pfSense

Use tabs consistently

(cherry picked from commit 553de3973dfdb0539a64510666976d523a21f2f9)

Re-enable executing the wifi mode command first. This fixes channel changing, which broke in d325e90818db2b22fc2562c38493769f217230f2.

(cherry picked from commit 8318da5192905a400076d5539ae86afeae82ee03)

Only configure wireless MAC address if a spoofed MAC address is set

(cherry picked from commit a6c4a66da2ee8b0d4d54480dd690700b8c16bb13)

Move copyright from ESF to Netgate

Force changes in routing to be detected by the system

When dhcp6c without RA is enabled, dhcp6c isn't killed and respawned, this causes the system not being able to pick up the routing changes. In this case, running the configuration script which fires rc.newwanipv6 solves the problem....

DHCP6 Before RA. Additions and ammendments

Replaced posix_kill() in kill_dhcp6client_process() with mwexec("kill -9 $pid"), this is because the posix_kill call was not reliably killing the dhcp6c process, kill -9 works every time.

Changes to the rtsold script creation. The script lines starting dhcp6c should not have be written to the script when dhcpwithoutra is true....

Improve dhcpd and dhcpleases reload

1) Avoid running services_dhcpd_configure() more times than needed.
2) Always restart dhcpleases after it's killed during interface recycle.
3) It's not necessary to restart dhcpdv4 when doing changes in ipv6 config.

(cherry picked from commit 509e9357df4755a4fe5d1d9b20eda65bafb855e7)

Text typo

Noted by Daryl Morse in comment on Redmine bug 6637.
(cherry picked from commit bf6ae0ad277710ffa64ee0967c5d4f717c2bc54b)

Merge pull request #3069 from phil-davis/LAGG-MTU

Remove '-x' flag from dhcpwithoutra launch of dhcp6c

This is the equivalent fix for the RELENG_2_3 branch to pull request #3078

Backport Fix issue with QinQ on LAGG interfaces where MTU doesn't apply to parent

Original pull request to master was #2905

Backport DHCP6 init before RA

Original PR to master was #3055

Review license / copyright on all files (final round)

Bring back the "set iface route default" used in 2.2.6 and earlier versions, which works around the root problem in Ticket #6495

Review license / copyright on all files (1st round)

simplify DHCP_Config_Option_Split, no loop needed for returning option array

(cherry picked from commit 264ca54e406eee7c01b01f748aabd4a29e9c4872)

Allow passing comma in string quoted DHCP client options

(cherry picked from commit 332a64d0c8a6486d40c6295351ca3068a77a216a)

Align GIF link options with those currently supported by the base OS. Ticket #6587

Remove GRE link flag options from GUI/backend. Keep link1 as it was also used to add an explicit static route, though the GUI previous GUI option name wasn't related to that at all. Updated GUI description to match reality. Ticket #6586

Properly follow style guidelines.

Include ignored DHCP server(s) config option even when advanced DHCP config is enabled

Fix Redmine #6011

This will add linklocal fallback where no IPv6 is otherwise configured to allow login using this IPs

Flush link-local IPs so we do not get stale data and do not waste time deleting IP if it is already set to fe80::1:1

Fix bad escapeshellarg logic on mpd execution

With this change single-quotes are applied in correct places

(cherry picked from commit 08cd022545be58a46b860500ff81bbe7438b6304)

Make QinQ interfaces work again

(cherry picked from commit 1322ee22354f1a6e184819fb7009a2996b63de97)

Only call interfaces_vips_configure once if it's needed, rather than doing the same thing over and over for every VIP on an interface. Ticket #6515

Fix style

Remove calls which rely in uninitizalized vars

After analyzing implementations in RELENG_2_1 and RELENG_2_2, this code seems to be a left over after code rewrite

Also remove a trailing space from comment

(cherry picked from commit 0120cad84ea4dddd0c1501ecb41f2a082ea1e7d9)

Fix script name in error log

Fix misspellings.

Unbound and dnsmasq can both be enabled so restart both if need be

Don't start unbound in track6 config if system is booting. Add dnsmasq here as well. Based on PR 2943. Ticket #6186

Merge pull request #2831 from NOYB/Includes_-_Remove_Personalizations

(cherry picked from commit 7a2c15f527459cc82aadaf23d06895174c008320)

After recreating the VLAN interface, run interface_configure on it if it's assigned and enabled. Ticket #5951

Revert "Always send a -HUP to dhcp6c in interfaces_track6_configure for DHCP6 WAN types. Doing so isn't any more disruptive than what it's already doing, and that ensures track6 interfaces don't lose their IPv6 IPs. Ticket #5945"

This reverts commit 471b81f8f7b0a5fe22019aed955e5374e3411dbf.

Always send a -HUP to dhcp6c in interfaces_track6_configure for DHCP6 WAN types. Doing so isn't any more disruptive than what it's already doing, and that ensures track6 interfaces don't lose their IPv6 IPs. Ticket #5945

Add function to return the physical interface(s) for a given parent. related to Ticket #5081

Makes interface_bring_down() remove all the CARP and IP aliases from interface.

Ticket #5913

Correct ALTQ interface test for ix interfaces, fixes #5923

Make link_interface_to_vips() also return the IP aliases on the specified vhid.

Fixes the handling of MTU with VLANs and PPPoE.

Ticket #5857

Makes get_real_interface() work with VLAN too.

This makes get_parent_interface() safe to use with VLANs.

replaced original copyright message from 2005 / Espen Johansen

Need $config here. No need to check count() as it'll always evaluate to true where is_array does. Ticket #4312

Set bridgeif for new bridge creation.

