(19)
(11)EP 3 120 247 B1

(12)EUROPEAN PATENT SPECIFICATION

(45)Mention of the grant of the patent:
04.12.2019 Bulletin 2019/49

(21)Application number: 15713085.7

(22)Date of filing:  12.03.2015
(51)International Patent Classification (IPC): 
G06F 11/07(2006.01)
G06F 11/36(2006.01)
(86)International application number:
PCT/US2015/020053
(87)International publication number:
WO 2015/142598 (24.09.2015 Gazette  2015/38)

(54)

FRAMEWORK FOR USER-MODE CRASH REPORTING

RAHMEN FÜR BENUTZERMODUS-ABSTURZBERICHTE

ENVIRONNEMENT POUR RAPPORT DE PLANTAGE EN MODE UTILISATEUR


(84)Designated Contracting States:
AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

(30)Priority: 17.03.2014 US 201414217030

(43)Date of publication of application:
25.01.2017 Bulletin 2017/04

(73)Proprietor: Microsoft Technology Licensing, LLC
Redmond, WA 98052-6399 (US)

(72)Inventors:
  • MUTHUKUMSARASAMY, Aruldevi
    Redmond, Washington 98052-6399 (US)
  • ZHANG, Hongwei
    Redmond, Washington 98052-6399 (US)
  • GUNAWAN, Aswin
    Redmond, Washington 98052-6399 (US)

(74)Representative: Grünecker Patent- und Rechtsanwälte PartG mbB 
Leopoldstraße 4
80802 München
80802 München (DE)


(56)References cited: : 
US-A1- 2003 005 414
US-A1- 2006 005 078
US-A1- 2013 080 502
US-A1- 2005 210 077
US-A1- 2006 253 837
  
  • YINGNONG DANG ET AL: "ReBucket: A method for clustering duplicate crash reports based on call stack similarity", SOFTWARE ENGINEERING (ICSE), 2012 34TH INTERNATIONAL CONFERENCE ON, IEEE, 2 June 2012 (2012-06-02), pages 1084-1093, XP032195284, DOI: 10.1109/ICSE.2012.6227111 ISBN: 978-1-4673-1066-6
  
Note: Within nine months from the publication of the mention of the grant of the European patent, any person may give notice to the European Patent Office of opposition to the European patent granted. Notice of opposition shall be filed in a written reasoned statement. It shall not be deemed to have been filed until the opposition fee has been paid. (Art. 99(1) European Patent Convention).


Description

BACKGROUND



[0001] Although commercially available computer programs are generally very reliable, a program can encounter problems that result in a malfunction that causes the program to stop functioning. Such a malfunction is referred to as a "crash." The problems that can cause a program to crash include hardware errors (e.g., defective memory location), logical errors in the program (e.g., using an invalid memory address), incompatible components (e.g., linking to an outdated function), and so on. When a program crashes, the operating system typically takes control as a result of a hardware or software exception being raised. A hardware exception can be raised for a variety of reasons, including attempted division by zero, use of an invalid memory address, attempted execution of an invalid instruction, and so on. The instruction set architecture of the processor on which the program executes defines different types of exceptions and where control is transferred when each type of exception is raised. The program or the operating system typically installs hardware exception handlers for each type of hardware exception. A software exception, in contrast, is raised by software detecting an unusual situation. For example, a software exception might be raised when a function is passed an invalid parameter, when a C++ exception is not caught, and so on. The program or operating system typically installs software exception handlers for each type of software exception. When an exception is raised, the exception handler for that type of exception is invoked to handle the exception. Each exception handler determines, based on the seriousness of the exception, whether the program should continue its execution or terminate its execution, resulting in a crash.

[0002] When a program crashes, the operating system typically collects information about the crash, generates a crash report based on the collected information, and makes the crash report available to the developer of the program that crashed. When a crash handler installed by the operating system is invoked, the crash handler typically executes in a supervisory privilege mode (e.g., supervisor mode or kernel mode) of the processor. Because the crash handler executes in a supervisory privilege mode, the crash handler may have access to all the data structures of the operating system and the data structures of the program that crashed. The access to these data structures allows the crash handler to collect any information it deems appropriate for a crash report. Once a crash report is generated, it can be made available to the developer. For example, the crash report may be transmitted to a crash reporting server where it is made available to the developer. generated, it can be made available to the developer. For example, the crash report may be transmitted to a crash reporting server where it is made available to the developer.

[0003] Although crash reports provided by an operating system provide valuable information to assist a developer in determining the cause of the crash of a program, the information in a crash report is not typically tailored to the program itself or the needs of the developer. Moreover, a crash reporting server may provide generic tools for use by developers of all types of programs. As a result, a developer of a program may not have program-specific tools available that would be helpful in determining the cause of the crash. Also, the operating system may control what crash reports to send and the timing of the sending. A developer may want such control to meet the developer's needs.

[0004] US 2006/005078 A1 discloses a method for identifying the source, i.e., program or plug-in module of a failure by analyzing memory content that existed at the time of the failure.

[0005] US 2006/0253837 A1 discloses a method comprising obtaining a trace of a call stack.

[0006] YINGNONG DANG ET AL: "ReBucket: A method for clustering duplicate crash reports based on call stack similarity", SOFTWARE ENGINEERING (ICSE), 2012 34TH INTERNATIONAL CONFERENCE ON, IEEE, 2 June 2012 (2012-06-02), pages 1084-1093, DOI: 10.1109/ICSE.2012.6227111ISBN: 978-1-4673-1066-6 discloses a bucketing method for clustering duplicate crash reports. A new metric for measuring similarity between two call stacks is proposed.

