DOCSLIB.ORG

User Memory Management

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
User Memory Management Memory management Types of information information stored in memory n Kernel management (next lecture) n By role in program: n User-process memory management u Program instructions u internal to process u Constants: F segments F pi, maxnum, strings used by printf/scanf F static allocation u Variables: F dynamic allocation F Locals, globals, function parameters, dynamic storage (from F allocation methods malloc or new) F reclaiming: garbage collection F Initialized or uninitialized u external n By protection status (important for sharing data and code): F stages in address binding u Readable and writable: variables F static relocation u Read-only: code, constants F dynamic relocation n Addresses vs. data: • swapping u Must modify addresses if program is moved (relocation, garbage • compaction collection) F paging (next lecture but one) 1 2 Segments Internal dynamic memory allocation n Static allocation is not sufficient for heaps and stacks n Process’ memory is divided into logical segments (text, data, bss, u Need dynamic storage — programmer may not know how heap, stack) much memory will be needed when program runs u Some are read-only, others read-write F Use malloc or new to get what’s necessary when it’s u Some are known at compile time, others grow dynamically as necessary program runs F For complex data structures (e.g., trees), allocate space n Who assigns memory to segments? for nodes on demand u Compiler and assembler generate an object file (containing u OS doesn’t know in advance which procedures will be called code and data segments) from each source file (would be wasteful to allocate space for every variable in u Linker combines all the object files for a program into a single every procedure in advance) executable object file, which is complete and self-sufficient u OS must be able to handle recursive procedures u Loader (part of OS) loads an executable object file into memory n Dynamic memory requires two fundamental operations: at location(s) determined by the operating system u Allocate dynamic storage u Program (as it runs) uses and to dynamically new malloc u Free memory when it’s no longer needed allocate memory, gets space on stack during function calls n Methods vary for stack and heap 3 4 Memory allocation methods Internal fragmentation n Stack (hierarchical) u Good when allocation and freeing are somewhat predictable n eventually end up with many small holes u Typically used: (fragments), each too small to be useful F to pass parameters to procedures u fragmentation, and it leads to wasted F for allocating space for local variables inside a procedure memory F for tree traversal, expression evaluation, parsing, etc. u Fragmentation wasn’t a problem with u Use stack operations: push and pop stack allocation, since we always add/delete from top of stack u Keeps all free space together in a structured organization u Solution goal: reuse the space in the u Simple and efficient, but restricted holes in such a way as to keep the n Heap number of holes small, and their size large u Used when allocation and freeing are not predictable n Compared to stack: more general, less u used for arbitrary list structures, complex data organizations, efficient, more difficult to implement etc. u Use new (malloc) to allocate space, use delete or free to release space u System memory consists of allocated areas and free areas 5 6 (holes) Heap memory allocation Reclaiming memory n Heap-based dynamic memory allocation techniques n typically maintain a free list, which keeps track of all the holes When can memory be freed? u Whenever programmer says to n Algorithms to manage the free list: u u Best fit Any way to do so automatically? F F Keep linked list of free blocks Difficult if that item is shared (i.e., if there are multiple pointers to it) F Search the whole list at each allocation n Potential problems in reclamation F Choose the hole that comes the closest to matching the request u is finished size; any unused space becomes a new (smaller) hole Dangling pointers — have to make sure that everyone using it F When freeing memory, combine adjacent holes u Memory leak — must not “lose” memory by forgetting to free it F Any way to do this efficiently? when appropriate u First fit n Implementing automatic reclamation: F Scan the list for the first hole that is large enough, choose that hole u Reference counts F Otherwise, same as best fit F Used by file systems u Worst fit F OS keeps track of number of outstanding pointers to each F scan for the largest hole (hoping that the remaining hole will be memory item large enough to be useful) F When count goes to zero, free the memory u Which is better? Why?? 7 8 Garbage collection Stages of address binding n u Garbage collection binding - assigning memory addresses to program objects (variables, pointers, etc.) F Used in LISP, Java n physical address - the “hardware” address F Storage isn’t explicitly freed by a free operation; programmer