If the MTU of one of an attached interface becomes too small to transmit
the local translation table then it must be resized to fit inside all
fragments (when enabled) or a single packet.
But if the MTU becomes too low to transmit even the header + the VLAN
specific part then the resizing of the local TT will never succeed. This
can for example happen when the usable space is 110 bytes and 11 VLANs are
on top of batman-adv. In this case, at least 116 byte would be needed.
There will just be an endless spam of
batman_adv: batadv0: Forced to purge local tt entries to fit new maximum fragment MTU (110)
in the log but the function will never finish. Problem here is that the
timeout will be halved all the time and will then stagnate at 0 and
therefore never be able to reduce the table even more.
There are other scenarios possible with a similar result. The number of
BATADV_TT_CLIENT_NOPURGE entries in the local TT can for example be too
high to fit inside a packet. Such a scenario can therefore happen also with
only a single VLAN + 7 non-purgable addresses - requiring at least 120
bytes.
While this should be handled proactively when:
* interface with too low MTU is added
* VLAN is added
* non-purgeable local mac is added
* MTU of an attached interface is reduced
* fragmentation setting gets disabled (which most likely requires dropping
attached interfaces)
not all of these scenarios can be prevented because batman-adv is only
consuming events without the the possibility to prevent these actions
(non-purgable MAC address added, MTU of an attached interface is reduced).
It is therefore necessary to also make sure that the code is able to handle
also the situations when there were already incompatible system
configuration are present.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The batadv_vlan proto doesn't need to do anything when it gets teared down.
But the scripts are still trying to call the teardown function of this
proto. This results in warnings like:
daemon.notice netifd: batmesh1 (18940): ./batadv_vlan.sh: eval: line 37: proto_batadv_vlan_teardown: not found
Just providing a stub function avoids this log spam.
Fixes: #1044
Reported-by: Rani Hod <rani.hod@gmail.com>
Fixes: f5205d7d24 ("batman-adv: upgrade package to latest release 2014.2.0")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The newest Linux stable kernel releases:
* v5.10.205
* v5.15.144
* v6.1.69
received a backported version of commit 1f5020acb33f ("net: vlan: introduce
skb_vlan_eth_hdr()"). batman-adv must therefore not provide this function
any longer for these kernel versions.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The automatic recalculation of the maximum allowed MTU is usually triggered
by code sections which are already rtnl lock protected by callers outside
of batman-adv. But when the fragmentation setting is changed via
batman-adv's own batadv genl family, then the rtnl lock is not yet taken.
But dev_set_mtu requires that the caller holds the rtnl lock because it
uses netdevice notifiers. And this code will then fail the check for this
lock:
RTNL: assertion failed at net/core/dev.c (1953)
Fixes: c738e4751e ("batman-adv: Merge bugfixes from 2023.2")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* Broken sync while rescheduling delayed work
* compat: Use native kstrtox.h for 5.10.185
* Do not get eth header before batadv_check_management_packet
* Trigger events for auto adjusted MTU
* Don't increase MTU when set by user
* Fix TT global entry leak when client roamed back
* Fix batadv_v_ogm_aggr_send memory leak
Signed-off-by: Sven Eckelmann <sven@narfation.org>
OpenWrt master is no longer providing any kernels older than 5.15. Just
drop the compat-hacks/patches for older kernels to make it easier to
maintain the package.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* support latest kernels (4.14 - 6.4)
* drop single unicast transfer optimization for unsnoopable IP addresses
* prepare infrastructure for multicast packets with multiple unicast destination
addresses
Signed-off-by: Sven Eckelmann <sven@narfation.org>
OpenWrt/packages removed AUTORELEASE treewide. Remove it also in the
routing feed.