[0007] US 2013/0080502 A1 discloses a diagnostic mechanism monitoring the execution of a user interface threat to detect the execution of a long running task that has caused the user interface threat to become unresponsive. An operating system contains methods that are used to create a dump file containing system information, such as a call stack. Additionally, delayed data is collected and stored in a delay file, wherein the collected data may contain a session ID.

[0008] US 2003/0005414 A1 discloses that a digital computer call stack is traced and analyzed following the occurrence of an unhandled exception or a crash. Stack signatures can be associated with program bugs, and stack signatures collected from the failure instance, at other times can be conveniently compared for recognizing particular types of failures.

[0009] WO 2005/043360 A1 discloses that a list of functions that a DLL exports is listed in the Export Address Table (EAT) which is part of the Portable Executable (PE) image header. Addresses listed in the EAT are relative addresses to code within the module.

SUMMARY



[0010] The object of the present invention provided improved computer-readable storage medium for generating a crash report for when an application crashes, and a corresponding computing system.

[0011] This object is solved by the subject matter of the independent claims.

[0012] Preferred embodiments are defined by the dependent claims.

[0013] References to embodiments in the following description which are not covered by the appended claims are considered as not being part of the invention according to this description.

[0014] A user-mode crash reporting system for generating a crash report when an application crashes is provided. The user-mode crash reporting system installs a crash handler to execute in the same process as the application when the application crashes. When the application crashes, the crash handler identifies a victim thread of the process that was executing at the time of a crash. The crash handler then walks a call stack of the victim thread to identify return addresses for returning from called functions of the application. The crash handler identifies offsets within the loaded modules corresponding to the return addresses and creates a call chain of functions using the identified offsets. The crash handler then generates a crash report based on the identified call chain.

BRIEF DESCRIPTION OF THE DRAWINGS



[0015] 

Figure 1 is a block diagram that illustrates an execution environment in which applications execute in some embodiments.

Figure 2 is a block diagram that illustrates a module table of the user-mode crash reporting system in some embodiments.

Figure 3 is a block diagram that illustrates the layout of a module in some 25 embodiments.

Figure 4 is a block diagram that illustrates a layout of the call stack in some embodiments.

Figure 5 is a block diagram that illustrates components of a user-mode crash reporting system in some embodiments.

Figure 6 is a flow diagram that illustrates the processing of a crash handler in some embodiments.

Fig 7 is a flow diagram that illustrates the processing of an add function to call chain component of the crash reporting system in some embodiments.

Figure 8 is a flow diagram that illustrates the processing of a find blamed function component of the crash reporting system in some embodiments.

Figure 9 is a flow diagram that illustrates the processing of a send crash report component of the user-mode crash reporting system in some embodiments.

Figure 10 is a flow diagram that illustrates the processing of an add module callback component of the user-mode crash reporting system in some embodiments.


DETAILED DESCRIPTION



[0016] A method and system for generating a user-mode crash report when an application crashes is provided. In some embodiments, a user-mode crash reporting system executes a crash handler in the same process (i.e., in-process) as the crashed application and generates a crash report based on information accessible to the process. The user-mode crash reporting system then controls the sending of the crash report to a crash reporting server. The user-mode crash reporting system installs a crash handler to execute in the same process in user privilege mode as the application that crashes. The user-mode crash reporting system installs crash handlers that are specific to the underlying operating system and programming environment. For example, if the operating system is Apple's iOS operating system, then crash handlers may be installed to handle Mach exceptions and BSD signals. If the programming environment is based on C++, then crash handlers may be installed to handle exceptions that are thrown, but not otherwise caught. When an application crashes, the crash handler may identify a "victim" thread of the process that was executing at the time of the crash and then generate a call chain of the functions that had been invoked (or called) at the time of the crash. To generate the call chain, the crash handler walks the call stack of the victim thread to identify return addresses for returning from called functions of the application. The code of the application is stored in one or more modules, each of which contains functions. The crash handler then identifies offsets within the module corresponding to the return addresses and identifies the functions within the module that includes those offsets. The crash handler may generate a call chain for each thread of the process and not just the victim thread. The user-mode crash reporting system then generates a crash report based on the functions in the identified call chain. Because the crash handler executes in the same process as the application, the crash handler has access to all the data of the application. Moreover, the crash handler stores the crash reports so they can be sent to a crash reporting server.

[0017] In some embodiments, the user-mode crash reporting system identifies a function of an application that is to blame for the crash of the application. The function to blame for the crash is the function that was likely executing on the victim thread at the time of the crash. The user-mode crash reporting system identifies a call chain of functions invoked at the time of the crash. The user-mode crash reporting system then accesses information identifying functions that are immune from being blamed for the crash. When a program is about to crash, the application may invoke various administrative functions in an attempt to mitigate the effects of the crash, to collect some crash-related information, and so on. These administrative functions are likely to be in the call chain but should not be blamed for the crash. These administrative and other functions that should not be blamed for a crash are designated as being immune from being blamed. The user-mode crash reporting system processes the call chain of functions to identify the last function invoked that is not immune from being blamed for the crash. The user-mode crash reporting system then designates that last function as being to blame for the crash. The information designating the functions that are immune may be specified in a configuration file and indicate the range of offsets of the functions within the module.

[0018] Figure 1 is a block diagram that illustrates an execution environment in which applications execute in some embodiments. An execution environment 100 includes operating system components 110, an application environment 120, and application sandboxes 130. The operating system components may include a Mach layer 111, a BSD layer 112, and a core services layer 113. The Mach layer provides kernel-based services such as task and thread management, virtual memory management, interprocess communication, thread scheduling, exception handling services (e.g., Mach exceptions), and so on. The BSD layer provides higher level services such as an POSIX API, a file system implementation, a signaling mechanism (e.g., BSD signals), and so on. The core services layer provides various services such as time and date management, stream-based I/O, and so on. The application environment may provide an API for a graphical user interface, event handling, I/O services, and so on. The application environment may also provide a sandbox in which the applications execute. Each application 131 executes in its own sandbox to prevent bad behavior of one application from adversely impacting other applications or the operating system. A sandbox provides an environment in which an application has access to a tightly controlled set of resources (e.g., network access, disk space). The crash handler executes in the same sandbox as the crashed application.

