summaryrefslogtreecommitdiff
path: root/server/RouteController.cpp
Commit message (Collapse)AuthorAgeFilesLines
* Add IPv4 link-local multicast range to local routing tableschiachangwang2022-09-211-0/+18
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This commit allows local multicast traffic to be sent locally instead of being sent through VPN when using a VPN automatic bypass for local traffic. Currently, the local network that is considered in VPN local exclusion mode is the same subnet of the network assigned address. If apps try to make some traffic to multicast range, it may be routed to VPN and block the traffic. E.g. If app connect a UDP socket to multicast range(224.0.0.x) and sends from the socket, or app send to 224.0.0.x from an unconnected socket. The traffic will send from VPN network. This traffic may not be well-routed in VPN network. So the case should be also considered to make the function work in the VPN bypass mode because it usually won't be the network assigned subnet range. Add the multicast range as a fixed range in the local exclusion table. The multicast range is 224.0.0.0/4 but only limit it to 224.0.0.0/24 since the IPv4 definitions are not as precise as for IPv6, it is the only range that the standards (RFC 2365 and RFC 5771) specify is link-local and must not be forwarded. Bug: 243200566 Test: cd system/netd ; atest Test: connect to Wifi or cellular network and check the routing Test: manually test with chromecast with local routes exclusion enabled Change-Id: I79fe499fb02a88ec687fadf3fad461c204fe3e47 (cherry picked from commit 5308c041c712b8cd2ecee04335c10d0aeb97d610) Merged-In: I79fe499fb02a88ec687fadf3fad461c204fe3e47
* Update methods namingchiachangwang2022-09-211-9/+9
| | | | | | | | | | | Update some methods naming to better reflect what the methods do as a clean up to address comments in previous commit. Bug: 184750836 Test: m Change-Id: I11861841b6099c82a5137c0ab045246fc15b859a (cherry picked from commit 2efd0ce2116a6f42474013b6271afb7852821592) Merged-In: I11861841b6099c82a5137c0ab045246fc15b859a
* Restrict the local network rangeChiachang2022-06-021-2/+20
| | | | | | | | | | | | | | If the network assigns a range that is not a defined local network range, it should not be considered as a local network range. Thus, intersect the network assigned range with RFC1918/ CGNAT/LINK LOCAL ranges to ensure it's an accepted local network range. Bug: 184750836 Test: cd system/netd ; atest Change-Id: I3ac6bba439986b72dbddec99c6aca3394c6d3235 (cherry picked from commit cff5e88c75a4d23d472544b7008a4e51af1381c4) Merged-In: I3ac6bba439986b72dbddec99c6aca3394c6d3235
* Update local routes based on network assigned rangeChiachang2022-06-021-10/+45
| | | | | | | | | | | | When the local routes are updated to netd, also update them to the local table for VPN local exclusion. This is specified by verifying nexthop available in the specific route. Bug: 184750836 Test: cd system/netd ; atest Change-Id: I793dd2e5dbe9fca0c0772814f5114ec98536fb4f (cherry picked from commit f9e81ac2e0629754a03d331a2a6e339b814204a0) Merged-In: I793dd2e5dbe9fca0c0772814f5114ec98536fb4f
* Add app default local rulechiachangwang2022-06-021-0/+33
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | Add an app default local rule prior to the VPN local route rule to route the per app default local traffic. If the routes setting for system default and app default are overlapped with each other, the traffic may be routed unexpectedly becuase the VPN local rules do not contain the uid range information. The rule will match first before app default rule. Thus, add an default local rule piror to the VPN local route rule to address the issue. Sample rule after applying the change: - App UID(99999) - Default(iface0), app default(iface1), vpn(tun0) 25000: ... 0x0/0x10000 iif lo uidrange 99998-99999 lookup iface1_local 26000: ... 0x0/0x10000 iif lo lookup iface0_local 27000: ... 0x0/0x30000 iif lo uidrange 99997-99998 lookup tun0 28000: ... 0xffdf/0xffff lookup iface0 29000: ... 0x0/0xffff iif lo uidrange 99998-99999 lookup iface1 30000: ... 0x0/0xffff iif lo lookup iface0 Bug: 184750836 Test: cd system/netd ; atest Change-Id: Ic092398a0d89b0104afcee8e1f22dfa93fa408ae (cherry picked from commit 0d5ae9805b1dcad074dd171dca62d5e3893d6a72) Merged-In: Ic092398a0d89b0104afcee8e1f22dfa93fa408ae