This is just copied from [0] with modification to the sed cmd because it
was not working for the routing feed:
The following temporary change was made to the core:
diff --git a/rules.mk b/rules.mk
index 57d7995d4fa8..f16367de87a8 100644
--- a/rules.mk
+++ b/rules.mk
@@ -429,7 +429,7 @@ endef
abi_version_str = $(subst -,,$(subst _,,$(subst .,,$(1))))
COMMITCOUNT = $(if $(DUMP),0,$(call commitcount))
-AUTORELEASE = $(if $(DUMP),0,$(call commitcount,1))
+AUTORELEASE = $(if $(DUMP),0,$(shell sed -i "s/\$$(AUTORELEASE)/$(call commitcount,1)/" $(CURDIR)/Makefile))
all:
FORCE: ;
And this command used to fix affected packages:
for i in $(cd feeds/routing; git grep -l PKG_RELEASE:=.*AUTORELEASE | \
sed 's/\/Makefile$//';);
do
make package/$i/download
done
[0] - 0c10c224be
Signed-off-by: Nick Hainke <vincent@systemli.org>
* support latest kernels (4.9 - 6.1)
* coding style cleanups and refactoring
* bugs squashed:
- limit the minimum MTU of hard-interface to avoid
"Forced to purge local tt entries" errors
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The buildsystem doesn't know that the KernelPackage/*/config kconfig
symbols are related to the batman-adv package build. It is necessary to
explicitly define it via PKG_CONFIG_DEPENDS.
Fixes: 522ce8dfdb ("batman-adv: rename folder name to match project & package name")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* support latest kernels (4.9 - 5.18)
* bugs squashed:
- resolve "time-of-check-time-of-use" race condition when checking the
network namespace of a lower device
- fix sanity check of network devices in different namespaces with
colliding IDs
- prevent transmission errors after splitting large GRO packets into
smaller fragments
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The build system of this package is written in a way that it is safe to run
the make steps in parallel. The build time can be reduced slightly on
modern systems.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Commit 97d35a552ec5b6ddf7923dd2f9a8eb973526acea of musl introduced the
macros __LITTLE_ENDIAN and __BIG_ENDIAN in alltypes.h. These are pulled
into the compilation of batman-adv. This has the side effect that the
function is_multicast_ether_addr of etherdevice.h in Linux kernel is
compiled as the big endian version and so fails to work properly on
little endian devices.
This commits prevents pulling in header files of musl libc similar to
OpenWRT commit 9ac47ee46918c45b91f4e4d1fa76b1e26b9d57fe
Signed-off-by: Hendrik Borghorst <hendrikborghorst@gmail.com>
* support latest kernels (4.4 - 5.11)
* coding style cleanups and refactoring
* drop support for sysfs+debugfs
* allow to select routing algorithm during creation of interface
* bugs squashed:
- allocate enough reserved room on fragments for lower devices
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The lowest kernel version supported by OpenWrt at the moment is Linux 5.4.
It is therefore not required to have these dead code sections anymore.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* support latest kernels (4.4 - 5.9)
* coding style cleanups and refactoring
* introduce a configurable per interface hop penalty
* bugs squashed:
- avoid uninitialized chaddr when handling DHCP
- fix own OGMv2 check in aggregation receive handling
- fix "NOHZ: local_softirq_pending 08" warnings caused by BLA
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* support latest kernels (4.4 - 5.8)
* coding style cleanups and refactoring
* dropped support for kernels < 4.4
* re-enabled link speed detection for interfaces without auto negotiation
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The mac80211 headers might depend on the actual configuration of mac80211
to generate an ABI compatible structure definition. batman-adv must
therefore include the header with these config definitions.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
It is easier to use the global define than to manually keep track of the
changes in the various kernel module makefiles in the main OpenWrt
repository and feeds.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The compat-sources are no longer needed for kernels >= 4.4. Older kernel
versions are not used since a while in OpenWrt.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The OpenWrt routing feed was tried to be merged together with the OpenWrt
package feed. But they ended up being rejected due to formalities like the
slightly different package template. Just moving to the OpenWrt package
based one should simplify similar approaches in the future.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* support latest kernels (3.16 - 5.6)
* coding style cleanups and refactoring
* use wifi tx rates as fallback for the B.A.T.M.A.N. V throughput estimation
* disable deprecated sysfs support by default
* bugs squashed:
- fix crash during the scheduling of OGMs for removed interfaces
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* support latest kernels (3.16 - 5.5)
* coding style cleanups and refactoring
* bugs squashed:
- fix DAT candidate selection on little endian systems
Signed-off-by: Sven Eckelmann <sven@narfation.org>
This changes the package version string so it does not start
with "openwrt", but with the base version we are modifying:
So far: openwrt-2019.4-1
Now: 2019.4-openwrt-1
Since it's us modifying version 2019.4 (in this case), this order
is more convenient (and also closer to what the kernel version
string does).
