Actions
Bug #325
closedBatman-adv module crashing kernel after setting throughput_override parameter
Added by Kirill Lukonin about 7 years ago. Updated about 6 years ago.
Start date:
02/17/2017
Due date:
02/20/2017
% Done:
0%
Estimated time:
12.00 h
Description
Batman module crashes when echoing to sysfs throughput_override file
For example, sysfs parameter is
/sys/class/net/mesh0/batman_adv/throughput_override
echo 0 > /sys/class/net/mesh0/batman_adv/throughput_override - nothing happens. File stays default (0.0 MBit)
echo 1 > /sys/class/net/mesh0/batman_adv/throughput_override - nothing happens. File stays default (0.0 MBit)
echo 10 > /sys/class/net/mesh0/batman_adv/throughput_override - nothing happens. File stays default (0.0 MBit)
echo 100 > /sys/class/net/mesh0/batman_adv/throughput_override - Kernel crashes.
Value can be from 0 to 99 to make it normal, but if it is more than 99 or it is a string, kernel crashes.
Updated by Sven Eckelmann about 7 years ago
- Status changed from New to Feedback
- Assignee changed from batman-adv developers to Kirill Lukonin
Just tried it and I can't reproduce it. This is what I did:
$ uname -srvmpio Linux 4.9.0-1-amd64 #1 SMP Debian 4.9.2-2 (2017-01-12) x86_64 unknown unknown GNU/Linux $ gcc --version gcc (Debian 6.3.0-5) 6.3.0 20170124 $ git describe v2016.5-17-g61bbb9 $ make CONFIG_BATMAN_ADV_BATMAN_V=y $ sudo modprobe bridge $ sudo modprobe libcrc32c $ sudo insmod build/net/batman-adv/batman-adv.ko $ sudo sh -c 'echo BATMAN_V > /sys/module/batman_adv/parameters/routing_algo' $ sudo sh -c 'echo 100 > /sys/class/net/enp0s31f6/batman_adv/throughput_override' $ sudo sh -c 'echo stringystring > /sys/class/net/enp0s31f6/batman_adv/throughput_override' $ sudo dmesg|tail [15098.806772] batman_adv: B.A.T.M.A.N. advanced 2016.5-17-g61bbb9a (compatibility version 15) loaded [15230.159871] batman_adv: bat0: Adding interface: enp0s31f6 [15230.159880] batman_adv: bat0: The MTU of interface enp0s31f6 is too small (1500) to handle the transport of batman-adv packets. Packets going over this interface will be fragmented on layer2 which could impact the performance. Setting the MTU to 1532 would solve the problem. [15230.159895] batman_adv: bat0: Interface activated: enp0s31f6 [15264.623886] batman_adv: enp0s31f6: throughput_override: Changing from: 0.0 MBit to: 0.1 MBit [15297.575130] batman_adv: enp0s31f6: Invalid throughput speed for throughput_override: stringystring
Can you please explain how I can reproduce it. The current instructions from you don't seem to be enough.
Updated by Kirill Lukonin about 7 years ago
Sven Eckelmann wrote:
Just tried it and I can't reproduce it. This is what I did:
[...]
Can you please explain how I can reproduce it. The current instructions from you don't seem to be enough.
Please try to reproduce it with LEDE kernel 4.4.42.
LEDE distro is
DISTRIB_ID='LEDE'
DISTRIB_RELEASE='SNAPSHOT'
DISTRIB_REVISION='r3157-2ef3810'
DISTRIB_CODENAME='reboot'
DISTRIB_TARGET='ar71xx/generic'
DISTRIB_ARCH='mips_24kc'
DISTRIB_DESCRIPTION='LEDE Reboot SNAPSHOT r3157-2ef3810'
Updated by Sven Eckelmann about 7 years ago
Just did it on an OM5P-ACv2 with LEDE 17.01-rc:
root@lede:/# modinfo batman-adv module: /lib/modules/4.4.42/batman-adv.ko alias: net-pf-16-proto-16-family-batadv alias: rtnl-link-batadv version: 2016.5 description: B.A.T.M.A.N. advanced author: Marek Lindner <mareklindner@neomailbox.ch>, Simon Wunderlich <sw@simonwunderlich.de> license: GPL depends: libcrc32c,crc16 root@lede:/# echo BATMAN_V > /sys/module/batman_adv/parameters/routing_algo root@lede:/# batctl if add eth0 root@lede:/# echo 100 > /sys/class/net/eth0/batman_adv/throughput_override root@lede:/# echo stringystring > /sys/class/net/eth0/batman_adv/throughput_override root@lede:/# dmesg|tail [ 95.462312] batman_adv: bat0: Interface activated: eth0 [ 95.556394] 8021q: adding VLAN 0 to HW filter on device bat0 [ 103.575532] batman_adv: bat0: Interface