* Add local rule only for default networkChiachang2022-06-021-10/+11
| | | | | | | | | | | | | | | | The existing flow would add local rules for each physical interface. The order of the rules in the routing table is depending on the order that interfaces were added. It may cause non-deterministic routing depending on the racing of registering networks. The rule should only be needed for default network, so update the flow to update rules with fall through rules updates while switching the default network. Bug: 184750836 Test: cd system/netd ; atest Change-Id: I632f249ead6b418df40fa9639104043a66726d23 (cherry picked from commit 2271c127ad8e8e99c675b38b1414ffa092726d25) Merged-In: I632f249ead6b418df40fa9639104043a66726d23
* Remove temporary hardcoded local exclusion routeschiachangwang2022-04-191-58/+3
| | | | | | | | | | | The local routes should be configured in the right tables depending on whether it's a "local" route or not based on the assigned ip ranges. This is a leading commit to remove the hardcoded ones on first for the following change. Bug: 184750836 Test: cd system/netd ; atest Change-Id: Ieead7f15f8d62166d2c101e5432ccc8b25555f55
* Stop setting v6 routes if the v6 is disabled on the interfaceChiachang Wang2022-03-091-12/+15
| | | | | | | | | | | | | | | | | | | | The v6 local exclude routes were added in every physical interface assigned to the network. For the clat interface, the v6 support is disabled, so setting a v6 route on the clat interface will cause a permission denial. A v6 route should not be set on those interfaces that disable v6 support. The permission denial causes no failure since the error was temporarily skipped, so remove the temporary workaround accordingly. Sample error log: E netd : Error adding route fe80::/10 -> (null) v4-wlan0 to table 1000000024: Permission denied Bug: 184750836 Test: cd system/netd ; atest Test: manually connect to v6 only wifi and check the log Change-Id: Id41fdad2593d80953f3202d91524e9742b5c2d4e
* Causing no error as removing routes on removed interfaceChiachang Wang2022-03-021-2/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | In an IPv6 only network with clatd started, clatd will be stopped first before destroying the network. The clat interface will be removed from kernel while clatd stops, but the clat interface was added with the network in the netd. Destroy the network will iterate the adding interfaces and try to remove all of them. The VPN local exclusion routes are added/removed with the existence of the interfaces. It will cause an ENODEV error while trying to find the interface index of the clat interface since the clat interface was removed. The failure was considered as an error and stop destroying the network. The routing will not be cleared as expected and cause the left over rules. Because stopping clat is always before destroying the network, it is expected to get such error since the clat interface exists in ConnectivityService if the network destroys with clat started. Thus, stop considering this failure an error to finish all works necessary for clearing interfaces in the network. Sample issue rule: 16000: from all fwmark 0x10063/0x1ffff iif lo lookup local_network 16000: from all fwmark 0x10064/0x1ffff iif lo lookup wlan0 16000: from all fwmark 0x10065/0x1ffff iif lo lookup wlan0 16000: from all fwmark 