Java IoT Authors: AppDynamics Blog, Dana Gardner, Automic Blog, Pat Romanski, Liz McMillan

Related Topics: Java IoT

Java IoT: Article

Calling Java From C

A framework for easier JNI

Though most Java developers think of the Java Native Interface (JNI) as a framework for developing native libraries that can be called from Java, relatively few know that JNI also supports communication in the reverse direction: it provides native programs written in C with the ability to call Java objects. However, the coding is thorny; logic that can be coded readily in a few lines of Java requires several times more lines of C, thanks to JNI's granular programming model and peculiar approaches to exception handling and garbage collection. This article explores the nature and typical use of the C-to-Java JNI interface and presents the design of a framework that eases the programming effort.

The JNI Architecture
As Figure 1 illustrates, JNI is actually a pair of APIs:

  • "JNI Proper" supports the manipulation of Java objects and classes, such as the ability to call object methods.
  • The Invocation API is a smaller C library that enables C programs to create and destroy a Java Virtual Machine (JVM).
A C-to-Java program (that is, a C program that uses Java) calls the Invocation API to create a JVM, and calls JNI Proper to use Java objects. As for the Java-to-C direction (not discussed in this article), Java code calls a native method, which is implemented as a C native library function; the C code uses JNI to interpret its Java input types and build its Java output types.

The JVM is packaged as a shared library ("jvm.dll" in the Sun SDK on Windows platforms and "libjvm.so" on Solaris) and exposes JNI Proper and the Invocation API as public exports. As a runtime entity, a JVM is really just the JVM library linked to an executable C program. Any Java developer can code such a program, e.g., "myCtoJProgram.exe", shown in Figure 2. Interestingly, the famed SDK C program known as the "launcher" (java.exe on Windows, Java on Solaris) is written in just the same way (for more, see sidebar JNI Case Study: Java Launcher). Completing the picture are native libraries, such as "myNative.dll", whose functions can be called from Java; these libraries are linked to the runtime process alongside the JVM.

A C-to-Java JNI Design: The Zip Example
The launcher is the best-known example of a C program that uses the JNI's C-to-Java interface; its purpose is to house a JVM and bootstrap a Java application on the JVM by calling the application's main method. Other, less obvious examples are C programs that require functionality whose best or only implementation is Java objects that must be called "in process." Examples include programs that:

  • Create, extract, or list the contents of zip files. The Java SDK java.util.zip package is the most suitable API available.
  • Transform XML to XML, HTML, or PDF. Though C XML APIs exist, Java's support for XML is vastly superior. Launching a JVM to do XML with Java is a plausible strategy for a C program.
  • Call an Enterprise JavaBean (EJB). The C program uses JNI to execute standard EJB client-side logic.
The example of listing the contents of a zip file highlights the coding challenge of C-to-Java JNI. (Source code for this article can be downloaded from www.sys-con.com/java/sourcec.cfm.) The Java code to perform this logic, shown in Listing 1, is trivial: line 4 instantiates the class ZipFile in java.util.zip, passing the zip file name to the constructor; lines 5-9 loop over a java.util.Enumeration of java.util.zip.ZipEntry objects getting, in line 8, the name of each entry in the zip file.

Developing the equivalent logic in C requires hundreds of lines of code. The main steps are summarized in Table 1. Listing 2 shows the code for steps 7 and 8; the JNI calls in lines 5, 10, 15, and 21 are followed by calls in lines 7, 11, 16, and 22 to the checkException() function (implementation not shown), which in turn calls the JNI exception handling functions ExceptionCheck(), ExceptionDescribe(), and ExceptionClear() to swallow and report Java exceptions triggered by the JNI calls.

