how can i write C++ code to check performance

Springer.The.Developer’s.Guide.to.Debugging.2008.pdf

1 You Write Software; You have Bugs . . . . . . . . . 1 2 A Systematic Approach to Debugging . . . . . . . 5 2.1 Why Follow a Structured Process? . . . . . . . 5 2.2 Making the Most of Your Opportunities . . . . . . 5 2.3 13 Golden Rules . . . . . . . . . . . . . 7 2.3.1 Understand the Requirements . . . . . . 8 2.3.2 Make it Fail . . . . . . . . . . . . 8 2.3.3 Simplify the Test Case . . . . . . . . . 9 2.3.4 Read the Right Error Message . . . . . . 9 2.3.5 Check the Plug . . . . . . . . . 9 2.3.6 Separate Facts from Interpretation . . . . . 10 2.3.7 Divide and Conquer . . . . . . . . 10 2.3.8 Match the Tool to the Bug . . . . . . . . . 12 2.3.9 One Change at a Time . . . . . . . . . 12 2.3.10 Keep an Audit Trail . . . . . . . . 12 2.3.11 Get a Fresh View . . . . . . . . . . . 13 2.3.12 If You Didn’t Fix it, it Ain’t Fixed . . . . . 13 2.3.13 Cover your Bugfix with a Regression Test . . . . . 13 2.4 Build a Good Toolkit . . . . . . . . . . . . 14 2.4.1 Your Workshop . . . . . . . . . 15 2.4.2 Running Tests Every Day Keeps the Bugs at Bay . . 15 2.5 Know Your Enemy – Meet the Bug Family . . . . . . 17 2.5.1 The Common Bug . . . . . . . . . . 17 2.5.2 Sporadic Bugs . . . . . . . . . . 18 2.5.3 Heisenbugs. . . . . . . . . . 18 2.5.4 Bugs Hiding Behind Bugs . . . . . . . . . 19 2.5.5 Secret Bugs – Debugging and Confidentiality . . . . . 20 2.5.6 Further Reading . . . . . . . . . 21 3 Getting to the Root – Source Code Debuggers . . . . . . 23 3.1 Visualizing Program Behavior . . . . . . . . 23 3.2 Prepare a Simple Predictable Example . . . . . . . 24 3.3 Get the Debugger to Run with Your Program . . . . . . . 24 3.4 Learn to do a Stack Trace on a Program Crash . . . . . . 27 3.5 Learn to Use Breakpoints . . . . . . . . . 28 3.6 Learn to Navigate Through the Program. . . . . . 28 3.7 Learn to Inspect Data: Variables and Expressions . . . . 29 3.8 A Debug Session on a Simple Example . . . . . . 30 4 Fixing Memory Problems . . . . . . . . . . . . 33 4.1 Memory Management in C/C++ – Powerful but Dangerous . . . . 33 4.1.1 Memory Leaks . . . . . . . . . 34 4.1.2 Incorrect Use of Memory Management . . . . 34 4.1.3 Buffer Overruns . . . . . . . . . . . 34 4.1.4 Uninitialized Memory Bugs . . . . . . . . 34 4.2 Memory Debuggers to the Rescue . . . . . . . . 35 4.3 Example 1: Detecting Memory Access Errors . . . . . . . 36 4.3.1 Detecting an Invalid Write Access . . . . . 36 4.3.2 Detecting Uninitialized Memory Reads . . . . 37 4.3.3 Detecting Memory Leaks . . . . . . . 38 4.4 Example 2: Broken Calls to Memory Allocation/Deallocation . . 38 4.5 Combining Memory and Source Code Debuggers . . . . . . 40 4.6 Cutting Down the Noise – Suppressing Errors . . . . . . . 40 4.7 When to Use a Memory Debugger . . . . . . . 41 4.8 Restrictions . . . . . . . . . . . . . . 42 4.8.1 Prepare Test Cases with Good Code Coverage . . . . 42 4.8.2 Provide Additional Computer Resources . . . . . . 42 4.8.3 Multi-Threading May Not be Supported . . . . . . 42 4.8.4 Support for Non-standard Memory Handlers . . . 42 5 Profiling Memory Use . . . . . . . . . . . . 45 5.1 Basic Strategy – The First Steps . . . . . . . . . 45 5.2 Example 1: Allocating Arrays . . . . . . . . 46 5.3 Step 1: Look for Leaks . . . . . . . . . . . 46 5.4 Step 2: Set Your Expectations . . . . . . . . 47 5.5 Step 3: Measure Memory Consumption . . . . . . 47 5.5.1 Use Multiple Inputs . . . . . . . . 48 5.5.2 Stopping the Program at Regular Intervals . . . . . 48 5.5.3 Measuring Memory Consumption with Simple Tools . . 49 5.5.4 Use top . . . . . . . . . . . . 49 5.5.5 Use the Windows Task Manager . . . . . . . 50 5.5.6 Select Relevant Input Values for testmalloc . . . 51 5.5.7 Determine how Memory is Deallocated on Your Machine . 51 5.5.8 Use a Memory Profiler . . . . . . . . . 53 5.6 Step 4: Identifying Greedy Data Structures . . . . . . 54 5.6.1 Instrumenting Data Structures . . . . . . 55 5.7 Putting it Together – The genindex Example. . . . . . 55 5.7.1 Check that There are No Major Leaks . . . . . 56 5.7.2 Estimate the Expected Memory Use . . . . . . . 56 5.7.3 Measure Memory Consumption . . . . . . . 57 5.7.4 Find the Data Structures that Consume Memory . . . 57 6 Solving Performance Problems . . . . . . . . . . 63 6.1 Finding Performance Bugs – A Step-by-Step Approach . . . . . 63 6.1.1 Do an Upfront Analysis . . . . . . . . 64 6.1.2 Use a Simple Method of Measuring Time . . . . . 64 6.1.3 Create a Test Case . . . . . . . . . . 65 6.1.4 Make the Test Case Reproducible . . . . . . 65 6.1.5 Check the Program for Correctness . . . . . . . 66 6.1.6 Make the Test Case Scalable . . . . . . . 66 6.1.7 Isolate the Test Case from Side Effects . . . . 67 6.1.8 Measurement with time can have Errors and Variations . 68 6.1.9 Select a Test Case that Exposes the Runtime Bottleneck . . 68 6.1.10 The Difference Between Algorithm and Implementation . . 70 6.2 Using Profiling Tools . . . . . . . . . . . . 72 6.2.1 Do Not Write Your Own Profiler . . . . . . . 72 6.2.2 How Profilers Work . . . . . . . . 73 6.2.3 Familiarize Yourself with gprof . . . . . . 74 6.2.4 Familiarize Yourself with Quantify . . . . . 79 6.2.5 Familiarize Yourself with Callgrind . . . . . . . 81 6.2.6 Familiarize Yourself with VTune . . . . . . 82 6.3 Analyzing I/O Performance . . . . . . . . . . 84 6.3.1 Do a Sanity Check of Your Measurements . . . . . 85 7 Debugging Parallel Programs . . . . . . . . . 87 7.1 Writing Parallel Programs in C/C++ . . . . . . . . . 87 7.2 Debugging Race Conditions . . . . . . . . . . 88 7.2.1 Using Basic Debugger Capabilities to Find Race Conditions . . . . . . . . . . 89 7.2.2 Using Log Files to Localize Race Conditions . . . . . 91 7.3 Debugging Deadlocks . . . . . . . . . . . . 93 7.3.1 How to Determine What the Current Thread is Executing . 94 7.3.2 Analyzing the Threads of the Program . . . . . 95 7.4 Familiarize Yourself with Threading Analysis Tools . . . . . 96 7.5 Asynchronous Events and Interrupt Handlers . . . . . . . 98 8 Finding Environment and Compiler Problems . . . . . . 101 8.1 Environment Changes – Where Problems Begin . . . . . 101 8.1.1 Environment Variables . . . . . . . . . 101 8.1.2 Local Installation Dependencies . . . . . . . 102 8.1.3 Current Working Directory Dependency . . . . . . 102 8.1.4 Process ID Dependency . . . . . . . . 102 8.2 How else to See what a Program is Doing . . . . . . . 103 8.2.1 Viewing Processes with top . . . . . . . 103 8.2.2 Finding Multiple Processes of an Application with ps . . . 103 8.2.3 Using /proc/<pid> to Access a Process . . . . 104 8.2.4 Use strace to Trace Calls to the OS . . . . . 104 8.3 Compilers and Debuggers have Bugs too . . . . . . . . 106 8.3.1 Compiler Bugs . . . . . . . . . 106 8.3.2 Debugger and Compiler Compatibility Problems . . 107 9 Dealing with Linking Problems . . . . . . . . . . 109 9.1 How a Linker Works . . . . . . . . . . 109 9.2 Building and Linking Objects . . . . . . . . 110 9.3 Resolving Undefined Symbols . . . . . . . . 111 9.3.1 Missing Linker Arguments . . . . . . . . 111 9.3.2 Searching for Missing Symbols . . . . . . . . 112 9.3.3 Linking Order Issues . . . . . . . . . . 113 9.3.4 C++ Symbols and Name Mangling . . . . . 114 9.3.5 Demangling of Symbols . . . . . . . . 115 9.3.6 Linking C and C++ Code . . . . . . . 115 9.4 Symbols with Multiple Definitions . . . . . . . 116 9.5 Symbol Clashes . . . . . . . . . . . 117 9.6 Identifying Compiler and Linker Version Mismatches . . . 118 9.6.1 Mismatching System Libraries . . . . . . . . 119 9.6.2 Mismatching Object Files . . . . . . . . . 119 9.6.3 Runtime Crashes . . . . . . . . . . . 120 9.6.4 Determining the Compiler Version . . . . . 120 9.7 Solving Dynamic Linking Issues . . . . . . . . . 122 9.7.1 Linking or Loading DLLs . . . . . . . . . 122 9.7.2 DLL Not Found . . . . . . . . . 124 9.7.3 Analyzing Loader Issues . . . . . . . 125 9.7.4 Setting Breakpoints in DLLs . . . . . . . 126 9.7.5 Provide Error Messages for DLL Issues . . . . 127 10 Advanced Debugging . . . . . . . . . . . . . 129 10.1 Setting Breakpoints in C++ Functions, Methods, and Operators . 129 10.2 Setting Breakpoints in Templatized Functions and C++ Classes . 131 10.3 Stepping in C++ Methods . . . . . . . . . 133 10.3.1 Stepping into Implicit Functions . . . . . . . 134 10.3.2 Skipping Implicit Functions with the Step-out Command . 135 10.3.3 Skipping Implicit Functions with a Temporary Breakpoint 136 10.3.4 Returning from Implicit Function Calls . . . . 136 10.4 Conditional Breakpoints and Breakpoint Commands . . . . 137 10.5 Debugging Static Constructor/Destructor Problems . . . . . 140 10.5.1 Bugs Due to Order-Dependence of Static Initializers . . 140 10.5.2 Recognizing the Stack Trace of Static Initializers . . 141 10.5.3 Attaching the Debugger Before Static Initialization . . . 142 10.6 Using Watchpoints . . . . . . . . . . . 143 10.7 Catching Signals . . . . . . . . . . . . . 144 10.8 Catching Exceptions . . . . . . . . . . 147 10.9 Reading Stack Traces . . . . . . . . . . . . 148 10.9.1 Stack Trace of Source Code Compiled with Debug Information . . . . . . . . 148 10.9.2 Stack Trace of Source Code Compiled Without Debug Information . . . . . . . . 149 10.9.3 Frames Without Any Debug Information . . . . . . 149 10.9.4 Real-Life Stack Traces . . . . . . . . . 150 10.9.5 Mangled Function Names . . . . . . . . . 151 10.9.6 Broken Stack Traces . . . . . . . . 151 10.9.7 Core Dumps . . . . . . . . . . . . 152 10.10 Manipulating a Running Program . . . . . . . . 153 10.10.1 Changing a Variable . . . . . . . . 156 10.10.2 Calling Functions . . . . . . . . . . 156 10.10.3 Changing the Return Value of a Function . . . . . . 157 10.10.4 Aborting Function Calls . . . . . . . . 157 10.10.5 Skipping or Repeating Individual Statements . . . . . 158 10.10.6 Printing and Modifying Memory Content . . . . . 159 10.11Debugging Without Debug Information . . . . . . 161 10.11.1 Reading Function Arguments From the Stack . . . . . 163 10.11.2 Reading Local/Global Variables, User-Defined Data Types 165 10.11.3 Finding the Approximate Statement in the Source Code . . 165 10.11.4 Stepping Through Assembly Code . . . . . 166 11 Writing Debuggable Code . . . . . . . . . . . . 169 11.1 Why Comments Count . . . . . . . . . . . 169 11.1.1 Comments on Function Signatures . . . . . 