OSDN Git Service

android-x86/system-netd.git
9 years agoAdd "scope link" when adding routes without a nexthop.
Sreeram Ramachandran [Fri, 18 Jul 2014 20:03:47 +0000 (13:03 -0700)]
Add "scope link" when adding routes without a nexthop.

This is consistent with what /sbin/ip does, and it makes
recursive lookups on secondary tables work even when the main
table is empty.

This was originally proposed by Sreeram as http://ag/506456 to
fix a VPN issue, but then abandoned because it did not fix that
particular problem.

Bug: 16628572
Change-Id: I85753389c683ae2127b21af722463a35a33b60eb

9 years agoMerge "Prohibit address families by default unless a VPN explicitly allows them....
Sreeram Ramachandran [Mon, 28 Jul 2014 18:53:50 +0000 (18:53 +0000)]
Merge "Prohibit address families by default unless a VPN explicitly allows them." into lmp-dev

9 years agoProhibit address families by default unless a VPN explicitly allows them.
Sreeram Ramachandran [Sun, 27 Jul 2014 01:43:25 +0000 (18:43 -0700)]
Prohibit address families by default unless a VPN explicitly allows them.

Bug: 15972465
Change-Id: I3278d94536fefacc86390c1ba4231680f7be8589

9 years agoImplement support for bypassable VPNs.
Sreeram Ramachandran [Wed, 23 Jul 2014 20:27:31 +0000 (13:27 -0700)]
Implement support for bypassable VPNs.

Bypassable VPNs grab all traffic by default (just like secure VPNs), but:
+ They allow all apps to choose other networks using the multinetwork APIs.
  If these other networks are insecure ("untrusted"), they will enforce that the
  app holds the necessary permissions, such as CHANGE_NETWORK_STATE.
+ They support consistent routing. If an app has an existing connection over
  some other network when the bypassable VPN comes up, it's not interrupted.

Bug: 15347374
Change-Id: Iaee9c6f6fa8103215738570d2b65d3fcf10343f3

9 years agoMerge "Fix fwmark handling for bypassable VPNs and DNS." into lmp-dev
Sreeram Ramachandran [Mon, 28 Jul 2014 18:52:15 +0000 (18:52 +0000)]
Merge "Fix fwmark handling for bypassable VPNs and DNS." into lmp-dev

9 years agoFix fwmark handling for bypassable VPNs and DNS.
Sreeram Ramachandran [Fri, 25 Jul 2014 02:04:32 +0000 (19:04 -0700)]
Fix fwmark handling for bypassable VPNs and DNS.

This is a significant change to the way fwmarks are handled for two purposes:

1. Bypassable VPN.

   This was introduced in http://ag/510058 and had an issue that if there's a
   default network, it would always be used in connect(), so the bypassable VPN
   wouldn't get any traffic. This CL fixes that issue by using the bypassable
   VPN's NetId in connect(). See the comments in the code for more details.

2. DNS.

   The previous DNS code (specifically, getNetworkForUser()) had two problems:

   + Even if a user asks for a NetId they have permission for, we'd always use
     the user's VPN if they were subject to one. So, for example, a system IMS
     app that brings up the mobile network in the presence of a VPN would still
     have its DNS queries sent over the VPN, instead of mobile as desired.

   + Any user could perform DNS over any valid network, even one they didn't
     have permissions for, as long as they weren't subject to a VPN. So, for
     example, an app could use the DNS servers of a different profile's VPN.

   This CL fixes those problems. See getNetworkForDns() for more details.

The two pieces above are inter-related. Previously, we never set the explicit
bit from the DNS code. But we need to do that if the user asks for a network
explicitly, for two reasons:

o So that the DNS query is really restricted to that network and doesn't
  fallthrough to the default network.

o So that the heuristic described in ON_CONNECT works in all cases. I.e., if the
  DNS proxy's connect() request comes in with the explicit bit NOT set, we know
  that the NetId can only be either the default network or a VPN.

This CL is not intended to be robust against race conditions. In general, very
little of the netd code is resilient. A separate effort needs to be undertaken
to carefully audit all the code and logic to guard against things like:

* A VPN being established between calls to getNetworkForDns() and connect().
* State changes between multiple calls to NetworkController from clients such as
  FwmarkServer and DnsProxyListener.
* Routing rules / iptables rules being set up in a less-than-ideal order.
* ... etc.

Bug: 15347374
Change-Id: I5baad9168c4f4f3ef4129e07234b4bf24b0d8ba2

9 years agoImplement the fallthrough rule to support split tunnel VPNs.
Sreeram Ramachandran [Wed, 23 Jul 2014 05:23:20 +0000 (22:23 -0700)]
Implement the fallthrough rule to support split tunnel VPNs.

Change-Id: Ibc48caedb5954c6b12bfa553d978bab56c4b09aa

9 years agoImplement the rule to forward traffic to local networks via VPNs.
Sreeram Ramachandran [Wed, 23 Jul 2014 01:51:06 +0000 (18:51 -0700)]
Implement the rule to forward traffic to local networks via VPNs.

Change-Id: I4bffb2ce854a6fa7d4c0d35b97f123d91c6a84d6

9 years agoCosmetic: Move some functions around to match rule priority order.
Sreeram Ramachandran [Wed, 23 Jul 2014 01:16:44 +0000 (18:16 -0700)]
Cosmetic: Move some functions around to match rule priority order.

Change-Id: I81af639b66ac66272e0fae2d53119de1bfa12e69

9 years agoRemove the oif=iface clause when tethering.
Sreeram Ramachandran [Tue, 22 Jul 2014 23:44:37 +0000 (16:44 -0700)]
Remove the oif=iface clause when tethering.

The rule is meant to be "iif=inputIface => lookup <table_for_outputIface>", not
"iif=foo oif=bar => lookup <table_for_bar>" (the latter would never match
because packets coming in on the input interface will not have any output
interface binding).