The C code in the zip example can be made easier and less cumbersome by using two design patterns:

  1. Java Proxy: Put the hard code where it belongs, on the Java side. Develop a Java object, called a proxy, that performs complex logic on behalf of the C code. The C code need only call the proxy.
  2. C Façade: Hide JNI's peculiar programming model in a C façade library. Have the C program call the façade rather than JNI directly. In addition, build proxy support into the façade; expose façade functions to call the proxy.
The proxy and façade constitute an abstract framework for use in any program resembling the zip program. Java proxies implement the interface shown in Listing 3; the execute() method is defined generically as accepting an input, performing some action or set of actions, and returning an output. Listing 4 shows the proxy implementation for the zip example. The execute() method expects as input a string specifying the name of the zip file (line 26); it implements logic similar to that in Listing 1 to enumerate the entries in the zip file (lines 26-33) and returns a string containing the name of the entries in a pipe-delimited list (see lines 30-32 and 35). The method could also have returned an array or Java collection type, but the calling C program is likely happier parsing a string than contending with JNI array or collection class iteration logic.

If the C zip program were to call the proxy using JNI directly, its length would be shorter but the complexity of JNI would remain. Using the façade reduces the length even further and shields the code from JNI oddities. A design for the façade is depicted in Figure 3. The façade consists of a set of data types, modeled as C structures, and a set of functions. The data types represent entities such as JVM (cjJVM_t), class (cjClass_t), method (cjMethod_t), and object (cjObject_t). The functions are operations performed on the entities (e.g., cjJVMConnect() and cjJVMDisconnect() performed on the JVM), and a special set of proxy operations. Table 2 describes and lists the JNI usage of each of the C functions.

Listing 5 is the complete source code of the C zip program that uses the proxy and façade. (Listing 5 can be downloaded from www.sys-con.com/java/sourcec.cfm.) The program launches a JVM (line 26), gets a reference to the proxy class (line 29), instantiates it (line 33), and calls the execute() method (line 36). The remaining code (lines 40-50) cleans up the proxy object and class and destroys the JVM.

Hosting the hard logic in a Java proxy and wrapping JNI calls to the proxy in a C façade reduces the complexity of C-to-Java programming with JNI.


  • "Java Native Interface Specification," Sun Microsystems: java.sun.com/j2se/1.5.0/guide/jni/spec/jniTOC.html
  • "Tutorial on JNI," Sun Microsystems: java.sun.com/docs/books/tutorial/native1.1/concepts/index.html


    JNI Case Study: Java Launcher
    Every Java developer who uses the Sun SDK is grateful for the C program known as the launcher (its executable name is "java"), which uses the JNI Invocation API to create a JVM, load a Java class into the JVM, and call its main() method, thereby launching a Java application on behalf of the caller. If the launcher did not exist, a good C developer with JNI knowledge could write an equivalent program in less than a week.

    The launcher's source code is available from Sun and is packaged with the SDK (version 1.4.2_04, which can be downloaded from java.sun.com/j2se/1.4.2/download.html). In the base directory of the installed SDK is a file called src.zip. If you extract that file, the exploded "launcher" directory contains the four source files that constitute the launcher program: java.h, java.c, java_md.h, and java_md.c. The launcher's source code is unmistakably C: murky, idiomatic, and circuitous. On the other hand, the end result is a functional program that has been run successfully innumerable times by innumerable users. Understanding how it works is a case study in the use of JNI.

    Suppose there is a class called Hi in the package com.mike that has a public main() method and thus can be started as a Java application through the launcher. The following is the source code:

    1	package com.mike;
    3	Public class Hi
    4	{
    5	   public static void main(String args[])
    6	   {
    7	      ...
    8	   }
    9	}

    The Hi application is started under SDK 1.4.2_04 on Windows 2000 with the following commands:

    1 set JAVA_HOME=c:\j2sdk1.4.2_04
    2 set _JAVA_LAUNCHER_DEBUG=true
    3 %JAVA_HOME%\bin\java -classpath src -Xms32m -Xmx64m -Dmy.property=1 com.mike.Hi arg1 arg2

    Line 3 calls the launcher executable Java in the bin directory of my SDK (which, as line 1 indicates, is c:\j2sdk1.4.2_04). The arguments passed to the launcher are:

    • -classpath src: Look for my "Hi" class in the directory src.
    • -Xms32m: Set its minimum heap size to 32MB.
    • -Xmx64m: Set the maximum heap size to 64MB.
    • -Dmy.property=1: Make a property available to the application with the key my.property and value 1.
    • com.mike.Hi: Run the application in this class.
    • arg1 arg2: Pass arguments "arg1" and "arg2" to the application's main method.
    Line 2 sets an environment variable called _JAVA_LAUNCHER_DEBUG, which causes the launcher to generate debugging output to the console at runtime:

    1	----_JAVA_LAUNCHER_DEBUG----
    2	JRE path is c:\j2sdk1.4.2_04\jre
    3	jvm.cfg[0] = ->-client<-
    4	jvm.cfg[1] = ->-server<-
    5	jvm.cfg[2] = ->-hotspot<-
    6	jvm.cfg[3] = ->-classic<-
    7	jvm.cfg[4] = ->-native<-
    8	jvm.cfg[5] = ->-green<-
    9	1306 micro seconds to parse jvm.cfg
    10	JVM path is c:\j2sdk1.4.2_04\jre\bin\client\jvm.dll
    11	5571 micro seconds to LoadJavaVM
    12	JavaVM args:
    13	    version 0x00010002, ignoreUnrecognized is JNI_FALSE, nOptions is 6
    14	    option[ 0] = '-Djava.class.path=.'
    15	    option[ 1] = '-Djava.class.path=src'
    16	    option[ 2] = '-Xms32m'
    17	    option[ 3] = '-Xmx64m'
    18	    option[ 4] = '-Dmy.property=1'
    19	    option[ 5] = '-Dsun.java.command=com.mike.Hi arg1 arg2'
    20	125113 micro seconds to InitializeJVM
    21	Main-Class is 'com.mike.Hi'
    22	Apps' argc is 2
    23	    argv[ 0] = 'arg1'
    24	    argv[ 1] = 'arg2'
    25	32937 micro seconds to load main class
    26	----_JAVA_LAUNCHER_DEBUG----

    The launcher begins by finding the JRE (line 2) and the right JVM (lines 3-10), and then loads the JVM dynamically (line 11); as we'll see, this logic is platform dependent. In this case, because we didn't name a specific JVM when calling the launcher, the launcher defaults to "client" (more on this below). Lines 12-19 show the arguments that the launcher will pass to the JVM; these correspond to the arguments passed to the launcher. In line 20, the launcher starts the JVM. The class whose main method that launcher will call is given on line 21; the arguments passed to it are shown in lines 22-24.

    The launcher's logical design (as of version 1.4.2_04) is depicted in Figure 4.

    The launcher consists of two modules: java.c, which contains the main function of the launcher program as well as platform-independent helper functions, and java_md.c, which houses platform-specific functions. The modules share a bidirectional dependency: func-tions in java.c call java_md.c and vice versa; for example, main() in java.c calls CreateExecutionEnvironment() in java_md.c, which in turn calls ReadKnownVMs() in java.c.

    The source code in java_md.c is different for each SDK platform release. For example, the Windows SDK has the Windows version of java_md.c but does not have the Solaris version. If you want to see both (as I did as I was writing this article), you must download both releases.

    The main steps in the launcher's processing are the following:

    1.  Create the execution environment: The CreateExecutionEnvironment() function, implemented in java_md.c, is a platform-specific search for the JRE path, JVM path, and JVM type for use by the launcher. The Windows version looks up the JRE path in the registry (on my machine, the registry key HKEY_LOCAL_MACHINE\Software\JavaSoft\Java Runtime Edition\1.4\JavaHome is C:\Program Files\Java\j2re1.4.2_04), and then checks whether the JVM type selected by the caller (corresponding to "-client" or "-server" launcher command-line options) is one of the allowable types listed in the file JREPath\lib\Arch\jvm.cfg (on my machine "Arch" is "i386"). If the caller did not specify a JVM type, the launcher defaults to the first type listed in jvm.cfg (on my machine it's "client"). The JVM path on Windows is JREPath\bin\JVMType\jvm.dll (e.g., JREPath\bin\client\jvm.dll for the client JVM).