170 11.1.2 Comments on Workarounds . . . . . . . . 171 11.1.3 Comments in Case of Doubt . . . . . . . 171 11.2 Adopting a Consistent Programming Style . . . . . . 171 11.2.1 Choose Names Carefully . . . . . . . 171 11.2.2 Avoid Insanely Clever Constructs . . . . . . 172 11.2.3 Spread Out Your Code . . . . . . . . . 172 11.2.4 Use Temporary Variables for Complex Expressions . . . 172 11.3 Avoiding Preprocessor Macros . . . . . . . . . . 173 11.3.1 Use Constants or Enums Instead of Macros . . . . 173 11.3.2 Use Functions Instead of Preprocessor Macros . . . . 175 11.3.3 Debug the Preprocessor Output . . . . . . . . 176 11.3.4 Consider Using More Powerful Preprocessors . . . . 177 11.4 Providing Additional Debugging Functions . . . . . . 179 11.4.1 Displaying User-Defined Data Types . . . . . . 179 11.4.2 Self-Checking Code . . . . . . . . 180 11.4.3 Debug Helpers for Operators . . . . . . . 181 11.5 Prepare for Post-Processing . . . . . . . . . . 181 11.5.1 Generate Log Files . . . . . . . . . 181 12 How Static Checking Can Help . . . . . . . . . . 183 12.1 Using Compilers as Debugging Tools . . . . . . . . 183 12.1.1 Do not Assume Warnings to be Harmless . . . . . 184 12.1.2 Use Multiple Compilers to Check the Code . . . . 186 12.2 Using lint . . . . . . . . . . . . . . 186 12.3 Using Static Analysis Tools . . . . . . . . . . 187 12.3.1 Familiarize Yourself with a Static Checker . . . . 187 12.3.2 Reduce Static Checker Errors to (Almost) Zero . . . 189 12.3.3 Rerun All Test Cases After a Code Cleanup . . . 190 12.4 Beyond Static Analysis . . . . . . . . . . . 190 13 Summary . . . . . . . . . . . . . . 191 A Debugger Commands . . . . . . . . . . . . . 193 B Access to Tools . . . . . . . . . . . . . 195 B.1 IDEs, Compilers, Build Tools . . . . . . . . 195 B.1.1 Microsoft Visual Studio . . . . . . . . 195 B.1.2 Eclipse . . . . . . . . . . . . . 196 B.1.3 GCC . . . . . . . . . . . . 196 B.1.4 GNU Make . . . . . . . . . . . . 196 B.2 Debuggers . . . . . . . . . . . . 196 B.2.1 dbx . . . . . . . . . . . . . 196 B.2.2 DDD . . . . . . . . . . . . 197 B.2.3 GDB . . . . . . . . . . . . 197 B.2.4 ARM RealView . . . . . . . . . 197 B.2.5 TotalView Debugger . . . . . . . . 197 B.2.6 Lauterbach TRACE32 . . . . . . . . . 197 B.3 Environments . . . . . . . . . . . . . 198 B.3.1 Cygwin . . . . . . . . . . . . . 198 B.3.2 VMware . . . . . . . . . . . . 198 B.4 Memory Debuggers . . . . . . . . . . . 198 B.4.1 Purify . . . . . . . . . . . 198 B.4.2 Valgrind . . . . . . . . . . . . 199 B.4.3 KCachegrind . . . . . . . . . . . 199 B.4.4 Insure++ . . . . . . . . . . . . 199 B.4.5 BoundsChecker . . . . . . . . . 200 B.5 Profilers . . . . . . . . . . . . . . 200 B.5.1 gprof . . . . . . . . . . . . 200 B.5.2 Quantify . . . . . . . . . . . . 200 B.5.3 Intel VTune . . . . . . . . . . . . 200 B.5.4 AQtime . . . . . . . . . . . . . 201 B.5.5 mpatrol . . . . . . . . . . . . . 201 B.6 Static Checkers . . . . . . . . . . . 201 B.6.1 Coverity . . . . . . . . . . . . 201 B.6.2 Lint . . . . . . . . . . . . . 201 B.6.3 Splint . . . . . . . . . . . 202 B.6.4 /analyze option in Visual Studio Enterprise Versions . . 202 B.6.5 Klocwork . . . . . . . . . . . 202 B.6.6 Fortify . . . . . . . . . . . . . 202 B.6.7 PC-lint/FlexeLint . . . . . . . . . . 203 B.6.8 QA C++ . . . . . . . . . . . . 203 B.6.9 Codecheck . . . . . . . . . . 203 B.6.10 Axivion Bauhaus Suite . . . . . . . . . 203 B.6.11 C++ SoftBench CodeAdvisor . . . . . . 203 B.6.12 Parasoft C++test . . . . . . . . . . . 204 B.6.13 LDRA tool suite . . . . . . . . . . . 204 B.6.14 Understand C++ . . . . . . . . . . . 204 B.7 Tools for Parallel Programming . . . . . . . . . . 204 B.7.1 Posix Threads . . . . . . . . . . 204 B.7.2 OpenMP . . . . . . . . . . . . 204 B.7.3 Intel TBB . . . . . . . . . . . 205 B.7.4 MPI . . . . . . . . . . . . . 205 B.7.5 MapReduce . . . . . . . . . . . . 205 B.7.6 Intel Threading Analysis Tools . . . . . . . . 205 B.8 Miscellaneous Tools . . . . . . . . . . 206 B.8.1 GNU Binutils . . . . . . . . . . . 206 B.8.2 m4 . . . . . . . . . . . . . . 206 B.8.3 ps . . . . . . . . . . . . . . 206 B.8.4 strace / truss . . . . . . . . . . 207 B.8.5 top. . . . . . . . . . . . . 207 B.8.6 VNC . . . . . . . . . . . . 207 B.8.7 WebEx . . . . . . . . . . . . . 207 C Source Code . . . . . . . . . . . . . . . 209 C.1 testmalloc.c . . . . . . . . . . . . . . 209 C.2 genindex.c . . . . . . . . . . . . 210 C.3 isort.c . . . . . . . . . . . . . 214 C.4 filebug.c . . . . . . . . . . . . . . 216 References . . . . . . . . . . . . . . 217 Index . . . . . . . . . . . . . . . . 219

Windows XP source code leaked version download

代码下载地址： https://pan.quark.cn/s/d04f5c0b822b Windows XP以及Windows 2003的源代码存在泄露情况。这些被泄露的版本，这些被泄露的版本。被泄露的版本，被泄露的版本，被泄露的版本，被泄露的版本，被泄露的版本，被泄露的版本，被泄露的版本。

Bochs - The cross platform IA-32 (x86) emulator

Changes in 2.4.6 (February 22, 2011): Brief summary : - Support more host OS to run on: - Include win64 native binary in the release. - Fixed failures on big endian hosts. - BIOS: Support for up to 2M ROM BIOS images. - GUI: select mouse capture toggle method in .bochsrc. - Ported most of Qemu's 'virtual VFAT' block driver (except runtime write support, but plus FAT32 suppport) - Added write protect option for floppy drives. - Bugfixes / improved internal debugger + instrumentation. Detailed change log : - CPU and internal debugger - Implemented Process Context ID (PCID) feature - Implemented FS/GS BASE access instructions support (according to document from http://software.intel.com/en-us/avx/) - Rewritten from scratch SMC detection algorithm - Implemented fine-grained SMC detection (on 128 byte granularity) - Bugfixes for CPU emulation correctness and stability - Fixed failures on Big Endian hosts ! - Print detailed page walk information and attributes in internal debugger 'page' command - Updated/Fixed instrumentation callbacks - Configure and compile - Bochs now can be compiled as native Windows x86-64 application (tested with Mingw gcc 4.5.1 and Microsoft Visual Studio Express 2010) - Added ability to configure CPUID stepping through .bochsrc. The default stepping value is 3. - Added ability to disable MONITOR/MWAIT support through .bochsrc CPUID option. The option is available only if compiled with --enable-monitor-mwait configure option. - Determine and select max physical address size automatically at configure time: - 32-bit physical address for 386/486 guests - 36-bit physical address for PSE-36 enabled Pentium guest - 40-bit physical address for PAE enabled P6 or later guests - Update config.guess/config.sub scripts to May 2010 revisions. - Update Visual Studio 2008 project files in build/win32/vs2008ex-workspace.zip - Added Bochs compilation timestamp after Bochs version string. - GUI and display libraries (Volker) - Added new .bochsrc option to select mouse capture toggle method. In addition to the default Bochs method using the CTRL key and the middle mouse button there are now the choices: - CTRL+F10 (like DOSBox) - CTRL+ALT (like QEMU) - F12 (replaces win32 'legacyF12' option) - display library 'x' now uses the desktop size for the maximum guest resolution - ROM BIOS - Support for up to 2M ROM BIOS images - I/O Devices - 3 new 'pseudo device' plugins created by plugin separation (see below) - Fixes for emulated DHCP in eth_vnet (patch from @SF tracker) - Added support for VGA graphics mode with 400 lines (partial fix for SF bug #2948724) - NE2K: Fixed "send buffer" command issue on big endian hosts - USB - converted common USB code plus devices to the new 'usb_common' plugin Now the USB device classes no longer exist twice if both HC plugins are loaded. - added 'pseudo device' in common USB code for the device creation. This makes the HCs independent from the device specific code. - USB MSD: added support for disk image modes (like ATA disks) - USB printer: output file creation failure now causes a disconnect - re-implemented "options" parameter for additional options of connected devices (currently only used to set the speed reported by device and to specify an alternative redolog file of USB MSD disk image modes) - hard drive - new disk image mode 'vvfat' - ported the read-only part of Qemu's 'virtual VFAT' block driver - additions: configurable disk geometry, FAT32 support, read MBR and/or boot sector from file, volatile write support using hdimage redolog_t class, optional commit support on Bochs exit, save/restore file attributes, 1.44 MB floppy support, set file modification date/time - converted the complete hdimage stuff to the new 'hdimage' plugin - new hdimage method get_capabilities() that can return special flags - vmware3, vmware4 and vvfat classes now return HDIMAGE_HAS_GEOMETRY flag - other disk image modes by default return HDIMAGE_AUTO_GEOMETRY if cylinder value is set to 0 - multiple sector read/write support for some image modes - new log prefix "IMG" for hdimage messages - floppy - added write protect option for floppy drives (based on @SF patch by Ben Lunt) - vvfat support - bugfix: close images on exit - SB16 - converted the sound output module stuff to the new 'soundmod' plugin - SF patches applied [3164945] hack to compile under WIN64 by Darek Mihocka and Stanislav [3164073] Fine grain SMC invalidation by Stanislav [1539417] write protect for floppy drives by Ben Lunt [2862322] fixes for emulated DHCP in eth_vnet - these S.F. bugs were closed/fixed [2588085] Mouse capture [3140332] typo in mf3/ps2 mapping of BX_KEY_CTRL_R [3111577] No "back" option in log settings [3108422] Timing window in NE2K emulation [3084390] Bochs won't load floppy plugin right on startup [3043174] Docbook use of '_' build failure [3085140] Ia_arpl_Ew_Rw definition of error [3078995] ROL/ROR/SHL/SHR modeling wrong when dest reg is 32 bit [2864794] BX_INSTR_OPCODE in "cpu_loop" causes crash in x86_64 host [2884071] [AIX host] prefetch: EIP [00010000] > CS.limit [0000ffff] [3053542] 64 bit mode: far-jmp instruction is error [3011112] error compile vs2008/2010 with X2APIC [3002017] compile error with vs 2010 [3009767] guest RFLAGS.IF blocks externel interrupt in VMX guest mode [2964655] VMX not enabled in MSR IA32_FEATURE_CONTROL [3005865] IDT show bug [3001637] CMOS MAP register meaning error [2994370] Cannot build with 3DNow support - these S.F. feature requests were closed/implemented [1510142] Native Windows XP x64 Edition