[0019] Figure 2 is a block diagram that illustrates a module table of the user-mode crash reporting system in some embodiments. A module table 210 maps module identifiers to the images of the modules 220 loaded into the address space of the application. Each module contains functions of the application. The module table includes an entry 211 for each module that is loaded into the process of an application. Each entry contains identification information about the module along with a reference to the loaded module. To generate the module table, the user-mode crash reporting system executing in the process of the application installs add and remove module callback components so that the system can be notified when a module is loaded into the process (i.e., address space of the process) and unloaded from the process, respectively. When a module is loaded, the add module callback is called. The add module callback is passed a reference to information about the module, including the address range into which the module is loaded, the name of the module, a unique build identifier of the module, and so on, and adds a corresponding entry to the module table. When a module is unloaded, the remove module callback is called. The remove module callback is passed a reference to the module and removes the corresponding entry from the module table.

[0020] Figure 3 is a block diagram that illustrates the layout of a module in some embodiments. A module 310 for an application includes various functions and may include a main function, which is passed control to start execution of the application. A module may also have an associated module function map 320 that contains an entry for each function of the module along with a reference to the starting point of that module in the address space of the application. A module function map may be provided along with the module, or the user-mode crash reporting system may generate the module function map based on analysis of other module-related information. The immune function table 330 identifies those functions of the module that are immune from being blamed for a crash. In this example, functions f4, f5, and f6 are immune from being blamed.

[0021] Figure 4 is a block diagram that illustrates a layout of the call stack in some embodiments. As illustrated by a call stack 400, the functions f1, f2, f3, f4, and f5 of module 310 (Figure 3) have been invoked as represented by their stack frames 401, 402, 403, 404, and 405, respectively. The stack pointer SP points to the stack frame at the top of the call stack. Each stack frame may contain a parameters area for passing parameters between the invoking function and the invoked function. Each stack frame also contains a reference SP to the stack frame of the invoking function. For example, stack frame 405 includes a reference that points to stack frame 404. The reference SP is used by a function to remove its stack frame as it returns to the invoking function. Thus, the reference SPs form a linked list of stack frames. Each stack frame also includes a reference RA to the return address within the invoking function. For example, stack frame 405 includes a return address within function f4 that is after the invocation of function f5. A reference RA is used by a returning function to return to the location after the invocation (or other location).

[0022] To generate a call chain, the crash handler of the user-mode crash reporting system walks the call stack to identify each function that has been invoked. The crash handler may be passed the thread identifier of the victim thread that was executing at the time of the crash. The crash handler may interact with the operating system to retrieve thread-related information such as the stack pointer for the thread and the instruction pointer at the time of the crash. The crash handler may use the instruction pointer to identify the function that was executing at the time of the crash (i.e., when the corresponding exception was raised) by using the module function map to identify the function corresponding to that address. To walk the call stack, the crash handler uses that stack pointer to retrieve a pointer to the next lower stack frame and the return address of the invoking function. The crash handler uses that return address to identify the function that was executing at the time of the crash by using the module function map. The crash handler continues walking the stack until it encounters the bottom of the stack. The bottom of the stack may be designated, for example, by a distinguished value (e.g., all ones) stored in the bottom frame of the call stack (e.g., reference SP == null). In some embodiments, the crash handler may walk the call stack of each thread of the process in a similar manner to generate a more comprehensive crash report.

[0023] In some embodiments, the bottom of the stack may not be readily ascertainable by the crash handler. For example, the stack frame 401 may contain a bad memory address for the reference SP, that is, where the pointer to the next lower stack frame would normally be stored. When the crash handler attempts to access memory using that bad address, a bad address exception may occur. To detect that the bottom of a call stack has been encountered, the crash handler may install a bad memory access handler that is invoked when a reference to a bad memory address is attempted. The bad memory access handler may simply set a global flag to indicate the bad memory access attempt and return to the crash handler to execute the instruction after the attempted memory access. The following instructions can check the global flag and use it as an indication that the bottom of the stack has been reached.

[0024] Figure 5 is a block diagram that illustrates components of a user-mode crash reporting system in some embodiments. A user crash reporting system 500 includes module data 510, data structures 520, and components 530. The module data, for each module, includes the module 511, a symbol table 512, a module function map 513, and an immune function table 514. The module data may be provided by the developer. The symbol table contains symbol-related information, which may be used to symbolicate references to functions and variables. The immune function table may be provided as configuration information stored in a configuration file of the application.

[0025] The data structures used by the user-mode crash reporting system include a call stack 521, a call chain 522, a module table 523, and a crash report store 524. In the following, the user-mode crash reporting system is described in reference to processing of the victim thread of a crash. The processing of other threads of the application that crashed would be performed in a similar manner to provide a more comprehensive crash report. The call stack represents the call stack of the victim thread. The call chain represents the chain of function calls as determined by the crash handler. The module table contains a mapping of the modules that are loaded into the address space of the application to their range of addresses. The crash report store stores the crash reports generated by the user-mode crash reporting system. The crash report store may store the crash report for each crash until it is uploaded to a crash reporting server and may persistently store other identifying information about a crash to optimize the uploading of subsequent crash reports.