Bug: 16242255
Change-Id: I98a2a8ab90765aee833134297b58d4f7a212036f

9 years agoCleanup: Delete dead code.
Sreeram Ramachandran [Sun, 20 Jul 2014 06:22:33 +0000 (23:22 -0700)]
Cleanup: Delete dead code.

Bug: 15413389
Change-Id: I315468832ef18ffc84174e54774ab63b86d284dc

9 years agoTry to undo iptables rules if a command fails.
Sreeram Ramachandran [Tue, 22 Jul 2014 19:40:36 +0000 (12:40 -0700)]
Try to undo iptables rules if a command fails.

As per review comments at:
https://googleplex-android-review.git.corp.google.com/#/c/504240/5/server/NatController.cpp@194

Bug: 15413694
Bug: 15413741
Change-Id: I99e2f83792fa81498447e4c03f5f8be1d2c43b7d

9 years agoFlush routing rules and add an "unreachable" rule on Init().
Sreeram Ramachandran [Sat, 19 Jul 2014 07:22:15 +0000 (00:22 -0700)]
Flush routing rules and add an "unreachable" rule on Init().

Without the flush, rules accumulate duplicates when netd is restarted due to a
runtime restart. Nothing functionally wrong with having duplicates; it just
makes the output of "ip rule" look as though something went wrong in the system.

Time to add the unreachable rule, to suss out issues with corner cases. With the
flush, there's no more a naked "from main" rule that we need to protect by
adding the unreachable rule. But it's a good idea to add the unreachable rule
anyway, in case somebody comes along and adds a rule below it later.

Change-Id: I975b2221868b7f5366bd7cf60937a82fb4b75913

9 years agoFix WiFi-Direct and Tethering.
Sreeram Ramachandran [Tue, 15 Jul 2014 23:20:28 +0000 (16:20 -0700)]
Fix WiFi-Direct and Tethering.

A LocalNetwork object now always exists in the NetworkController, with a fixed
NetId that's guaranteed not to collide with NetIds created by the framework.

When routes are added on an interface tracked by the LocalNetwork, they are
added to a fixed "local_network" table.

When NAT is enabled, we add a special "iif -> oif" tethering rule.

Bug: 15413694
Bug: 15413741

Change-Id: I36effc438d5ac193a77174493bf196cb68a5b97a

9 years agoDon't reset a VPN's NetId in the connect() shim.
Sreeram Ramachandran [Sat, 12 Jul 2014 00:06:12 +0000 (17:06 -0700)]
Don't reset a VPN's NetId in the connect() shim.

Change-Id: I0cc6c0e221a40c9100c8f4c0c5e761fce3f9b0ae

9 years agoIntroduce a SELECT_FOR_USER fwmarkd command.
Sreeram Ramachandran [Fri, 11 Jul 2014 23:27:02 +0000 (16:27 -0700)]
Introduce a SELECT_FOR_USER fwmarkd command.

This lets SYSTEM apps mark a socket with the network that would apply to an
arbitrary UID. I.e., either the VPN that applies to that user, or the default
network, if there's no such VPN.

This command will be used by system apps that proxy stuff for a user, so that
they can route the same way that user would have. Examples of such system apps
are the DnsProxyListener, MediaServer and DownloadManager.

The "explicit" bit is NOT set, so that if the user's VPN is a split tunnel, the
route lookup will fall-through to the default network as desired.

The "protect" bit is set, so that the socket bypasses any VPN applicable to the
system app itself. Note that even if the uid being proxied for is also subject
to the same VPN, this still works because the relevant rule doesn't care about
the protected bit (see modifyVpnSystemPermissionRule() in RouteController.cpp).

Change-Id: I4d501e5214b127f4ae9eaeb7befb1751cd102308

9 years agoCosmetic: Remove unnecessary constructor.
Sreeram Ramachandran [Fri, 11 Jul 2014 23:01:33 +0000 (16:01 -0700)]
Cosmetic: Remove unnecessary constructor.

Change-Id: Ib1de6e8988c4728e962da8607d0ecd98d6121e0c

9 years agoMerge "Add default routing tables to rt_tables."
Sreeram Ramachandran [Fri, 11 Jul 2014 22:29:49 +0000 (22:29 +0000)]
Merge "Add default routing tables to rt_tables."

9 years agoAdd default routing tables to rt_tables.
Sreeram Ramachandran [Fri, 11 Jul 2014 18:45:14 +0000 (11:45 -0700)]
Add default routing tables to rt_tables.

This is so that these tables also get dumped by bugreport / dumpstate.

Change-Id: Ia5ed8a23911d7b34ba964934dec42849f4a73824

9 years agoFix tethering in the case of a regular upstream connection.
Sreeram Ramachandran [Fri, 11 Jul 2014 16:10:20 +0000 (09:10 -0700)]
Fix tethering in the case of a regular upstream connection.

Fixes tethering via Ethernet, Bluetooth and WiFi (hotspot).

Tethering when the upstream has a DUN-specific APN is likely still broken
(untested).

For now, assign a fixed NetId (a hack) until we can change the framework to
create a valid NetworkAgent and all that jazz.

Bug: 15968336
Bug: 14988803
Change-Id: Idcf4d492d9329a9c87913e27be6dd835a792bea2

9 years agoCosmetic: Move constants into file scope.
Sreeram Ramachandran [Wed, 9 Jul 2014 14:39:30 +0000 (07:39 -0700)]
Cosmetic: Move constants into file scope.

Change-Id: Ic1398a5867693bb8f1e588e727869ec6a4574432

9 years agoAdd symbolic table names for routing table numbers.
Sreeram Ramachandran [Tue, 8 Jul 2014 05:11:37 +0000 (22:11 -0700)]
Add symbolic table names for routing table numbers.

This makes the output from "ip rule" much more readable.

Companion changes are in AOSP.

Change-Id: I69deb1a64d5d6647470823405bf0cc55b24b22de

9 years agoAdd full support for UIDs in VPNs.
Sreeram Ramachandran [Sun, 6 Jul 2014 00:15:14 +0000 (17:15 -0700)]
Add full support for UIDs in VPNs.