binary [1062553] select mouse (de)activation in bochsrc [2930633] legacy mouse capture key : not specific enough [2930679] Let user change mouse capture control key [2803538] Show flags for pages when using "info tab" ------------------------------------------------------------------------- Changes in 2.4.5 (April 25, 2010): Brief summary : - Major configure/cpu rework allowing to enable/disable CPU options at runtime through .bochsrc (Stanislav) - Bugfixes for CPU emulation correctness and stability - Implemented X2APIC extensions (Stanislav) - Implemented Intel VMXx2 extensions (Stanislav) - Extended VMX capability MSRs, APIC Virtualization, X2APIC Virtualization, Extended Page Tables (EPT), VPID, Unrestricted Guests, new VMX controls. - Implemented PCLMULQDQ AES instruction - Extended Bochs internal debugger functionality - USB HP DeskJet 920C printer device emulation (Ben Lunt) Detailed change log : - Configure rework - Deprecate --enable-popcnt configure option. POPCNT instruction will be enabled automatically iff SSE4_2 is supported (like in hardware). - Make --ignore-bad-msrs runtime option in .bochsrc. Old --ignore-bad-msrs configure option is deprecated and should not be used anymore. - Enable changing part of CPU functionality at runtime through .bochsrc. - Now you could enable/disable any of SSEx/AES/MOVBE/SYSENTER_SYSEXIT/XSAVE instruction sets using new CPUID option in .bochsrc. - When x86-64 support is compiled in, you could enable/disable long mode 1G pages support without recompile using new CPUID option in .bochsrc. Configure options: --enable-mmx, --enable-sse, --enable-movbe, --enable-xsave, --enable-sep, --enable-aes, --enable-1g-pages are deprecated and should not be used anymore. - Local APIC configure option --enable-apic is deprecated and should not be used anymore. The LAPIC option now automatically determined from other configure options. XAPIC functionality could be enabled using new CPUID .bochsrc option. - Changed default CPU configuration (generated by configure script with default options) to BX_CPU_LEVEL=6 with SSE2 enabled. - CPU - Implemented PCLMULQDQ AES instruction - Implemented X2APIC extensions / enable extended topology CPUID leaf (0xb), in order to enable X2APIC configure with --enable-x2apic - Implemented Intel VMXx2 extensions: - Enabled extended VMX capability MSRs - Implemented VMX controls for loading/storing of MSR_PAT and MSR_EFER - Enabled/Implemented secondary proc-based vmexec controls: - Implemented APIC virtualization - Implemented Extended Page Tables (EPT) mode - Implemented Descriptor Table Access VMEXIT control - Implemented RDTSCP VMEXIT control - Implemented Virtualize X2APIC mode control - Implemented Virtual Process ID (VPID) - Implemented WBINVD VMEXIT control - Implemented Unrestricted Guest mode In order to enable emulation of VMXx2 extensions configure with --enable-vmx=2 option (x86-64 must be enabled) - Bugfixes for CPU emulation correctness - Fixed Bochs crash when accessing the first byte above emulated memory size - Internal Debugger - Introduced range read/write physical watchpoints - Allow reloading of segment registers from internal debugger - Improved verbose physical memory access tracing - BIOS - Fix MTRR configuration (prevented boot of modern Linux kernels) - Fix interrupt vectors for INT 60h-66h (reserved for user interrupt) by setting them to zero - Fix BIOS INT13 function 08 when the number of cylinders on the disk = 1 - I/O Devices - USB HP DeskJet 920C printer device emulation (Ben Lunt) - Misc - Updated Bochs TESTFORM to version 0.5 - SF patches applied [2864402] outstanding x2apic patches by Stanislav [2960379] Fix build with -Wformat -Werror=format-security by Per Oyvind Karlsen [2938273] allow instrumentation to change execute by Konrad Grochowski [2926072] Indirection operators in expressions by Derek Peschel [2914433] makesym.perl misses symbols by John R. Jackson [2908481] USB Printer by Ben Lunt - these S.F. bugs were closed/fixed [2861662] dbg_xlate_linear2phy needs to be updated [2956217] INT13 AH=8 returns wrong values when cylinders=1 [2981161] Allow DMA transfers to continue when CPU is in HALT state [2795115] NX fault could be missed [2964824] bad newline sequence in aspi-win32.h [913419] configure options and build process needs some work [2938398] gdbstub compile error with x86_64 enabled [2734455] shutdown/reset type 05 should reinit the PICs [1921294] extended memory less than 1M wrong size [1947249] BX_USE_EBDA_TABLES and MP table placement [1933859] BX_USE_EBDA_TABLES and memory overlapping [2923680] "help dregs" is a syntax error [2919661] CPU may fail to do 16bit near call [2790768] Memory corruption with SMP > 32, Panic BIOS Keyboard Error [2902118] interrupts vectors 0x60 to 67 should be NULL ! [2912502] Instruction Pointer behaving erratically [2901047] Bochs crashed, closed by guest os [2905385] Bochs crash [2901481] Instruction SYSRET and SS(PL) [2900632] Broken long mode RETF to outer priviledge with null SS [1429011] Use bx_phyaddr_t for physaddr vars and bx_adress for lin adr - these S.F. feature requests were closed/implemented [2955911] RPM preuninstall scriptlet removes /core [2947863] don't abort on unrecognised options [2878861] numerics in the disassembler output [2900619] make more CPU state changeable ------------------------------------------------------------------------- Changes in 2.4.2 (November 12, 2009): - CPU and internal debugger - VMX: Implemented TPR shadow VMEXIT - Bugfixes for CPU emulation correctness (mostly for VMX support). - Bugfixes and updates for Bochs internal debugger - On SMP system stepN command now affects only current processor - Memory - Bugfixes for > 32-bit physical address space. - Allow to emulate more physical memory than host actually could or would like to allocate. For more details look for new .bochsrc 'memory' option. - Cleanup configure options - All paging related options now will be automatically determined according to --enable-cpu-level option. Related configure options --enable-global-pages, --enable-large-pages, --enable-pae, --enable-mtrr are deprecated now. Only 1G paging option still remaining unchanged. - Deprecate --enable-daz configure option. Denormals-are-zeros MXCSR control will be enabled automatically iff SSE2 is supported (like in hardware). - Deprecate --enable-vme configure option, now it will be supported iff CPU_LEVEL >= 5 (like in hardware). - I/O Devices - Bugfixes for 8254 PIT, VGA, Cirrus-Logic SVGA, USB UCHI - SF patches applied [2817840] Make old_callback static by Mark Marshall [2874004] fix for VMWRITE instruction by Roberto Paleari [2873999] fix CS segment type during fast syscall invocation by Roberto Paleari [2864389] Debugger gui maximize on startup by Thomas Nilsen [2817868] Rework loops in the memory code by Mark Marshall [2812948] PIT bug by Derek - these S.F. bugs were closed/fixed [2833504] GUI debugger bug-about GDT display [2872244] BIOS writes not allowed value to MTRR MSR causing #GP [2885383] SDL GUI memory leak [2872290] compilation in AIX5.3 ML10 failes [2867904] crash with cirrus bx_vga_c::mem_write [2851495] BIOS PCI returns with INT flag = 0 [2860333] vista 64 guest STOP 109 (GDT modification) [2849745] disassembler bug for 3DNow and SSE opcodes [1066748] Wrong registers values after #RESET, #INIT [2836893] Regression: Windows XP installer unable to format harddrive [2812239] VMX: VM-Exit: Incorrect instruction length on software int [2814130] bx_debug lex/yacc files incorrectly generated [2813199] MP Tables Missing From BIOS [2824093] VMX exception bug [2811909] VMX : CS Access-rights Type.Accessed stays 0 [2810571] Compile Errors on OSX [2823749] GCC regression or VM_EXIT RDMSR/WRMSR bug [2815929] Vista/XP64 unnecessary panic [2803519] Wrong example in man page bochsrc - these S.F. feature requests were closed/implemented [422766] Large Memory configurations [1311287] Idea for a better GUI [455971] USB support [615363] debugger shortcut for repeat last cmd ------------------------------------------------------------------------- Changes in 2.4.1 (June 7, 2009): - Fixed bunch of CPUID issues - Bochs is now able to install and boot 64-bit Windows images! (special thanks to Mark Ebersole for his patch) - Several bugfixes in CPU emulation (mostly for x87 instructions) - Fixed two critical deadlock bugs in the Win32 gui (patches from @SF tracker) - Fixes related to the 'show ips' feature - removed conflicting win32-specific alarm() functions ('win32' and 'sdl' gui) - feature now works in wx on win32 - Added support for gdb stub on big endian machine (patch by Godmar Back) - Rewritten obsolete hash_map code in dbg symbols module (patch from @SF) - BIOS: implemented missing INT 15h/89h (patch by Sebastian Herbszt) ------------------------------------------------------------------------- Changes in 2.4 (May 3, 2009): Brief summary : - Added graphical Bochs debugger frontend for most of the supported platforms. - Thanks for Chourdakis Michael and Bruce Ewing. - Many new CPU features in emulation - Support for > 32 bit physical address space and configurable MSRs - VMX, 1G pages in long mode, MOVBE instruction - Bugfixes for CPU emulation correctness, debugger and CPU instrumentation. - New config interface 'win32config' with start and runtime menu - USB: added OHCI support, external hub and cdrom - Added user plugin interface support. Detailed change log : - CPU and internal debugger - Support for VMX hardware emulation in Bochs CPU, to enable configure with --enable-vmx option Nearly complete VMX implementation, with few exceptions: - Dual-monitor treatment of SMIs and SMM not implemented yet - NMI virtualization, APIC virtualization not implemented yet - VMENTER to not-active state not supported yet - No advanced features like Extended Page Tables or VPID - Support for configurable MSR registers emulation, to enable configure with --enable-configurable-msrs option Look for configuration example in .bochsrc and msrs.def - Support new Intel Atom(R) MOVBE instruction, to enable configure with --enable-movbe option - Support for 1G pages in long mode, to enable configure with --enable-1g-pages option - Support for > 32 bit physical address space in CPU. Up to 36 bit could be seen in legacy mode (PAE) and up to 40 bit in x86-64 mode. Still support the same amount of the