[0026] The components of the user-mode crash reporting system include a crash handler 531, an add function to call chain component 532, a find blamed function component 533, a store crash report component 534, an add module callback 535, a remove module callback 536,and a call stack bad address handler 537. The crash handler, which may be installed to handle various types of crashes, walks the call stack and generates a crash report. The add function to call chain component identifies the function associated with a return address and adds an identifier of that function to the call chain. The find blamed function component identifies a function to blame for the crash. The store crash report component controls the storing of the crash report for sending to a crash reporting server when the application next executes. The add module callback and the remove module callback are installed when the application is initialized to track the modules that are loaded and unloaded. The call stack bad address handler is installed by the crash handler to handle bad address references resulting from reaching the bottom of a call stack.

[0027] The computing devices and systems on which the user-mode crash reporting system may be implemented may include a central processing unit, input devices, output devices (e.g., display devices and speakers), storage devices (e.g., memory and disk drives), network interfaces, graphics processing units, accelerometers, cellular radio link interfaces, global positioning system devices, and so on. The input devices may include keyboards, pointing devices, touch screens, gesture recognition devices (e.g., for air gestures), head and eye tracking devices, microphones for voice recognition, and so on. The computing devices may include desktop computers, laptops, tablets, e-readers, personal digital assistants, smartphones, gaming devices, servers, and computer systems such as massively parallel systems. The computing devices may access computer-readable media that include computer-readable storage media and data transmission media. The computer-readable storage media are tangible storage means that do not include a transitory, propagating signal. Examples of computer-readable storage media include memory such as primary memory, cache memory, and secondary memory (e.g., DVD) and include other storage means. The computer-readable storage media may have recorded upon or may be encoded with computer-executable instructions or logic that implements the user-mode crash reporting system. The data transmission media is used for transmitting data via transitory, propagating signals or carrier waves (e.g., electromagnetism) via a wired or wireless connection.

[0028] The user-mode crash reporting system may be described in the general context of computer-executable instructions, such as program modules and components, executed by one or more computers, processors, or other devices. Generally, program modules or components include routines, programs, objects, data structures, and so on that perform particular tasks or implement particular data types. Typically, the functionality of the program modules may be combined or distributed as desired in various embodiments. Aspects of the user-mode crash reporting system may be implemented in hardware using, for example, an application-specific integrated circuit ("ASIC").

[0029] Figure 6 is a flow diagram that illustrates the processing of a crash handler in some embodiments. A crash handler 600 is passed an indication of a victim thread, generates a call chain, finds a function to blame, and sends a crash report. In block 601, the component retrieves a thread control block or other information related to the victim thread. In block 602, the component determines the instruction pointer at the time of the crash. In block 603, the component invokes the add function to call chain component to add a function associated with the instruction pointer to the call chain. In block 604, the component retrieves the stack pointer at the time of the crash. In block 605, the component installs a call stack bad address handler to detect when a bad address is encountered when walking the call stack. In blocks 606-609, the component loops, walking the call stack. In block 606, the component retrieves the return address referenced by the current stack pointer. If the stack pointer SP points to a bad memory address, then the use of the stack pointer to access the return address will cause a bad address exception. In decision block 607, if the bottom of the stack frame has been reached (e.g., as indicated by a global flag indicating a bad address exception), then the component continues at block 610, else the component continues at block 608. In block 608, the component invokes the add function to call chain component, passing the return address. In block 609, the component retrieves a reference to the next stack frame and loops to block 606 to process the next stack frame. In block 610, the component invokes the find blamed function component to identify a function to blame for the crash. In block 611, the component invokes the store crash report component and then completes.

[0030] Figure 7 is a flow diagram that illustrates the processing of an add function to call chain component of the crash reporting system in some embodiments. An add function to call chain component 700 is passed an address, identifies the function within a module corresponding to that address, and adds that function to the call chain. The function may initially identify the module using the module table. In block 701, the component selects the next entry in the module function map. In decision block 702, if all the entries have already been selected, then the function corresponding to the passed address has not been found and the component reports an error, else the component continues at block 703. In decision block 703, if the address is within the range of the selected entry, then the component continues at block 704, else the component loops to block 701 to select the next entry. In block 704, the component adds an entry corresponding to the identified function to the call chain and then returns.

[0031] Figure 8 is a flow diagram that illustrates the processing of a find blamed function component of the crash reporting system in some embodiments. A find blamed function component 800 processes the call chain to identify a function to blame for the crash that is not immune. In block 801, the component selects the next function in the call chain. In decision block 802, if the end of the call chain has been reached, then the component returns an indication that no function to blame has been identified, else the component continues at block 803. In decision block 803, if the selected function is immune as indicated by the immune function table, then the component loops to block 801 to select the next function, else the component returns an indication of the selected function to blame.

[0032] Figure 9 is a flow diagram that illustrates the processing of a send crash report component of the user-mode crash reporting system in some embodiments. A store crash report component 900 is invoked to send a crash report. The component may generate a crash identifier and stores the crash report for sending to a crash reporting server when the application is next launched. The component may use the crash identifier to identify whether similar crashes have been reported. In block 901, the component generates the crash identifier, for example, as a combination of the application name and its version, the blamed module name (i.e., module that contains the blamed function) and its version, and the offset of the crash within the blamed module. In block 902, the component collects various crash report information that may include the call chains of each thread, the functions to blame, symbolication of variables and function names (e.g., the function to blame), dumps of various portions of the address space of the application, and so on. In block 903, the component adds a session identifier to the crash report. The session identifier uniquely identifies an execution of an application from start to termination (e.g., crash or normal termination). In certain environments, an application (or application monitor) may report activity or other information of the application to a server, such as a customer experience server, and identify the activity by a session identifier. The use of the session identifier in the crash report allows the activity information to be correlated to the crash report. In this way, a developer may be able to take advantage of having both the crash report and the activity information when analyzing the cause of the crash. In block 904, the component adds the crash report to the crash report repository and then returns. When the application next executes, if an upload criterion is satisfied, the application uploads the crash report to the crash report server. The application may use various upload criteria to determine whether a crash report should be uploaded. For example, the upload criteria may be based on the current communication bandwidth (e.g., cellular versus Wi-Fi), whether a certain number of the same type of crash reports have already been uploaded, and so on.

