Wind River customers often ask questions about our Device Management products ranging from product structure and feature availability to deeply technical issues. Answers are based on the 2.2 release.
Do Wind River Field Diagnostics and Wind River Lab Diagnostics require the device application to have the symbol table embedded in it for Sensorpoint instrumentation?
No, not as a turnkey solution. However, these products are designed to be portable so that Wind River Professional Services can port to the operating system (OS) platform of your choice in a timely manner.
Can Sensorpoints be applied to instrument Java applications?
Not now, but this is being considered for a future release.
Can Field Diagnostics and Lab Diagnostics be applied to an application that doesn’t run VxWorks or Wind River Linux?
No, not as a turnkey solution. These products are designed to be portable so that Wind River Professional Services can port to the OS platform of your choice in a timely manner.
How does Wind River differentiate diagnostics from debugging with a debugger?
We differentiate between freeze-mode debugging and dynamic diagnostics. Where freeze-mode debug requires part of the system to be frozen, changing the timing characteristics of the system, dynamic analysis with Lab Diagnostics enables developers to access the application in a minimally intrusive manner, preserving real-world system timing. Also, the Sensorpoint technology allows many capabilities that debuggers do not, such as patching a live system without rebuilding.
What does Wind River mean by the term “device”?
It refers to an embedded device or an embedded system.
Can the Lab Diagnostics and Field Diagnostics data logging framework interface with customers’ existing data logging framework?
No, but this is being considered for a future release.
How much will the kernel footprint increase when the run-time components (Device Management agent and Sensorpoint Manager) are added?
The run-time components must be included in the kernel, so the kernel footprint will increase by 100KB to 200KB depending on configuration.
What kind of networking stack enables communication between devices and Lab Diagnostics or Field Diagnostics tools?
TCP/UDP/IP over Ethernet enables communication.
What language are Sensorpoints written in?
Sensorpoints are written in ANSI-C with custom directives. In order to create a Sensorpoint binary, the Sensorpoint toolchain has a two-step compile process. The first step translates custom directives into C; the second step generates the binary via GNU or Wind River Compiler.
Can Sensorpoints have negative impacts?
Sensorpoints are developed by developers; therefore the developer has complete control of what Sensorpoints do. Sensorpoint binaries have to be rigorously tested, especially if they are going to be used in the field.
Can Sensorpoints be used to force execution of exception handlers or branches that rarely execute?
Yes. Sensorpoints enable engineers to completely control the execution of the device application.
Can testers and support engineers deploy Sensorpoint binaries safely?
With Lab Diagnostics and Field Diagnostics, domain experts can safely use Sensorpoint binaries (created and tested by developers) in the lab and field environments.
How does Field Diagnostics secure data exchange between the end user site and OEM site?
Security happens at three levels: technology, policies, and personnel. Wind River enables Transport Layer Security (TLS) and Secure Sockets Layer (SSL) for point-to-point encrypted data exchange. For policies, the server application requires login and password, and end user control of data exchange. For personnel, the server apps only permit access to those who are authorized.
Can Field Diagnostics transfer data if the end user location has a firewall?
Yes. Field Diagnostics server applications are firewall-friendly. They work with the user’s existing IT infrastructure.
Do Lab Diagnostics and Field Diagnostics support a data format that helps the customer organize and exchange information?
Yes. Sensorpoint logs can be converted to an XML-based file format. To achieve performance and data storage requirements, Sensorpoint logs contained within a device are in binary form. The binary to XML file conversion takes place in Workbench Diagnostics, not on the device.
What is the overhead?
There is a performance benchmark. Please request it from your Wind River technical account manager.
Can you use Sensorpoints to skip a function?
Yes. There is a macro to skip the execution of a function.
What is the Site Manager?
The Site Manager is a server application for managing data flow to and from devices. In a Lab Diagnostics environment, it would be installed in the QA lab if there are multiple labs. If only one lab exists, only the Device Management server is needed. In a Field Diagnostics implementation, it would be installed at an end customer site.
In some cases, the end user will not allow high bandwidth connection between their site and the OEM site. Will Field Diagnostics work in that situation?
Field Diagnostics can be configured without high bandwidth connectivity. In that situation, data can be transferred onto DVD-ROM or a CD-ROM, or transferred by another secure mechanism.
What are some example use cases for Sensorpoints?
These white papers contain extensive information about using Sensorpoints:
Mapping Wind River Lab Diagnostics to the Device Software Verification and Validation Process
Mapping Field Diagnostics to the Customer Support Process
Can you place multiple Sensorpoints at the same point in the application binary?
Yes.
What if the end user does not allow the OEM to install the Field Diagnostics server application onsite?
In this case the Site Manager application can be installed on the laptop of the OEM’s field engineer and can be temporarily brought onsite when deep diagnostics are needed.
Can logging calls from Sensorpoints be preempted by higher priority tasks?
No. Unlike printf(), the logging call from the Sensorpoint is non-preemptable.
Do Sensorpoints persist once they’re on a device and the device reboots?
Yes. They can persist, or not, by your choice.
What logs are supported with Lab Diagnostics and Field Diagnostics?
Sensorpoint logs, VxWorks 6 kernel exception logs, and VxWorks 6 core files are supported.
What is the difference between the Device Management agent and the Workbench Debugger agent?
The Device Management agent is a new target agent specifically designed for diagnostics of running devices or systems. The Workbench Debugger agent is designed for standard freeze-mode debugging.
Can the Device Management agent be dynamically downloaded to the target?
Yes, with Lab Diagnostics 2.2, users can update the Device Management agent.
When is the Device Management agent initialized?
Partial initialization is performed during kernel initialization when other components are initialized. Final initialization is performed after VxWorks or Wind River Linux is up.
Is the Device Management agent aware that VxWorks 6 core images exist on the target?
Yes. The Device Management agent manages VxWorks 6 core images, EDnR logs, and Sensorpoint logs.
Can users dynamically instrument C++ applications?
Yes. Instrumentation of C++ applications is fully supported.
Can users dynamically instrument VxWorks 6 RTP applications?
No. In version 2.2, users cannot instrument applications in Real-Time Transport Protocols (RTPs). However, this feature is being considered for a future release.
Can users dynamically instrument the VxWorks kernel and kernel applications?
Yes. Sensorpoints can be used to instrument kernel application functions, kernel functions, interrupt service routines (ISRs), and device drivers.
Can Sensorpoints be used to instrument the boot sequence or board support package (BSP) initialization?
No. The Device Management agent is not initialized until later in the boot sequence.
Can users simultaneously use debugger breakpoint and Sensorpoint instrumentation?
Yes. Users must follow a specific sequence that is described in the product documentation.
After a crash and reboot sequence, will Sensorpoints that were enabled before the crash automatically enable themselves?
Yes. In Lab Diagnostics 2.2, persistent Sensorpoints are supported.
Does Workbench Debugger fully support core file analysis?
Yes. With Workbench Diagnostics, users can attach the Workbench Debugger to a core file that exists on the host file system. Users can analyze a core file with the debugger in read-only mode.
Is a target required when analyzing a core file?
Once a core file is extracted from the target to the host, a target is not needed to analyze the core file.
Can users configure core dump generation to accommodate targets with limited resources?
Yes. Users can filter out memory regions, set the number of core images that are stored locally on the target, and/or compress core images to manage the size of core files. In addition, users can configure the target to stream the core image off of the target eliminating the need for local storage.