of a memory word (0xb0000) just deletes the pointers and doesn’t worry about what it’s n pointing at relocatable address - relative address (14 bytes from the beginning of this module) F When OS needs more storage space, it recursively searches n absolute code - all addresses are physical through all the active pointers and reclaims memory that no one is using n relocatable code - all addresses are relative F Makes life easier for application programmer, but is difficult to n linking - preparing compiled program to run program the garbage collector u static - all lib-functions included in the F Often expensive — may use 20% of CPU time in systems that code use it u dynamic - lib-functions can be hooked • May spend as much as 50% of time allocating and up at load-time automatically freeing memory n loading - a program into memory u static - all at once u 9 dynamic - on demand 10 Linking Why is linking difficult n Functions of a linker: n When assembler assembles a file, it may find external references u Combine all files and libraries of a program — symbols it doesn’t know about (e.g., printf, scanf) u Regroup all the segments from each file together (one big data u Compiler just puts in an address of 0 when producing the object segment, etc.) code u Adjust addresses to match regrouping u Compiler records external symbols and their location (in object file) in a cross-reference list, and stores that list in the object file u Result is an executable program u Linker must resolve those external references as it links the n Contents of object files: files together u File header — size and starting address (in memory) of each n segment Compiler doesn’t know where program will go in memory (if u multiprogramming, always 0 for uniprogramming) Segments for code and initialized data u u Compiler just assumes program starts at 0 Symbol table (symbols, addresses) u u Compiler records relocation information (location of addresses External symbols (symbols, location) to be adjusted later), and stores it in the object file u Relocation information (symbols, location) n for details type man ld u Debugging information u For UNIX details, type man a.out 11 12 Loading Goals in memory usage for multiprogramming n The loader loads the completed program into memory where it can be executed u Loads code and initialized data segments into memory at n Transparency: specified location u Multiple processes must coexist in memory u Leaves space for uninitialized data (bss) u No process should be aware that the memory is shared u Returns value of start address to operating system u Each process should execute regardless of where it is n Alternatives in loading located in memory u Absolute loader — loads executable file at fixed location n Safety: u Relocatable loader — loads the program at an arbitrary u Processes must not be able to corrupt each other, or the OS memory location specified by OS u Protection mechanisms are used to enforce safety u Assembler and linker provide relocatable addreses, loader n Efficiency: translates them to physical addresses u The performance of the CPU and memory should not F When program is loaded, loader modifies all addresses degrade very much as a result of sharing by adding the real start location to those addresses 13 14 Static external relocation Static external relocation (cont.) n Problems with static relocation: u Safety — not satisfied — one process can access / corrupt n Put the OS in the highest another’s memory, can even corrupt OS’s memory main main memory u Processes can not change size (why…?) address memory address memory 2400 2400 n Compiler and linker u Processes can not move after beginning to run (why would they OS OS assume each process want to?) 2200 2200 1900 starts at address 0 u Used by MS-DOS, Windows, Mac OS n B At load time, the OS: n An alternative: dynamic relocation u Allocates the process u The basic idea is to change each memory address dynamically as 1200 1200 a segment of memory the process runs in which it fits u This translation is done by hardware — between the CPU and the completely A A memory is a memory management unit (MMU) (also called a u Adjusts the translation unit ) that converts virtual addresses to physical addresses in the addresses 0 0 processes to reflect F This translation happens for every memory reference the its assigned location process makes in memory 15 16 Dynamic relocation Dynamic relocation (cont.) n There are now two different views of the address space: u The physical address space — seen only by the OS — is as large as there is physical memory on the machine u The virtual
Load more

Recommended publications
  • Ben Livshits 1 Basic Instrumentation