0x10066/0x1ffff iif lo lookup wlan0 17000: from all iif lo oif dummy0 lookup dummy0 17000: from all iif lo oif wlan0 lookup wlan0 17000: from all iif lo oif wlan0 lookup wlan0 17000: from all iif lo oif wlan0 lookup wlan0 18000: from all fwmark 0x0/0x10000 lookup legacy_system 19000: from all fwmark 0x0/0x10000 lookup legacy_network 20000: from all fwmark 0x0/0x10000 lookup local_network 23000: from all fwmark 0x64/0x1ffff iif lo lookup wlan0 23000: from all fwmark 0x65/0x1ffff iif lo lookup wlan0 23000: from all fwmark 0x66/0x1ffff iif lo lookup wlan0 Bug: 184750836 Bug: 220997151 Test: cd system/netd ; atest Test: Manually connect/disconnect IPv6 only WiFi and observe the routing is removed as expected Change-Id: Ia981535a61dcc18aa25e7f35133c0fd822239b67
* Remove redundant "using" statementsYi Kong2022-02-181-1/+0
| | | | | | | | They are already under the same namespace. Test: presubmit Bug: 219872355 Change-Id: I05f95ca05b164b193be4f58053fe670eebb8df19
* Merge "[ELR#6] Add rules into local exclusion table"Chiachang Wang2022-01-271-9/+55
|\
| * [ELR#6] Add rules into local exclusion tableChiachang Wang2022-01-271-9/+55
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The local table will match the locally connected routes. Add those rules when the VPN network is created. The local connected rules are the link-local address for v6 and the link-local (169.254.0.0/16) for v4. These rules are hardcoded but it should depend on what actual subnet the network is which will be addressed in the following patches. Sample rule output: 24000: from all fwmark 0xc0066/0xcffff lookup ipsec1 25000: from all fwmark 0x0/0x10000 iif lo lookup wlan0_local 27000: from all fwmark 0x66/0xffff lookup wlan0 $ adb shell ip ro sh table wlan0_local adb shell ip ro sh table wlan0_local 169.254.0.0/16 dev wlan0 proto static scope link $ adb shell ip -6 ro sh table wlan0_local fd00::/10 dev wlan0 proto static metric 1024 pref medium Bug: 184750836 Test: cd system/netd ; atest Test: atest HostsideVpnTests Change-Id: Idb2188b05c2568c72c155a39d3c9f1cb6e3fa150
* | Merge "[ELR#4] Install/Remove local routes rules with the interfaces"Chiachang Wang2022-01-261-0/+31
|\|
| * [ELR#4] Install/Remove local routes rules with the interfacesChiachang Wang2022-01-261-0/+31
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Install/Remove the local routes rules with the life time of the interfaces. The default networks will break into two tables: the original one and the one for placing the local rules. The local routing table is still unfunctional since it's empty and no routes matched. Sample routings(VPN connected with default network on WiFi): 20000: from all fwmark 0x0/0x10000 lookup local_network 23000: from all fwmark 0x64/0x1ffff iif lo lookup wlan0 24000: from all fwmark 0x0/0x30000 iif lo uidrange 0-99999 lookup ipsec8 24000: from all fwmark 0xc0066/0xcffff lookup ipsec8 25000: from all fwmark 0x0/0x10000 lookup iif lo wlan0_local 27000: from all fwmark 0x66/0xffff lookup wlan0 Bug: 184750836 Test: cd system/netd ; atest Change-Id: Ia67ec04e8c133e832dd39c10977f542d9c5b9cda
* | Add special subpriority that does not set default network for uidsPatrick Rohr2022-01-261-3/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This CL changes the valid subPriority range from 0-999 to 0-998 and uses 999 as a special value that does not set the network as the default for the given uids. We have evaluated adding a boolean to the UidRangesParcel, but that would require us to keep track of it in mUidRangeMap and separating Network::appliesToUser into two functions (isUsersDefaultNetwork and doesUserHaveAccess). In addition, per uid deny rules are not supported, so there is really no benefit to the use of multiple subPriorities in explicit and implicit rules. Test: atest PerAppNetworkPermissionsTest Change-Id: I7522de13e36f2bdc3d192264d78b96423d76c607