    2.  Load the JVM dynamically: Whereas most programs let the operating system implicitly link shared libraries to their processes, the launcher, which allows the user to specify at runtime which version of the JVM library to use (the "-client" or "-server" command-line arguments to the launcher), explicitly loads the JVM library using a platform-specific interface. The logic resides in java_md.c's LoadJavaVM() function. On Windows, this function calls the Win32 LoadLibrary() to load the JVM DLL and link it to the launcher process, and then calls the Win32 GetProcAddress() function to get a pointer to the invocation API function JNI_CreateJavaVM() used in step 4.

    3.  Prepare JVM runtime options based on command-line options passed to the launcher: Command-line options such as -D, -X, and -classpath are assembled into an array to be passed to the JVM. The launcher adds an additional property for use by the JVM of the form -Djava.sun.command=class arg1 arg2 ..., where class is the fully qualified name of the target class and "arg1 arg2 ..." is the list of command-line arguments to be passed to its main method.

    4.  Create the JVM: The launcher calls the JVM's JNI_Create-JavaVM() function, passing the options prepared above.

    5.  Load the target class into the JVM by calling the JNI FindClass() method: The launcher first replaces dots with slashes in the class name (e.g., it converts com.mike.Hi to com/mike/Hi) because the JVM expects slashes instead of dots.

    6.  Call the main method of the target class: First, the launcher gets a reference to the main() method by calling the JNI GetStaticMethodID() function, passing the class reference acquired in step 5, the method name ("main"), and the signature ("([Ljava/lang/String;)V", the JVM's peculiar representation of a void method that accepts an array of java.lang.String objects). Second, the launcher calls the method via CallStaticVoidMethod(). The launcher prepares the string array method input using some of the JNI's array functions (NewObjectArray(), SetObjectArrayElement()), and handles exceptions in the main method using JNI's ExceptionOccurred(), ExceptionDescribe(), and ExceptionClear().

    7.  Shutdown the JVM: This is done by calling the JNI DetachCurrent-Thread() and DestroyJavaVM() functions.

    A mystery to many Java developers, the launcher is nothing more than a little C program that uses the JNI to initiate a Java application.

  • More Stories By Michael Havey

    Michael Havey is a Chordiant consultant with 10 years of industry experience, mostly with application integration. Michael's book Essential Business Process Modeling was published by O'Reilly in August 2005.

    Comments (5) View Comments

    Share your thoughts on this story.

    Add your comment
    You must be signed in to add a comment. Sign-in | Register

    In accordance with our Comment Policy, we encourage comments that are on topic, relevant and to-the-point. We will remove comments that include profanity, personal attacks, racial slurs, threats of violence, or other inappropriate material that violates our Terms and Conditions, and will block users who make repeated violations. We ask all readers to expect diversity of opinion and to treat one another with dignity and respect.

    Most Recent Comments
    Bubb 07/26/07 08:47:36 PM EDT

    I got file not found when I accessed

    DHAYANITHI R 03/23/05 03:01:47 AM EST

    Your matirial is good, But still now I can't understand how i call java function from C. Can u give some model source code, it is very usefull for me

    Michael 10/13/04 09:30:30 AM EDT

    It is really great article!!
    Is it possible to run on the Unix?


    pete 09/22/04 04:02:15 AM EDT

    it indeed is a very nice article.it helped me a lot for my assignment

    Jesse Barnum 08/21/04 10:45:44 PM EDT

    Thanks for this tutorial - it's exactly what I've been looking for. Everything that I see on JNI is for calling C from Java - this article is the other way around, which is just what I need to do.

