Remote Execution – Arma 3

From Bohemia Interactive Community
Revision as of 13:31, 3 July 2015 by Warka (talk | contribs)
Jump to navigation Jump to search

-wrong parameter ("A3") defined!-[[:Category:Introduced with A3 version 1.48|1.48]]

⚠
This system is not considered done and it is still subject to change. It may change rapidly, use it carefully on your own risk.

Overview

In A3 1.48 the 2 new script commands were added to natively handle execution of code and functions on remote machines - remoteExec and remoteExecCall.

Links

Motivation

Until A3 1.48 there was no engine based remote execution. The only officially supported way how to execute code or command on non-local machine was by using scripted remote execution framework - BIS_fnc_MP. This scripted framework, however working, suffered from several issues, mainly poor network traffic optimization and insufficient security control. The network traffic was optimized in 1.46 but the security issues could not be properly fix in the scripted framework.

To address those security issues, in A3 1.48 the remote execution was implemented into the engine. There are now 2 new script commands - remoteExec and remoteExecCall, that allow for fully controlled and secure remote execution from client machines(s).

Syntax

Both script commands have same syntax but differ in way how the code is executed on target machine(s). For details and explanation of the difference, see the next section. 

<params> remoteExec [<function>,(<target>,<isPersistent>)];
<params> remoteExecCall [<function>,(<target>,<isPersistent>)];
  • params: ARRAY - parameters passed to the function
  • function: STRING - function name. Commands and functions defined in CfgRemoteExec are whitelisted.
  • target: - Optional. [default: 0]
    OBJECT - function will be executed only where unit is local
    NUMBER - the function will be executed only on machine with the given owner ID.
    0 - the function will be executed on each machine including the one where remoteExec was called from.
    2 - it will be executed only by server.
    negative value - it will be executed everywhere except for machines with with the given owner ID.
    SIDE - the function will be executed only on machines where the player is on the specified side
    GROUP - the function will be executed only on machines where the player is in the specified group
    ARRAY - array of any of types listed above
  • isPersistent: STRING or BOOL - Optional [default: false].
    If true, function generates an unique ID for the message and the message itself is added to the JIP queue with the unique ID.
    If a non-empty string, it is a custom ID of the message and the message itself is added to the JIP queue overriding any remoteExec message with the same ID.
    Otherwise, no ID is generated and no message is placed into the JIP queue (default state).
Return Values
NIL in case of error.
STRING otherwise. If JIP is not requested this is an empty string. Otherwise this is an unique JIP ID.

Using remoteExec and remoteExecCall

remoteExec

Content of MP message sent by remoteExec command is executed on target machine in scheduled environment and as such abides to the same rules and limitations as other scheduled code.

If you want to execute more complex code that needs to be processed as separate spawn use remoteExec command.

More specifically:

  1. If your code includes any delays (commands like sleep or waitUntil) you MUST use remoteExec.
  2. If your code contains more CPU demanding operations that will take some time for the game to process, you SHOULD use remoteExec, otherwise you might experience performance drops.

remoteExecCall

Unlike remoteExec, MP messages send by remoteExecCall command are executed sequentially outside of the scheduled environment. This makes the remoteExecCall command very useful, if you need to to execute multiple codes/commands in a set sequence.

The code sent by remoteExecCall MUST NOT contain any delays and SHOULD NOT be too complex and CPU demanding.

Security

The biggest issue of previous scripted remote execution was the lack of control over what can and cannot be executed from local clients. To address this we’ve created system where content authors can define through CfgRemoteExec config how the remote execution should operate on clients.

Note: Server doesn’t have any limitations at place, everything is enabled and opened for him. All limitations and rules apply only for client(s).

Because remoteExec and remoteExecCall can be used to remotely execute script commands (like for e.g. setDamage) as well as scripted functions (e.g. BIS_fnc_setRank) we have separated the security settings into two groups Functions and Commands. This allows people to quickly set different rules for functions and commands separately.

The security rules consist of 3 security settings - operation mode, allowed targets and jip. The operation mode is very important as it defines if functions/commands can be executed remotely from client or not and ev. allows you to whitelist specific functions/commands. The allowed targets and jip are very optional and you really do not need to use them, unless you want to be super safe.

To get more info about those settings and the CfgRemoteExec config, check the subsections below.

Operation modes

Operation mode is numeric value describing how functions or commands should be treated on server on client. 

