This backports a commit from upstream dnsmasq to fix CVE-2022-0934.
CVE-2022-0934 description:
A single-byte, non-arbitrary write/use-after-free flaw was found in
dnsmasq. This flaw allows an attacker who sends a crafted packet
processed by dnsmasq, potentially causing a denial of service.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
(cherry picked from commit 002a99eccd)
Fix 'serversfile' option not being jail_mounted by the init script.
Signed-off-by: Bruno Victal <brunovictal@outlook.com>
(cherry picked from commit 0276fab649)
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
If logfacility is a path to a file it needs to be r/w mounted in the
sandbox as well for dnsmasq to work.
Reported-by: @iointerrupt
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
(cherry picked from commit 2b5fa44f60)
Problem exist when dnsmasq is exclusively bind to particular interface.
After reconfiguring or restarting this interface, its index changes, but
dnsmasq uses the old one. When this problem occurs, dnsmasq does not
listen on the correct interface so DHCP does not work, and clients do not
get an IP address. Procd netdev param can be added to restart dnsmasq when
the interface index is changed.
Signed-off-by: Valentyn Datsko <valikk.d@gmail.com>
[combined into a single &&-connected statement]
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
(cherry picked from commit 76f55e3c3f)
Fix the return value, shell return codes should be 0 to indicate success
(i.e. mount point found), 1 should be failure (i.e. mount point not-found).
Fixes: ac4e8aa ("dnsmasq: fix more dnsmasq jail issues")
Signed-off-by: Oldřich Jedlička <oldium.pro@gmail.com>
We can't use booleans, since we're not including stdbool.h. Use integers
instead.
Fixes: 0b79e7c01e ("dnsmasq: generate the dns object name dynamically")
Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
dnsmasq may call hotplug.dhcp, hotplug.neigh and hotplug.tftp.
Only the first two callees were listed in the ACL, so add missing
hotplug.tftp.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
A set of tags can be specified for --dhcp-host option to restrict the
assignment to the requests which match all the tags.
Example usage:
config vendorclass
option networkid 'udhcp'
option vendorclass 'udhcp'
config host
option mac '*:*:*:*:*:*'
list match_tag 'switch.10'
list match_tag 'udhcp'
option ip '192.168.25.10'
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Commit a2fcd3900c ("dnsmasq: improve init script") broke the existing
handling for hosts_dir. Remove the redundant mount again to fix it.
Reported-by: Hartmut Birr <e9hack@gmail.com>
Fixes: a2fcd3900c ("dnsmasq: improve init script")
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
* fix restart in LuCI (inherited umask was to restrictive)
* make directory of hosts-file (!= /tmp) accessible in ujail
Reported-by: Hannu Nyman <hannu.nyman@iki.fi>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Bring the usage in line with the dnsmasq man page and the other options
where set: is mandatory.
No functional change.
Signed-off-by: Paul Fertser <fercerpav@gmail.com>
Summary of upstream CHANGELOG:
* Handle DHCPREBIND requests in the DHCPv6 server code.
* Fix bug which caused dnsmasq to lose track of processes forked.
* Major rewrite of the DNS server and domain handling code.
* Revise resource handling for number of concurrent DNS queries.
* Improve efficiency of DNSSEC.
* Connection track mark based DNS query filtering.
* Allow smaller than 64 prefix lengths in synth-domain.
* Make domains generated by --synth-domain appear in replies
when in authoritative mode.
* Ensure CAP_NET_ADMIN capability is available when
conntrack is configured.
* When --dhcp-hostsfile --dhcp-optsfile and --addn-hosts are
given a directory as argument, define the order in which
files within that directory are read.
* Support some wildcard matching of input tags to --tag-if.
Signed-off-by: Etan Kissling <etan.kissling@gmail.com>
EXTRA_MOUNT variable should be reset in dnsmasq_start() rather than
just once at the beginning of the script.
