NightStar

NightView allows users to simultaneously debug multiple, time-critical processes. With NightView, a programmer can change program execution and modify or display data without stopping or interrupting the program. Eventpoint conditions, such as hit and ignore counts, are patched directly into an application and execute at full application speed. NightView provides fine-grained control without adversely affecting application deadlines.

NightView monitorpoints can display expressions, including function calls, at user-selected locations without stopping a process, thus providing data displays that are synchronized with the application's algorithms. Watchpoints utilize hardware address trap features that cause an application to stop when user-specified variables or memory locations are selectively read or modified.

Language-sensitive Debugging
NightView supports the debugging of multiple applications written in any combination of C/C++, Ada and Fortran. All variables and expressions in each program are referenced in its own language syntax. NightView is fully integrated with the NightTrace event analyzer. NightView can insert tracepoints at user-specified locations for live or post-execution analysis with NightTrace.

More Powerful Than The Gnu Debugger
NightView offers many features not available in gnu debuggers (gdb/ddd). NightView advantages include the ability to debug multiple processes from a single session and processes started from scripts. With NightView, patched-in code runs at full speed. While a process is executing, hot patching can modify variables, add eventpoints and add or replace subprograms. Monitorpoints can display expressions and stack variables, and signals can be delivered directly to the process, bypassing the debugger.

Complete Memory Debugging
NightView includes an interactive memory debugger that helps find and eliminate problems during the debug process without code recompilation or relinking. NightView watches for heap memory leaks, monitors the amount of memory an application uses, and tracks how it allocates and frees memory. With its memory debugger enabled, NightView lets users track heap transactions in real-time, thus allowing for more efficient debugging than post-run analysis. Programmers can stop execution, check for problems, test patches and then continue debugging.

NightView can detect common heap programming errors, such as uninitialized use of allocated areas, double-frees, use of freed pointers and writing past the end of allocated areas. No recompilation or relinking is required.

NightTrace is a tool for displaying and analyzing the dynamic behavior of applications, the Linux operating system and the interaction between them. NightTrace can log events from multiple processes executing simultaneously on multiple CPUs or systems. NightTrace can also combine user-defined application events with OS events to present a synchronized view of the entire system. NightTrace allows users to zoom, search, filter, summarize and analyze events. Tracing analysis can be performed live or post-execution.

NightTrace was specifically designed to meet the most stringent requirements of time-critical applications. Using synchronized, fast-access hardware clocks and kernel-free primitives, NightTrace tracepoints are logged with minimal overhead. Tracepoints can be inserted into device drivers, interrupt level code and any user application. Tracepoints can be left in production applications even when not collecting trace data. In distributed applications, multiple systems can be traced when connected via Concurrent's Real-Time Clock & Interrupt Module (RCIM).

NightTrace also provides instrumentation for user programs to automatically display and log the details of their activity including function call parameters, entry, return and return values.

Graphical and Interactive
NightTrace graphically displays requested events and states along a timeline graph to clearly show the relative timing of events and provide an overall picture of application and operating system activity. NightTrace can locate specific events and zoom in on them with a fine degree of granularity for precise timing observation.

The NightTrace graphical display is completely user-configurable and customizable. Configurations can be saved and later recalled, and different configurations can be viewed simultaneously.

Automatic Function Call Tracing
NightTrace’s application illumination feature allows programmers to automatically trace application function calls and examine the values of parameters passed and returned. Function call tracing is fully customizable and can provide a complete view of glibc activities.

Kernel Trace Support
By combining system event information such as interrupts, exceptions, context switches, system calls and device accesses together with event information from user applications, NightTrace provides a clear picture of the activities of the kernel at any point during the application's run.

NightTrace provides statistical performance data about events and states, including frequency, time of occurrence, duration, gap and minimum and maximum times. Users can create state definitions and qualify events by specifying the applicable process, thread, CPU, system and event content. Conditional tracing can be expressed using C syntax. Displays can be customized to yield insight into operating system and application performance and behavior.

NightTrace's Analysis API allows users to easily create custom programs that monitor or analyze application or system activity.

NightProbe is a tool for monitoring, modifying and recording data values from multiple, independent application resources, including programs, shared memory segments, memory mapped files and PCI devices. NightProbe can be used in development for debugging, analysis, prototyping and fault injection, or in a production environment to create a GUI control panel for program input and output.