    Runtime monitoring CO444H Ben Livshits 1 Basic Instrumentation • Insert additional code into the program • This code is designed to record important events as they occur at runtime • Some examples • A particular function is being hit or a statement is being hit • This leads to function-level or line-level coverage • Each allocation to measure overall memory allocation 2 Levels of Instrumentation • Native code • Instrument machine code • Tools like LLVM are often used for rewriting • Bytecode • Common for languages such as Java and C# • A variety of tools are available for each bytecode format • JoeQ is in this category as well, although it’s a lot more general • Source code • Common for languages like JavaScript • Often the easiest option – parse the code and add more statements 3 Runtime Code Monitoring •Three major examples of monitoring • Purify/Valgrind • Detecting data races • Detecting memory leaks 4 5 Memory Error Detection Purify • C and C++ are not type-safe • The type system and the runtime fail to enforce type safety • What are some of the examples? • Possible to read and write outside of your intended data structures • Write beyond loop bounds • Or object bounds • Or overwrite the code pointer, etc. 6 Track Each Byte of Memory • Three states for every byte of tracker memory • Unallocated: cannot be read or written • Allocated but not initialized: cannot be read • Allocated and initialized: all operations are allowed 7 Instrumentation for Purify • Check the state of each byte at every access • Binary instrumentation: • Add
    [Show full text]
  • Memory Leak Or Dangling Pointer
    Modern C++ for Computer Vision and Image Processing Igor Bogoslavskyi Outline Using pointers Pointers are polymorphic Pointer “this” Using const with pointers Stack and Heap Memory leaks and dangling pointers Memory leak Dangling pointer RAII 2 Using pointers in real world Using pointers for classes Pointers can point to objects of custom classes: 1 std::vector<int> vector_int; 2 std::vector<int >* vec_ptr = &vector_int; 3 MyClass obj; 4 MyClass* obj_ptr = &obj; Call object functions from pointer with -> 1 MyClass obj; 2 obj.MyFunc(); 3 MyClass* obj_ptr = &obj; 4 obj_ptr->MyFunc(); obj->Func() (*obj).Func() ↔ 4 Pointers are polymorphic Pointers are just like references, but have additional useful properties: Can be reassigned Can point to ‘‘nothing’’ (nullptr) Can be stored in a vector or an array Use pointers for polymorphism 1 Derived derived; 2 Base* ptr = &derived; Example: for implementing strategy store a pointer to the strategy interface and initialize it with nullptr and check if it is set before calling its methods 5 1 #include <iostream > 2 #include <vector > 3 using std::cout; 4 struct AbstractShape { 5 virtual void Print() const = 0; 6 }; 7 struct Square : public AbstractShape { 8 void Print() const override { cout << "Square\n";} 9 }; 10 struct Triangle : public AbstractShape { 11 void Print() const override { cout << "Triangle\n";} 12 }; 13 int main() { 14 std::vector<AbstractShape*> shapes; 15 Square square; 16 Triangle triangle; 17 shapes.push_back(&square); 18 shapes.push_back(&triangle); 19 for (const auto* shape : shapes) { shape->Print(); } 20 return 0; 21 } 6 this pointer Every object of a class or a struct holds a pointer to itself This pointer is called this Allows the objects to: Return a reference to themselves: return *this; Create copies of themselves within a function Explicitly show that a member belongs to the current object: this->x(); 7 Using const with pointers Pointers can point to a const variable: 1 // Cannot change value , can reassign pointer.
    [Show full text]
  • Effective and Efficient Memory Protection Using Dynamic Tainting
    1 Effective and Efficient Memory Protection Using Dynamic Tainting Ioannis Doudalis, Student Member, IEEE, James Clause, Member, IEEE, Guru Venkataramani, Member, IEEE, Milos Prvulovic, Senior Member, IEEE, and Alessandro Orso, Member, IEEE, ! Abstract—Programs written in languages allowing direct access to pointer p or a pointer derived from p and 2) if the access memory through pointers often contain memory-related faults, which occurs during the interval when p is valid, (i.e. between the cause non-deterministic failures and security vulnerabilities. We present allocation and deallocation of m). All other accesses to m are a new dynamic tainting technique to detect illegal memory accesses. illegal memory accesses (IMAs), where a pointer is used to When memory is allocated, at runtime, we taint both the memory and the corresponding pointer using the same taint mark. Taint marks access memory outside the bounds of the memory area with are then propagated and checked every time a memory address m which it was originally associated, or outside the time period is accessed through a pointer p; if the associated taint marks differ, during which the pointer is valid. an illegal access is reported. To allow always-on checking using a IMAs are especially relevant for several reasons. First, they low-overhead, hardware-assisted implementation, we make several key are caused by typical programming errors, such as array-out- technical decisions. We use a configurable, low number of reusable taint of-bounds accesses and stale pointer dereferences, and are thus marks instead of a unique mark for each allocated area of memory, reducing the performance overhead without losing the ability to target widespread and common.
