See bad free.
A relation between two entities in a program depicted graphically as lines between rectangles (nodes). For example, arcs can represent function calls, file dependency, or inheritance.
A Debugger view that displays the values of an array in a spreadsheet format and can also depict them graphically in a 3D rendering.
A problem that occurs when a program frees a malloced piece of memory that it had already freed (also referred to as an anti-leak condition or double free).
A display mode of Tester that shows a summary of coverage information in a bar graph.
A block of machine-level instructions used as a metric in Performance Analyzer and Tester experiments. A basic block is the largest set of consecutive machine instructions that can be formed with no branches into or out of them.
A problem that occurs when a program writes beyond a specified region, for example overwriting the end of an array or a malloced structure.
A problem that occurs when a program writes in front of a specified region, for example, writing ahead of the first element in an array or a malloced structure.
See trap (breakpoint) and watchpoint (data-breakpoint).
A facility within the Static Analyzer for viewing structural and relationship information in C++ programs. It provides three views: Browser View for displaying member and class information; Class Graph for displaying inheritance, containment, interaction, and friend relationships in the hierarchy; and Call Graph for displaying the calling relationships of methods, virtual methods, and functions.
A tool that displays a graph of program files (source and object) indicating build dependencies and provides access to the source files.
A tool for recompiling programs within WorkShop. The Build Manager has two windows: Build Analyzer and Build View.
A view that lets you run compiles. In addition, Build View displays compile errors and provides access to the code containing the errors.
See time line.
A generic term for views used in several tools (Static Analyzer, C++ Browser, Performance Analyzer, and Tester) that display a graph of the calling hierarchy of functions. Double-clicking a function in a call graph causes the Source View window to be displayed showing the function's source code.
A display mode of the C++ Browser that shows methods and their calls. See also call graph and C++ Browser.
A Performance Analyzer view that shows functions, their calls, and associated performance data. See also call graph and C++ Browser.
A view that displays the call stack at the current context. In the Debugger this means where the process is stopped; in the Performance Analyzer this means sample traps and other events where data was written out to disk. Each frame in the Call Stack view can show the function; argument names, values, and types; the function's source file and line number; and the PC (program counter). Double-clicking a frame in the Call Stack view causes the Source View window to be displayed showing the corresponding source code.
A Static Analyzer view that displays the results of function queries as a call graph. See also call graph and Static Analyzer.
A Tester view that displays function coverage information in a call graph. See also Tester.
A C++ Browser view for displaying member and class information. See also C++ Browser.
A C++ Browser view for displaying inheritance, containment, interaction, and friend relationships in the class hierarchy.
A Static Analyzer view that displays the results of class queries as a class hierarchy. See also Static Analyzer.
A field in the Debugger Main View that lets you enter a set of commands similar to dbx commands.
A system command used to rearrange procedures in an executable file to reduce paging and achieve better instruction cache mapping. The Cord Analyzer and Working Set View let you analyze the effectiveness of an arrangement and try out new arrangements to improve efficiency.
A tool that lets you analyze the paging efficiency of your executable's working sets, that is, the executable code brought into memory during a particular phase or operation. It also calculates an optimized ordering and lets you try out different working set configurations to reduce paging problems. The Cord Analyzer works with the Working Set View, a part of the Performance Analyzer. See also cord, working set, and Working Set View.
The number of times a piece of code (function, line, instruction, or basic block) was executed as listed by Tester or the Performance Analyzer.
A term used in Tester. Coverage means a test has exercised a particular unit of source code, such as functions, individual source lines, arcs, blocks, or branches. In the case of branches, coverage means the branch has been executed under both true and false conditions.
A performance analysis term for a condition in which a process spends its time in the CPU and is limited by CPU speed and availability.
A performance analysis metric approximating the time spent in the CPU. CPU time is calculated by multiplying the number of times a PC appears in the profile of a function, source line, or instruction by 10 ms.
The name of the Cord Analyzer executable. See also Cord Analyzer.
The name of the Tester command line interface executable. See also Tester.