    @ThingsExpo Stories
    Extracting business value from Internet of Things (IoT) data doesn’t happen overnight. There are several requirements that must be satisfied, including IoT device enablement, data analysis, real-time detection of complex events and automated orchestration of actions. Unfortunately, too many companies fall short in achieving their business goals by implementing incomplete solutions or not focusing on tangible use cases. In his general session at @ThingsExpo, Dave McCarthy, Director of Products...
    Ask someone to architect an Internet of Things (IoT) solution and you are guaranteed to see a reference to the cloud. This would lead you to believe that IoT requires the cloud to exist. However, there are many IoT use cases where the cloud is not feasible or desirable. In his session at @ThingsExpo, Dave McCarthy, Director of Products at Bsquare Corporation, will discuss the strategies that exist to extend intelligence directly to IoT devices and sensors, freeing them from the constraints of ...
    WebRTC is bringing significant change to the communications landscape that will bridge the worlds of web and telephony, making the Internet the new standard for communications. Cloud9 took the road less traveled and used WebRTC to create a downloadable enterprise-grade communications platform that is changing the communication dynamic in the financial sector. In his session at @ThingsExpo, Leo Papadopoulos, CTO of Cloud9, discussed the importance of WebRTC and how it enables companies to focus...
    The best-practices for building IoT applications with Go Code that attendees can use to build their own IoT applications. In his session at @ThingsExpo, Indraneel Mitra, Senior Solutions Architect & Technology Evangelist at Cognizant, provided valuable information and resources for both novice and experienced developers on how to get started with IoT and Golang in a day. He also provided information on how to use Intel Arduino Kit, Go Robotics API and AWS IoT stack to build an application tha...
    With an estimated 50 billion devices connected to the Internet by 2020, several industries will begin to expand their capabilities for retaining end point data at the edge to better utilize the range of data types and sheer volume of M2M data generated by the Internet of Things. In his session at @ThingsExpo, Don DeLoach, CEO and President of Infobright, discussed the infrastructures businesses will need to implement to handle this explosion of data by providing specific use cases for filterin...
    Is your aging software platform suffering from technical debt while the market changes and demands new solutions at a faster clip? It’s a bold move, but you might consider walking away from your core platform and starting fresh. ReadyTalk did exactly that. In his General Session at 19th Cloud Expo, Michael Chambliss, Head of Engineering at ReadyTalk, will discuss why and how ReadyTalk diverted from healthy revenue and over a decade of audio conferencing product development to start an innovati...
    So, you bought into the current machine learning craze and went on to collect millions/billions of records from this promising new data source. Now, what do you do with them? Too often, the abundance of data quickly turns into an abundance of problems. How do you extract that "magic essence" from your data without falling into the common pitfalls? In her session at @ThingsExpo, Natalia Ponomareva, Software Engineer at Google, provided tips on how to be successful in large scale machine learning...
    Early adopters of IoT viewed it mainly as a different term for machine-to-machine connectivity or M2M. This is understandable since a prerequisite for any IoT solution is the ability to collect and aggregate device data, which is most often presented in a dashboard. The problem is that viewing data in a dashboard requires a human to interpret the results and take manual action, which doesn’t scale to the needs of IoT.
    What does it look like when you have access to cloud infrastructure and platform under the same roof? Let’s talk about the different layers of Technology as a Service: who cares, what runs where, and how does it all fit together. In his session at 18th Cloud Expo, Phil Jackson, Lead Technology Evangelist at SoftLayer, an IBM company, spoke about the picture being painted by IBM Cloud and how the tools being crafted can help fill the gaps in your IT infrastructure.
    "C2M is our digital transformation and IoT platform. We've had C2M on the market for almost three years now and it has a comprehensive set of functionalities that it brings to the market," explained Mahesh Ramu, Vice President, IoT Strategy and Operations at Plasma, in this SYS-CON.tv interview at @ThingsExpo, held June 7-9, 2016, at the Javits Center in New York City, NY.