physical memory in the memory object, so system with > 4Gb of RAM yet cannot be emulated. To enable configure with --enable-long-phy-address option. - Implemented modern BIOSes mode limiting max reported CPUID function to 3 using .bochsrc CPU option. The mode is required in order to correctly install and boot WinNT. - Added ability to configure CPUID vendor/brand strings through .bochsrc (patch from @SF by Doug Reed). - Many bugfixes for CPU emulation correctness (both x86 and x86-64). - Updated CPU instrumentation callbacks. - Fixed Bochs internal debugger breakpoints/watchpoints handling. - Configure and compile - Added ability to choose Bochs log file name and Bochs debugger log file name from Bochs command line (using new -log and -dbglog options) - Removed Peter Tattam's closed source external debugger interface from the code. - Removed --enable-guest2host-tlb configure option. The option is always enabled for any Bochs configuration. - Removed --enable-icache configure option. The option is always enabled for any Bochs configuration. Trace cache support still remains optional and could be configured off. - Added configure option to compile in GUI frontend for Bochs debugger, to enable configure with --enable-debugger-gui option. The GUI debugger frontend is enabled by default with Bochs debugger. - Removed --enable-port-e9-hack configure option. The feature now could be configured at runtime through .bochsrc. - Added configure option to enable/disable A20 pin support. Disabling the A20 pin support slightly speeds up the emulation. - reduced dependencies between source files for faster code generation - BIOS - Added S3 (suspend to RAM) ACPI state to BIOS (patch by Gleb Natapov) - Implemented MTRR support in the bios (patches by Avi Kivity and Alex Williamsion with additions by Sebastian Herbszt) - Bug fixes - I/O Devices - Added user plugin support - remaining devices converted to plugins: pit, ioapic, iodebug - added 'plugin_ctrl' bochsrc option to control the presence of optional device plugins without a separate option. By default all plugins are enabled. - added register mechanism for removable mouse and keyboard devices - Hard drive / cdrom - PACKET-DMA feature now supported by all ATAPI commands - ATAPI command 0x1A added (based on the Qemu implementation) - sb16 - Added ALSA sound support on Linux (PCM/MIDI output) - FM synthesizer now usable with MIDI output (simple piano only) - Fixed OPL frequency to MIDI note translation - Fixed MIDI output command - keyboard - added keyboard controller commands 0xCA and 0xCB - USB - USB code reorganized to support more HC types and devices - added USB OHCI support written by Ben Lunt - added external USB hub support (initial code ported from Qemu) - added USB cdrom support (SCSI layer ported from Qemu) - added status bar indicators to show data transfer - VGA - VBE video memory increased to 16 MB - implemented changeable VBE LFB base address (PCI only, requires latest BIOS and VGABIOS images) - I/O APIC - implemented I/O APIC device hardware reset - Config interface - new config interface 'win32config' with start and runtime menu is now the default on Windows ('textconfig' is still available) - win32 device config dialogs are now created dynamicly from a parameter list (works like the wx ParamDialog) - changes in textcofig and the wx ParamDialog for compatibility with the new win32 dialog behaviour - Bochs param tree index keys are case independent now - some other additions / bugfixes in the simulator interface code - Misc - updated LGPL'd VGABIOS to version 0.6c - Updated Bochs TESTFORM to version 0.4 - SF patches applied [2784858] IO Handler names are not compared properly [2712569] Legacy bios serial data buffer timeout bug by grybranix [2655090] 64 bit BSWAP with REX.W broken by M. Eby [2645919] CR8 bug when reading by M. Eby [1895665] kvm: bios: add support to memory above the pci hole by Izik Eidus [2403372] rombios: check for valid cdrom before using it by Sebastian [2307269] acpi: handle S3 by Sebastian [2354134] TAP networking on Solaris/Sparc repaired [2144692] The scsi device can not complete its writing data command by naiyue [1827082] [PATCH] Configurable CPU vendor by Marcel Sondaar [2217229] Panic on EBDA overflow in rombios32 by Sebastian [2210194] Log pci class code by Sebastian [1984662] red led for disk write and titlebar mod by ggbsf [2142955] Fix for monitor/mwait by Doug Gibson [2137774] Patch to fix bug: cdrom: read_block: lseek returned error by Gabor Olah [2134642] Fix scan_to_scanascii table for F11 and F12 by Ben Guthro & Steve Ofsthun [2123036] sdl fullscreen fix by ggbsf [2073039] Remove CMOS accsess from AML code by Gleb Natapov [2072168] smbios: add L1-L3 cache handle to processor information by Sebastian [2055416] bochsrc cpu options for cpuid vendor and brand string by Doug Reed [2035278] rombios: Fix return from BEV via retf by Sebastian [2035260] rombios: El Torito load segment fix by Sebastian [2031978] Fix VMware backdoor command 0Ah by Jamie Lokier [2015277] Remove obsolete comment about DATA_SEG_DEFS_HERE hack by Sebastian [2011268] Set new default format and unit only if both are supported by Sebastian [2001919] gdbstub: fix qSupported reply by Sebastian [2001912] gdbstub: enclose packet data by apostrophes by Sebastian [1998071] fix missing SIGHUP and SIGQUIT with term ui on mingw by Sebastian [1998063] fix wrong colors with term ui by Sebastian [1995064] Compile fix needed for --enable-debugger and gcc 4.3 by Hans de Goede [1994564] Fix typo in RDMSR BX_MSR_MTRRFIX16K_A0000 by Sebastian [1994396] Change hard_drive_post #if by Sebastian [1993235] TESTFORM email address update by Sebastian [1992322] PATCH: fix compilation of bochs 2.3.7 on bigendian machines by Hans de Goede [1991280] Shutdown status code 0Ch handler by Sebastian [1990108] Shutdown status code 0Bh handler by Sebastian [1988907] Shutdown status code 0Ah handler by Sebastian [1984467] two typos in a release! (2.3.7) [1981505] Init PIIX4 PCI to ISA bridge and IDE by Sebastian - these S.F. bugs were closed/fixed [2784148] an integer overflow BUG of Bochs-2.3.7 source code [2695273] MSVC cpu.dsp failure in 2.3.7.zip [616114] Snapshot/Copy crash on Win2K [2628318] 'VGABIOS-latest' bug [1945055] can't 'make install' lastest bochs on loepard [2031993] Mac OS X Makefile bug [1843199] install error on mac osx [2710931] Problem compiling both instrumentation and debugger [2617003] ExceptionInfo conflicts with OS X api [2609432] stepping causes segfault (CVS) [2605861] compile error with --enable-smp [1757068] current cvs(Jul19, 07) failed to boot smp [2426271] cannot get correct symbol entry [2471982] VGA character height glitches [1659659] wrong behaviour a20 at boot [1998027] minwg + --with-term + --with-out-win32 = link failure [1871936] bochs-2.3.6 make fails on wx.cc [1684666] info idt for long mode [2105989] could not read() hard drive image file at byte 269824 [1173093] Debugger totally not supports x86-64 [1803018] new win32debug dialog problems [2141679] windows vcc build broken [2162824] latest cvs fails to compile [2164506] latest bochs fails to start [2129223] MOV reg16, SS not working in real mode due to dead code [2106514] RIS / startrom.com install ALMOST works [2123358] SMP (HTT): wbinvd executed by CPU1 crashes CPU0 [2002758] Arch Linux: >>PANIC<< ATAPI command with zero byte count [2026501] El Torito incorrect boot segment:offset [2029758] BEV can return via retf instead of int 18h [2010173] x command breaks after one error about x/s or x/i [1830665] harddrv PANIC: ATAPI command with zero byte count [1985387] fail to make using gcc4 with --enable-debugger [1990187] testform feedback [1992138] Misspell in cpu/ia_opcodes.h - these S.F. feature requests were closed/implemented [2175153] Update MSVC project files [658800] front end program and bios [1883370] Make cd and floppy images more usable [422783] change floppy size without restarting [2552685] param tree names should be case insensitive [1214659] PC Speaker emu turnoff. Plugin Controll. [1977045] support 40 bit physical address [1506385] Intel Core Duo VT features [1429015] Support for user plugins [1488136] debugger access to floppy controller [1363136] Full debugger SMP and 64 bit support [2068304] Support for ACPI [431032] debugger "x" command [423420] profiling ideas (SMF) [445342] Add FM support? [928439] alsa ------------------------------------------------------------------------- Changes in 2.3.7 (June 3, 2008): Brief summary : + More optimizations in CPU code - Bochs 2.3.7 is more than 2x faster than Bochs 2.3.5 build ! - Implemented LBA48 support in BIOS - Added memory access tracing for Bochs internal debugger - Implemented Intel(R) XSAVE/XRSTOR and AES instruction set extensions - Many fixes in CPU emulation and internal debugger - MenuetOS64 floppy images booting perfect again ! - updated LGPL'd VGABIOS to version 0.6b Detailed change log : - CPU - Support of XSAVE/XRSTOR CPU extensions, to enable configure with --enable-xsave option - Support of AES CPU extensions, to enable configure with --enable-aes option - Fixed Bochs failure on RISC host machines with BxRepeatSpeedups optimization enabled - Implemented SYSENTER/SYSEXIT instructions in long mode - More than 100 bugfixes for CPU emulation correctness (both x86 and x86-64) - MenuetOS64 floppy images booting perfect again ! - Updated CPU instrumentation callbacks - Bochs Internal Debugger and Disassembler - Added memory access tracing for Bochs internal debugger, enable by typing 'trace-mem on' in debugger command line - Many bug fixes in Bochs internal debugger and disassembler - System BIOS (Volker) - Implemented LBA48 support - Added generation of SSDT ACPI table that contains definitions for available processors - Added RTC device to ACPI DSDT table - Added implementation of SMBIOS - I/O devices (Volker) - VGA - Implemented screen disable bit in sequencer register #1 - Implemented text mode cursor blinking - Serial - new serial modes 'pipe-server' and 'pipe-client' for win32 - new serial mode 'socket-server' - Configure and compile - Fixed configure bug with enabling of POPCNT instruction, POPCNT instruction should be enabled by default when SSE4.2 is enabled. - Removed --enable-magic-breakpoint configure option. The option is automatically enabled if Bochs internal debugger is compiled