* | rename subpriority constants to make more readablePatrick Rohr2022-01-251-6/+6
|/ | | | | | | | Rename DEFAULT_SUB_PRIORITY to SUB_PRIORITY_HIGHEST and LOWEST_SUB_PRIORITY to SUB_PRIORITY_LOWEST. Test: builds Change-Id: Ic62ad37d8bb2fafa488589e2e25f8c890c5dd649
* [ELR#3] Create <iface>_local table to place the local routesChiachang Wang2022-01-251-22/+50
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The new local exclusion rules for the VPN networks will be in the new set of rules above BYPASSABLE_VPN_LOCAL_EXCLUSION for the local exclusion VPN network. The rules for the local IP rules will need to be in new tables separated from the existing interfaces tables. Create <iface>_local table based on the existing interfaces. The new routing table will only be written in the rt_tables instead of open a new device in /dev since it's only required for supporting routing, such as legacy_network or legacy_system tables. The new <iface>_local tables share the same life time of the specific interface tables which means these new tables will also be removed when the specific interfaces tables are gone. These new tables are unfunctional now but only created in the rt_tables for the mapping between interface name and the index. Sample content in rt_table before the patch: 255 local 254 main 97 local_network 98 legacy_network 99 legacy_system 1003 dummy0 1010 rmnet_data0 Sample content in the rt_table after the patch: 255 local 254 main 97 local_network 98 legacy_network 99 legacy_system 1003 dummy0 1000000003 dummy0_local 1010 rmnet_data0 1000000010 rmnet_data0_local Bug: 184750836 Test: cd system/netd ; atest Change-Id: I13e1efa73a7145c22970880d8b72cbbd7366276c
* [ELR#2] Split RULE_PRIORITY_BYPASSABLE_VPNChiachang Wang2022-01-141-16/+29
| | | | | | | | | | | | | | | | | | Split RULE_PRIORITY_BYPASSABLE_VPN into two depending on whether the VPN excludes local routes: RULE_PRIORITY_BYPASSABLE_VPN_NO_LOCAL_EXCLUSION (24000) and RULE_PRIORITY_BYPASSABLE_VPN_LOCAL_EXCLUSION (26000). Only one of them will exist at any given time determined by whether the VPN excludes local routes. Local exclusion will not work until the new priority rules for local routes are added. The new sets of rules have priority above VPN_FALLTHROUGH, so update the rule priority of VPN_FALLTHROUGH and rules after it. Bug: 184750836 Test: cd system/netd ; atest Change-Id: I23caa68c61276c5e59dc4a85a60be054f1ccc15a
* Keep subPriority in intKen Chen2021-12-131-11/+16
| | | | | | | | | Signed integer gives us flexibility to use negative number (like -1) for special purposes in the future. Bug: N/A Test: atest Change-Id: I1e930459c9ed9d3834613473430c9570dc1a302a
* Merge changes Idd57eb85,I4d457152Taras Antoshchuk2021-10-191-31/+25
|\ | | | | | | | | | | * changes: Add "throw" and "unreachable" routes to NetdBinderTest Use route priority only for route cache invalidation
| * Use route priority only for route cache invalidationTaras Antoshchuk2021-10-111-31/+25
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | For routes cache invalidation we add and remove default throw routes with low priority. Initial implementation set low priority for all default throw routes, which didn't matter at the time, since throw routes were not commonly used. Now, throw routes are going to become more common (e.g. used in VPN routing tables). This CL makes sure we only set low priority on default throw routes if they are used for route cache invalidation. Bug: 186082280 Test: atest NetdBinderTest Change-Id: I4d457152c4177528ca9766a7909bc3ee51319a33
* | rename OffloadUtils to TcUtilsPatrick Rohr2021-10-151-1/+1
|/ | | | | Test: m Change-Id: I40082f7d56f4d3ba088ebcab9417b4d2da0d6ba2
* Support subsidiary priority on UID rangesKen Chen2021-07-011-89/+114