   0: remote execution is blocked
   1: only whitelisted functions/commands are allowed for remote execution
   2: remote execution fully opened

Allowed targets

In addition to the operation mode, in client subclass, allowed targets can be defined for individual whitelisted commands and functions. This adds another layer of security and control to the system, as it allows you to define not only who can send the execution request, but also where it can be executed. 

   0: can target all machines (default)
   1: can target only clients, execution on server is denied
   2: can target only server, execution on clients is denied

JIP

To control who can add JIP message into JIP queue, we have added an optional parameter jip. This parameter can be defined in whitelisted function/command class, at the same place as allowed targets are defined or in the Functions and/or Commands classes. This parameter affects only clients (no effect if defined within Server subclass).

If it is defined on both levels, the more local definition takes precedence (which is the value defined in the whitelisted function/command). 

   0: JIP flag cannot be set
   1: JIP flag can be set (default)

Config location

The CfgRemoteExec class can be defined in mission description.ext, campaign description.ext or global (addon) config. As usual the more local config takes precedence. In case of more global configs exist, the mode attribute will be overridden by the last parsed config and whitelisted commands and functions will be merged.

Sample CfgRemoteExec definition: 

class CfgRemoteExec
{
        class Commands
        {
                mode = 1;

                class setFuel {allowedTargets = 2;};           //execute only on server
                class hint {jip = 0;};                         //jip is not allowed for this command
        };
        class Functions
        {
                mode = 0;
                jip = 0;                                       //no functions can use jip

                class BIS_fnc_setRank {allowedTargets = 1;};   //execute only on clients, server execution denied
        };
};

The BIS_fnc_MP backward compatibility

By implementation of the engine based remote execution, the scripted solution becomes deprecated. As a result for A3 1.50 it's planned to implement remoteExec and remoteExecCall into the BIS_fnc_MP. Although the insides of the BIS_fnc_MP function will be completely rewritten, function syntax and functionality will remain 100% same. There is no need to worry about backward compatibility.

In addition from A3 1.50 the BIS_fnc_MP function will abide the same CfgRemoteExec rules for remote execution as remoteExec and remoteExecCall commands do. Because of that the content authors that were using the BIS_fnc_MP will be able to set the security permissions for remote execution and will make their content more resilient to hacking.

Advanced Techniques & Functionality Insight

If you want to know more about how the remote execution works and handles some more complicated issues and edge cases continue reading, otherwise feel free to skip.

JIP queue

When command or function is executed through remoteExec or remoteExecCall, it can be flagged as persistent through the isPersistent attribute. If it is flagged, the request is being stored on server in the JIP queue under the unique JIP id. When new player joins the MP session that is currently running (we call this Join-In-Progress, JIP) all the MP messages that were stored in the JIP queue are executed on him.

JIP id

The JIP id is a unique key under which the JIP MP message is being stored in the JIP queue on the server. When you set isPersistent flag to true, the JIP id is auto-generated on the machine the remote execution was initiated. If you set the isPersistent flag to specific string, that string is considered to be the JIP id.

The auto-generated JIP is always unique. If the JIP is manually supplied, the content author needs to make sure it is unique, otherwise he/she will overwrite another MP msg that is currently stored in the queue under the JIP id.

Overwriting JIP message in the queue

Every JIP message is stored in the queue. If you need to overwrite the JIP message with another message, use the same JIP id for the new message. This way the new message will be stored in the queue under the provided id and will overwrite the previously stored message that used the same id.

Deleting JIP message from the queue

To remove a specific JIP message from the queue call the remoteExec with function/command name set to an empty string, the JIP id of the message you want to remove and no other params. 

remoteExec ["", "JIPid"];

Validity verification

Validity of remote execution request is verified in 2 steps: when the request is initiated (issued from a client machine) when the server is going to broadcast the request. If the request is initialized directly from the server, step 1 is skipped (this applies for hosted server admins as well).

For a remote execution request to be valid following criteria need to be met:

  1. input parameters must be properly defined
  2. function/command must exist on the machine
  3. function/command remote execution request must be allowed in CfgRemoteExec
    • remote execution must be either fully open (mode=2)
    • or the particular function/command must be whitelisted
  4. if JIP is used (isPersistent flag is not false), JIP must be allowed in CfgRemoteExec (jip=1)

If any of the aforementioned is not met remote execution is blocked.

Retrieved from "https://community.bistudio.com/wiki?title=Arma_3:_Remote_Execution&oldid=90524"