Fixes: ac4e8aa2f8 ("dnsmasq: fix more dnsmasq jail issues")
Reported-by: Hartmut Birr <e9hack@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
* remove superflus mounts of /dev/null and /dev/urandom
* reset EXTRA_MOUNTS at the beginning of the script
* add mount according to ignore_hosts_dir
* don't add mount for file which is inside a directory already in the
EXTRA_MOUNTS list
Fixes: 59c63224e1 ("dnsmasq: rework jail mounts")
Reported-by: Hartmut Birr <e9hack@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
* split into multiple lines to improve readability
* use EXTRA_MOUNT for addnhosts instead of blindly adding /tmp/hosts
* remove no longer needed mount for /sbin/hotplug-call
* add dhcp-script.sh dependencies (jshn, ubus)
Fixes: 3a94c2ca5c ("dnsmasq: add /tmp/hosts/ to jail_mount")
Fixes: aed95c4cb8 ("dnsmasq: switch to ubus-based hotplug call")
Reported-by: Stijn Tintel <stijn@linux-ipv6.be>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Programs like the olsr-name-plugin write hostname files to "/tmp/hosts/".
If you don't add this to the jail_mount, dnsmasq can't read it anymore.
Signed-off-by: Nick Hainke <vincent@systemli.org>
'--local' is a synonym for '--server' so let's use '--local' in the
resultant config file for uci's 'local' instead of uci's local
parameter being turned into '--server'. Slightly less confusion all
round.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Currently, when using multiple dnsmasq instances they are all assigned
to the same Ubus instance name. This does not work, as only a single
instance can register with Ubus at a time. In the log, this leads to
`Cannot add object to UBus: Invalid argument` error messages.
Furthermore, upstream 3c93e8eb41952a9c91699386132d6fe83050e9be changes
behaviour so that instead of the log, dnsmasq exits at start instead.
With this patch, all dnsmasq instances are assigned unique names so that
they can register with Ubus concurrently. One of the enabled instances
is always assigned the previous default name "dnsmasq" to avoid breaking
backwards compatibility with other software relying on that default.
Previously, a random instance got assigned that name (while the others
produced error logs). Now, the first unnamed dnsmasq config section is
assigned the default name. If there are no unnamed dnsmasq sections the
first encountered named dnsmasq config section is assigned instead.
A similar issue exists for Dbus and was similarly addressed.
Signed-off-by: Etan Kissling <etan.kissling@gmail.com>
[tweaked commit message] dnsmasq was not crashing it is exiting
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
This reverts commit 3628870015.
dnsmasq v2.86test3 has some issues with ubus, so is being reverted.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
This reverts commit dea4bae7c2.
dnsmasq v2.86test3 has some issues with ubus and needs reverting, hence
this needs reverting.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Need this version to add config option for connmark DNS filtering.
Summary of upstream CHANGELOG:
* Handle DHCPREBIND requests in the DHCPv6 server code.
* Fix bug which caused dnsmasq to lose track of processes forked.
* Major rewrite of the DNS server and domain handling code.
* Revise resource handling for number of concurrent DNS queries.
* Improve efficiency of DNSSEC.
* Connection track mark based DNS query filtering.
Signed-off-by: Etan Kissling <etan.kissling@gmail.com>
Fixes issue with merged DNS requests in 2.83/2.84 not being
retried on the firsts failed request causing lookup failures.
Also fixes the following security problem in dnsmasq:
* CVE-2021-3448:
If specifiying the source address or interface to be used
when contacting upstream name servers such as:
server=8.8.8.8@1.2.3.4, server=8.8.8.8@1.2.3.4#66 and
server=8.8.8.8@eth0 then all would use the same socket
bound to the explicitly configured port. Now only
server=8.8.8.8@1.2.3.4#66 will use the explicitly
configured port and the others random source ports.
Remove upstreamed patches and update remaining patch.
Signed-off-by: Alan Swanson <reiver@improbability.net>
[refreshed old runtime support patch]
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
When running multiple instances of dnsmasq, for example one being for the lan
and another for a guest network, it might not be desirable to have the same dns names
configured in both networks
Signed-off-by: João Henriques <joaoh88@gmail.com>
dnsmasq v2.84rc2 has been promoted to release.
No functional difference between v2.83test3 and v2.84/v2.84rc2
Backport 2 patches to fix the version reporting
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Use new ubus-based hotplug call in dhcp-script.sh
As sysntpd now makes use of the new ubus-based hotplug calls, dnsmasq
no longer needs to ship ACL to cover ntpd-hotplug.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
dnsmasq v2.83 has a bug in handling duplicate queries which means it may
try to reply using the incorrect network socket. This is especially
noticeable in dual stack environments where replies may be mis-directed to
IPv4 addresses on an IPv6 socket or IPv6 addresses on an IPv4 socket.