in. It is still possible to turn on/off the feature through .bochsrc. - Allow boot from network option in .bochsrc - Added Bochs version info for Win32 - Display libraries - implemented text mode character blinking in some guis - improved 'X' gui runtime dialogs - SF patches applied [1980833] Fix shutdown status code 5h handler by Kevin O'Connor [1928848] "pipe" mode for serial port (win32 only) by Eugene Toder [1956843] Set the compatible pci interrupt router back to PIIX by Sebastian [1956366] Do not announce C2 & C3 cpu power state support by Igor Lvovsky [1921733] support for LBA48 by Robert Millan [1938185] Fix link problem with --enable-debugger by Sebastian [1938182] Makefile.in - use @IODEV_LIB_VAR@ by Sebastian [1928945] fix for legacy rombios - e820 map and ACPI_DATA_SIZE by Sebastian [1925578] rombios32.c - fix ram_size in ram_probe for low memory setup by Sebastian [1908921] rombios32.c - move uuid_probe() call by Sebastian [1928902] improvements to load-symbols by Eugene Toder [1925568] PATCH: msvc compilation by Eugene Toder [1913150] rombios.c - e820 cover full size if memory <= 16 mb by Alexander van Heukelum [1919804] rombios.c - fix and add #ifdef comments by Sebastian [1909782] rombios.c - remove segment values from comment by Sebastian [1908918] SMBIOS - BIOS characteristics fix by Sebastian [1901027] BIOS boot menu support (take 3) [1902579] rombios32.c - define pci ids by Sebastian [1859447] Pass segment:offset to put_str and introduce %S by Sebastian [1889057] rombios.c - boot failure message by Sebastian [1891469] rombios.c - print BEV product string by Sebastian [1889851] Win32 version information FILEVERSION for bochs.exe by Sebastian [1889042] rombios.c - fix comment by Sebastian [1881500] bochsrc, allow boot: network by Sebastian [1880755] Win32 version information for bochs.exe by Sebastian [1880471] SMBIOS fix type 0 by Sebastian [1878558] SMBIOS fixes by Sebastian [1864692] SMBIOS support by Filip Navara [1865105] Move bios_table_area_end to 0xcc00 by Sebastian [1875414] Makefile.in - change make use by Sebastian [1874276] Added instrumentation for sysenter/sysexit by Lluis [1873221] TLB page flush: add logical address to instrumentation by Lluis [1830626] lba32 support by Samuel Thibault [1861839] Move option rom scan after floppy and hard drive post by Sebastian [1838283] Early vga bios init by Sebastian [1838272] rom_scan range parameter by Sebastian [1864680] Save CPUID signature by Filip Navara - these S.F. bugs were closed [1976171] Keyboard missing break code for enter (0x9C) [666433] physical read/write breakpoint sometimes fails [1744820] info gdt and info idt shows the entire tables [1755652] graphics: MenuetOS64 shows black screen [1782207] Windows Installer malfunction, Host=Linux, Guest=Win98SE [1697762] OS/2 Warp Install Failed [1952548] String to char * warnings [1940714] SYSENTER/SYSEXIT doesn't work in long mode [1422342] SYSRET errors [1923803] legacy rombios - e820 map and ACPI_DATA_SIZE [1936132] Link problem with --enable-debugger & --enable-disasm [1934477] Linear address wrap is not working [1424984] virtual machine freezes in Bochs 2.2.6 [1902928] with debugger cpu_loop leaves CPU with unstable state [1898929] Bochs VESA BIOS violates specs (banks == 1) [1569256] bug in datasegment change in long mode [1830662] ACPI: no DMI BIOS year, acpi=force is required [1868806] VGA blink enable & screen disable [1875721] Bit "Accessed" in LDT/GDT descriptors & #PF [1874124] bx_Instruction_c::ilen() const [1873488] bochs-2.3.6 make fails on dbg_main.cc - these S.F. feature requests were implemented [1422769] SYSENTER/SYSEXIT support in x86-64 mode [1847955] Version information for bochs(dbg).exe [939797] SMBIOS support ------------------------------------------------------------------------- Changes in 2.3.6 (December 24, 2007): Brief summary : + More than 25% emulation speedup vs Bochs 2.3.5 release! - Thanks to Darek Mihocka (http://www.emulators.com) for providing patches and ideas that made the speedup possible! + Up to 40% speedup vs Bochs 2.3.5 release with trace cache optimization! - Lots of bugfixes in CPU emulation - Bochs benchmarking support - Added emulation of Intel SSE4.2 instruction set Detailed change log : - CPU - Added emulation of SSE4.2 instruction set, to enable use --enable-sse=4 --enable-sse-extension configure options to enable POPCNT instruction only use configure option --enable-popcnt - Implemented MTRR emulation, to enable use --enable-mtrr configure option. MTRRs is enabled by default when cpu-level >= 6. - Implemented experimental MONITOR/MWAIT support including optimized MWAIT CPU state and hardware monitoring of physical address range, to enable use --enable-monitor-mwait configure option. - Removed hostasm optimizations, after Bochs rebenchmarking it was found that the feature bringing no speedup or even sometimes slows down emulation! - Merged trace cache optimization patch, the trace cache optimization is enabled by default when configure with --enable-all-optimizations option, to disable trace cache optimization configure with --disable-trace-cache - Many minor bugfixes in CPU emulation (both ia32 and x86-64) - Updated CPU instrumentation callbacks - Bochs Internal Debugger and Disassembler - Many fixes in Bochs internal debugger and disassembler, some debugger interfaces significantly changed due transition to the param tree architecture - Added support for restoring of the CPU state from external file directly from Bochs debugger - Configure and compile - Renamed configure option --enable-4meg-pages to --enable-large-pages. The option enables page size extensions (PSE) which refers to 2M pages as well. - Removed --enable-save-restore configure option, save/restore feature changed to be one of the basic Bochs features and compiled by default for all configurations. - Added new Bochs benchmark mode. To run Bochs in benchmark mode execute it with new command line option 'bochs -benchmark time'. The emulation will be automatically stopped after 'time' millions of emulation cycles executed. - Another very useful option for benchmarking of Bochs could be enabled using new 'print_timestamps' directive from .bochsrc: print_timestamps: enable=1 - Added --enable-show-ips option to all configuration scripts used to build release binaries, so all future releases will enjoy IPS display. - Enable alignment check in the CPU and #AC exception by default for --cpu-level >= 4 (like in real hardware) - SF patches applied [1491207] Trace Cache Speedup patch by Stanislav [1857149] Define some IPL values by Sebastian [1850183] Get memory access mode in BX_INSTR_LIN_READ by Lluis Vilanova [1841421] pic: keep slave_pic.INT and master_pic.IRQ_in bit 2 in sync by Russ Cox [1841420] give segment numbers in exception logs by Russ Cox [1801696] Allow Intel builds on Mac OS X [1830658] Fix >32GB disk banner by Samuel Thibault [1813314] Move #define IPL_* and typedef ipl_entry by Sebastian [1809001] Save PnP Option ROM Product Name string in IPL Boot Table by Sebastian [1821242] Fix for #1801285, Niclist.exe broken by Sebastian [1819567] Code warning cleanup [1816162] Update comment on bios_printf() by Sebastian [1811139] Trivial Fix when BX_PCIBIOS and BX_ROMBIOS32 not defined by Myles Watson [1811190] Improve HD recognition and CD boot by Myles Watson [1811860] Implement %X in bios_printf by Sebastian [1809649] printf %lx %ld %lu by Myles Watson [1809651] move BX_SUPPORT_FLOPPY by Myles Watson [1809652] dpte and Int13DPT fixes by Myles Watson [1809669] clip cylinders to 16383 in hard drive by Myles Watson [1799903] Build BIOS on amd64 by Robert Millan [1799877] Fix for parallel build (make -j2) by Robert Millan - these S.F. bugs were closed [1837354] website bug: View the Source link broken [1801268] Reset from real mode no longer working [1843250] Using forward slashes gives invalid filename [1823446] BIOS bug, local APIC #0 not detected [1801285] Niclist.exe broken [1364472] breakpoints sometimes don't work [994451] breakpoint bug [1801295] NSIS installer vs Windows Notepad [1715328] Unreal mode quirk [1503972] debugger doesn't debug first instruction on exception [1069071] div al, byte ptr [ds:0x7c18] fails to execute [1800080] Wrong "BX_MAX_SMP_THREADS_SUPPORTED" assertion - these S.F. feature requests were implemented [1662687] Download for Win32-exe with x64 Mode and debugging [604221] Debugger command: query lin->phys mapping ------------------------------------------------------------------------- Changes in 2.3.5 (September 16, 2007): Brief summary : - Critical problems fixed for x86-64 support in CPU and Bochs internal debugger - ACPI support - The release compiled with x86-64 and ACPI - Hard disk emulation supports ATA-6 (LBA48 addressing, UDMA modes) - Added emulation of Intel SSE4.1 instruction set Detailed change log : - CPU - Fixed critical bug with 0x90 opcode (NOP) handling in x86-64 mode - implied stack references where the stack address is not in canonical form should causes a stack exception (#SS) - Added emulation of SSE4.1 instruction set (Stanislav) - Do not save and restore XMM8-XMM15 registers when not in x86-64 mode - Fixed zero upper 32-bit part of GPR in x86-64 mode - CMOV_GdEd should zero upper 32-bit part of GPR register even if the 'cmov' condition was false ! - Implemented CLFLUSH instruction, report non-zero cache size in CPUID - Fixed PUSHA/POPA instructions behavior in real mode - Fixed detection of inexact result by FPU - Fixed denormals-are-zero (DAZ) handling by SSE convert instructions - Implemented Misaligned Exception Mask support for SSE (MXCSR[17]) - Implemented Alignment Check in the CPU and #AC exception, to enable use --enable-alignment-check configure option - General - 2nd simulation support in wxBochs now almost usable (simulation cleanup code added and memory leaks fixed) - Configure and compile - several fixes for MacOSX, OpenBSD and Solaris 10 - enable save/restore feature by default for all configurations - reorganized SSE configure options to match Intel(R) Programming Reference Manual, new option introduced for SSE extensions enabling. To enable Intel Core Duo 2 new instructions use --enable-sse=3 --enable-sse-extension enabling of SSE4.1 (--enable-sse=4) will enable SSE3 extensions as well - removed old PIT, always use new PIT written by Greg Alexander, removed