NightProbe utilizes a non-intrusive technique of mapping the target resource's address space into its own. Subsequent direct memory reads and writes by NightProbe allow it to sample and modify data without interrupting or otherwise affecting resources.

Synchronized and Asynchronous Logging
NightProbe can perform synchronous logging of data via a simple API. Asynchronous logging can be performed via on-demand sampling or a cyclic clock rate.

NightProbe allows logging of data items using tracepoints for simultaneous analysis by the NightTrace event analyzer. Sampled data can be combined with kernel trace and additional user trace data to obtain a synchronized picture of application and OS behavior. NightProbe can also record data to disk files for subsequent playback.

Interactive Sampling and Modification
NightProbe offers a variety of data display options including lists, tables, spreadsheets and graphs. Direct modification of user data is accomplished by simply typing new data item values into cells that display the variables. NightProbe allows sampled data to be passed off to user applications for live customized display or for post-run analysis.

NightProbe supports scalar and structured data types in C/C++, Ada and Fortran that have static base addresses. Pointer indirection is also supported. NightProbe scans the symbol table and debug information of user programs allowing the user to browse for data items or specifically entering the names of data items to be monitored. Any application that contains symbol table and debug information may be used with NightProbe. No application source code changes are required.

NightSim is a tool for scheduling and monitoring time-critical applications that require predictable, cyclic process execution. Ideal for simulation applications, NightSim allows developers to dynamically adjust the execution of multiple, coordinated processes, their priorities, scheduling policies and CPU assignments. Users can monitor the performance of applications by displaying period execution times, minimums and maximums, and optionally pause all application execution when any process overruns its allocated frame.

NightSim provides a GUI to the operating system's Frequency-Based Scheduler (FBS), a high-resolution task scheduler that enables processes to run cyclically. NightSim lets users easily configure groups of processes to run on local or distributed systems, and save the resulting configurations for reuse. A performance monitor gathers CPU utilization data for processes running under the FBS.

NightSim may be used during the development, debug and production phases of a simulation model. Configurations can be saved as a script and then re-executed to repeat a simulation. NightSim scripts are useful in target environments where GUI processing is prohibited or undesired.

Synchronized Distributed Scheduling
In addition to symmetric multiprocessors, NightSim supports multiple systems connected via a Concurrent RCIM. NightSim simplifies the creation of distributed scheduling and provides a single-point-of-control for individual schedulers distributed across multiple target systems. NightSim handles the interface to hardware such as real-time clocks and distributed interrupt sources. Users don't need to interface with the underlying OS for scheduling operations.

Extensive Performance Statistics
NightSim monitors up to 14 different performance statistics for each process, such as minimum and maximum cycle times. Using these statistics, users can optimize CPU utilization by balancing their load across multiple processors. NightSim displays are customizable, allowing users to select specific statistics and processes to monitor and sorting criteria for weighted display.

NightTune provides a GUI for monitoring and tuning application and system performance. Users can monitor the priority, scheduling policy, CPU assignment and CPU usage of user applications. NightTune also monitors system CPU usage, context switches, interrupts, memory utilization, and disk and network activity.

NightTune can monitor processes individually or in user-specified groups or by CPU. NightTune also displays information about individual threads or tasks within a process. Available information includes library and system call tracking (strace), file descriptor usage and connections, signal states, and detailed memory usage with page-level statistics on residency, locking and NUMA pools. Multiple frames and windows are used to display information allowing users to customize the desired display.

System Tuning by GUI
NightTune's GUI can change the process attributes of an individual thread, task, process or group of processes as a whole. For example, dragging a process icon to a CPU icon binds the process to that processor. The user instantly views the results of the tuning effort both graphically and as text. NightTune provides a hierarchial display of logical CPUs showing multiple cores and hyperthreading. Users can also easily change CPU interrupt assignments.

NightTune optionally provides a textual log of all application and system tuning actions taking during a NightTune session

NightStar is available for Concurrent RedHawk Linux, Novell SUSE Linux Enterprise Real Time, SUSE Linux Enterprise Server, SUSE Linux Enterprise Desktop and Red Hat Enterprise Linux. Not all features are supported on all Linux distributions. Please consult NightStar Versions for details.