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | Network preference per-profile and OEM network preferences can't be set at the same time, because it is unclear what should happen if both preferences are active for one given UID. Therefore, it needs a parameter for ConnectivityService to specify which preference is prior to others. In this commit: 1. Adds a pair of methods with parcelable parameter, which currently includes netId, UID range array and subsidiary priority. 2. The subsidiary priority will be used to adjust the original IP rule priority. UID ranges can applies to different network with different subsidiary priority. But a single UID should not apply to multiple networks with the same subsidiary priority. 3. The possible value of subsidiary priority for physical and unreachable networks is 0-999. 0 is the highest priority. 0 is also the default value. Virtual network supports only the default value. 4. Netd and its tests reference to latest AIDL version (unstable). Bug: 182460808 Test: m; flash; cd system/netd/; atest Test: atest FrameworksNetTests Test: atest HostsideVpnTests Change-Id: I94e8830d0a21ffcca17757fe4783a4be9438c8b4
* PANS - Support unreachable default networkKen Chen2021-03-261-0/+73
| | | | | | | | | | | | | | | | | | | | | | | | | | | Framework provides several preferences in PANS feature. To meet those preferences, Netd needs to support two operations for framework: (1) Set OEM-paid network as default network for apps. (2) Prohibit apps to use default network if it is not explicitly selected. The #1 is supported by previous commit already. This commit implements the #2, which adds a new IP rule priority for unconnected socket, reuses existing IP rule priorities in explicit and implicit network selection. Rules are looks like: 15000: from all fwmark 0x10034/0x1ffff iif lo uidrange x-y unreachable ... 22000: from all fwmark 0x34/0x1ffff iif lo uidrange x-y unreachable ... 27000: from all fwmark 0x0/0xffff iif lo uidrange x-y unreachable An UNREACHABLE network (netId 52) is created for framework to specify that the default network is unavailable for designated apps. Bug: 181579204 Test: atest Change-Id: I21530928a85870df673e2d1387fde130fe5a0104
* Unify similar functionsKen Chen2021-03-251-32/+14
| | | | | | | | | There are only slight differences between the implementations of modifyUidExplicitNetworkRule and modifyUidImplicitNetworkRule. Unity them into one function. No functionality changes. Test: atest netd_integration_test Change-Id: I552e870b32459ada24c84fb908fea188f47c510c
* Merge "Configurable per application default network"Maciej Żenczykowski2021-02-051-7/+113
|\
| * Configurable per application default networkKen Chen2021-02-021-7/+113
| | | | | | | | | | | | | | | | | | | | | | | | | | Extend networkAddUidRanges and networkRemoveUidRanges from virtual-network-only to physical network. With this change, the ConnectivityService can replace the default physical network for specified applications without changing applications' code. Bug: 176507580 Test: cd system/netd; atest Test: atest HostsideVpnTests Test: atest FrameworksNetTests Change-Id: I556043f4401746bcf844a0c15a7d92aec12faad3
* | bpf is always supportedMaciej Żenczykowski2021-01-191-2/+0
|/ | | | | | | Test: builds, atest, TreeHugger Bug: 167500195 Signed-off-by: Maciej Żenczykowski <maze@google.com> Change-Id: Ia1a45de523bb20d451df2041a9cc3fe9930f6686
* Move IP rule priorities to header fileKen Chen2021-01-121-17/+0
| | | | | | | Make constants visible for tests, rather than have copies. Test: atest Change-Id: Ibb70ae5c719e174ace4278d617564dfd48d1455d
* Add clsact filters to local_network interfaces as well.Lorenzo Colitti2021-01-061-1/+4
| | | | | | | | | | This is required for TC-BPF offload of upstream traffic. Bug: 176857251 Test: new tests in netd_integration_test Test: enabled wifi-to-wifi tethering Test: adb shell tc qdisc show 2>&1 | grep clsact Change-Id: I091d183682170d1ddf19ab9f04aa88413769de79
