always true

7 years agoFix Expression '!iface' is always true 219/head
Alexis La Goutte [Sun, 15 Jan 2017 21:14:27 +0000 (22:14 +0100)] 
Fix Expression '!iface' is always true

7 years agonetlink: minimize code duplication when handling buffer sizes feature/netlink-autoadapt 218/head
Vincent Bernat [Sun, 15 Jan 2017 19:44:09 +0000 (20:44 +0100)] 
netlink: minimize code duplication when handling buffer sizes

7 years agonetlink: by default, don't change netlink buffer size
Vincent Bernat [Sat, 14 Jan 2017 10:51:58 +0000 (11:51 +0100)] 
netlink: by default, don't change netlink buffer size

We can't really guess what the current maximum is. On my system, the
default max is quite low (~200k). We don't want everybody to get a
warning. Also, only warn people if we got a smaller size that we
requested.

7 years agonetlink: automatically increase netlink receive buffer on errors
Vincent Bernat [Sat, 14 Jan 2017 10:46:28 +0000 (11:46 +0100)] 
netlink: automatically increase netlink receive buffer on errors

7 years agonetlink: set size to the correct value for getsockopt()
Vincent Bernat [Sat, 14 Jan 2017 10:30:52 +0000 (11:30 +0100)] 
netlink: set size to the correct value for getsockopt()

Also, don't check if the returned size is correct. This is not done in
other parts of the code and it seems a bit useless to do it here.

7 years agonetlink: make change of buffer size optional
Vincent Bernat [Sat, 14 Jan 2017 10:12:45 +0000 (11:12 +0100)] 
netlink: make change of buffer size optional

We don't really need to change the send buffer size. Put it to 0 by
default.

7 years agonetlink: make the values for buffer size configurable through ./configure
Vincent Bernat [Sat, 14 Jan 2017 10:11:58 +0000 (11:11 +0100)] 
netlink: make the values for buffer size configurable through ./configure

7 years agodoc: fix name of new ethtool command
Vincent Bernat [Sun, 1 Jan 2017 12:43:04 +0000 (13:43 +0100)] 
doc: fix name of new ethtool command

7 years agodaemon: config netlink socket buf sizes at build time
Alexandru Ardelean [Thu, 12 Jan 2017 12:46:05 +0000 (14:46 +0200)] 
daemon: config netlink socket buf sizes at build time

I admit it would be nicer to have this configurable at run-time
or better yet, introduce a mechanism for recovering from
overrun situations via polling.

We're starting to see in certain situations that on devices
with many ports (i.e. switches with around ~50 ports) lldpd
info is not always consistent.

I'm not entirely sure that the issue is with netlink
sock buffer sizes.

But we've also recentl addressed some issues for the mstpd
daemon with netlink sock buffer sizes.
So, I'm tempted to think that it's the same issue.

We [especially I] prefer to bump sock buffer sizes where needed
and not via sysctl `net.core.rmem_default` & `net.core.wmem_default`.

Signed-off-by: Alexandru Ardelean <ardeleanalex@gmail.com>
7 years agotests: avoid race condition when linking two namespaces
Vincent Bernat [Sun, 1 Jan 2017 12:26:08 +0000 (13:26 +0100)] 
tests: avoid race condition when linking two namespaces

When creating a link between two namespaces, one lldpd will see the link
and send a LLDPDU before the other one. There is a slight delay for
lldpd to wait before sending the LLDPDU, but when the load is heavy, it
can be expected that the other side didn't configure the interface when
the LLDPDU is received. To solve this, just wait a bit before putting up
the second interface.

7 years agobuild: use @mkdir_p@ instead of @MKDIR_P@
Vincent Bernat [Sun, 1 Jan 2017 07:52:38 +0000 (08:52 +0100)] 
build: use @mkdir_p@ instead of @MKDIR_P@

Uppercase "variables" are otherwise only used for lldpd-specific stuff.

7 years agodaemon: suggests additional sandboxing with systemd
Vincent Bernat [Sun, 1 Jan 2017 07:24:38 +0000 (08:24 +0100)] 
daemon: suggests additional sandboxing with systemd

This is only suggested because they are not supported by all systemd
versions (and may therefore emit warnings).

See:
 - https://lwn.net/Articles/709764/
 - https://lwn.net/Articles/709350/

7 years agointerfaces: on Linux, clear socket error queue on error fix/epollerrr 216/head
Vincent Bernat [Fri, 30 Dec 2016 22:03:34 +0000 (23:03 +0100)] 
interfaces: on Linux, clear socket error queue on error

With AF_PACKET, it is possible to get an error in the error queue even
when we didn't ask to receive them (in contrast with AF_INET where we
have to set IP_RECVERR socket option). Currently, the only error we can
receive is SOL_PACKET/PACKET_TX_TIMESTAMP, but we just want to clear any
error in the socket queue to be able to read packets (otherwise, poll()
will always return POLLERR).

Fix #215.

7 years agointerfaces: refactor generic ethernet/bond reception for Linux
Vincent Bernat [Fri, 30 Dec 2016 21:14:41 +0000 (22:14 +0100)] 
interfaces: refactor generic ethernet/bond reception for Linux

7 years agoRevert "tests: mark tests spawning a new port as flaky"
Vincent Bernat [Mon, 26 Dec 2016 09:55:52 +0000 (10:55 +0100)] 
Revert "tests: mark tests spawning a new port as flaky"

This reverts commit 0efde559064fc3b842d06b56bafa1c457b1a8fb4. This
isn't compatible with xdist...