configure option --enable-new-pit - I/O devices (Volker) - Floppy - partial non-DMA mode support (patch by John Comeau) - Hard drive / cdrom - hard disk emulation now supports ATA-6 (LBA48 addressing, UDMA modes) - VMWare version 4 disk image support added (patch by Sharvil Nanavati) - PCI - initial support for the PIIX4 ACPI controller - Serial - added support for 3-button mouse with Mousesystems protocol - USB - experimental USB device change support added - rewrite of the existing USB devices code - new USB devices 'disk' and 'tablet' (ported from the Qemu project) - Bochs internal debugger - fixed broken debugger "rc file" option (execute debugger command from file) - implementation of a gui frontend ("windebug") for win32 started - gdbstub now accepts connection from any host - several documentation updates - a lot of disasm and internal debugger x86_64 support fixes - Configuration interface - fixes and improvements to the save state dialog handling - Display libraries - text mode color handling improved in some guis - win32 fullscreen mode (patch by John Comeau) - System BIOS (Volker) - 32-bit PM BIOS init code for ACPI, PCI, SMP and SMM (initial patches by Fabrice Bellard) - PCI BIOS function "find class code" implemented - SF patches applied [1791000] 15h 8600h is reading the wrong stack frame by Sebastian [1791016] rombios32.c, ram_probe(), BX_INFO missing value by Sebastian [1786429] typo in bochsrc.5 by Sebastian [1785204] Extend acpi_build_table_header to accept a revision number by Sebastian [1766536] Partial Patch for Bug Report 1549873 by Ben Lunt [1763578] ACPI Table Revision 0 -> 1 [1642490] implement alignment check and #AC exception by Stanislav Shwartsman [1695652] [PATCH] .pcap pktlog and vnet PXE boot by Duane Voth [1741153] Add expansion-ROM boot support to the ROMBIOS [1734159] Implemented INT15h, fn 0xC2 (mouse), subfn 3, set resolution [1712970] bios_printf %s fix [1573297] PUSHA/POPA real mode fix by Stanislav Shwartsman [1641816] partial support for non-DMA access to floppy by John Comeau [1624032] shows where write outside of memory occurred by John Comeau [1607793] allow fullscreen when app requests it by John Comeau [1603013] Bugfix for major NOP problem on x64 by mvysin [1600178] Make tap and tuntap compile on OpenBSD by Jonathan Gray [1149659] improve gdbstub network efficiency by Avi Kivity [1554502] Trivial FPU exception handling fix - these S.F. bugs were closed [1316008] Double faults when it shouldn't - gcc 4.0.2 [1787289] broken ABI for redolog class when enable-compressed-hd [1787500] tftp_send_optack not 64bit clean [1264540] Security issue with Bochs website [1767217] Debugger Faults including ud2 [1729822] Various security issues in io device emulation [1675202] mptable hosed (bad entry count in header) [1197141] 'make install' installs to bad location [1157623] x86Solaris10 cannot recoginize ACPI RSD PTR [1768254] large HDD in Bochs/bximage [1496157] Windows Vista Beta2 dosn't boot [1755915] Illegal Hard Disk Signature Output [1717790] info gdt and info idt scrolls away, too long result [1726640] Debugger displays incorrect segment for mov instruction [1719156] Typo in misc_mem.cpp [1715270] Debugger broken in/beyond 2.3 [1689107] v8086 mode priviledge check failed [1704484] A few checks when CPU_LEVEL < 4 [1678395] Problem with zero sector... [876990] SA-RTL OS fails on PIC configuration [1673582] save/restore didn't restore simulation correctly [1586662] EDD int 13h bug, modify eax [666618] POP_A Panic in DOS EMU [1001485] panic: not enough bytes on stack [1667336] delay times an order of magnitude slow [1665601] crash disassembling bootcode [1657065] CVS sources won't compile [1653805] bochs's gdbstub uses incorrect protocol [1640737] ASM sti command frezzes guest OS [1636439] latest CVS sources don't compile under Cygwin [1634357] disasm incorrect (no sign ext) displacement in 64-bit mode [1376453] pcidev segfaults bochs [1180890] IOAPIC in BOCHS - WinXP 64 in MP version [1597528] 2.3 fails to compile on amd64 [1526255] FLD1 broken when compaling with gcc 4.0.x [1597451] eth_fbsd is broken under FreeBSD [1571949] Bochs will not compile under Solaris [1500216] Bochs fails to boot BeOs CD [1458339] bochs-2.2.6 WinXP Binary ACPI error installing FreeBSD 6.0 [1440011] patches needed for FreeBSD 6.0 to compile Bochs [431674] some devices don't have a prefix [458150] QNX demo disk crashes with new pit [818322] Bochs 2.1 cvs: OS/2 - read verify on non disk [906840] KBD: bogus scan codes generated in set 3 [1005053] No keyboard codes translation [1109374] Problem with Scancodeset 2 [1572345] Bochs won't continue [1568153] Bochs looks for (and loads?) unspecified display libraries [1563462] Errors in /iodev/harddrv.h [1562172] TLB_init() fails to initialize priv_check array if USE_TLB 0 [1385303] debugger crashes after panic [1438227] crc.cpp missing in bx_debug version 2.2.6 [1501825] debugger crashes on to high input [1420959] Memory leak + buffer overflow in Bochs debugger [1553289] Error in Dis-assembler [542464] I cannot use FLAT [1548270] Bochs won't die with its pseudo terminal [1545588] roundAndPackFloatx80 does not detect round up correctly ------------------------------------------------------------------------- Changes in 2.3 (August 27, 2006): Brief summary : - limited save/restore support added (config + log options, hardware state) - configuration parameter handling rewritten to a parameter tree - lots of cpu and internal debugger fixes - hard disk geometry autodetection now supported by most of the image types - hard disk emulation now supports ATA-3 (multiple sector transfers) - VBE memory size increased to 8MB and several VGA/VBE fixes - updated LGPL'd VGABIOS to version 0.6a Detailed change log : - CPU and internal debugger fixes - Fixed bug in FSTENV instruction (Stanislav Shwartsman) - Recognize #XF exception (19) when SSE is enabled - Fixed bug in PSRAW/PSRAD MMX and SSE instructions - Save and restore RIP/RSP only for FAULT-type exceptions, not for traps - Correctly decode, disassemble and execute multi-byte NOP '0F F1' opcode - Raise A20 line after system reset (Stanislav Shwartsman) - Implemented SMI and NMI delivery (APIC) and handling in CPU (Stanislav) - Experimental implementation of System Management Mode (Stanislav) - Added emulation of SSE3E instructions (Stanislav Shwarstman) - Save and restore FPU opcode, FIP and FDP in FXSAVE/FRSTOR instructions - Fixed bug in MOVD_EdVd opcode (always generated #UD exception) - Fixed critical issue, Bochs was not supporting > 16 bit LDT.LIMIT values - Many fixes in Bochs internal debugger and disassembler - CPU x86-64 fixes - Fixed SYSRET instruction implementation - Fixed bug in CALL/JMP far through 64-bit callgate in x86-64 mode - Correctly decode, disassemble and execute 'XCHG R8, rAX' instruction - Correctly decode and execute 'BSWAP R8-R15' instructions - Fixed ENTER and LEAVE instructions in x86-64 mode (Stanislav) - Fixed CR4 exception condition (No Name) - Fixed x86 debugger to support x86-64 mode (Stanislav) - APIC and SMP - Support for Dual Core and Intel(R) HyperThreading Technology. Now you could choose amount of cores per processor and amount of HT threads per core from .bochsrc for SMP simulation (Stanislav Shwartsman) - Allow to control SMP quantum value through .bochsrc CPU option parameter. Previous Bochs versions used hardcoded quantum=5 value. - Fixed interrupt priority bug in service_local_apic() - Fixed again reading of APIC IRR/ISR/TMR registers. Finally it becomes fully correct :-) - Configure and compile - Moved configure time --enable-reset-on-triple-fault option to runtime, the 'cpu' option in .bochsrc is extended and the old configure option is deprecated (Stanislav Shwartsman) - Removed --enable-pni configure option, to compile with PNI use --enable-sse=3 instead (Stanislav Shwartsman) - enable SEP (SYSENTER/SYSEXIT) support by default for Penitum II+ processor emulation (i.e. if cpu-level >= 6 and MMX is enabled) - general - Limited save/restore support added. The state of CPU, memory and all devices can be saved now (state of harddisk images not handled yet). - Fixed several memory leaks - configuration interface - Configuration parameter handling rewritten to a parameter tree. This is required for dynamic menus/dialogs, user-defined options and save/restore. - Support for user-defined bochsrc options added - help support at the parameter prompt in textconfig added - I/O devices (Volker) - Floppy - partial sector transfers fixed - Hard drive / cdrom - several fixes to the IDE register behaviour (e.g. in case of a channel with only one drive connected) - fixed data alignment of 'growing' hard drive images (sharing images between Windows and Linux now possible) - disk geometry autodetection now supported by most of the image types (unsupported: external, dll and compressed modes) - multi sector read/write commands implemented - hard disk now reporting ATA-3 supported - ATAPI 'inquiry' now returns a unique device name - Keyboard - reset sent to keyboard has no effect on the 8042 (scancode translation) - PCI - forward PIRQ register changes to the I/O APIC (if present) - attempt to fix and update the emulation part of 'pcidev' (untested) - VGA - VBE memory size increased to 8MB and several VBE fixes - VGA memory read access fixed (bit plane access and read mode) - VGA memory is now a part of the common video memory - System BIOS (Volker) - enable interrupts before executing INT 19h - fixed ATA device detection in case of one drive only connected to controller - improved INT 15h function AX=E820h - real mode PCI BIOS now returns IRQ routing information (function 0Eh) - keyboard LED flags handling fixed and improved - fixed handling of extended keys in INT 09h - Updated LGPL'd VGABIOS to version 0.6a - SF patches applied [1340111] fixes and updates to usb support by Ben Lunt [1539420] minor addition to pci_usb code by Ben Lunt [1455958] call/jmp through call gate in 64-bit mode [1433107] PATCH: fix compile with wxwindows 2.6 (unicode / utf8) by jwrdegoede [1386671] Combined dual core and hyper-threading patch - these S.F. bugs were closed [833927] TTD: System Error TNT.40025: Unexpected processor exception [789230] Sending code that shows lock up when setting idt [909670] Problems with Symantec Ghost [1540241] include missing in osdep.cc [1539373] Incorrect