deactivated: eth0 [ 103.581066] batman_adv: bat0: Removing interface: eth0 [ 131.313908] batman_adv: bat0: Adding interface: eth0 [ 131.319010] batman_adv: bat0: The MTU of interface eth0 is too small (1500) to handle the transport of batman-adv packets. Packets going over this interface will be fragmented on layer2 which could impact the performance. Setting the MTU to 1532 would solve the problem. [ 131.343265] batman_adv: bat0: Interface activated: eth0 [ 131.449751] 8021q: adding VLAN 0 to HW filter on device bat0 [ 159.156106] batman_adv: eth0: throughput_override: Changing from: 0.0 MBit to: 0.1 MBit [ 181.683486] batman_adv: eth0: Invalid throughput speed for throughput_override: stringystring root@lede:/# iw phy phy1 interface add mesh0 type ibss root@lede:/# ip link set up dev mesh0 root@lede:/# iw dev mesh0 ibss join foobarland 2412 root@lede:/# batctl if add mesh0 root@lede:/# echo 100 > /sys/class/net/mesh0/batman_adv/throughput_override root@lede:/# echo stringystring > /sys/class/net/mesh0/batman_adv/throughput_override root@lede:/# dmesg|tail [ 486.051379] mesh0: Trigger new scan to find an IBSS to join [ 486.748138] mesh0: Creating new IBSS network, BSSID e2:10:dc:81:2f:ff [ 486.778907] IPv6: ADDRCONF(NETDEV_CHANGE): mesh0: link becomes ready [ 509.147694] batman_adv: bat0: Adding interface: mesh0 [ 509.152838] batman_adv: bat0: The MTU of interface mesh0 is too small (1500) to handle the transport of batman-adv packets. Packets going over this interface will be fragmented on layer2 which could impact the performance. Setting the MTU to 1532 would solve the problem. [ 509.177219] batman_adv: bat0: Interface activated: mesh0 [ 516.785439] mesh0: No active IBSS STAs - trying to scan for other IBSS networks with same SSID (merge) [ 523.951626] batman_adv: mesh0: throughput_override: Changing from: 0.0 MBit to: 0.1 MBit [ 543.481081] batman_adv: mesh0: Invalid throughput speed for throughput_override: stringystring [ 547.505999] mesh0: No active IBSS STAs - trying to scan for other IBSS networks with same SSID (merge)
So no crash happened for me
Updated by Sven Eckelmann about 7 years ago
Repeated the same steps again with current LEDE master
$ ./scripts/diffconfig.sh CONFIG_TARGET_ar71xx=y CONFIG_TARGET_ar71xx_generic=y CONFIG_TARGET_ar71xx_generic_DEVICE_OM5PAC=y # CONFIG_FEED_luci is not set # CONFIG_FEED_packages is not set # CONFIG_FEED_routing is not set # CONFIG_FEED_telephony is not set CONFIG_KMOD_BATMAN_ADV_BATMAN_V=y CONFIG_KMOD_BATMAN_ADV_BLA=y CONFIG_KMOD_BATMAN_ADV_DAT=y CONFIG_KMOD_BATMAN_ADV_DEBUGFS=y CONFIG_KMOD_BATMAN_ADV_MCAST=y CONFIG_PACKAGE_ALFRED_VIS=y CONFIG_PACKAGE_alfred=y CONFIG_PACKAGE_batctl=y CONFIG_PACKAGE_kmod-batman-adv=y CONFIG_PACKAGE_kmod-crypto-crc32c=y CONFIG_PACKAGE_kmod-crypto-hash=y CONFIG_PACKAGE_kmod-lib-crc16=y CONFIG_PACKAGE_kmod-lib-crc32c=y CONFIG_PACKAGE_kmod-nls-base=y CONFIG_PACKAGE_kmod-usb-core=y CONFIG_PACKAGE_kmod-usb2=y CONFIG_PACKAGE_librt= $ git describe reboot-3519-g27040dbf89 $ git -C feeds/routing rev-parse HEAD f15b8246fabbf042cf520b40876247e654396a54
root@LEDE:/# cat /etc/openwrt_release DISTRIB_ID='LEDE' DISTRIB_RELEASE='SNAPSHOT' DISTRIB_REVISION='r3519-27040dbf89' DISTRIB_CODENAME='reboot' DISTRIB_TARGET='ar71xx/generic' DISTRIB_ARCH='mips_24kc' DISTRIB_DESCRIPTION='LEDE Reboot SNAPSHOT r3519-27040dbf89' DISTRIB_TAINTS='no-all' root@LEDE:/# modinfo batman-adv module: /lib/modules/4.4.49/batman-adv.ko alias: net-pf-16-proto-16-family-batadv alias: rtnl-link-batadv version: 2016.5 description: B.A.T.M.A.N. advanced author: Marek Lindner <mareklindner@neomailbox.ch>, Simon Wunderlich <sw@simonwunderlich.de> license: GPL depends: libcrc32c,crc16 root@lede:/# echo BATMAN_V > /sys/module/batman_adv/parameters/routing_algo root@lede:/# batctl if add eth0 root@lede:/# echo 100 > /sys/class/net/eth0/batman_adv/throughput_override root@lede:/# echo stringystring > /sys/class/net/eth0/batman_adv/throughput_override