This results in system log spam such as:
dnsmasq[16020]: failed to send packet: Network unreachable
dnsmasq[16020]: failed to send packet: Address family not supported by protocol
dnsmasq v2.84test3 resolves these issues.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
This fixes the following security problems in dnsmasq:
* CVE-2020-25681:
Dnsmasq versions before 2.83 is susceptible to a heap-based buffer
overflow in sort_rrset() when DNSSEC is used. This can allow a remote
attacker to write arbitrary data into target device's memory that can
lead to memory corruption and other unexpected behaviors on the target
device.
* CVE-2020-25682:
Dnsmasq versions before 2.83 is susceptible to buffer overflow in
extract_name() function due to missing length check, when DNSSEC is
enabled. This can allow a remote attacker to cause memory corruption
on the target device.
* CVE-2020-25683:
Dnsmasq version before 2.83 is susceptible to a heap-based buffer
overflow when DNSSEC is enabled. A remote attacker, who can create
valid DNS replies, could use this flaw to cause an overflow in a heap-
allocated memory. This flaw is caused by the lack of length checks in
rtc1035.c:extract_name(), which could be abused to make the code
execute memcpy() with a negative size in get_rdata() and cause a crash
in Dnsmasq, resulting in a Denial of Service.
* CVE-2020-25684:
A lack of proper address/port check implemented in Dnsmasq version <
2.83 reply_query function makes forging replies easier to an off-path
attacker.
* CVE-2020-25685:
A lack of query resource name (RRNAME) checks implemented in Dnsmasq's
versions before 2.83 reply_query function allows remote attackers to
spoof DNS traffic that can lead to DNS cache poisoning.
* CVE-2020-25686:
Multiple DNS query requests for the same resource name (RRNAME) by
Dnsmasq versions before 2.83 allows for remote attackers to spoof DNS
traffic, using a birthday attack (RFC 5452), that can lead to DNS
cache poisoning.
* CVE-2020-25687:
Dnsmasq versions before 2.83 is vulnerable to a heap-based buffer
overflow with large memcpy in sort_rrset() when DNSSEC is enabled. A
remote attacker, who can create valid DNS replies, could use this flaw
to cause an overflow in a heap-allocated memory. This flaw is caused
by the lack of length checks in rtc1035.c:extract_name(), which could
be abused to make the code execute memcpy() with a negative size in
sort_rrset() and cause a crash in dnsmasq, resulting in a Denial of
Service.
Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Allow configuring ipsets with dedicated config sections:
config ipset
list name 'ss_rules_dst_forward'
list name 'ss_rules6_dst_forward'
list domain 't.me'
list domain 'telegram.org'
instead of current, rather inconvenient syntax:
config dnsmasq
...
list ipset '/t.me/telegram.org/ss_rules_dst_forward,ss_rules6_dst_forward'
Current syntax will still continue to work though.
With this change, a LuCI GUI for DNS ipsets should be easy to implement.
Signed-off-by: Aleksandr Mezin <mezin.alexander@gmail.com>
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com> [PKG_RELEASE increase]
/etc/hotplug.d/ntp/25-dnsmasqsec is being sourced by /sbin/hotplug-call
running as ntpd user. For that to work the file needs to be readable by
that user.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
For IPv6 native connections when using IPv6 DNS lookups, there is no
valid default resolver if ignoring WAN DHCP provided nameservers.
This uses a runtime check to determine if IPv6 is supported on the host.
Signed-off-by: Joel Johnson <mrjoel@lixil.net>
ntpd in packages feed had already a user 'ntp' with UID 123 declared.
Rename the username of busybox-ntpd to be 'ntp' instead of 'ntpd' so
it doesn't clash.
Reported-by: Etienne Champetier <champetier.etienne@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
It should return false to indicate that the option should not be ignored
Fixes 064dc1e8 ("dnsmasq: abort when dnssec requested but not
available")
Reported-by: Sami Olmari <sami@olmari.fi>
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
The TFTP server provided by dnsmasq supports serving a select boot image
based on the client's MAC or IP address. This allows an administrator
to activate this feature in /etc/config/dhcp. Here is an example
/etc/config/dhcp that configures dnsmasq with --tftp-unique-root=mac:
...
config dnsmasq
option enable_tftp 1
option tftp_root /usr/libexec/tftpboot
option tftp_unique_root mac
config boot router
option serveraddress 192.168.1.1
option servername tftp.example.com
option filename openwrt-initramfs-kernel.bin
...