disasm for "mov moffset,bla" in 64bit [1538419] incorrect disassembly of [rip+disp] with rex.b [1535432] shift+cursor key maps to a digit [1504891] Knoopix 5.0.1 error [1424355] bochs-2.2.6 ata failure in windoze 98se [1533979] wrong disassembly of IN instruction [620059] paste won't stop [1164904] status bar doesn't show num/caps/scroll lock status [1061720] ATA Support level for HD [1522196] Broken CHANGES link in main page [1438415] crash if screen scrolled downwards [778441] Shouldn't interrupts be enable after BIOS? [1514949] I got a problem with the 8253 timer [1513544] disasm of 0xec (in AL,DX) returns ilen of 2 instead of 1 [1508947] APIC interrupt priority checking and interrupt delivery [766286] Debugger halts after any GPF exception [639143] va_list is not a pointer on linuxppc [1501815] debugger examines memory over page-boundary wrong [1503978] movsb/w/d doesn't work when direction is stored [1499405] WinPCap has changed URL hosting [1498519] APIC IRR bits not set while interrupts disabled [1498193] Bochs segfaults on LTR instruction [787140] Guest2HostTLB optimization bug [1492070] instrument stop [1487772] No SEP on P4 [1488335] Growing hard disk images severe interoperability errors! [1076312] Shadow RAM and TLB [1282249] The real i440FX chipset Award bios hangs [1479763] mistake "mov ax,[es:di]" for "mov ax,[ds:di]" [1453575] Misconfigured floppy DMA transfers do not terminate. [1460068] Incorrect handling for the Options Menu Item [910203] bochs-2.1.1 wx.lo failed [1438654] PANIC when trying to run install-amd64-minimal-2005.0.iso [1458320] compile hdimage.h fails [1455880] bochs-2.2.6,2: make error on FreeBSD [696890] Network wouldn't run under W2k hosting MSDOS [673391] SMP timer problems [1291059] wxWindows GUI on non-windows/configure issue [1356450] bochs 2.2.1 errors-omittions [1178017] Win98 guest cannot receive network packets from host [1076315] a20_mask after restarting [1436323] real hw does not panic when bad Ib in CMPSS_VssWssIb [1435269] cdrom_amigaos is not compilable [1433314] disasm issues [1170614] relative jumps/calls wrong in debugger [758121] user might get confused when interrupt handler invoked [1170622] You cannot toggle OFF "show" flags [1406387] JMP instruction should display absolute address [1428813] PANIC: ROM address space out of range [1426288] DR-DOSs EMM386 problem [1412036] Bochs cannot recognize PCI NIC correctly [435115] dbg: modebp broken and no docs [1419366] disasm cs:eip does not work anymore [1419393] SSE's #XF exception -> "exception(19): bad vector" [1419429] disassembly of "260f6f00" show DS: instead of ES: prefix [1417583] Interrupt behaviour changed from 2.2.1 to 2.2.5 [1418281] 'push' (6A) incorrectly disassembled [1417791] FLDENV generating exception when real hw does not. [1264583] OS/2 1.1 doesn't run ------------------------------------------------------------------------- Changes in 2.2.6 (January 29, 2006): - First major SMP release ! - several APIC and I/O APIC fixes make SMP Bochs booting Windows NT4.0 or Knoppix 4.0.2 without noapic kernel option in SMP configuration. - critical APIC timer bug fixed - obsolete SMP BIOS images removed (MP tables created dynamicaly) - determine number of processors in SMP configuration through .bochsrc new .bochsrc option 'CPU' allows to choose number of processors to emulate - new configure option --enable-smp to configure Bochs for SMP support, the old --enable-processors=N option is deprecated - CPU and internal debugger fixes - enabled #PCE bit in CR4 register, previosly setting of this bit generated #GP(0) fault - enabled LAHF/SAHF instructions in x86-64 mode - fixed bug in PMULUDQ SSE2 instruction - fixes in Bochs debugger - Configure and compile - enable VME (virtual 8086 mode extensions) by default if cpu-level >= 5 - enable Bochs disassembler by default for all configurations - win32 installer script improvements - ips parameter moved to new 'CPU' option - show IPS value in status bar if BX_SHOW_IPS is enabled - Other - several fixes in the hard drive, keyboard, timer, usb and vga code - new user button shortcut "bksl" (backslash) - updated Bochs instrumentation examples - user and development documentation improved ------------------------------------------------------------------------- Changes in 2.2.5 (December 30, 2005): Brief summary : - added virtual 8086 mode extensions (VME) implementation - several fixes/improvements in x86-64 emulation, debugger and disassembler - new serial mode 'socket' connects a network socket - IDE busmaster DMA feature for harddisks and cdroms completed and enabled - many improvements in Bochs emulated I/O devices (e.g. floppy, cdrom) - Updated LGPL'd VGABIOS to version 0.5d Detailed change log : - CPU - fixed XMM registers restore in FXRSTOR instruction (Andrej Palkovsky) - print registers dump to the log if tripple fault occured - fixed PANIC in LTR instruction (Stanislav) - added virtual 8086 mode extensions (VME) implementation, to enable configure with --enable-vme (Stanislav) - flush caches and TLBs when executing WBINVD and INVD instructions - do not modify segment limit and AR bytes when modifying segment register in real mode (support for unreal mode) - fixed init/reset values for LDTR and TR registers - reimplemented hardware task switching mechanism (Stanislav) - generate #GP(0) when fetching instruction cross segment boundary - CPU (x86-64) (Stanislav Shwartsman) - implemented call_far/ret_far/jmp_far instructions in long mode - fixed IRET operation in long mode - fixed bug prevented setting of NXE/FFXSR bits in MSR.EFER register - implemented RDTSCP instruction - do not check CS.limit when prefetching instructions in long mode - fixed masked write instructions (MASKMOVQ/MASKMOVDQU) in long mode - fetchdecode fixes for x86-64 - APIC - Fixed bug in changing local APIC id (Stanislav) - Fixed reading of IRR/ISR/TMR registers (patch by wmrieker) - Implemented spurious interrupt register (Stanislav, patch by wmrieker) - Fixed interrupt delivery bug (anonymous #SF patch) - Correctly implemented ESR APIC register (Stanislav) - Bochs debugger - Fixed bug in bochs debugger caused breakpoints doesn't fire sometimes (Alexander Krisak) - watchpoints in device memory fixed (Nickolai Zeldovich) - new debug interface to access Bochs CPU general purpose registers with support for x86-64 - Disassembler (Stanislav Shwartsman) - Fixed disassembly for FCOMI/FUCOMI instructions - Full x86-64 support in disassembler. The disassembler module extended to support x86-64 extensions. Still limited by Bochs debugger which is not supporting x86-64 at all ;( - I/O devices (Volker) - general - memory management prepared for large BIOS images (up to 512k) - slowdown timer sleep rate fixed (now using 1 msec on all platforms) - some device specific parameter handlers moved into the device code - serial - new serial mode 'socket' connects a network socket (#SF patch by Andrew Backer) - hard drive / cdrom - assign a unique serial number to each drive (fixes harddrive detection problems with Linux kernels 2.6.x: "ignoring undecoded slave") - geometry autodetection for 'flat' hard disk images added. Works with images created with bximage (heads = 16, sectors per track = 63) - ATAPI command 'read cd' implemented, some other commands improved - cdrom read block function now tries up to 3 times before giving up - emulation of raw cdrom reads added, some other lowlevel cdrom fixes - IDE busmaster DMA feature for harddisks and cdroms completed and enabled - disk image size limit changed from 32 to 127 GB - split ATA/ATAPI emulation code and image handling code - floppy - fixes for OS/2 (patch by Robin Kay) - disk change line behaviour fixed (initial patch by Ben Lunt) - end-of-track (EOT) condition handling implemented - more accurate timing for read/write data and format track commands using a motor speed of 300 RPM - timing of recalibrate and seek commands now depends on the step rate, date rate and the steps to do - floppy controller type changed to 82077AA - cmos - RTC 12-hour and binary mode implemented - number of CMOS registers changed from 64 to 128 - bochsrc option 'cmosimage' improved - save cmos image on exit if enabled - speaker - simple speaker support for OS X added (patch by brianonn@telus.net) - pci - BeOS boot failure fix in the PCI IDE code - don't register i/o and memory regions during PCI probe - vga - memory allocation for vga extensions fixed - usb - some bugfixes by Ben Lunt (mouse and keypad are usable now) - networking modules - VDE networking module now enabled on Linux - display libraries - general - new syntax for the userbutton shortcut string and more keys supported - win32 - fixed keycode generation for right alt/ctrl/shift keys - runtime dialog is now a property sheet - x11 - simple dialog boxes for the "ask" and "user shortcut" feature implemented - Slovenian keymap added (contributed by Mitja Ursic) - configuration interface - ask dialog is now enabled by default for win32, wx and x display libraries - bochsrc option floppy_command_delay is obsolete now (floppy timing now based on hardware specs) - floppy image size detection now available in the whole config interface - some device specific parameter handlers moved into the device code - calculate BIOS ROM start address from image if not specified - System BIOS (Volker) - PCI i/o and memory base address initialization added - several keyboard interrupt handler fixes (e.g. patch by japheth) - several floppy fixes (e.g. OS/2 works with patch by Robin Kay) - some more APM functions added - Updated LGPL'd VGABIOS to version 0.5d - generate SMP specific tables dynamicly by the Bochs memory init code - SF patches applied [1389776] Disk sizes over 64 Gbytes by Andrzej Zaborowski [1359162] disasm support for x86-64 by Stanislav Shwartsman [857235] task priority and other APIC bugs, etc by wmrieker [1359011] build breaks for 386 + debugger + disasm by shirokuma [1352761] Infinite loop when trying to debug a triple exception [1311170] small APIC bug fix (interrupt sent to the wrong CPU) [1309763] Watchpoints don't work in device memory by Nickolai Zeldovich [1294930] change line status on floppy by Ben Lunt [1282033] SSE FXRESTORE not working correctly by Ondrej Palkovsky [816979] wget generalizations by Lyndon Nerenberg [1214886] No more pageWriteStamp / unified icache by