[0033] Figure 10 is a flow diagram that illustrates the processing of an add module callback component of the user-mode crash reporting system in some embodiments. An add module callback component 1000 is installed when the user-mode crash reporting system initializes and records those modules whose images have been loaded into the address space of the application. The application environment may invoke the add module callback passing an indication of a module identifier to notify the application that a module has been loaded. In blocks 1001-1003, the component may retrieve various information about the module that has been loaded, such as its address in memory, its name, its build identifier, and so on. In block 1004, the component adds an entry to the module table and then completes.

[0034] Although the subject matter has been described in language specific to structural features and/or acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims. Accordingly, the invention is not limited except as by the appended claims.


Claims

1. A computer-readable storage medium storing computer-executable instructions for controlling a computing device to generate a crash report when an application crashes, the application being organized as a module (310) with functions, the computer-executable instructions comprising instructions that:

install a crash handler (600) to execute in the same process as the application when the application crashes, wherein the application and the crash handler execute in a user mode within the process; and

under control of the crash handler executing in the process of the application after the application crashes, identify a call chain of functions using offsets, comprising:

identify a victim thread of the process that was executing at the time of a crash,

retrieve a stack pointer for the victim thread and an instruction pointer to an instruction at the time of the crash,

identify a function that was executing at the time of the crash based on said instruction pointer by using a module function map, said module function map containing an entry for each function of the module along with a reference to the starting point of that module in the address space of the application,

walk a call stack of the victim thread to identify return addresses for returning from called functions of the application,

identify the offsets within the module, the offsets corresponding to the return addresses and identify the functions within the module that include the identified offsets; and

create a crash report based on the identified call chain.


 
2. The computer-readable storage medium of claim 1, wherein the computing device is executing under control of an operating system with a Mach-based kernel and a BSD-based kernel and multiple crash handlers are installed to handle Mach exceptions and BSD signals generated when the application crashes.
 
3. The computer-readable storage medium of claim 1, wherein the computer-executable instructions further comprise instructions that identify a function to blame for the crash based on analysis of the call chain.
 
4. The computer-readable storage medium of claim 3, wherein a function that is designated as immune is excluded from being blamed for a crash.
 
5. The computer-readable storage medium of claim 1, wherein the computer-executable instructions further comprise instructions to identify a function of the application to blame for a crash of the application, the computer-executable instructions comprising instructions performed by the crash handler, when the application crashes:

identify offsets within the module corresponding to return addresses and identify the functions within the module that include the identified offsets;

identify a call chain of functions invoked at the time of the crash using the identified offsets;

access information identifying one or more functions as being immune from being blamed for the crash;

process the call chain of functions to identify a last function invoked that is not immune from being blamed for the crash; and

designate that last function as being to blame for the crash,

wherein the crash handler executes in the same process as the application.


 
6. The computer-readable storage medium of claim 5, wherein the functions of the application are stored in the module (310) and a function that is immune is specified by its offset within the module.
 
7. The computer-readable storage medium of claim 6, wherein the instructions that identify a call chain of functions walk a call stack of the application and map return addresses for called functions to offsets within the module (310).
 
8. The computer-readable storage medium of claim 5, wherein the computer-executable instructions further comprise instructions that identify a name of the designated function to blame using symbol table information associated with the application.
 
9. The computer-readable storage medium of claim 5, wherein the computer-executable instructions further comprise instructions that upload a crash report to a crash report server, the crash report identifying the function to blame for the crash.
 
10. A computing system for generating a crash report when an application crashes, the application being organized as a module (310) with functions, the computing system comprising:

a computer-readable storage medium storing computer-executable instructions of:

the module; and

a crash handler (600) for executing in the same process as the application when the application crashes, the application and the crash handler executing in a user mode within the process, the crash handler identifying a call chain of functions using offsets by being adapted to:

identify a victim thread of the process, the victim thread being a thread of the process that was executing at the time of a crash,

retrieve a stack pointer to a call stack of the victim thread and an instruction pointer to an instruction execution at the time of the crash,

identify a function that was executing at the time of the crash based on said instruction pointer by using a module function map, said module function map containing an entry for each function of the module along with a reference to the starting point of that module in the address space of the application,

walk the call stack of the victim thread to identify return addresses for called functions, the return addresses indicating a return location within a calling function,

identify offsets within the module, the offsets corresponding to the return addresses and the instruction pointer;

the crash handler further comprising:

a component adapted to identify a function to blame for the crash, the function to blame for the crash being the last function called before the crash that is not immune from being blamed for a crash as indicated by an immune function table;

a component adapted to create a crash report that identifies the function to blame for the crash and a session identification of an application activity log for the execution of the application; and

a component adapted to upload the crash report to a crash report server, when an upload criterion is satisfied; and

wherein the crash handler is further adapted to identify the functions within the module that include the identified offsets; and

a processor adapted to execute the computer-executable instructions stored in the computer-readable storage medium.


 
11. The computing system of claim 10, wherein the crash handler (600) further comprises a component adapted to identify a name of the function to blame using symbol table information associated with the application.
 
12. The computing system of claim 10, wherein the computing system is adapted to execute under control of an operating system with a Mach-based kernel and a BSD-based kernel and multiple crash handlers are installed to handle one or more Mach exceptions and BSD signals generated when the application crashes.
 