The name of the Debugger executable file. cvd has options for attaching the Debugger to a running process (-pid ), examining core files (executable), and running from a remote host (-host). See also Debugger.
The name of the executable file that calls the Performance Analyzer. cvperf has an option (-exp) for designating the name of the experiment directory. See also Performance Analyzer.
The name of the executable file that brings up the Performance Panel, a window for setting up Performance Analyzer experiments. See also Performance Panel.
The name of the executable file that calls the Static Analyzer. See also Static Analyzer.
The name of the executable file that calls the graphical interface of Tester. See also Tester.
The specified number of times to hit a breakpoint before stopping the process, it defaults to one. The cycle count for any trap can be set through the Trap Manager view in the Debugger.
A Debugger view that graphically displays data structures including data values and pointer relationships.
A tool in the ProDev WorkShop toolkit used for analyzing general software problems using a live process. The Debugger lets you stop the process at specific locations in the code by setting breakpoints (referred to as traps) or by clicking the Stop button. At each trap, you can examine data by displaying special windows called views. See also cvd.
A view that lets you see the program's machine-level code. The Debugger version shows you the code; the Performance Analyzer version additionally displays performance data for each line.
See bad free.
An ELF (Executable and Linking Format) format object file, similar in structure to an executable program but with no main. It has a shared component, consisting of shared text and read-only data; a private component, consisting of data and the GOT (Global Offset Table); several sections that hold information necessary to load and link the object; and a liblist, the list of other shared objects referenced by this object. Most of the libraries supplied by Silicon Graphics are available as dynamic shared objects.
A problem that occurs when a program calls free() on addresses that were not returned by malloc, such as static, global, or automatic variables, or other invalid expressions.
An action that takes place during a process, such as a function call, signal, or a form of user interaction. The Performance Analyzer uses event tracing in experiments to help you correlate measurements to points in the process where events occurred.
Performance Analyzer data collected for a function without including the data for any functions it calls. See also inclusive performance data.
A Debugger view that serves as a simple shell to provide access outside of the WorkShop environment. It is typically used to set environment variables, inspect error messages, and conduct I/O with the program being debugged.
The model for using the Performance Analyzer and Tester. The steps in creating an experiment are (1) creating a directory to hold the results, (2) instrumenting the executable (instrumentation is recompiling with special libraries for collecting data), (3) running the instrumented executable as a test, and (4) analyzing the results using the views in the tools. The first two steps are done automatically when you use the Performance Panel and select a performance task (performance experiments only). The term experiment can also refer to the actual data itself that was saved.
A Debugger view that lets you specify one or more expressions to be evaluated whenever the process stops or the callstack context is changed. Expression View lets you save sets of expressions for subsequent reuse, specify the language of the expression (Ada, Fortran 77, Fortran 90, C, or C++), and specify the format for the resulting values.
A Static Analyzer view that displays the results of queries in a graph indicating file dependency relationships. See also Static Analyzer.
A window for specifying a fileset, that is, the set of files to be used in creating a database for Static Analyzer queries. The Fileset Editor also lets you specify whether a file is to be analyzed using scanner mode or parser mode. See also parser mode, scanner mode, and Static Analyzer.
A technique in performance analysis that captures resource usage data between sample traps.
A feature in the Debugger that lets you make source level changes and continue debugging without having to perform a full compile and relinking.
A problem that occurs when a program cannot complete a numerical calculation due to division by zero, overflow, underflow, inexact result, or invalid operand. Floating-point exceptions can be captured by the Performance Analyzer and can also be identified in the Array Browser.
Freed memory is memory that was originally malloced and has been returned for general use by calling free(). Accessing freed memory is a problem that occurs when a program attempts to read or write this memory, possibly corrupting the free list maintained by malloc .
A generic type of view used in several tools (Static Analyzer, Performance Analyzer, Tester, and Cord Analyzer) to list functions and related information, such as location, experiment data, and executable code size. Double-clicking a function displays its source code in Source View.
A graphical software tool for debugging application programs that use the IRIS Graphics Library (GL). GLDebug locates programming errors in executables when GL calls are used incorrectly. GLDebug is not part of WorkShop but is accessible from the Admin menu in Main View.