    "delaPlex is a software development company. We do team-based outsourcing development," explained Mark Rivers, COO and Co-founder of delaPlex Software, in this SYS-CON.tv interview at 18th Cloud Expo, held June 7-9, 2016, at the Javits Center in New York City, NY.
    Whether your IoT service is connecting cars, homes, appliances, wearable, cameras or other devices, one question hangs in the balance – how do you actually make money from this service? The ability to turn your IoT service into profit requires the ability to create a monetization strategy that is flexible, scalable and working for you in real-time. It must be a transparent, smoothly implemented strategy that all stakeholders – from customers to the board – will be able to understand and comprehe...
    Traditional IT, great for stable systems of record, is struggling to cope with newer, agile systems of engagement requirements coming straight from the business. In his session at 18th Cloud Expo, William Morrish, General Manager of Product Sales at Interoute, outlined ways of exploiting new architectures to enable both systems and building them to support your existing platforms, with an eye for the future. Technologies such as Docker and the hyper-convergence of computing, networking and sto...
    SYS-CON Events announced today that LeaseWeb USA, a cloud Infrastructure-as-a-Service (IaaS) provider, will exhibit at the 19th International Cloud Expo, which will take place on November 1–3, 2016, at the Santa Clara Convention Center in Santa Clara, CA. LeaseWeb is one of the world's largest hosting brands. The company helps customers define, develop and deploy IT infrastructure tailored to their exact business needs, by combining various kinds cloud solutions.
    The cloud market growth today is largely in public clouds. While there is a lot of spend in IT departments in virtualization, these aren’t yet translating into a true “cloud” experience within the enterprise. What is stopping the growth of the “private cloud” market? In his general session at 18th Cloud Expo, Nara Rajagopalan, CEO of Accelerite, explored the challenges in deploying, managing, and getting adoption for a private cloud within an enterprise. What are the key differences between wh...
    It’s 2016: buildings are smart, connected and the IoT is fundamentally altering how control and operating systems work and speak to each other. Platforms across the enterprise are networked via inexpensive sensors to collect massive amounts of data for analytics, information management, and insights that can be used to continuously improve operations. In his session at @ThingsExpo, Brian Chemel, Co-Founder and CTO of Digital Lumens, will explore: The benefits sensor-networked systems bring to ...
    SYS-CON Events announced today the Enterprise IoT Bootcamp, being held November 1-2, 2016, in conjunction with 19th Cloud Expo | @ThingsExpo at the Santa Clara Convention Center in Santa Clara, CA. Combined with real-world scenarios and use cases, the Enterprise IoT Bootcamp is not just based on presentations but with hands-on demos and detailed walkthroughs. We will introduce you to a variety of real world use cases prototyped using Arduino, Raspberry Pi, BeagleBone, Spark, and Intel Edison. Y...
    Large scale deployments present unique planning challenges, system commissioning hurdles between IT and OT and demand careful system hand-off orchestration. In his session at @ThingsExpo, Jeff Smith, Senior Director and a founding member of Incenergy, will discuss some of the key tactics to ensure delivery success based on his experience of the last two years deploying Industrial IoT systems across four continents.
    Much of IT terminology is often misused and misapplied. Modernization and transformation are two such terms. They are often used interchangeably even though they mean different things and have very different connotations. Indeed, it is somewhat safe to assume that in IT any transformative effort is likely to also have a modernizing effect, and thus, we can see these as levels of improvement efforts. However, many businesses are being led to believe if they don’t transform now they risk becoming ...
    SYS-CON Events announced today that Venafi, the Immune System for the Internet™ and the leading provider of Next Generation Trust Protection, will exhibit at @DevOpsSummit at 19th International Cloud Expo, which will take place on November 1–3, 2016, at the Santa Clara Convention Center in Santa Clara, CA. Venafi is the Immune System for the Internet™ that protects the foundation of all cybersecurity – cryptographic keys and digital certificates – so they can’t be misused by bad guys in attacks...