7 years agotests: mark tests spawning a new port as flaky
Vincent Bernat [Mon, 26 Dec 2016 09:52:57 +0000 (10:52 +0100)] 
tests: mark tests spawning a new port as flaky

They will be retried once.

7 years agotests: fix when Dot3 is not enabled
Vincent Bernat [Mon, 26 Dec 2016 09:38:29 +0000 (10:38 +0100)] 
tests: fix when Dot3 is not enabled

7 years agoclient: display current MAU type even when autoneg is disabled
Vincent Bernat [Mon, 26 Dec 2016 09:16:37 +0000 (10:16 +0100)] 
client: display current MAU type even when autoneg is disabled

7 years agolldpd: don't make lldpcli complain on missing conf files
Vincent Bernat [Mon, 26 Dec 2016 09:03:48 +0000 (10:03 +0100)] 
lldpd: don't make lldpcli complain on missing conf files

Only complain if the user has directly provided additional configuration
files. This is done through an hidden option (-C) for lldpcli.

7 years agotests: minimize file descriptor leaks
Vincent Bernat [Mon, 26 Dec 2016 08:56:06 +0000 (09:56 +0100)] 
tests: minimize file descriptor leaks

This is not perfect as we still rely on garbage collection to do its job
and it seems it is not running often enough.

7 years agolib: include MAU type map only when Dot3 support is enabled
Vincent Bernat [Mon, 26 Dec 2016 08:23:13 +0000 (09:23 +0100)] 
lib: include MAU type map only when Dot3 support is enabled

7 years agoMerge pull request #212 from vincentbernat/feature/ethtool-linksettings
Vincent Bernat [Mon, 26 Dec 2016 08:17:25 +0000 (09:17 +0100)] 
Merge pull request #212 from vincentbernat/feature/ethtool-linksettings

linux: add support for ethtool GLINKSETTINGS

7 years agotests: wait even more when spawning a new interface while lldpd is running
Vincent Bernat [Sun, 25 Dec 2016 21:21:06 +0000 (22:21 +0100)] 
tests: wait even more when spawning a new interface while lldpd is running

7 years agolinux: make ethtool stuff run as root again feature/ethtool-linksettings 212/head
Vincent Bernat [Sun, 25 Dec 2016 07:46:56 +0000 (08:46 +0100)] 
linux: make ethtool stuff run as root again

This is needed for ETHTOOL_GLINKSETTINGS.

7 years agolib: add more MAU types
Vincent Bernat [Wed, 30 Nov 2016 20:56:20 +0000 (21:56 +0100)] 
lib: add more MAU types

7 years agolinux: add support for ethtool GLINKSETTINGS
Vincent Bernat [Wed, 30 Nov 2016 20:37:36 +0000 (21:37 +0100)] 
linux: add support for ethtool GLINKSETTINGS

With a recent enough kernel, one can use GLINKSETTINGS instead of
GSET. This is needed to overcome the limitation on the number of
supported/advertising modes supported by GSET (which is a fixed-length
bitset). We use the same kind of compatibility structure than for
ethtool.

Also, update the list of supported MAU types to support higher speed
interfaces. Unfortunately, while Linux reports the exact
advertised/supported modes, the operational mode of the interface is
only reported with its speed and its kind (copper, fiber, ...). When
this is fiber, we don't know if this is MMF or SMF or the supported
distance. Therefore, we use our best guess (this was already the case,
nothing new here).

7 years agoseccomp: add more syscall (sendmmsg and recvfrom)
Vincent Bernat [Wed, 7 Dec 2016 19:55:18 +0000 (20:55 +0100)] 
seccomp: add more syscall (sendmmsg and recvfrom)

7 years agoMerge pull request #213 from chutz/add-mprotect-to-seccomp-whitelist
Vincent Bernat [Sat, 3 Dec 2016 07:25:07 +0000 (08:25 +0100)] 
Merge pull request #213 from chutz/add-mprotect-to-seccomp-whitelist

seccomp: add mprotect to seccomp whitelist

7 years agoseccomp: add mprotect to seccomp whitelist 213/head
Patrick McLean [Sat, 3 Dec 2016 00:16:42 +0000 (16:16 -0800)] 
seccomp: add mprotect to seccomp whitelist

7 years agotests: more fix for pcap test
Vincent Bernat [Fri, 25 Nov 2016 23:21:52 +0000 (00:21 +0100)] 
tests: more fix for pcap test

7 years agotests: fix previous commit by making LLDP-MED part optional
Vincent Bernat [Fri, 25 Nov 2016 22:44:06 +0000 (23:44 +0100)] 
tests: fix previous commit by making LLDP-MED part optional

7 years agotests: use a PCAP file to do some integration tests
Vincent Bernat [Fri, 25 Nov 2016 20:09:15 +0000 (21:09 +0100)] 
tests: use a PCAP file to do some integration tests

7 years agomed: fix parsing of LLDP-MED LCI when TLV size exceeds addr size
Vincent Bernat [Fri, 25 Nov 2016 19:17:58 +0000 (20:17 +0100)] 
med: fix parsing of LLDP-MED LCI when TLV size exceeds addr size

Some equipments are sending an address whose size is less than the
declared TLV size. The TLV is padded with 0. We rejected this because of
a size mismatch.

Fix #210.

7 years agolinux: comment out part of ethtool.h
Vincent Bernat [Thu, 24 Nov 2016 16:54:51 +0000 (17:54 +0100)] 
linux: comment out part of ethtool.h

It relies too much on kernel.h.

7 years agogithub: shows a tcpdump output
Vincent Bernat [Thu, 24 Nov 2016 15:48:23 +0000 (16:48 +0100)] 
github: shows a tcpdump output