A memory problem that may be due to boundary overrun or underrun, accessing uninitialized memory, accessing freed memory, freeing a memory location twice, or attempting to free a memory location erroneously. See also malloc debugging library.
A Performance Analyzer view that displays a map of memory indicating how blocks of memory were used in the time interval set by the time line calipers.
A performance analysis metric that assumes that each instruction takes one cycle of the particular machine's time. It is then useful to compare the ideal time with the actual time in an experiment.
Performance Analyzer data collected for a function where the total includes data for all of the called functions. See also exclusive performance data.
See experiment.
A performance analysis term for a condition in which a process has to wait for I/O to complete and may be limited by disk access speeds or memory caching.
A Performance Analyzer view that displays a chart devoted to I/O system calls. I/O View can identify up to ten files involved in I/O.
A user interface toolkit on Silicon Graphics systems based on X/Motif.
A Debugger view for debugging X/Motif applications. The IRIS IM Analyzer lets you look at object data, set breakpoints at the object or X protocol level, trace X and widget events, and tune performance.
A toolkit that provides predefined widgets and classes for building applications.
A Performance Analyzer view that displays each memory leak that occurred in your experiment, its size, the number of times the leak occurred at that location during the experiment, and the call stack corresponding to the selected leak.
A path you may need to specify when debugging executables or core files to indicate which DSOs (dynamic shared objects) are required for debugging. See also DSO.
The main window of the Debugger. The MainView provides access to other tools and views, process controls, a source code display, and a command line for entering a set of commands similar to dbx. You can also add custom buttons to Main View using the command line.
A Performance Analyzer view that displays each malloc error (leaks and bad frees) that occurred in an experiment, the number of times the malloc occurred (a count is kept of mallocs with identical call stacks), and the call stack corresponding to the selected malloc error.
A special library (libmalloc_cv.a) for detecting heap corruption problems. Relinking your executable with the malloc library sets up mechanisms for trapping memory problems.
A Performance Analyzer view that displays each malloc (whether or not it caused a problem) that occurred in your experiment, its size, the number of times the malloc occurred (a count is kept of mallocs with identical call stacks), and the call stack corresponding to the selected malloc.
The package name for a set of advanced tools for the development of C and C++ applications.
A performance analysis term for a condition in which a process continuously needs to swap out pages of memory.
A problem when a program dynamically allocates memory and fails to deallocate that memory when it is through with the space.
A Debugger view that lets you see or change the contents of memory locations.
A Debugger view that lets you manage the debugging of a multiprocess executable. For example, you can set traps in individual processes or across groups of processes.
The rectangles in graphical views. A node may represent a function, class, or file depending on the type of graph.
A window in graphical views that displays the current graph at a reduced scale and lets you navigate to different parts of the graph.
A method of extracting Static Analyzer data from source files. Parser mode uses the compiler to build the Static Analyzer database. It is language-specific and very thorough; as a result, it is slower than scanner mode. See also scanner mode and Static Analyzer.
A dialog box that lets you set mappings to redirect filenames used in building your executable to their actual locations in the filesystem.
The current line in a stopped process, indicated by a right-pointing arrow with a highlight in the source code display areas and by a highlighted frame in the Call Stack views.
A tool in the ProDev WorkShop toolkit used for measuring the performance of an application. To use the tool, you select one of the predefined analysis tasks, run an experiment, and examine the results in one of the Performance Analyzer views. See alsocvperf.
A window for setting up Performance Analyzer experiments. The panel displays toggles and fields for specifying data to be captured. As a convenience, you can select performance tasks (such as Determine bottlenecks... or Find memory leaks) from a menu that specifies the data automatically. See also cvspeed(1).
A performance analysis term for a period in an experiment covering a single activity. In a phase, there is one limiting resource that controls the speed of execution.
A technique in performance analysis that captures performance data, such as resource usage or event tracing, at regular intervals.
A view that monitors the resource usage of a running process without saving the data. See also Performance Analyzer and Performance Panel.
The package name for the core WorkShop tools.