Ansprüche

1. Computerlesbares Speichermedium, das computerausführbare Anweisungen zur Steuerung einer Rechenvorrichtung speichert, um einen Absturzbericht zu erzeugen, wenn eine Anwendung abstürzt, wobei die Anwendung als ein Modul (310) mit Funktionen organisiert wird, und die computerausführbaren Anweisungen Anweisungen umfassen, die:

einen Crash-Handler (600) installieren, um in dem gleichen Prozess wie die Anwendung ausgeführt zu werden, wenn die Anwendung abstürzt, wobei die Anwendung und der Crash-Handler in einem Benutzermodus in dem Prozess ausgeführt werden; und

unter Steuerung des Crash-Handlers, der in dem Verfahren der Anwendung, nachdem die Anwendung abstürzt, ausgeführt wird, eine Aufrufkette aus Funktionen unter Anwendung von Offsets identifizieren, umfassend:

Identifizieren eines Opfer-Threads des Prozesses, der zur Zeit eines Absturzes ausgeführt wurde;

Abrufen eines Stapelzeigers für den Opfer-Thread und eines Anweisungszeigers auf eine Anweisung zum Zeitpunkt des Absturzes,

Identifizieren einer Funktion, die zum Zeitpunkt des Absturzes ausgeführt wurde, basierend auf dem Anweisungszeiger unter Verwendung einer Modulfunktionskarte, wobei die Modulfunktionskarte einen Eintrag für jede Funktion des Moduls zusammen mit einer Referenz auf den Anfangspunkt des Moduls in dem Adressraum der Anwendung enthält,

Durchlaufen eines Aufrufstapels des Opfer-Threads, um Rückgabeadressen für die Rückgabe von aufgerufenen Funktionen der Anwendung zu identifizieren,

Identifizieren der Offsets in dem Modul, wobei die Offsets den Rückgabeadressen entsprechen, und Identifizieren der Funktionen in dem Modul, die die identifizierten Offsets einschließen; und

Erzeugen eines Absturzberichts auf Grundlage der identifizierten Aufrufkette.


 
2. Computerlesbares Speichermedium nach Anspruch 1, wobei die Rechenvorrichtung unter Steuerung eines Betriebssystems mit Mach-basiertem Kernel und einem BSD-basierten Kernel ausgeführt wird, und mehrere Crash-Handlers installiert werden, um Mach-Ausnahmen und BSD-Signale zu behandeln, die erzeugt werden, wenn die Anwendung abstürzt.
 
3. Computerlesbares Speichermedium nach Anspruch 1, wobei die computerausführbaren Anweisungen des weiteren Anweisungen umfassen, die basierend auf der Analyse der Aufrufkette eine Funktion identifizieren, die für den Absturz verantwortlich ist.
 
4. Computerlesbares Speichermedium nach Anspruch 3, wobei eine Funktion, die als immun bezeichnet wird, davon ausgeschlossen ist, für einen Absturz verantwortlich gemacht zu werden.
 
5. Computerlesbares Speichermedium nach Anspruch 1, wobei die computerausführbaren Anweisungen des Weiteren Anweisungen umfassen, um eine Funktion der Anwendung zu identifizieren, die für einen Absturz der Anwendung verantwortlich ist, wobei die computerausführbaren Anweisungen Anweisungen umfassen, die von dem Crash-Handler durchgeführt werden, wenn die Anwendung abstürzt;
Identifizieren von Offsets in dem Modul, die den Rückgabeadressen entsprechen, und Identifizieren der Funktionen in dem Modul, die die identifizierten Offsets einschließen;
Identifizieren einer Aufrufkette aus Funktionen, die zum Zeitpunkt des Absturzes aufgerufen wurden, unter Anwendung der identifizierten Offsets;
Zugangsinformation, die eine oder mehrere Funktionen als immun in Bezug auf die Verantwortlichkeit für den Absturz identifiziert;
Verarbeiten der Aufrufkette aus Funktionen, um mindestens eine aufgerufene Funktion zu identifizieren, die in Bezug auf die Verantwortlichkeit für den Absturz nicht immun ist; und
Bezeichnen der letzten Funktion als verantwortlich für den Absturz,
wobei der Crash-Handler in dem gleichen Prozess wie die Anwendung ausgeführt wird.
 
6. Computerlesbares Speichermedium nach Anspruch 5, wobei die Funktionen der Anwendungen in dem Modul (310) gespeichert werden und eine Funktion, die immun ist, durch ihren Offset in dem Modul spezifiziert wird.
 
7. Computerlesbares Speichermedium nach Anspruch 6, wobei die Anweisungen, die eine Aufrufkette aus Funktionen identifizieren, einen Aufrufstapel der Anwendung durchlaufen und Rückgabeadressen für aufgerufene Funktionen den Offsets in dem Modul (310) zuordnen.
 
8. Computerlesbares Speichermedium nach Anspruch 5, wobei die computerausführbaren Anweisungen des Weiteren Anweisungen umfassen, die einen Namen der bezeichneten verantwortlichen Funktion unter Anwendung von Symboltabelleninformation, die mit der Anwendung assoziiert ist, identifizieren.
 
9. Computerlesbares Speichermedium nach Anspruch 5, wobei die computerausführbaren Anweisungen des Weiteren Anweisungen umfassen, die einen Absturzbericht auf einen Absturzbericht-Server hochladen, wobei der Absturzbericht die für den Absturz verantwortliche Funktion identifiziert.
 