Major:
+ Implement the functions mentioned in http://go/android-multinetwork-routing
  correctly, including handling accept(), connect(), setNetworkForSocket()
  and protect() and supporting functions like canUserSelectNetwork().
+ Eliminate the old code path of getting/setting UID ranges through
  SecondaryTableController (which is currently unused) and mUidMap.

Minor:
+ Rename some methods/variables for clarity and consistency.
+ Moved some methods in .cpp files to match declaration order in the .h files.

Bug: 15409918
Change-Id: Ic6ce3646c58cf645db0d9a53cbeefdd7ffafff93

9 years agoUpdate routing rules.
Sreeram Ramachandran [Thu, 3 Jul 2014 19:20:48 +0000 (12:20 -0700)]
Update routing rules.

As per the latest changes to: http://go/android-multinetwork-routing

Functional changes:
+ Add explicit=NO to the implicit network rules, though it's a no-op.
+ Remove most of the UID=0 (kernel access) rules since they are no longer
  needed, except in one case to allow access to a VPN.
+ Add the explicit, protect and permissions bits to the incoming packet mark.
+ VPNs now don't need an implicit network rule.
+ Modifying network permissions now modifies the incoming packet mark as well.

Cosmetic changes:
+ Renamed the legacy tables to match their permissions (SYSTEM and NETWORK).
+ Renamed most functions and methods for clarity and consistency.
+ Renamed and adjusted some ule priorities.
+ Move most rule modifications into their own functions, to prevent brittle
  reliance on the previous state of the fwmark/mask variables.

Change-Id: I958a7e158ee918d5254de606fcfa55fe23327438

9 years agoFix permissions handling.
Sreeram Ramachandran [Sat, 5 Jul 2014 19:31:05 +0000 (12:31 -0700)]
Fix permissions handling.

+ Rename the permissions as per: http://go/android-multinetwork-routing
+ Make the SYSTEM permission explicitly include NETWORK.
+ Grant the SYSTEM permission to system UIDs by default, but allow the framework
  to override them if necessary.
+ Move the "string to permission" parsing to CommandListener.cpp, thus allowing
  us to get rid of Permission.cpp.
+ There's no need to support multiple permissions string arguments, so tighten
  that up.

Change-Id: I73d51b5e2f44a97e6d5ab5943ff198cebfbcc0c4

9 years agoOnly add permissions in accept(); do not try to enforce anything.
Sreeram Ramachandran [Sat, 5 Jul 2014 18:00:55 +0000 (11:00 -0700)]
Only add permissions in accept(); do not try to enforce anything.

Previously, we were enforcing that the user has access to the network over which
the socket is accepted. This has two problems:

1. We weren't handling the loopback interface ('lo') correctly. It's not part of
   any network, so the NetId in the fwmark is NETID_UNSET. In
   NetworkController::isUserPermittedOnNetwork(), we would fail to find a valid
   network, and so we would return false.

2. We have decided that in fact, we don't want to enforce this even for other
   interfaces, due to the SYN-ACK problem. See the comments and discussion at:
   http://go/android-multinetwork-routing

Bug: 16079376
Change-Id: I3f07f67d06dd7b48bfdfa5616ee22e098e31f8af

9 years agoFix regressions in flushRoutes().
Sreeram Ramachandran [Thu, 3 Jul 2014 01:06:34 +0000 (18:06 -0700)]
Fix regressions in flushRoutes().

http://ag/486276 left out a comma, causing the command to be "routeflush" (which
is invalid), instead of the correct "route flush".

http://ag/495068 changed the order, causing the interface to be removed from the
interfaceToIndex map before we are done using it in modifyPerNetworkRules().

Change-Id: I65af0b5763b13e47b48e7e2b81d243dc19e1a03b

9 years agoMinor style change.
Sreeram Ramachandran [Wed, 2 Jul 2014 21:49:33 +0000 (14:49 -0700)]
Minor style change.

Change-Id: Ib1f89bff55506cf3881047dd876adbf97c789423

9 years agoKeep track of users allowed to call protect() explicitly.
Sreeram Ramachandran [Wed, 2 Jul 2014 17:09:49 +0000 (10:09 -0700)]
Keep track of users allowed to call protect() explicitly.

This is an API change between ConnectivityService and Netd.

The ownerUid was meant for this purpose, but it's insufficient, as apps need to
call protect() _before_ they create a VPN.

Bug: 15409918
Change-Id: If804aa106002e96d5ffb623d32db35fd76928367

9 years agoCosmetic: Add a way to query the type of a network.
Sreeram Ramachandran [Wed, 2 Jul 2014 02:01:56 +0000 (19:01 -0700)]
Cosmetic: Add a way to query the type of a network.

This is a cosmetic change, i.e., there's no change in functionality.

This is the poor man's RTTI. It turns out that maintaining separate lists (or
maps) of the different types of networks gets burdensome pretty quickly
(especially in an upcoming CL where we add functions like
canUserSelectNetwork()).

Change-Id: If5250c0fc106045f681d0fd71278b793addbe1e3

9 years agoFix routing rules.
Sreeram Ramachandran [Tue, 1 Jul 2014 21:30:30 +0000 (14:30 -0700)]
Fix routing rules.

Fix the rules in RouteController as per the latest iteration of the routing
design: http://go/android-multinetwork-routing. Changes include:

+ The legacy tables have to be global. So remove the UID being passed in to
  modifyRoute() and remove the associated TODOs.

+ Add UID=0 rules to let the kernel access routes on privileged networks.

+ Add a UID=0 clause to the directly-connected hack, thus fixing the TODO.

+ Add the privileged_legacy table just above the legacy table, when overriding
  the default network. (The same table remains added at the top of the rule
  chain, to override VPNs, but only for CONNECTIVITY_INTERNAL-privileged apps).

Other cosmetic changes:
+ Update the names and values of the rule priorities.
+ Move the legacy table IDs to the .h file in anticipation of using them from
  bugreport / dump commands.
+ Make 'action' the first parameter consistently.