* Track local network ifindices in sInterfaceToTable as well.Lorenzo Colitti2020-04-261-4/+24
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | RouteController tracks in the sInterfaceToTable map the the interface indices of every interface used in physical and virtual networks. This ensures that when an interface is removed from a network (by passing in the interface name), the rules (which specify interface indices) are correctly deleted even if the interface has been deleted or has been deleted and re-added with a new interface index. Currently this does not happen for interfaces added to the local network. That means that when those interfaces are deleted, the rules might not be deleted. It also results in spurious messages logged by NetworkController such as: 03-24 00:38:47.553 16612 16635 E Netd : getIfIndex: cannot find interface testtap1 03-24 00:38:47.553 16612 16635 E Netd : inconceivable! added interface testtap1 with no index Note that since P this map is read by RouteController::getIfIndex and used by code that assumes this will always return an interface index. In the case of an interface in the local network, this is not possible because the map stores mappings between ifindex and routing table. In the case of the local network, the routing table is always ROUTE_TABLE_LOCAL_NETWORK, so there is no way to get the interface index. Fix these callers so they do not attempt do to this. Bug: 150644681 Test: TetheringIntegrationTests:EthernetTetheringTest Change-Id: I8042e5b91bcb3175d9ad540526df396a139976f0
* Move attaching and detaching clsact from ClatdController to RouteControllerHungming Chen2020-02-171-1/+38
| | | | | | | | | | | | | | | | | | | | | | | | | The clsact attaching and detaching control plane is moved to: RouteController::addInterfaceToPhysicalNetwork - add clsact RouteController::removeInterfaceFromPhysicalNetwork - del clsact The above change implies that the clsact lifetime for each interface has been extended from clat enabled time to interface lifetime. The only exception is that attaching clsact to v4- tun interface still lives in ClatdController. The reason is that clat is started before the v4- tun interface is added to the network and clat has already needed to add the bpf filters. After all, keep attaching and detaching clat {in, e}gress bpf filters in ClatdController. Test: manual clatd test 1. Connect to IPv6-Only WiFi hotspot 2. Browse 172.217.0.46 (google.com) successfully 3. Disconnect from WiFi Repeat the above steps three times. Change-Id: I971e105484c7678ac304788e5ffff7cc709c400d
* Tweak RTA_METRICS size calculation.Lorenzo Colitti2020-01-241-1/+2
| | | | | | | | | | Use the more correct RTA_SPACE macro instead of manually calculating the size, and the more appropriate size_t type instead of int. Bug: 142892223 Test: covered by existing unit tests Change-Id: I74dd5c912e2a13721e1bd6c90df4a579e826805f
* netd: Route MTUTyler Wear2020-01-231-24/+43
| | | | | | | | | | - Route may include optional MTU parameter - Change route is added so routes don't need to be deleted then re-added - Add/Del/Change functions to pass route info as parcel Bug: 142892223 Test: new unit tests Change-Id: Idc32ecb0520b1f4136b3fe0e3f7b6800fb3005a6
* [netd] fix -Wreorder-init-listNick Desaulniers2019-10-111-5/+5
| | | | | | | | | | C++20 will require members in a designated initializer to be in order unlike C99. Bug: 139945549 Test: mm Change-Id: I4b856942f5c323898cf572dc60622d62c6ffed94 Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
* Use C++17's [[nodiscard]] instead of WARN_UNUSED_RESULTBernie Innocenti2019-08-211-58/+53
| | | | | | | No functionality change. Also remove some header guards along the way. Test: m Change-Id: I1afdcaea95a3dd56f392c4e61d7670f43615792a
* Remove unused deps on liblogwrapBernie Innocenti2019-06-051-1/+0