10. Rechnersystem zur Erzeugung eines Absturzberichts, wenn eine Anwendung abstürzt, wobei die Anwendung als ein Modul (310) mit Funktionen organisiert wird, wobei das Rechensystem umfasst:

ein computerlesbares Speichermedium, das computerlesbare Anweisungen speichert, von:

dem Modul; und

einem Crash-Handler (600) zur Ausführung in dem gleichen Prozess wie die Anwendung, wenn die Anwendung abstürzt, wobei die Anwendung und der Crash-Handler in einem Benutzermodus in dem Prozess ausgeführt werden, der Crash-Handler eine Aufrufkette aus Funktionen unter Anwendung von Offsets identifiziert, durch Anpassung an:

Identifizieren eines Opfer-Threads des Prozesses, wobei der Opfer-Thread ein Thread des Prozesses ist, der zur Zeit eines Absturzes ausgeführt wurde;

Abrufen eines Stapelzeigers auf einen Abrufstapel des Opfer-Threads und eines Anweisungszeigers auf eine Anweisungsausführung zum Zeitpunkt des Absturzes,

Identifizieren einer Funktion, die zum Zeitpunkt des Absturzes ausgeführt wurde, basierend auf dem Anweisungszeiger unter Verwendung einer Modulfunktionskarte, wobei die Modulfunktionskarte einen Eintrag für jede Funktion des Moduls zusammen mit einer Referenz auf den Anfangspunkt des Moduls in dem Adressraum der Anwendung enthält, Durchlaufen des Aufrufstapels des Opfer-Threads, um Rückgabeadressen für aufgerufene Funktionen zu identifizieren, wobei die Rückgabeadressen eine Rückgabeposition in einer Aufruffunktion anzeigen,

Identifizieren von Offsets in dem Modul, wobei die Offsets den Rückgabeadressen und dem Anweisungszeiger entsprechen;

wobei der Crash-Handler des Weiteren umfasst:

eine Komponente, daran angepasst, eine für den Absturz verantwortliche Funktion zu identifizieren, wobei die für den Absturz verantwortliche Funktion die letzte Funktion ist, die vor dem Absturz aufgerufen wurde, und die nicht dagegen immun ist, für einen Absturz verantwortlich gemacht zu werden, wie er durch eine Immunfunktionstabelle angegeben wird;

eine Komponente, daran angepasst, einen Absturzbericht, der die für den Absturz verantwortliche Funktion identifiziert, und eine Sitzungsidentifikation eines Anwendungsaktivitätsprotokolls für die Ausführung der Anwendung zu erzeugen; und

eine Komponente, daran angepasst, den Absturzbericht auf einen Absturzbericht-Server hochzuladen, wenn ein Hochladekriterium erfüllt wird; und

wobei der Crash-Handler des Weiteren daran angepasst wird, die Funktionen in dem Modul zu identifizieren, die die identifizierten Offsets einschließen; und

einen Prozessor, daran angepasst, die in dem computerlesbaren Speichermedium gespeicherten computerausführbaren Anweisungen auszuführen.


 
11. Rechensystem nach Anspruch 10, wobei der Crash-Handler (600) des Weiteren eine Komponente umfasst, die daran angepasst ist, einen Namen der verantwortlichen Funktion unter Anwendung von Symboltabelleninformation, die mit der Anwendung assoziiert ist, zu identifizieren.
 
12. Rechensystem nach Anspruch 10, wobei das Rechensystem daran angepasst ist, unter Steuerung eines Betriebssystems mit Mach-basiertem Kernel und einem BSD-basierten Kernel ausgeführt zu werden, und mehrere Crash-Handlers installiert werden, um eine oder mehrere Mach-Ausnahmen und BSD-Signale, die erzeugt werden, wenn die Anwendung abstürzt, zu behandeln.
 


Revendications

1. Support de stockage pouvant être lu par ordinateur stockant des instructions exécutables par ordinateur permettant de piloter un dispositif informatique pour qu'il génère un rapport de plantage lorsqu'une application se bloque, l'application étant organisée sous forme d'un module (310) comportant des fonctions, les instructions exécutables par ordinateur comprenant des instructions qui :

installent un gestionnaire de plantage (600) à exécuter dans le même traitement que l'application lorsque l'application se bloque, l'application et le gestionnaire de plantage s'exécutant dans un mode utilisateur à l'intérieur du traitement, et

identifient une chaîne d'appels de fonctions utilisant des décalages, sous le contrôle du gestionnaire de plantage, s'exécutant dans le traitement de l'application après que l'application s'est bloquée, ce qui comprend :

identifier un processus victime du traitement qui s'exécutait à l'instant d'un plantage,

récupérer un pointeur de pile pour le processus victime et un pointeur d'instruction vers une instruction à l'instant du plantage,

identifier une fonction qui s'exécutait à l'instant du plantage sur la base dudit pointeur d'instruction en utilisant une mappe de fonctions de modules, ladite mappe de fonctions de modules contenant une entrée pour chaque fonction du module en même temps qu'une référence vers le point de démarrage de ce module dans l'espace d'adressage de l'application,

parcourir une pile d'appels du processus victime afin d'identifier des adresses de renvoi permettant de revenir des fonctions appelées de l'application,

identifier les décalages à l'intérieur du module, les décalages correspondant aux adresses de renvoi, et identifier les fonctions à l'intérieur du module qui incluent les décalages identifiés, et

créent un rapport de plantage fondé sur la chaîne d'appels identifiée.


 
2. Support de stockage pouvant être lu par ordinateur selon la revendication 1, dans lequel le dispositif informatique s'exécute sous le contrôle d'un système d'exploitation comportant un noyau sur base Mach et un noyau sur base BSD, et de multiples gestionnaires de plantage sont installés pour prendre en charge les exceptions de type Mach et les signaux BSD générés lorsque l'application se bloque.
 
3. Support de stockage pouvant être lu par ordinateur selon la revendication 1, dans lequel les instructions exécutables par ordinateur comprennent en outre des instructions qui identifient une fonction à rendre responsable du plantage sur la base de l'analyse de la chaîne d'appels.
 