Change-Id: I6634a19ddc8062b2ef55d926c7892fff8c586106

9 years agoDon't use %zu for uid_t. It's always unsigned int, so %u is correct.
Sreeram Ramachandran [Tue, 1 Jul 2014 22:49:20 +0000 (15:49 -0700)]
Don't use %zu for uid_t. It's always unsigned int, so %u is correct.

Change-Id: I5be1d479b524495037c2aedc8336c794d2698914

9 years agoUse %zu to printf size_t correctly on both 32-bit and 64-bit platforms.
Sreeram Ramachandran [Tue, 1 Jul 2014 22:06:05 +0000 (15:06 -0700)]
Use %zu to printf size_t correctly on both 32-bit and 64-bit platforms.

Change-Id: I5223e574084fca47606b844d74a99a642c7d66be

9 years agoAdd UID range support to VPNs.
Sreeram Ramachandran [Tue, 24 Jun 2014 01:54:27 +0000 (18:54 -0700)]
Add UID range support to VPNs.

This adds the necessary routing rules.

Future CLs will add the ability to select the right netId for connect(),
setNetworkForSocket(), DNS resolutions, etc.

Bug: 15409918
Change-Id: I88a67660d49cecda834dd72ab947fbfed250f09d

9 years agoIntroduce VPN support.
Sreeram Ramachandran [Mon, 23 Jun 2014 19:41:37 +0000 (12:41 -0700)]
Introduce VPN support.

This change sets up the basic routing rules for VPNs. It doesn't yet handle UID
ranges (that are meant to apply to the VPN) correctly. That's forthcoming in
other CLs.

Bug: 15409918
Change-Id: I284de04f176dcf6ba702361de6a614266256d04e

9 years agoRefactor: Encapsulate permissions and interfaces into a Network class.
Sreeram Ramachandran [Mon, 23 Jun 2014 16:54:06 +0000 (09:54 -0700)]
Refactor: Encapsulate permissions and interfaces into a Network class.

Currently, there's a lot of logic in NetworkController surrounding events such
as interface addition/removal, network creation/destruction and default network
change, because these events are interwined. For example, adding an interface
means also adding a corresponding default network rule if the interface is being
added to the current default network.

When we introduce VPNs into this mix, things will get hairy real quick for all
this logic in NetworkController.

In this refactor, we introduce an abstract base class Network which supports
adding and removing interfaces. The main concrete implementation of this is
PhysicalNetwork, which allows setting permissions and "default network" state.

Since we've moved network permissions into the above class, and user permissions
into NetworkController, PermissionsController is unused and has been removed.

Also fix a few bugs in RouteController:
+ Use uidEnd correctly.
+ Check for all error cases in inet_pton.
+ Check the return value of android_fork_execvp() correctly.
+ The "return cmd1() && cmd2()" pattern is wrong. Rewrite that code.

Also (non-functional changes):
+ Remove instantiations of RouteController. It has static methods only.
+ Reorder some blocks in CommandListener so that the most frequent commands are
  checked first.
+ Remove unused paramError() and clearNetworkPreference().
+ Change all return codes to int (negative errno) wherever applicable.
+ Add WARN_UNUSED_RESULT everywhere.
+ Cleanup some style in RouteController and NetworkController.
+ Use uid_t instead of unsigned for user IDs.
+ Add clearer log messages at the source of failures.
+ Add a check for when fwmark bits are set without corresponding mask bits.

Bug: 15409918

Change-Id: Ibba78b0850160f9f3d17d476f16331a6db0025d1

9 years agoDon't fail when adding a duplicate legacy route.
Sreeram Ramachandran [Fri, 27 Jun 2014 13:42:11 +0000 (06:42 -0700)]
Don't fail when adding a duplicate legacy route.

This only affects calls to requestRouteToHost() by apps. We still fail if the
framework itself tries to add a non-legacy duplicate route, since the framework
should know better (we can consider relaxing that too in the future).

Bug: 15925532
Change-Id: I9ee434277e462d570f88e6fe63a096e5ae41eee9

9 years agoConvert rta_* and fra_* variable names to camelCase.
Lorenzo Colitti [Thu, 26 Jun 2014 04:51:10 +0000 (13:51 +0900)]
Convert rta_* and fra_* variable names to camelCase.

Change-Id: I0ca539ac4c54bb71b033f288fb4229afd71b7989

9 years agoSupport adding and removing UID rules via netlink.
Lorenzo Colitti [Tue, 24 Jun 2014 18:20:29 +0000 (03:20 +0900)]
Support adding and removing UID rules via netlink.

Change-Id: Idae13bceda869261689260759084b8d6ef1ff639

9 years agoPass rule modification errors back to CommandListener.
Lorenzo Colitti [Mon, 23 Jun 2014 06:09:54 +0000 (15:09 +0900)]
Pass rule modification errors back to CommandListener.

Change-Id: If01334dccad8b6230648713a57fd58be180ac66b

9 years agoUse netlink to add/delete rules as well as routes.
Lorenzo Colitti [Fri, 20 Jun 2014 14:03:29 +0000 (23:03 +0900)]
Use netlink to add/delete rules as well as routes.

Also change the indentation of the rtattrs used in modifyIpRoute
to make it easier to see what attributes are being used and in
what sequence.

This change does not yet pass the errors back to CommandListener;
that is done in the next change in the series.

Change-Id: Ib2e174386c63cb0647d838d9c7d731cd6df39c4f

10 years agoUnrevert the 'revert the "talk to netlink directly" change.'
Sreeram Ramachandran [Tue, 24 Jun 2014 23:09:21 +0000 (16:09 -0700)]
Unrevert the 'revert the "talk to netlink directly" change.'

http://ag/486277 was reverted in http://ag/491263 and is being unreverted in
this change. The bug in the original CL was a typo ("interface" instead of the
desired "ifindex"), which is now fixed.