With this configuration, dnsmasq will serve
/usr/libexec/tftpboot/00-11-22-33-44-55/openwrt-initramfs-kernel.bin to
the client with MAC address 00:11:22:33:44:55.
Signed-off-by: W. Michael Petullo <mike@flyn.org>
Abort the dhcp-check based on the interface instead of the carrier
state. In cases where the interface is up but the carrier is down,
netifd won't cause a dnsmasq reload, thus dhcp won't become active
on this interface.
Signed-off-by: David Bauer <mail@david-bauer.net>
Before this commit, if uci option "dnssec" was set, we pass "--dnssec"
and friends to dnsmasq, let it start and decide whether to quit and
whether to emit message for diagnosis
# dnsmasq --dnssec; echo $?
dnsmasq: DNSSEC not available: set HAVE_DNSSEC in src/config.h
1
DNSSEC as a feature is different from others like dhcp, tftp in that
it's a security feature. Better be explicit. With this change
committed, we make it so by not allowing it in the first in the
initscript, should dnsmasq later decides to not quit (not likely) or
quit without above explicit error (unlikely but less so ;)
So this is just being proactive. on/off choices with uci option
"dnssec" are still available like before
Link: https://github.com/openwrt/openwrt/pull/3265#issuecomment-667795302
Signed-off-by: Yousong Zhou <yszhou4tech@gmail.com>
This fixes a nasty problem introduced in 2.81 which causes random
crashes on systems where there's significant DNS activity over TCP. It
also fixes DNSSEC validation problems with zero-TTL DNSKEY and DS
records.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Hotplug scripts are sourced so the #!/bin/sh is superfluous/deceptive.
Re-arrange script to only source 'procd' if we get to the stage of
needing to signal the process, reduce hotplug processing load a little.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Init script checks for an already active DHCP server on the interface
and if such DHCP server is found, then it logs "refusing to start DHCP"
message, starts dnsmasq without DHCP service unless `option force 1` is
set and caches the DHCP server check result.
Each consecutive service start then uses this cached DHCP server check
result, but doesn't provide log feedback about disabled DHCP service
anymore.
So this patch ensures, that the log message about disabled DHCP service
on particular interface is always provided.
Acked-by: Hans Dedecker <dedeckeh@gmail.com>
Signed-off-by: Petr Štetiar <ynezz@true.cz>
Bump to 2.81rc5 and re-work ipset-remove-old-kernel-support.
More runtime kernel version checking is done in 2.81rc5 in various parts
of the code, so expand the ipset patch' scope to inlude those new areas
and rename to something a bit more generic.:wq
Upstream changes from rc4
532246f Tweak to DNSSEC logging.
8caf3d7 Fix rare problem allocating frec for DNSSEC.
d162bee Allow overriding of ubus service name.
b43585c Fix nameserver list in auth mode.
3f60ecd Fixed resource leak on ubus_init failure.
0506a5e Handle old kernels that don't do NETLINK_NO_ENOBUFS.
e7ee1aa Extend stop-dns-rebind to reject IPv6 LL and ULA addresses. We also reject the loopback address if rebind-localhost-ok is NOT set.
Signed-off-by: Kevin Darbyshire-Bryant <ldir@darbyshire-bryant.me.uk>
Fix the test for an enabled sysntp initscript in dnsmasq.init, and get
rid of "test -o" while at it.
Issue reproduced on openwrt-19.07 with the help of pool.ntp.br and an
RTC-less ath79 router. dnssec-no-timecheck would be clearly missing
from /var/etc/dnsmasq.conf.* while the router was still a few days in
the past due to non-working DNSSEC + DNS-based NTP server config.
The fix was tested with the router in the "DNSSEC broken state": it
properly started dnsmasq in dnssec-no-timecheck mode, and eventually ntp
was able to resolve the server name to an IP address, and set the system
time. DNSSEC was then enabled by SIGINT through the ntp hotplug hook,
as expected.
A missing system.ntp.enabled UCI node is required for the bug to show
up. The reasons for why it would be missing in the first place were not
investigated.
Signed-off-by: Henrique de Moraes Holschuh <henrique@nic.br>
Signed-off-by: Hans Dedecker <dedeckeh@gmail.com> [PKG_RELEASE increase]
Refer to shellcheck SC2166. There are just too many caveats that are
shell-dependent on test -a and test -o to use them.
Signed-off-by: Henrique de Moraes Holschuh <henrique@nic.br>