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
The '-m' option to select the mesh interface or vlan interfaces was
replaced with device type specific subcommand prefixes:
* meshif <netdev>
* vlan <vdev>
* meshif <netdev> vid <vid>
* hardif <netdev>
This change should also be made in the proto script to allow batctl to drop
the support of '-m' completely in the future.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The mac80211 package version was updated to a 5.2 based one in commit
0b2c42ced21a ("mac80211: Update to version 5.2-rc7"). The batman-adv module
must therefore no longer the pre-5.2 API.
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The ap_isolation setting only supports boolean values. So setting an empty
string as ap_isolation is not supported by batctl.
Fixes: f5205d7d24 ("batman-adv: upgrade package to latest release 2014.2.0")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
B.A.T.M.A.N. V support additional settings which are hardif specific. The
batadv_hardif proto has to expose them to allow automatic configuration.
The default configuration would be:
config interface 'bat0_hardif_eth0'
option proto 'batadv_hardif'
option master 'bat0'
option ifname 'eth0'
option mtu '1536'
option 'elp_interval' 500
# string like '1mbit' is accepted instead of kbit for override
option 'throughput_override' '0'
Signed-off-by: Sven Eckelmann <sven@narfation.org>
The state of slave interfaces are handled differently depending on whether
the interface is up or not. All active interfaces (IFF_UP) will transmit
OGMs. But for B.A.T.M.A.N. IV, also non-active interfaces are scheduling
(low TTL) OGMs on active interfaces. The code which setups and schedules
the OGMs must therefore already be called when the interfaces gets added as
slave interface and the transmit function must then check whether it has to
send out the OGM or not on the specific slave interface.
But v2016.3 moved the setup code from the enable function to the activate
function. The latter is called either when the added slave was already up
when batadv_hardif_enable_interface processed the new interface or when a
NETDEV_UP event was received for this slave interfac. As result, each
NETDEV_UP would schedule a new OGM worker for the interface and thus OGMs
would be send a lot more than expected.
Fixes: 549909f89d ("batman-adv: upgrade package to latest release 2016.3")
Reported-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* support latest kernels (3.16 - 5.2)
* coding style cleanups and refactoring
* implement multicast-to-unicast support for multiple targets
* deprecate sysfs support and make it optional
* refresh DAT entry timeouts on incoming ARP Replies
* bugs squashed:
- fix multicast tt/tvlv worker locking
- drop roam tvlv handler when unregistering mesh interface
Signed-off-by: Sven Eckelmann <sven@narfation.org>
* support latest kernels (3.16 - 5.1)
* coding style cleanups and refactoring
* implement meshif, hardif and vlan config interface via netlink
* snoop DHCP ACKs to improve DAT cache efficiency
* bugs squashed:
- avoid potential reference counter underflows and use-after free errors for
several objects
- fix GCC warning when B.A.T.M.A.N. V is build in but cfg80211 is not
available
- reject too small VLAN packets before they are processed further
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Although batman-adv config is cycled in config_foreach, changes
are made in the network config. Thus, this one has to be committed.
Signed-off-by: Adrian Schmutzler <freifunk@adrianschmutzler.de>