    [Show full text]
  • Tutorial on Debugging, Memory Leaks and Profiling
    High Performance Computing for Science and Engineering I Tutorial on debugging, memory leaks and profiling Michalis Chatzimanolakis Computational Science & Engineering Laboratory OUTLINE - Approaches to debug your code •Using assertions •Using a debugger •Memory sanitation •Using a profiler Assertions in C++ An assertion is a logical statement that will cause your code to abort execution, if it is false. Example: Assertions can affect performance, Assertions perform checks that make sure that as they are essentially ‘if’ statements. some conditions are met. However, it is good practice to add assertions in a code. A simple compilation flag can ignore all assertions, so it is generally a good idea to use them. It is useful to combine assertions with the std functions isinf and isnan They can also be used with screen output. Compiler warnings Compiler warnings are compilation flags that you can (should) add to your Makefiles. Enabling warnings makes the compiler check for several issues in your source file, such as: • Forgetting to initialize a variable • Not using a declared variable • Using a function before declaring it • Having screen/file output with incorrect format (for example, treating doubles as integers) Generally, you should add warnings and try to fix all of them, before proceeding with further debugging your code. Note that different compilers may produce different warnings. The most common compiler warning flags are listed below: -Wall : enables several construction/declaration warnings -Wextra : more warnings than Wall including warnings for unused variables -Werror: will treat all warnings as errors -Wpedantic: warnings related to the use of non standard C++ extensions Using a debugger What is a debugger? A debugger is a program.
    [Show full text]
  • Memory Vulnerability Diagnosis for Binary Program
    ITM Web of Conferences 7 03004 , (2016) DOI: 10.1051/itmconf/20160703004 ITA 2016 Memory Vulnerability Diagnosis for Binary Program Feng-Yi TANG, Chao FENG and Chao-Jing TANG College of Electronic Science and Engineering National University of Defense Technology Changsha, China Abstract. Vulnerability diagnosis is important for program security analysis. It is a further step to understand the vulnerability after it is detected, as well as a preparatory step for vulnerability repair or exploitation. This paper mainly analyses the inner theories of major memory vulnerabilities and the threats of them. And then suggests some methods to diagnose several types of memory vulnerabilities for the binary programs, which is a difficult task due to the lack of source code. The diagnosis methods target at buffer overflow, use after free (UAF) and format string vulnerabilities. We carried out some tests on the Linux platform to validate the effectiveness of the diagnosis methods. It is proved that the methods can judge the type of the vulnerability given a binary program. 1 Introduction (or both) the memory area. These headers can contain some information like size, used or not, and etc. Memory vulnerabilities are difficult to detect and diagnose Therefore, we can monitor the allocator so as to especially for those do not crash the program. Due to its determine the base address and the size of the allocation importance, vulnerability diagnosis problem has been areas. Then, check all STORE and LOAD accesses in intensively studied. Researchers have proposed different memory to see which one is outside the allocated area. techniques to diagnose memory vulnerabilities.
    [Show full text]
  • A Statistical Approach for Identifying Memory Leaks in Cloud Applications
    A STATISTICAL APPROACH FOR IDENTIFYING MEMORY LEAKS IN CLOUD APPLICATIONS Vladimir Sorˇ and Satish Narayana Srirama Institute of Computer Science, University of Tartu, J. Liivi 2, Tartu, Estonia Keywords: Troubleshooting, JavaTM Virtual Machine, Byte code instrumentation, Cloud computing, Tools. Abstract: This position paper describes the attempt to automate the statistical approach for memory leak detection in JavaTM applications. Proposed system extends the basic statistical memory leak detection method with further intelligence to pinpoint the source of the memory leak in the source code. As the method adds only small overhead in runtime it is designed to be used in production systems and will help detecting memory leaks in production environments without constraint to the source of the leak. Architecture of the proposed approach is intended to use in cloud applications. 1 INTRODUCTION in production under very specific circumstances (of- ten hardly specifiable) can be very hard to find and fix Memory leaks can be a major problem in distributed in development and test environments. applications, depleting their performance, even if they This is the area we think we can improve by de- run on platforms with automatic memory manage- veloping the solution that uses efficient statistical al- ment like Java Virtual Machine. Finding memory gorithm to detect memory leaks in advance, imposes leaks is coveredby many researches and there are sev- low overhead in production system and would help eral tools and methodologies to find memory leaks. tracking down the source of the leak. The rest of the However, these tools are incorporated in profilers and paper is organized as follows.