Bug: 15840054
Change-Id: If66987c74cc86e9ba4f7a35d36f0a39afe939a68

10 years agoTemporarily revert the "talk to netlink directly" change.
Sreeram Ramachandran [Tue, 24 Jun 2014 22:03:52 +0000 (15:03 -0700)]
Temporarily revert the "talk to netlink directly" change.

The change being reverted is http://ag/486277. Instead, use /sbin/ip again. The
code to talk to netlink fails on volantis. I.e., instead of this:
    $ ip route show table 1006
    default via 100.110.191.254 dev wlan0
    100.110.128.0/18 dev wlan0  scope link
we end up with this:
    $ ip route show table 1006
    default dev wlan0  proto static
    100.110.128.0/18 dev wlan0  proto static
Notice the lack of the nexthop and the addition of "proto static". I think the
netlink message is somehow not properly aligned on volantis, causing the kernel
to misinterpret it.

Bug: 15840054
Change-Id: Ief60473e337410f7cb35890de0a5a74a21723a41

10 years agoNetd: Interface controller: Demote a log message severity from warning to verbose.
Sasha Levitskiy [Mon, 23 Jun 2014 17:51:05 +0000 (10:51 -0700)]
Netd: Interface controller: Demote a log message severity from warning to verbose.

Change-Id: Ia8b2e6a0b039c2bda418a5f7138c0aac6b6c643b
Signed-off-by: Sasha Levitskiy <sanek@google.com>
10 years agoMerge "Pass route add/delete errors back to CommandListener."
Lorenzo Colitti [Mon, 23 Jun 2014 16:41:13 +0000 (16:41 +0000)]
Merge "Pass route add/delete errors back to CommandListener."

10 years agoReturn negative errno (instead of positive) on failure.
Sreeram Ramachandran [Sun, 22 Jun 2014 18:02:57 +0000 (11:02 -0700)]
Return negative errno (instead of positive) on failure.

http://ag/489245 changed some return values from bools to errno values. However,
in forthcoming CLs, @lorenzo uses the convention of negative errno to indicate
failure. So, be consistent with that style.

Change-Id: I3eac8f142c36a2e779cda289c07ee374c49e2f6b

10 years agoPass route add/delete errors back to CommandListener.
Lorenzo Colitti [Tue, 17 Jun 2014 15:41:58 +0000 (00:41 +0900)]
Pass route add/delete errors back to CommandListener.

Change-Id: Id1d6d578963080e141f71bc1303801fc53bce40a

10 years agoUse native netlink code instead of /sbin/ip to manipulate routes
Lorenzo Colitti [Tue, 17 Jun 2014 15:22:17 +0000 (00:22 +0900)]
Use native netlink code instead of /sbin/ip to manipulate routes

Shelling out to /sbin/ip is slow, and more importantly it does
not preserve the error messages returned by the kernel when
adding or deleting a route fails.  Instead, use netlink directly.

This change does not yet pass the errors back to CommandListener;
that is done in the next change in the series.

Change-Id: I5ad3c8583580857be6386a620ff5c4f3872d685b

10 years agoNotify network observers of route changes.
Lorenzo Colitti [Thu, 12 Jun 2014 04:51:05 +0000 (13:51 +0900)]
Notify network observers of route changes.

This reverts commit 20d3f5e856a38ef22851e32d7f9b6a7cd02eb459.

Bug: 9180552
Change-Id: I004448e2f0c731f6ca2c6d3535dc4dd51a622293

10 years agoAdd getNetworkForSocket().
Sreeram Ramachandran [Fri, 20 Jun 2014 18:59:40 +0000 (11:59 -0700)]
Add getNetworkForSocket().

This doesn't require a roundtrip to netd, since anybody can read the SO_MARK.

Change-Id: I51dd17725c4534cb0d5dbc8e93e844e6a7847959

10 years agoReturn errors explicitly instead of using errno.
Sreeram Ramachandran [Fri, 20 Jun 2014 18:51:48 +0000 (11:51 -0700)]
Return errors explicitly instead of using errno.

Change-Id: Ia29f500e747a8c72d13a8f38c3b08c319c8c029a

10 years agoMerge "Add a client API to protect a socket from VPNs."
Sreeram Ramachandran [Fri, 20 Jun 2014 17:23:46 +0000 (17:23 +0000)]
Merge "Add a client API to protect a socket from VPNs."

10 years agoMerge "Revert "Notify network observers of route changes.""
Lorenzo Colitti [Fri, 20 Jun 2014 07:03:19 +0000 (07:03 +0000)]
Merge "Revert "Notify network observers of route changes.""

10 years agoRevert "Notify network observers of route changes."
Lorenzo Colitti [Fri, 20 Jun 2014 07:02:36 +0000 (07:02 +0000)]
Revert "Notify network observers of route changes."

This reverts commit d2f23fb6285000831f14cacc16881d96f6046c33.

Change-Id: Ifabbc6848ab33842461465fccb9886863a08aec6

10 years agoNotify network observers of route changes.
Lorenzo Colitti [Thu, 12 Jun 2014 04:51:05 +0000 (13:51 +0900)]
Notify network observers of route changes.

Bug: 9180552
Change-Id: Ic23025e4d7cb4e58ae59017777267ac45f00b434

10 years agoMerge "Flush both IPv4 and IPv6 routes."
Lorenzo Colitti [Fri, 20 Jun 2014 06:02:48 +0000 (06:02 +0000)]
Merge "Flush both IPv4 and IPv6 routes."

10 years agoFlush both IPv4 and IPv6 routes.
Lorenzo Colitti [Tue, 17 Jun 2014 07:14:17 +0000 (16:14 +0900)]
Flush both IPv4 and IPv6 routes.

The current code unintentionally flushes only IPv4 routes
because it reuses a function that is normally used to add and
remove routes (where the IP version is implicitly specified by
the route to add or remove).

Instead of fixing the runIpRouteCommand function, add a new
flushIpRoutes function because runIpRouteCommand will be
replaced by a netlink implementation in an upcoming CL.