Merge pull request #2660 from phil-davis/gtetc1

Internationalize etc inc a-i files

Fix printf(printf())

This looked odd having nested calls to printf()
Should it actually log_error()?

Separate bridge application of STP and advanced parameters into their own functions so they can be applied separately from creation of the entire bridge. Apply those after adding a bridge member. Ticket #4312

Clean up logic here. It makes no sense to proceed if bridgeif isn't defined but platform_booting is true. In interfaces_bridge_configure before it gets to this point, it's setting bridgeif if it's not set, and doing so in a way that won't potentially create conflicting bridgeifs (which this removed code didn't).

Remove variables deprecated some time ago.

Fix link_interface_to_vips() to account for IP alias over CARP.

Simplify the code with the use the newly introduced function to list CARP addresses.

Add support for multiple span ports on bridge.

While here, fix and improve the error detection for the selected ports.

Inspired by pull request #2613 from heper/patch-1

Ticket #5871

Logic correction

From my understanding ppp['mtu'] and ppp['ports'] are directly related with each other and they correlate information about mtu and ports. This specific commit doesn't introduce anything new other than simplifying the code slightly

Make a couple changes to PPP default gateway handling to address issues with default gateway switching. Fixes #1837
Tested with default gateway switching on and off, by rebooting, manually disconnecting the PPP link, and manually disconnecting the NIC link, worked each time, though still possible some issues remain. Feedback appreciated.

Return loopback description for friendly interface. noticed as part of Ticket #4450, applicable across a variety of things.

Fix style issues.

Fix get_interface_ip[v6]() to try first the IP from our configuration before ask the kernel. The OS will return the first IP it find for the interface and not necessarily the main IP address.

This can cause conflicts where get_interface_ip[v6]() returns a CARP or IP alias for the interface (missing the real IP and generating a duplicated IP address).

Review of CARP uniqid changes.

It turns out that current CARP implementation is not much different from an IP alias.

This commit converts the IP alias to also use the CARP uniqid scheme, this simplify the code in all other places because now we have only two different cases to deal with:...

get_failover_interface returns the parent physical NIC for _lloc interfaces, so make sure it's returning null before it gets to that point.

Fix get_interface_ip to return correct IP for CARP VIPs. Ticket #4858

Fix find_interface_ip for gateway groups with VIPs. Ticket #4858

Merge pull request #2585 from k-paulius/patch-pkg-syslog-v2

Set gif interface MTU in interface_gif_configure if it's not already correct. Ticket #5842

Adding ability to run dhcp6c in debug mode.

Import David W's patch fixing issues with dhcp6c being launched multiple times in some circumstances. Ticket #5621

handle bridge interface IPs so they don't get added in the wrong order. Matches behavior to what "apply changes" on interfaces.php does. Ticket #3997

Simplify logic and fixes on interface_vlan_adapt_mtu()

- Remove some redundant code and define $if_mtu respecting the order
- Assinged interface
- PPP
- default
- When interface is a VLAN and parent is LAGG, it must be re-created

When parent of a VLAN is LAGG and need to set a different MTU, interface must be re-created

Use proper variable ($wantedmtu) to get vlan MTU

All lagg members are exclusive, set the same MTU defined on interface instead of consider what is currently set on members

Rename variables to make more sense

Simplify logic

Prevent GIF/GRE from making dummy/unusable gateways that show up for monitoring/routing/etc. Fixes #5766

Update copyright notice

Fix attribution. Spotted-by BBcan177

Add missing spaces

Rework a bit interface_mtu_wanted_for_pppoe(), no functional changes

Merge remote-tracking branch 'upstream/master' into rfc4638, fix conflict with c399d0d

Restore default MTU when field is cleared. Fixes #3926

Remove all pfSense_MODULE and pfSense_BUILDER_BINARIES definitions, whatever was the reason they were added, it was never finished and it's not being used

Code style and white space in etc

Merge pull request #2091 from mb300sd/patch-2

Make interface_vlan_configure() compatible with the new pfSense PHP module.

Allow the VLAN pcp setting.

While here, fix a bug when editing a VLAN. Make sure it is destroyed before it can be recreated.

Pull Request #1757. Ticket #4133

Convert CARP interface name to uniqid notation. Ticket #4858

Only add the parentheses when there is a valid description.

Pass the original interface and the correct address family to identify the real carp interface and retrieves its IP address. Ticket #4021

Fix get_interface_ip() to not return an IPv4 for a link-local request.

Fix get_interface_ipv6() to check for link-local prefix in original interface.

This fixes the generated unbound config, now it is possible to specify link-local addresses. Ticket #4021

Fix x-in-ipv6 tunnels for gre and gif

interfaces.inc - use full paths to executables

Fixed #5123

Catch one more possibility for the IPSEC interface name, this match the code used in other places.

Remove config stuff for WEP. on upgrade, disable WEP on interfaces that have it configured and disable those interfaces and generate a notice. Fixes #5123.

Coding style fixes in interface_mtu_wanted_for_pppoe()

Limit interface MTU to 1500 if no explicit MTU configured

Without this change, setting the MTU of a PPPoE interface with a
VLAN parent interface above 1492 will result in the MTU of other
VLANs on the same physical NIC that are not explicitly configured
to rise above 1500. It seems unwise - and a Principle of Least...

Add RFC4638 support (MTU > 1492) to the PPPoE client

Adjust interface MTUs for PPPoE but do not override explicitly configured MTUs

Add interface_mtu_wanted_for_pppoe()

Returns the MTU wanted if an interface or vlan is used as a PPPoE port.

get_parent_interface() - add support for vlans configured as a vlan but not an interface

One scenario when this happens is a vlan used solely as a PPPoE port.