    [Show full text]
  • Using Cyclic Memory Allocation to Eliminate Memory Leaks
    Using Cyclic Memory Allocation to Eliminate Memory Leaks Huu Hai NGUYEN1 and Martin RINARD2 1Singapore-MIT Alliance, National University of Singapore 2CSAIL, Massachusetts Institute of Technology Abstract— We present and evaluate a new memory man- I. INTRODUCTION agement technique for eliminating memory leaks in programs with dynamic memory allocation. This technique observes the A program that uses explicit allocation and deallocation has execution of the program on a sequence of training inputs to a memory leak when it fails to free objects that it will no longer find m-bounded allocation sites, which have the property that at access in the future. A program that uses garbage collection any time during the execution of the program, the program only has a memory leak when it retains references to objects that accesses the last m objects allocated at that site. The technique it will no longer access in the future. Memory leaks are an then transforms the program to use cyclic memory allocation at that site: it preallocates a buffer containing m objects of the issue since they can cause the program to consume increasing type allocated at that site, with each allocation returning the amounts of memory as it runs. Eventually the program may next object in the buffer. At the end of the buffer the allocations exhaust the available memory and fail. Memory leaks may wrap back around to the first object. Cyclic allocation eliminates therefore be especially problematic for server programs that any memory leak at the allocation site — the total amount of must execute for long (and in principle unbounded) periods of memory required to hold all of the objects ever allocated at the site is simply m times the object size.
    [Show full text]
  • Jamaicavm 8.1 — User Manual
    JamaicaVM 8.1 — User Manual Java Technology for Critical Embedded Systems aicas GmbH 2 JamaicaVM 8.1 — User Manual: Java Technology for Critical Embedded Systems JamaicaVM 8.1, Release 1. Published May 31, 2017. c 2001–2017 aicas GmbH, Karlsruhe. All rights reserved. No licenses, expressed or implied, are granted with respect to any of the technology described in this publication. aicas GmbH retains all intellectual property rights associated with the technology described in this publication. This publication is intended to assist application developers to develop applications only for the Jamaica Virtual Machine. Every effort has been made to ensure that the information in this publication is accurate. aicas GmbH is not responsible for printing or clerical errors. Although the information herein is provided with good faith, the supplier gives neither warranty nor guarantee that the information is correct or that the results described are obtainable under end-user conditions. aicas GmbH phone +49 721 663 968-0 Emmy-Noether-Straße 9 fax +49 721 663 968-99 76131 Karlsruhe email [email protected] Germany web http://www.aicas.com aicas incorporated phone +1 203 359 5705 6 Landmark Square, Suite 400 Stamford CT 06901 email [email protected] USA web http://www.aicas.com aicas GmbH phone +33 1 4997 1762 9 Allee de l’Arche fax +33 1 4997 1700 92671 Paris La Defense email [email protected] France web http://www.aicas.com This product includes software developed by IAIK of Graz University of Technology. This software is based in part on the work of the Independent JPEG Group.
    [Show full text]
  • Efficient High Frequency Checkpointing for Recovery and Debugging Vogt, D
    VU Research Portal Efficient High Frequency Checkpointing for Recovery and Debugging Vogt, D. 2019 document version Publisher's PDF, also known as Version of record Link to publication in VU Research Portal citation for published version (APA) Vogt, D. (2019). Efficient High Frequency Checkpointing for Recovery and Debugging. General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal ? Take down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. E-mail address: [email protected] Download date: 30. Sep. 2021 Efficient High Frequency Checkpointing for Recovery and Debugging Ph.D. Thesis Dirk Vogt Vrije Universiteit Amsterdam, 2019 This research was funded by the European Research Council under the ERC Advanced Grant 227874. Copyright © 2019 by Dirk Vogt ISBN 978-94-028-1388-3 Printed by Ipskamp Printing BV VRIJE UNIVERSITEIT Efficient High Frequency Checkpointing for Recovery and Debugging ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad Doctor aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr.