Change-Id: Ie96ae4124baca3edb8e0d0841e7abadb6b3ee9ab

10 years agoMerge "Notify network observers of route changes."
Lorenzo Colitti [Fri, 20 Jun 2014 05:52:15 +0000 (05:52 +0000)]
Merge "Notify network observers of route changes."

10 years agoserver: check interface names in RPC arguments for validity
JP Abgrall [Fri, 20 Jun 2014 01:35:24 +0000 (18:35 -0700)]
server: check interface names in RPC arguments for validity

This patch introduces a method isIfaceName that checks interface
names from various RPCs for validity before e.g. using them as
part of iptables arguments or in filenames.

All of these RPC calls can only be called from applications
with at least the CONNECTIVITY_INTERNAL permission in recent
Android versions, so the impact of the missing checks luckily
isn't very high.

Orig-Author: Jann Horn <jann@thejh.net>

Change-Id: I80df8d745a3de99ad02d6649f0d10562c81f6b98
Signed-off-by: JP Abgrall <jpa@google.com>
10 years agoAdd a client API to protect a socket from VPNs.
Sreeram Ramachandran [Thu, 19 Jun 2014 17:03:07 +0000 (10:03 -0700)]
Add a client API to protect a socket from VPNs.

The server side hasn't been implemented yet (see FwmarkServer.cpp).

A UID can only be in a single VPN at any time, so there's no need
to specify a netId or vpnId.

Change-Id: Ie9c4590a9900e1ebf28418c4b9c4760cc0a5501a

10 years agoCache interface indices in case interfaces go away.
Paul Jensen [Thu, 12 Jun 2014 20:46:37 +0000 (16:46 -0400)]
Cache interface indices in case interfaces go away.

Without caching them netd will fail to remove rules and routes,
for example, when the Bluetooth reverse-tether interface ("bt-pan")
goes away.

bug:15407087

Change-Id: I99fcf00f9645a0b029455516a705b70110f62ff6

10 years agoMerge "Pass a socket mark to clatd so it can bind to its network."
Lorenzo Colitti [Fri, 13 Jun 2014 02:35:29 +0000 (02:35 +0000)]
Merge "Pass a socket mark to clatd so it can bind to its network."

10 years agoPass a socket mark to clatd so it can bind to its network.
Lorenzo Colitti [Mon, 9 Jun 2014 05:09:20 +0000 (14:09 +0900)]
Pass a socket mark to clatd so it can bind to its network.

Bug: 15340961
Change-Id: If15e90cbd5526f6c8fd839d4d009846d64d9e77a

10 years agoam b72f6d71: (-s ours) DO NOT MERGE: Fix use-after-free of interface name during...
Paul Jensen [Thu, 12 Jun 2014 19:56:50 +0000 (19:56 +0000)]
am b72f6d71: (-s ours) DO NOT MERGE: Fix use-after-free of interface name during network destruction.

* commit 'b72f6d71840afeb1923f64f8c44417f4f208c282':
  DO NOT MERGE: Fix use-after-free of interface name during network destruction.

10 years agoDO NOT MERGE: Fix use-after-free of interface name during network destruction.
Paul Jensen [Wed, 11 Jun 2014 14:58:16 +0000 (10:58 -0400)]
DO NOT MERGE: Fix use-after-free of interface name during network destruction.

bug:15560990

Change-Id: I899827c4f50847a3a60b6359f829bee5d6eb4f00
(cherry picked from commit 6ff16785ae1f67f44a73ad00d6c86690af6772d2)

10 years agoFix use-after-free of interface name during network destruction.
Paul Jensen [Wed, 11 Jun 2014 14:58:16 +0000 (10:58 -0400)]
Fix use-after-free of interface name during network destruction.

bug:15560990

Change-Id: I899827c4f50847a3a60b6359f829bee5d6eb4f00

10 years agoam 41417388: (-s ours) Use AF_INET sockets when checking netIds in setNetworkFor...
Sreeram Ramachandran [Fri, 6 Jun 2014 15:50:13 +0000 (15:50 +0000)]
am 41417388: (-s ours) Use AF_INET sockets when checking netIds in setNetworkFor{Process,Resolv}(). DO NOT MERGE

* commit '414173882a87ded2fdd4f5181cd7dff3842a8917':
  Use AF_INET sockets when checking netIds in setNetworkFor{Process,Resolv}(). DO NOT MERGE

10 years agoam d42cdf50: am 437eb9c3: Merge "Changes uses of index() to strchr()"
Dan Albert [Thu, 5 Jun 2014 20:33:08 +0000 (20:33 +0000)]
am d42cdf50: am 437eb9c3: Merge "Changes uses of index() to strchr()"

* commit 'd42cdf50ab883acb74b17d63646f0228d881df3c':
  Changes uses of index() to strchr()

10 years agoam 437eb9c3: Merge "Changes uses of index() to strchr()"
Dan Albert [Thu, 5 Jun 2014 20:29:33 +0000 (20:29 +0000)]
am 437eb9c3: Merge "Changes uses of index() to strchr()"

* commit '437eb9c3092c96c1828e326edb6574090fe37aaa':
  Changes uses of index() to strchr()

10 years agoMerge "Changes uses of index() to strchr()"
Dan Albert [Thu, 5 Jun 2014 19:55:52 +0000 (19:55 +0000)]
Merge "Changes uses of index() to strchr()"

10 years agoChanges uses of index() to strchr()
Dan Albert [Thu, 5 Jun 2014 18:35:41 +0000 (11:35 -0700)]
Changes uses of index() to strchr()

Bionic is geting rid of index() because it was removed from POSIX 2008.

Change-Id: I182523734f372d1c6ed052b1373422332587f8ff

10 years agoUse std::atomic<> instead of volatile sig_atomic_t.
Sreeram Ramachandran [Tue, 3 Jun 2014 19:51:08 +0000 (12:51 -0700)]
Use std::atomic<> instead of volatile sig_atomic_t.

std::atomic<> comes with true load/store barriers, so it's preferable.