| | | | | | | Nothing seems to be calling into it, so it can probably go. Test: rebuild everything, then run atest Change-Id: I6e446f98decd708f59e5994fa10f77be4476d02f
* Remove explicitlySelected bit for vpn fallthrough ruleLuke Huang2019-05-231-5/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Previously, netd use explicitlySelected bit to ensure that the VPN fallthrough rule does not match if a non-zero NetId was explicitly specified. This is inconsistent with the normal case of an application that is subject to the VPN, because in that case, the fallthrough rule does match. This commit removes the explicitlySelected bit and relax the guarantee. This ensure that the behaviour of an app that selects the VPN network is the same as the behaviour of an app that specifies no network but is subject to the VPN. VPN traffic will fallthrough default network if no route is hit in VPN route table. Assume vpn netId is 102, default network interface is wlan0 with no permisiion. Below is the result of route rule before/after this commit. Before: 21000: from all fwmark 0x10066/0x1ffff lookup wlan0 After: 21000: from all fwmark 0x66/0xffff lookup wlan0 Bug: 119216095 Test: built, flashed, booted system/netd/tests/runtests.sh Change-Id: I03411644dc82cabcaf1f3274a17f36ec4e173c2e
* netd: reorder the operations in getRouteTableForInterfaceLocked.mtk137992019-04-021-10/+18
| | | | | | | | | | | when network was switched quickly between wifi and cellular, netd should always use old ifindex to delete ip rule/route, and new ifindex to add ip rule/route. BUG: 128805131 Test: built, booted, datausage, Phone call, Internet under wifi and cellular data Change-Id: I2d88709a00d50e318b02362ffac543a1e7e40a81
* IPv6: allow the same prefix routing entries for link localJimmy Chen2019-03-081-0/+7
| | | | | | | | | | | | Allowing creating multiple link-local routes in the same table, so we can make IPv6 work on all interfaces in the local_network table. Bug: 126063997 Test: * enable hotspot & usb tethering * use ping6 to verify link local in both side of hotspot/usb tethering Change-Id: Ibe897004bd72bc88744c4bbac44ca512c0482f3b
* Refactor UidRange by using stable aidl structureLuke Huang2018-10-251-14/+13
| | | | | | | | | Use stable aidl generated structure instead of UidRange Test: built, flashed, booted system/netd/tests/runtests.sh passes Change-Id: I136afcff167eac7d3c4774b9e60f607e086488bf
* Move resolver public headers to separate include pathBernie Innocenti2018-10-181-2/+2
| | | | | | | | | | Additionally rename functions and structs to eliminate those ugly underscores from the public API (with the notable exception of __res_params, which will come in its own separate change to keep this one from growing too large). Test: build, flash, then 'atest netd_integration_test' Change-Id: I7e9b0ee6cdbec889e9c35b5a17b5daa65533686e
* Let lock_guard deduce its template argumentBernie Innocenti2018-08-101-4/+4
| | | | | | | | | | | | | | | No functional change, this is a cleanup. With C++17, it's no longer necessary to specify the teplate argument when it can be deduced from the types of constructor arguments. This allows de-cluttering our locking statements. To avoid typos, this patch was mechanically generated: perl -p -i -e 's/std::lock_guard<std::mutex>/std::lock_guard/g' \ $(find . -name '*.cpp' -o -name '*.h') Change-Id: Ibb15d9a6c5b1c861d81353e47d25474eb1d4c2df
* Modernize codebase by replacing NULL with nullptrYi Kong2018-07-251-8/+8
| | | | | | | | Fixes -Wzero-as-null-pointer-constant warning. Test: m Bug: 68236239 Change-Id: I226a0599db4f7c3557e55cade7869d00bd314949
* Refine lock of RouteController and add annotationsLuke Huang2018-07-111-5/+5
| | | | | | | Test: built, flashed, booted system/netd/tests/runtests.sh passes Change-Id: I855e044a2b7c9aae795edbcee717d61f58a7a21d