root@lede:/# dmesg|tail [ 43.076680] done. [ 43.078678] jffs2: notice: (975) jffs2_build_xattr_subsystem: complete building xattr subsystem, 0 of xdatum (0 unchecked, 0 orphan) and 0 of xref (0 dead, 0 orphan) found. [ 70.056721] random: nonblocking pool is initialized [ 76.983307] device eth0 left promiscuous mode [ 76.987852] br-lan: port 1(eth0) entered disabled state [ 77.007025] batman_adv: bat0: Adding interface: eth0 [ 77.012078] batman_adv: bat0: The MTU of interface eth0 is too small (1500) to handle the transport of batman-adv packets. Packets going over this interface will be fragmented on layer2 which could impact the performance. Setting the MTU to 1532 would solve the problem. [ 77.036363] batman_adv: bat0: Interface activated: eth0 [ 81.022029] batman_adv: eth0: throughput_override: Changing from: 0.0 MBit to: 0.1 MBit [ 86.349553] batman_adv: eth0: Invalid throughput speed for throughput_override: stringystring root@lede:/# iw phy phy1 interface add mesh0 type ibss root@lede:/# ip link set up dev mesh0 root@lede:/# iw dev mesh0 ibss join foobarland 2412 root@lede:/# batctl if add mesh0 root@lede:/# echo 100 > /sys/class/net/mesh0/batman_adv/throughput_override root@lede:/# echo stringystring > /sys/class/net/mesh0/batman_adv/throughput_override root@lede:/# dmesg|tail [ 154.022117] IPv6: ADDRCONF(NETDEV_UP): mesh0: link is not ready [ 158.947145] mesh0: Trigger new scan to find an IBSS to join [ 160.743327] batman_adv: bat0: Adding interface: mesh0 [ 160.748511] batman_adv: bat0: The MTU of interface mesh0 is too small (1500) to handle the transport of batman-adv packets. Packets going over this interface will be fragmented on layer2 which could impact the performance. Setting the MTU to 1532 would solve the problem. [ 160.772863] batman_adv: bat0: Interface activated: mesh0 [ 161.711919] mesh0: Trigger new scan to find an IBSS to join [ 163.964707] mesh0: Creating new IBSS network, BSSID 6e:ee:83:1f:ff:77 [ 163.995248] IPv6: ADDRCONF(NETDEV_CHANGE): mesh0: link becomes ready [ 164.443552] batman_adv: mesh0: throughput_override: Changing from: 0.0 MBit to: 0.1 MBit [ 171.010890] batman_adv: mesh0: Invalid throughput speed for throughput_override: stringystring
The crash was unfortunately also not reproducible this way.
Updated by Sven Eckelmann about 7 years ago
Just in case somebody thinks that I don't have a second node in my test: I am using a OM2P (only connected via mesh0). And the OM5P-ACv2 is also connected via ethernet connected to a Laptop. Here is the "batctl o" output after setting bat0 up on all three devices:
root@LEDE:/# batctl o [B.A.T.M.A.N. adv 2016.5, MainIF/MAC: eth0/ac:86:74:01:b5:48 (bat0/7e:ae:57:48:85:14 BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * ac:86:74:61:6c:20 0.960s ( 0.1) ac:86:74:61:6c:22 [ mesh0]
root@LEDE:/# batctl n [B.A.T.M.A.N. adv 2016.5, MainIF/MAC: eth0/ac:86:74:61:6c:20 (bat0/02:b2:6c:a3:6d:1f BATMAN_V)] IF Neighbor last-seen 50:7b:9d:ce:26:83 0.410s ( 0.1) [ eth0] ac:86:74:01:b5:4a 0.300s ( 0.1) [ mesh0]
$ sudo batctl o [B.A.T.M.A.N. adv 2016.5-17-g61bbb9a, MainIF/MAC: enp0s31f6/50:7b:9d:ce:26:83 (bat0/12:ff:74:98:00:5c BATMAN_V)] Originator last-seen ( throughput) Nexthop [outgoingIF] * ac:86:74:61:6c:20 0.820s ( 0.1) ac:86:74:61:6c:20 [ enp0s31f6]
Updated by Kirill Lukonin about 7 years ago
Sven Eckelmann wrote:
Just in case somebody thinks that I don't have a second node in my test: I am using a OM2P (only connected via mesh0). And the OM5P-ACv2 is also connected via ethernet connected to a Laptop. Here is the "batctl o" output after setting bat0 up on all three devices:
[...]
[...]
[...]
Thank you very much.
I don't know really what is it.
I'll try to debug through netconsole, but for now looks like it's not a bug of batman.
Updated by Sven Eckelmann about 6 years ago
- Status changed from Feedback to Rejected
Closing it because reporter is MIA and we cannot reproduce it.
Actions