Change-Id: Ife47b0e404f41aa74aeb168befd1ff4a6603f3ed

10 years agoRemove unnecessary includes.
Sreeram Ramachandran [Tue, 3 Jun 2014 00:47:36 +0000 (17:47 -0700)]
Remove unnecessary includes.

Neither of these actually does anything.

Change-Id: I82ee09a4565a8e8e4dc5f54f278bf83e002d7387

10 years agoMerge "Pass NetID to clatd as a command line argument."
Paul Jensen [Tue, 3 Jun 2014 17:24:46 +0000 (17:24 +0000)]
Merge "Pass NetID to clatd as a command line argument."

10 years agoPass NetID to clatd as a command line argument.
Paul Jensen [Fri, 30 May 2014 17:29:41 +0000 (13:29 -0400)]
Pass NetID to clatd as a command line argument.

Change-Id: I77a47b24e68c7786f790974f05787a40a4934af5

10 years agoUse AF_INET sockets when checking netIds in setNetworkFor{Process,Resolv}(). DO NOT...
Sreeram Ramachandran [Sat, 31 May 2014 02:59:51 +0000 (19:59 -0700)]
Use AF_INET sockets when checking netIds in setNetworkFor{Process,Resolv}(). DO NOT MERGE

AF_UNIX sockets don't need to be marked, so we don't give netd the permission to
operate on them (cf: netd.te). I.e., netd doesn't expect to receive them.

Make sure that the creation of the AF_INET socket doesn't trigger another
wasteful check with netd by calling the libc version directly.

Bug: 13885501
Change-Id: I6b549232e57cacd47501edcefa4c0b4b79df9da0
(cherry picked from commit 2756045bebaac342f7cb70dad11519f896d44833)

10 years agoUse AF_INET sockets when checking netIds in setNetworkFor{Process,Resolv}().
Sreeram Ramachandran [Sat, 31 May 2014 02:59:51 +0000 (19:59 -0700)]
Use AF_INET sockets when checking netIds in setNetworkFor{Process,Resolv}().

AF_UNIX sockets don't need to be marked, so we don't give netd the permission to
operate on them (cf: netd.te). I.e., netd doesn't expect to receive them.

Make sure that the creation of the AF_INET socket doesn't trigger another
wasteful check with netd by calling the libc version directly.

Bug: 13885501
Change-Id: I6b549232e57cacd47501edcefa4c0b4b79df9da0

10 years agoFix build.
Sreeram Ramachandran [Fri, 30 May 2014 22:11:54 +0000 (15:11 -0700)]
Fix build.

Change-Id: If170e46ff92c6a972bc1c2d838b1ac0eea6e23de

10 years agoam 82eab785: Support legacy routes added by apps via ensureRouteToHost().
Sreeram Ramachandran [Fri, 30 May 2014 21:33:07 +0000 (21:33 +0000)]
am 82eab785: Support legacy routes added by apps via ensureRouteToHost().

* commit '82eab785bd5cb2eff0a263f5b0dcde13e9139588':
  Support legacy routes added by apps via ensureRouteToHost().

10 years agoSupport legacy routes added by apps via ensureRouteToHost().
Sreeram Ramachandran [Thu, 22 May 2014 21:21:49 +0000 (14:21 -0700)]
Support legacy routes added by apps via ensureRouteToHost().

This adds the routes to two fixed tables:
+ LEGACY, which has higher priority than other non-explicit lookup tables
  (per-network and default network).
+ PRIVILEGED_LEGACY, available only to system apps and has higher priority than
  VPNs (system apps are those with the CONNECTIVITY_INTERNAL permission).

This will be changed to per-UID tables once the kernel supports UID-based
routing, so that these legacy routes are scoped to each app and not global.

Also, fix a TODO: The framework (as of http://ag/471599) will not set the
gateway argument if it's actually a direct-connected route.

Change-Id: I0ee1ca89fdc859d75a89021ca8c1902811b1e4a9
(cherry picked from commit 38b7af1f2cb9579895465fabc37865f5dadcac25)

10 years agoam ce8f583f: Set errno in NetworkController commands.
Lorenzo Colitti [Fri, 30 May 2014 15:06:46 +0000 (15:06 +0000)]
am ce8f583f: Set errno in NetworkController commands.

* commit 'ce8f583ff6620a19602d3c4604557e1b1501cafa':
  Set errno in NetworkController commands.

10 years agoam a5d68758: merge in master-release history after reset to ec00884cac216d1cb79556ca2...
The Android Automerger [Fri, 30 May 2014 15:06:46 +0000 (15:06 +0000)]
am a5d68758: merge in master-release history after reset to ec00884cac216d1cb79556ca23b21ce55e35af3e

* commit 'a5d6875887a63797711093a169e7ef19c5a1ab3b':
  Fix for DNS resolutions when there is no default network set yet. We need to determine the actual netId to use prior to comparing it against the default netId.  Also initialize the default network variable.

10 years agoTurn on C++11 and make all warnings into errors.
Sreeram Ramachandran [Wed, 28 May 2014 22:07:00 +0000 (15:07 -0700)]
Turn on C++11 and make all warnings into errors.

As a consequence:
+ Comment out the names of all unused parameters.
+ Remove all unused variables and functions.

In server/Android.mk, there are a couple of non-trivial changes:
+ Use libcxx instead of stlport. This is needed to fix a bunch of errors due to
  specifying -std=c++11.
+ LOCAL_SHARED_LIBRARIES is sorted. Technically, the order in which libraries
  are listed has an effect on linking, but nobody should be doing such brittle
  things anyway.

Change-Id: I0aff5b745e04609da23144d0e8be4c5694321b8b

10 years agoSupport legacy routes added by apps via ensureRouteToHost().
Sreeram Ramachandran [Thu, 22 May 2014 21:21:49 +0000 (14:21 -0700)]
Support legacy routes added by apps via ensureRouteToHost().