4. Support de stockage pouvant être lu par ordinateur selon la revendication 3, dans lequel une fonction qui est désignée comme insensible est exclue du rôle de responsable pour un plantage.
 
5. Support de stockage pouvant être lu par ordinateur selon la revendication 1, dans lequel les instructions exécutables par ordinateur comprennent en outre des instructions pour identifier une fonction de l'application à rendre responsable d'un plantage de l'application, les instructions exécutables par ordinateur comprenant des instructions, exécutées par le gestionnaire de plantage lorsque l'application se bloque, pour :

identifier des décalages à l'intérieur du module correspondant à des adresses de renvoi et identifier les fonctions à l'intérieur du module qui incluent les décalages identifiés,

identifier une chaîne d'appels de fonctions invoquées à l'instant du plantage en utilisant les décalages identifiés,

accéder à des informations identifiant une ou plusieurs fonctions comme étant préservées de se voir rendues responsables du plantage,

traiter la chaîne d'appels de fonctions pour identifier une dernière fonction invoquée qui n'est pas préservée de se voir rendue responsable du plantage, et

désigner cette dernière fonction comme responsable du plantage,

dans lequel le gestionnaire de plantage s'exécute dans le même traitement que l'application.


 
6. Support de stockage pouvant être lu par ordinateur selon la revendication 5, dans lequel les fonctions de l'application sont stockées dans le module (310), et une fonction qui est préservée est spécifiée par son décalage à l'intérieur du module.
 
7. Support de stockage pouvant être lu par ordinateur selon la revendication 6, dans lequel les instructions qui identifient une chaîne d'appels de fonctions parcourent une pile d'appels de l'application et mappent des adresses de renvoi pour les fonctions appelées sur des décalages à l'intérieur du module (310).
 
8. Support de stockage pouvant être lu par ordinateur selon la revendication 5, dans lequel les instructions exécutables par ordinateur comprennent en outre des instructions qui identifient le nom de la fonction désignée comme responsable en utilisant des informations de la table de symboles associées à l'application.
 
9. Support de stockage pouvant être lu par ordinateur selon la revendication 5, dans lequel les instructions exécutables par ordinateur comprennent en outre des instructions qui téléchargent vers l'amont un rapport de plantage jusqu'à un serveur de rapports de plantage, le rapport de plantage identifiant la fonction reconnue responsable du plantage.
 
10. Système informatique destiné à générer un rapport de plantage lorsqu'une application se bloque, l'application étant organisée sous forme d'un module (310) comportant des fonctions, le système informatique comprenant :

un support de stockage pouvant être lu par ordinateur stockant des instructions exécutables par ordinateur concernant :

le module, et

un gestionnaire de plantage (600) destiné à s'exécuter dans le même traitement que l'application lorsque l'application se bloque, l'application et le gestionnaire de plantage s'exécutant dans un mode utilisateur à l'intérieur du traitement, le gestionnaire de plantage identifiant une chaîne d'appels de fonctions utilisant des décalages en étant conçu pour :

identifier un processus victime du traitement, le processus victime étant un processus du traitement qui s'exécutait à l'instant d'un plantage,

récupérer un pointeur de pile sur une pile d'appels du processus victime et un pointeur d'instruction vers une instruction à l'instant de plantage,

identifier une fonction qui s'exécutait à l'instant de plantage sur la base dudit pointeur d'instruction en utilisant une mappe de fonctions de modules, ladite mappe de fonctions de modules contenant une entrée pour chaque fonction du module en même temps qu'une référence au point de démarrage de ce module dans l'espace d'adressage de l'application,

parcourir la pile d'appels du processus victime afin d'identifier des adresses de renvoi pour les fonctions appelées, les adresses de renvoi indiquant un emplacement de retour à l'intérieur d'une fonction appelante,

identifier les décalages à l'intérieur du module, les décalages correspondant aux adresses de renvoi et au pointeur d'instruction,

le gestionnaire de plantage comprenant en outre :

un composant conçu pour identifier une fonction comme responsable du plantage, la fonction à rendre responsable du plantage étant la dernière fonction appelée avant le plantage qui n'est pas préservée de se voir rendue responsable d'un plantage, comme indiqué par une table de fonctions insensibles,

un composant conçu pour créer un rapport de plantage qui identifie la fonction à rendre responsable du plantage et une identification de session d'un journal d'activité d'application pour l'exécution de l'application, et

un composant conçu pour télécharger en amont le rapport de plantage jusqu'à un serveur de rapports de plantage lorsqu'un critère de téléchargement montant est satisfait, et

dans lequel le gestionnaire de plantage est en outre conçu pour identifier les fonctions à l'intérieur du module qui incluent les décalages identifiés, et

un processeur conçu pour exécuter les instructions exécutables par ordinateur stockées sur le support de stockage pouvant être lu par ordinateur.


 
11. Système informatique selon la revendication 10, dans lequel le gestionnaire de plantage (600) comprend en outre un composant conçu pour identifier le nom de la fonction à rendre responsable en utilisant des informations de table de symboles associées à l'application.
 
12. Système informatique selon la revendication 10, dans lequel le système informatique est conçu pour s'exécuter sous le contrôle d'un système d'exploitation comportant un noyau sur base Mach et un noyau sur base BSD, et de multiples gestionnaires de plantage sont installés pour prendre en charge une ou plusieurs exceptions de type Mach et les signaux BSD générés lorsque l'application se bloque.
 




Drawing



































Cited references

REFERENCES CITED IN THE DESCRIPTION



This list of references cited by the applicant is for the reader's convenience only. It does not form part of the European patent document. Even though great care has been taken in compiling the references, errors or omissions cannot be excluded and the EPO disclaims all liability in this regard.

Patent documents cited in the description




Non-patent literature cited in the description