    [Show full text]
  • Xilinx XAPP1137 Linux Operating System Software Debugging
    Application Note: Embedded Processing Linux Operating System Software R Debugging Techniques with Xilinx Embedded Development Platforms XAPP1137 (v1.0) June 9, 2009 Author: Brian Hill Abstract This application note discusses Linux Operating System debugging techniques. Debugging boot issues, kernel panics, software and hardware debuggers, driver <-> application interaction, and various other tools are discussed. Included Included with this application note is one reference system built for the Xilinx ML507 Rev A Systems board. The reference system is available in the following ZIP file available at: https://secure.xilinx.com/webreg/clickthrough.do?cid=125262 Introduction This application note discusses Linux Operating System debugging techniques. Software applications and drivers are included to present debugging material. General Topics The application note discusses debugging Linux boot issues, and the use of various Linux debugging facilities such as strace. Included Software buggy_core The buggy_core application is used to demonstrate techniques for debugging application crashes. The use of gdb, gdbserver, and coredump analysis are discussed. buggy_memleak The buggy_memleak application allocates memory with malloc() and never frees it. This application is used to discuss memory leak detection and debugging techniques with the mtrace facility and the Linux /proc filesystem. buggy_drv A kernel driver ‘buggy_drv’ is provided as an example of debugging kernel bugs. buggy_allstrings The buggy_allstrings application operates with the buggy_drv driver. buggy_choosestring The buggy_choosestring application operates with the buggy_drv driver. Intended This application note is best suited to users who are comfortable configuring, building, and Audience booting Linux on a Xilinx embedded platform. © 2009 Xilinx, Inc. All rights reserved. XILINX, the Xilinx logo, and other designated brands included herein are trademarks of Xilinx, Inc.
    [Show full text]
  • Cork: Dynamic Memory Leak Detection for Garbage-Collected Languages £
    Cork: Dynamic Memory Leak Detection for Garbage-Collected Languages £ Maria Jump and Kathryn S. McKinley Department of Computer Sciences, The University of Texas at Austin, Austin, Texas 78712, USA g fmjump,mckinley @cs.utexas.edu Abstract agement. For example, C and C++ memory-related errors include A memory leak in a garbage-collected program occurs when the (1) dangling pointers – dereferencing pointers to objects that the program inadvertently maintains references to objects that it no program previously freed, (2) lost pointers – losing all pointers to longer needs. Memory leaks cause systematic heap growth, de- objects that the program neglects to free, and (3) unnecessary ref- grading performance and resulting in program crashes after per- erences – keeping pointers to objects the program never uses again. haps days or weeks of execution. Prior approaches for detecting Garbage collection corrects the first two errors, but not the last. memory leaks rely on heap differencing or detailed object statis- Since garbage collection is conservative, it cannot detect or reclaim tics which store state proportional to the number of objects in the objects referred to by unnecessary references. Thus, a memory leak heap. These overheads preclude their use on the same processor for in a garbage-collected language occurs when a program maintains deployed long-running applications. references to objects that it no longer needs, preventing the garbage This paper introduces a dynamic heap-summarization technique collector from reclaiming space. In the best case, unnecessary ref- based on type that accurately identifies leaks, is space efficient erences degrade program performance by increasing memory re- (adding less than 1% to the heap), and is time efficient (adding 2.3% quirements and consequently collector workload.
    [Show full text]
  • Purify: Fast Detection of Memory Leaks and Access Errors
    Purify: Fast Detectionof Memory Leaksand AccessErrors ReedHastings and Bob Joyce Pure Sofware Inc. Abstract This paper describesPurifyru, a software testing and quality assuranceOol that detects memory leaks and accesserors. Purify inserts additional checking instructions directly into the object code produced by existing compilers. These instructions check every memory read and write performedby the program-under-testand detect severaltypes of accesserrors, such as reading uninitialized memory or witing to freed memory.Purify inserts checkinglogic into all of the code in a program,including third-party and vendor object-codelibraries, and verifiessystem call interfaces.In addition,Purify tracksmemory usage and identifies individual memory leals using a novel adaptation of garbage collectiontechniques. Purify producesstandard executable files compatiblewith existing debuggers,and currentlyruns on Sun Microsystems'SPARC family of workstations. Purify's neafly-comprehensivememory accesschecking slows the target program down typically by less than a facor of three and has resulted in significantly more reliable softwarefor severaldevelopment goups. L. Introduction A single memory accesserror, suchas readingfrom uninitialized memoryor writing to freed memory can causea progam to act unpredicAbly or evencrash. Yet, it is nearly impossibleto eliminateall sucherrors from a non-trivial progam For one thing, theseerors may produce observableeffects infrequently and intermittently. Even when p-gru.r are testedintensively for extendedperiods, errors can and do escapedetection. The uniquecombination of litco*stanos required for an error to occur andfor its symptomslo becomcvisible may be virtually impossibleto createin the developmentor test environment.As a result, proglammersspend much time looking for theseerrors, but end-usersmay experiencetlem first. [Miller90] empiricallyshows the continuingprevalence of accesserrors in manywidely-used Unix Programs.
    [Show full text]