This adds the routes to two fixed tables:
+ LEGACY, which has higher priority than other non-explicit lookup tables
  (per-network and default network).
+ PRIVILEGED_LEGACY, available only to system apps and has higher priority than
  VPNs (system apps are those with the CONNECTIVITY_INTERNAL permission).

This will be changed to per-UID tables once the kernel supports UID-based
routing, so that these legacy routes are scoped to each app and not global.

Also, fix a TODO: The framework (as of http://ag/471599) will not set the
gateway argument if it's actually a direct-connected route.

Change-Id: I0ee1ca89fdc859d75a89021ca8c1902811b1e4a9

10 years agoSet errno in NetworkController commands.
Lorenzo Colitti [Thu, 29 May 2014 03:20:55 +0000 (12:20 +0900)]
Set errno in NetworkController commands.

Set errno in the NetworkController functions that are called by
CommandListener and where failure causes an operationError. This
is because operationError always appends errno to its messages,
and

"400 481 addInterfaceToNetwork() failed (Device or resource busy)"

is more useful than:

"400 481 addInterfaceToNetwork() failed (Success)"

We already call ALOGE to put descriptive messages in the system
log, but the bugreport already conveniently aggregates all netd
commands in one place in the network_management service dump, and
it would help if the errno in those messages were accurate.

Unfortunately many of the route and iptables commands shell out
to /sbin/ip and we can't return meaningful values, but we'll
fix that when (if?) we replace that with a proper netlink
implementation.

Bug: 15316764
Change-Id: Ia47f451029d611491aa72bca602de77333f2c6a0

10 years agomerge in master-release history after reset to ec00884cac216d1cb79556ca23b21ce55e35af3e
The Android Automerger [Thu, 22 May 2014 14:22:41 +0000 (07:22 -0700)]
merge in master-release history after reset to ec00884cac216d1cb79556ca23b21ce55e35af3e

10 years agomerge in master-release history after reset to ec00884cac216d1cb79556ca23b21ce55e35af3e
The Android Automerger [Thu, 22 May 2014 12:51:13 +0000 (05:51 -0700)]
merge in master-release history after reset to ec00884cac216d1cb79556ca23b21ce55e35af3e

10 years agoImplement the "select network" fwmark command.
Sreeram Ramachandran [Wed, 21 May 2014 21:01:16 +0000 (14:01 -0700)]
Implement the "select network" fwmark command.

Change-Id: Id4a49eb288b18022d53014d1ae2211ed7d1099a6

10 years agoRework the determination of a "valid network".
Sreeram Ramachandran [Wed, 21 May 2014 20:19:43 +0000 (13:19 -0700)]
Rework the determination of a "valid network".

+ isNetIdValid() doesn't make much sense. What we want is whether the netId has
  actually been created (via createNetwork()).
+ It isn't an error to call deleteNetwork() or setDefaultNetwork() even when
  there are no interfaces assigned to the network.
+ Secure all accesses to the maps in PermissionsController with locks; they are
  called from many threads (CommandListener, DnsProxyListener and FwmarkServer).
+ Remove the redundant mIfaceNetidMap.
+ Minor cosmetic changes to things such as #includes and log messages.

Change-Id: Ieb154589b24f00ba8067eaaec4def3534aec4923

10 years agoMark sockets on creation (socket()) and accept4().
Sreeram Ramachandran [Wed, 21 May 2014 20:08:34 +0000 (13:08 -0700)]
Mark sockets on creation (socket()) and accept4().

Continued from: https://android-review.git.corp.google.com/#/c/95094/

Change-Id: Ib0b8f5d7c5013b91eae6bbc3847852eb355c7714

10 years agoNew network selection APIs.
Sreeram Ramachandran [Wed, 21 May 2014 18:41:39 +0000 (11:41 -0700)]
New network selection APIs.

Continued from: https://android-review.git.corp.google.com/#/c/94977/

Change-Id: Ie0576888f50a8ce91bbb0a4794708b406eb0aa35

10 years agoMove netd_client into netd.
Sreeram Ramachandran [Wed, 21 May 2014 15:54:07 +0000 (08:54 -0700)]
Move netd_client into netd.

Change-Id: Ie4b6b303225c93f2448a503d6ea9cebb552cbad5

10 years agomerge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31
The Android Automerger [Wed, 21 May 2014 12:51:08 +0000 (05:51 -0700)]
merge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31

10 years agomerge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31
The Android Automerger [Tue, 20 May 2014 12:51:24 +0000 (05:51 -0700)]
merge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31

10 years agomerge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31
The Android Automerger [Mon, 19 May 2014 12:50:25 +0000 (05:50 -0700)]
merge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31

10 years agomerge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31
The Android Automerger [Sun, 18 May 2014 12:50:12 +0000 (05:50 -0700)]
merge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31

10 years agomerge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31
The Android Automerger [Sat, 17 May 2014 12:51:51 +0000 (05:51 -0700)]
merge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31

10 years agomerge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31
The Android Automerger [Thu, 15 May 2014 12:51:28 +0000 (05:51 -0700)]
merge in master-release history after reset to 5ff58d494a5ddf7e544adc77590300f04daf8e31

10 years agoFix build by using the right type (socklen_t instead of int).
Sreeram Ramachandran [Wed, 14 May 2014 16:57:31 +0000 (09:57 -0700)]
Fix build by using the right type (socklen_t instead of int).

Change-Id: Ia4a56ef679269755a3787408821c0cda6aec9420

10 years agoMark sockets on accept().
Sreeram Ramachandran [Tue, 13 May 2014 20:25:34 +0000 (13:25 -0700)]
Mark sockets on accept().

The kernel marks the netId into the accepted socket, so we just set the socket
owner's permission bits here in userspace.

We also check the permissions required by the network and return an error if
they are not satisfied. This avoids the "half-open" problem where an app listens
on all networks (interfaces) with bind(0), and receives an incoming connection
on a privileged interface that it can't send responses over.

(cherry picked from commit 920747a009efbb815465d980ec6a918531e74e9f)

Change-Id: I5d09be413cf720fbed905f96313b007997ada76c