A record of a program's PC (program counter), call stack, and resource consumption over time, used in performance analysis.
A Debugger view for managing the ProDev WorkShop toolkit and MegaDev tools operating on a common target.
The term for a search through a Static Analyzer database to locate elements in your program. Queries are similar to the IRIX grep(1) command but provide a more specific search. For example, you can perform a query to find where a method is defined. See also Static Analyze
A Debugger view that lets you see or change the contents of the machine registers.
A dialog box that lets you limit the scope of Static Analyzer queries. See also query and Static Analyzer.
Similar to a stop trap except that instead of stopping the process, performance data is written out to disk and the process continues running. See also trap.
In performance analysis, the capture of performance data, such as resource usage or event tracing, at points in an experiment so that a graph of usage over time can be created.
A method of extracting Static Analyzer data from source files. Scanner mode is fast but not language-specific so that the source code need not be compliable. Results may have minor inaccuracies. See also parser mode and Static Analyzer.
A dialog box for specifying signals to trap.
An option to the compiler where only those files that must be recompiled are recompiled.
A window for viewing or editing source code. Source View is an alternative editing window to Main View. If you have conducted Performance Analyzer or Tester experiments, you can view the results in the column to the left of the source code display area.
See Call Stack.
A tool in the ProDev WorkShop toolkit used for viewing the structure of a program at different levels and locating where elements of the program are used or defined. The Static Analyzer works by extracting structure and location information from files that you specify and storing the information in a database for subsequent analysis. You can view the analysis as a text list or graphically. See alsocvstatic(1), Call Tree View, Class Tree View, File Dependency View, and Text View.
A breakpoint. See also trap.
A dialog box for specifying system calls to trap. You can designate whether to trap the system calls at the entry or exit from the call.
A grouping of experiments in Tester used to test a common DSO (dynamic shared object).
A group of experiments in Tester used to test a common executable file.
A tool in the ProDev WorkShop toolkit used for measuring dynamic coverage over a set of tests. It tracks the execution of functions, individual source lines, arcs, blocks, and branches. Tester has both a command line and a graphical interface.
A Static Analyzer view that displays the results of queries as a scrollable text list. See also Static Analyzer.
A Tester view that displays function coverage information in a report form. See also Tester.
A feature in the main Performance Analyzer window that shows where events occurred in an experiment and provides calipers for controlling the scope of analysis for the Performance Analyzer views.
A record of a specified type of event (such as reads and writes, system calls, page faults, floating-point exceptions, and mallocs, reallocs, and frees) over time, used in performance analysis.
A mechanism to allow the debugger to get control at specified points and conditions in a live process. More commonly referred to as a breakpoint (either a code breakpoint or a data-breakpoint [watchpoint]).
There are two types of traps: stop traps are used in debugging to halt a process, and sample traps are used in performance analysis to collect data while halting the process only briefly (and continuing execution automatically). See also watchpoint.
A window for managing traps. It lets you set simple or conditional traps, browse (or modify) a list of traps, and save or load a set of traps.
Memory that is allocated but not assigned any specific contents. Accessing uninitialized memory is a problem that occurs when a program attempts to read memory that has not yet been initialized with valid information.
A Performance Analyzer view that contains charts indicating resource usage and the occurrence of events, corresponding to time intervals set by the time line calipers.
A Performance Analyzer report that displays the actual resource usage values corresponding to time intervals set by the time line calipers.
A Debugger view that displays the local variables valid in the current context and their values (or addresses). The Variable Browser also lets you view the previous value at the breakpoint. You can enter a new value directly if you wish.
A window that lets you analyze data.
See IRIS ViewKit.
Commonly referred to as a data-breakpoint. A trap that fires when a specified variable or address is read or written.
The set of executable pages, functions, and instructions brought into memory during a particular phase or operation. See also Working Set View.
A Performance Analyzer view that lets you measure the coverage of the dynamic shared objects (DSOs) that make up your executable. It indicates instructions, functions, and pages that were not used in a particular phase or operation in an experiment. Working Set View works with the Cord Analyzer. See also working set and Cord Analyzer.