* Replace RW lock with mutex, shared_mutexLuke Huang2018-07-111-6/+6
| | | | | | Test: built, flashed, booted system/netd/tests/runtests.sh passes Change-Id: I42b52d815b6ba0ba6f93dc27e83a900d2abec715
* Do not destroy socket when VPN interface address is still in useRubin Xu2018-06-071-0/+12
| | | | | | | | | | | | | | | | | | | | Normally when an IP address is removed, all sockets associated with the addresses are destroyed. This patchset changes this behavior such that if the address in question is still being used by another interface that belongs to the same underlying virtual network, the destroy operation is skipped. This change is needed to support VPN seamless handover where the VPN app will establish a second TUN interface (with different config) before tearing down the existing interface. The intention is that during this handover existing socket connections should not be disturbed. There is a companion change in the framework side to make sure during such handover, the VPN netId remains unchanged so routing still works. Bug: 64692591 Test: cts-tradefed run commandAndExit cts-dev -m CtsHostsideNetworkTests -t com.android.cts.net.HostsideVpnTests Test: system/netd/tests/runtests.sh Change-Id: I02c6b0db5f15cd1aef3e3fa6f0c36e86b4f427fd Merged-In: I02c6b0db5f15cd1aef3e3fa6f0c36e86b4f427fd (cherry picked from commit acbb6b7bbea17c5653929ee5224bd4f8e16c0f69)
* Add "iif lo" to all IP rules for originated traffic.Lorenzo Colitti2018-03-151-6/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This ensures that these rules are not mistakenly used by forwarded traffic. Forwarded traffic should only use rules that specify an explicit iif. The rules change as follows: 0: from all lookup local 10000: from all fwmark 0xc0000/0xd0000 lookup legacy_system 10500: from all {+iif lo+} oif dummy0 uidrange 0-0 lookup dummy0 10500: from all {+iif lo+} oif wlan0 uidrange 0-0 lookup wlan0 10500: from all {+iif lo+} oif v4-wlan0 uidrange 0-0 lookup v4-wlan0 10500: from all {+iif lo+} oif rmnet_data0 uidrange 0-0 lookup rmnet_data0 13000: from all fwmark 0x10063/0x1ffff {+iif lo+} lookup local_network 13000: from all fwmark 0x10065/0x1ffff {+iif lo+} lookup wlan0 13000: from all fwmark 0x10065/0x1ffff {+iif lo+} lookup v4-wlan0 13000: from all fwmark 0x50064/0x5ffff {+iif lo+} lookup rmnet_data0 14000: from all {+iif lo+} oif dummy0 lookup dummy0 14000: from all {+iif lo+} oif wlan0 lookup wlan0 14000: from all {+iif lo+} oif v4-wlan0 lookup v4-wlan0 14000: from all fwmark 0x40000/0x40000 {+iif lo+} oif rmnet_data0 lookup rmnet_data0 15000: from all fwmark 0x0/0x10000 lookup legacy_system 16000: from all fwmark 0x0/0x10000 lookup legacy_network 17000: from all fwmark 0x0/0x10000 lookup local_network 19000: from all fwmark 0x65/0x1ffff {+iif lo+} lookup wlan0 19000: from all fwmark 0x65/0x1ffff {+iif lo+} lookup v4-wlan0 22000: from all fwmark 0x0/0xffff {+iif lo+} lookup v4-wlan0 22000: from all fwmark 0x0/0xffff {+iif lo+} lookup wlan0 32000: from all unreachable Bug: 64976379 Bug: 73642792 Bug: 73032258 Test: marlin builds, boots, networking works Test: IPv4/v6 USB tethering works concurrently with httpurl --nethandle <foo> Test: USB tethering correctly able to use non-default-network upstream Test: T-Mobile wifi calling works on walleye internal build Change-Id: I9383a7ea54c60b0f33db8de8c6331e2f820539e1
* Merge changes Icc35c917,I12899e03,Iff5a202cLorenzo Colitti2018-01-101-28/+44
|\ | | | | | | | | | | | | * changes: Tighten up locking in NetworkController. Add locking to RouteController. Change RouteController from free functions/members to class functions.
| * Add locking to RouteController.Lorenzo Colitti2017-11-241-3/+14
| | | | | | | | | | Test: netd_{unit,integration}_test passes Change-Id: I12899e0304d266b25b0b021ae28f9073c8b42604
generated by cgit v1.2.3 (git 2.25.1) at 2025-12-20 04:41:33 +0000