H. Johansson [1107945] com->socket redirection support by Andrew Backer - these S.F. bugs were closed [669180] win95 install : unknown SET FEATURES subcommand 0x03 [1346692] bochs 2.2.1 VGA BIOS error [1354963] floppy in KolibriOS [1378204] error: bochs-2.2.1, --enable-sb16, --disable-gameport [1368412] VDE problems in BOCHS [533446] CPU and APIC devices appear twice [1000796] bximage fails to create image of specified size [1170793] Quarterdeck QEMM doesn't work [923704] Multiple opcode prefixes don't reflect Trap 13 [1166392] DocBook/documentation issues [1368239] broken grater than 4GB size of sparse type hd image [1365830] i386 compile breaks on paging [427550] Incomplete IRETD implementation [1215081] MSVC workspace STILL not fixed [736279] Jump to Task [1356488] FD change fail & occur error [957615] [CPU ] prefetch: RIP > CS.limit [1353866] not booting linux-2.6.14 [1351667] load32bitOSImage does not work with --enable-x86-debugger [1217476] Incorrect (?) handling of segment registers in real mode [1184711] OS2 DOS crash [2.2.pre2] [624330] support for disks > 32GiB [1348368] bochs 2.2.1 bximage error [1342081] Configuration Menu option failed [1138616] OS/2 Warp 4 hangs when booting [1049840] mouse and video conflict [1164570] Unable to perform Fedora Core 4 test 1 installation [1183201] Windows 2000 (MSDN build 2150?) does not completely install [1194284] Can't boot from CD-ROM (Windows NT) [962969] Windows NT crashes while trying to intall them. [1054594] WinXP install halts (redo) [1153107] Windows XP fails with BSOD on 'vga' [938518] Win XP installation fails [645420] getHostMemAddr vetoed direct read [1179985] MS XENIX: >>PANIC<< VGABIOS panic at vgabios.c, line 0 [1329600] WBINVD and INVD should flush caches and TLB [638924] eliminate BX_USE_CONFIG_INTERFACE [1048711] Funny behaviour with CTRL [1288450] keyboard BIOS error [1310706] Keyboard - about key SHIFT [1295981] Ubuntu 5.04 Live-CD won't boot in Bochs [879047] APIC timer behavior different before reset and after [1188506] I still can't install the german Windows XP! [1301847] Windows XP dosn't boot - FXRSTOR problem ? [661259] does not boot QNX under WinX [924412] Keyboard lock states all whacked [681127] MIPSpro compiler (IRIX) is allergic to ^M [1285923] BIOS keyboard handler [516639] ATA controller revisited... [657918] does not boot BeOS under WinX [649245] BeOS CD locks halfway on boot [1094385] Attachment for bug 1090339 (beos failure) [1183196] BeOS 4.5 developer CD does not install [1090339] BeOS fails to boot [639484] panics when int 13 is called [711701] divide by zero [704295] ATAPI/BIOS call missing [682856] hard drive problems [627691] Cursor keys problem [588011] keyboard not working [542260] os/2 warp crashes with floppy handling [1273878] SB16 doesn't work in pure DOS [542254] OS/2 FDC driver dies [1099610] Windows 98 SE Does not install [875479] cr3 problem on task switch [731423] NE2000 causing PANIC on Win2K detection [1156155] bochs fails to boot plan9 iso [1251979] --enable-cpu-level=3 should assume --without-fpu [1257538] Interupt 15h 83h - set wait event interval [658396] Panic for DR DOS emm386 [679339] /? doesn't divulge Bochs command-line syntax [1167016] call/jump/return_protected doesn't support x86-64 [1252432] Mac OS X compile bug [881442] Bochs 2.1 PANIC when loading DOS Turbo Pascal protected mode [1249324] Boch2.2.1 Buffer Overfollow in void bx_local_apic_c::init () [1197144] 'make install' has dependency on wget [1079595] LTR:386TSS: loading tr.limit < 103 [1244070] Compilation Error in gui/rfb.cc [761707] CPU error when trying to start Privateer [517281] Crash running Privateer in DOS... ------------------------------------------------------------------------- Changes in 2.2.1 (July 8, 2005): - Fixed several compilation warnings and errors for different platforms (Volker) - Fixed FPU tag word restore in FXRSTOR instruction (Stanislav) - Added missing scancodes for F11 and F12 to BIOS translation table (Volker) - Bochs disassembler bugfixes (h.johansson) - About 5% emulation speed improvement (h.johansson) - Handle writing of zero to APIC timer initial count register (Stanislav) - Enable Idle-Hack for 'TERM' GUI (h.johansson) - Reduced overhead of BX_SHOW_IPS option to minimum. Now every simulation could run with --enable-show-ips without significant performance penalty. (Stanislav) - Fixed pcipnic register access (Volker) - Limited write support for TFTP server in 'vnet' networking module added (Volker) - Changed some timing defaults to more useful values (Volker) - WinXP/2003 style common controls now supported (Vitaly Vorobyov) - Updated LGPL'd VGABIOS to version 0.5c (Volker) - Added new BX_INSTR_HLT callback to instrumentation (Stanislav) ------------------------------------------------------------------------- Changes in 2.2 (May 28, 2005): Brief summary : - New floating point emulator based on SoftFloat floating point emulation library. - improved x86-64 emulation - Cirrus SVGA card emulation added - status bar with indicators for keyboard, floppy, cdrom and disk (gui dependant) - many improvements in Bochs emulated I/O devices (e.g. PCI subsystem) Detailed change log : - CPU - fixes for booting OS/2 by Dmitri Froloff - fixed v8086 priveleged instruction processing bug (was also reported by LightCone Aug 7 2003) - exception process bug (was reported by Diego Henriquez Sat Nov 15 01:16:51 CET 2003) - segment validation with IRET instruction - CS segment not present exception processing with IRET - several fixes by Kevin Lawton - add MSVC host asm instructions (patch by suzu) - fixed bug in HADDPD/HSUBP

au3反编译源码

au3反编译源码 myAut2Exe - The Open Source AutoIT Script Decompiler 2.9 ======================================================== *New* full support for AutoIT v3.2.6++ :) ... mmh here's what I merely missed in the 'public sources 3.1.0' This program is for studying the 'Compiled' AutoIt3 format. AutoHotKey was developed from AutoIT and so scripts are nearly the same. Drag the compiled *.exe or *.a3x into the AutoIT Script Decompiler textbox. To copy text or to enlarge the log window double click on it. Supported Obfuscators: 'Jos van der Zande AutoIt3 Source Obfuscator v1.0.14 [June 16, 2007]' , 'Jos van der Zande AutoIt3 Source Obfuscator v1.0.15 [July 1, 2007]' , 'Jos van der Zande AutoIt3 Source Obfuscator v1.0.20 [Sept 8, 2007]' , 'Jos van der Zande AutoIt3 Source Obfuscator v1.0.22 [Oct 18, 2007]' , 'Jos van der Zande AutoIt3 Source Obfuscator v1.0.24 [Feb 15, 2008]' , 'EncodeIt 2.0' and 'Chr() string encode' Tested with: AutoIT : v3. 3. 0.0 and AutoIT : v2.64. 0.0 and AutoHotKey: v1.0.48.5 The options: =========== 'Force Old Script Type' Grey means auto detect and is the best in most cases. However if auto detection fails or is fooled through modification try to enable/disable this setting 'Don't delete temp files (compressed script)' this will keep *.pak files you may try to unpack manually with'LZSS.exe' as well as *.tok DeTokeniser files, tidy backups and *.tbl (<-Used in van Zande obfucation). If enable it will keep AHK-Scripts as they are and doesn't remove the linebreaks at the beginning Default:OFF 'Verbose LogOutput' When checked you get verbose information when decompiling(DeTokenise) new 3.2.6+ compiled Exe Default:OFF 'Restore Includes' will separated/restore includes. requires ';<AUT2EXE INCLUDE-START' comment to be present in the script to work Default:ON 'Use 'normal' Au3_Signature to find start of script' Will uses the normal 16-byte start signature to detect the start of a

操作系统(内存管理)

文将对 Linux™ 程序员可以使用的内存管理技术进行概述，虽然关注的重点是 C 语言，但同样也适用于其他语言。文中将为您提供如何管理内存的细节，然后将进一步展示如何手工管理内存，如何使用引用计数或者内存池来半手工地管理内存，以及如何使用垃圾收集自动管理内存。为什么必须管理内存内存管理是计算机编程最为基本的领域之一。在很多脚本语言中，您不必担心内存是如何管理的，这并不能使得内存管理的重要性有一点点降低。对实际编程来说，理解您的内存管理器的能力与局限性至关重要。在大部分系统语言中，比如 C 和 C++，您必须进行内存管理。本文将介绍手工的、半手工的以及自动的内存管理实践的基本概念。追溯到在 Apple II 上进行汇编语言编程的时代，那时内存管理还不是个大问题。您实际上在运行整个系统。系统有多少内存，您就有多少内存。您甚至不必费心思去弄明白它有多少内存，因为每一台机器的内存数量都相同。所以，如果内存需要非常固定，那么您只需要选择一个内存范围并使用它即可。不过，即使是在这样一个简单的计算机中，您也会有问题，尤其是当您不知道程序的每个部分将需要多少内存时。如果您的空间有限，而内存需求是变化的，那么您需要一些方法来满足这些需求：确定您是否有足够的内存来处理数据。从可用的内存中获取一部分内存。向可用内存池（pool）中返回部分内存，以使其可以由程序的其他部分或者其他程序使用。实现这些需求的程序库称为分配程序（allocators），因为它们负责分配和回收内存。程序的动态性越强，内存管理就越重要，您的内存分配程序的选择也就更重要。让我们来了解可用于内存管理的不同方法，它们的好处与不足，以及它们最适用的情形。回页首 C 风格的内存分配程序 C 编程语言提供了两个函数来满足我们的三个需求： malloc：该函数分配给定的字节数，并返回一个指向它们的指针。如果没有足够的可用内存，那么它返回一个空指针。 free：该函数获得指向由 malloc 分配的内存片段的指针，并将其释放，以便以后的程序或操作系统使用（实际上，一些 malloc 实现只能将内存归还给程序，而无法将内存归还给操作系统）。物理内存和虚拟内存要理解内存在程序中是如何分配的，首先需要理解如何将内存从操作系统分配给程序。计算机上的每一个进程都认为自己可以访问所有的物理内存。显然，由于同时在运行多个程序，所以每个进程不可能拥有全部内存。实际上，这些进程使用的是虚拟内存。只是作为一个例子，让我们假定您的程序正在访问地址为 629 的内存。不过，虚拟内存系统不需要将其存储在位置为 629 的 RAM 中。实际上，它甚至可以不在 RAM 中 —— 如果物理 RAM 已经满了，它甚至可能已经被转移到硬盘上！由于这类地址不必反映内存所在的物理位置，所以它们被称为虚拟内存。操作系统维持着一个虚拟地址到物理地址的转换的表，以便计算机硬件可以正确地响应地址请求。并且，如果地址在硬盘上而不是在 RAM 中，那么操作系统将暂时停止您的进程，将其他内存转存到硬盘中，从硬盘上加载被请求的内存，然后再重新启动您的进程。这样，每个进程都获得了自己可以使用的地址空间，可以访问比您物理上安装的内存更多的内存。在 32-位 x86 系统上，每一个进程可以访问 4 GB 内存。现在，大部分人的系统上并没有 4 GB 内存，即使您将 swap 也算上，每个进程所使用的内存也肯定少于 4 GB。因此，当加载一个进程时，它会得到一个取决于某个称为系统中断点（system break）的特定地址的初始内存分配。该地址之后是未被映射的内存 —— 用于在 RAM 或者硬盘中没有分配相应物理位置的内存。因此，如果一个进程运行超出了它初始分配的内存，那么它必须请求操作系统“映射进来（map in）”更多的内存。（映射是一个表示一一对应关系的数学术语 —— 当内存的虚拟地址有一个对应的物理地址来存储内存内容时，该内存将被映射。）基于 UNIX 的系统有两个可映射到附加内存中的基本系统调用： brk： brk() 是一个非常简单的系统调用。还记得系统中断点吗？该位置是进程映射的内存边界。 brk() 只是简单地将这个位置向前或者向后移动，就可以向进程添加内存或者从进程取走内存。 mmap： mmap()，或者说是“内存映像”，类似于 brk()，但是更为灵活。首先，它可以映射任何位置的内存，而不单单只局限于进程。其次，它不仅可以将虚拟地址映射到物理的 RAM 或者 swap，它还可以将它们映射到文件和文件位置，这样，读写内存将对文件中的数据进行读写。不过，在这里，我们只关心 mmap 向进程添加被映射的内存的能力。 munmap() 所做的事情与 mmap() 相反。如您所见， brk() 或者 mmap() 都可以用来向我们的进程添加额外的虚拟内存。在我们的例子中将使用 brk()，因为它更简单，更通用。实现一个简单的分配程序如果您曾经编写过很多 C 程序，那么您可能曾多次使用过 malloc() 和 free()。不过，您可能没有用一些时间去思考它们在您的操作系统中是如何实现的。本节将向您展示 malloc 和 free 的一个最简化实现的代码，来帮助说明管理内存时都涉及到了哪些事情。要试着运行这些示例，需要先复制本代码清单，并将其粘贴到一个名为 malloc.c 的文件中。接下来，我将一次一个部分地对该清单进行解释。在大部分操作系统中，内存分配由以下两个简单的函数来处理： void *malloc(long numbytes)：该函数负责分配 numbytes 大小的内存，并返回指向第一个字节的指针。 void free(void *firstbyte)：如果给定一个由先前的 malloc 返回的指针，那么该函数会将分配的空间归还给进程的“空闲空间”。 malloc_init 将是初始化内存分配程序的函数。它要完成以下三件事：将分配程序标识为已经初始化，找到系统中最后一个有效内存地址，然后建立起指向我们管理的内存的指针。这三个变量都是全局变量：清单 1. 我们的简单分配程序的全局变量 int has_initialized = 0; void *managed_memory_start; void *last_valid_address; 如前所述，被映射的内存的边界（最后一个有效地址）常被称为系统中断点或者当前中断点。在很多 UNIX® 系统中，为了指出当前系统中断点，必须使用 sbrk(0) 函数。 sbrk 根据参数中给出的字节数移动当前系统中断点，然后返回新的系统中断点。使用参数 0 只是返回当前中断点。这里是我们的 malloc 初始化代码，它将找到当前中断点并初始化我们的变量：清单 2. 分配程序初始化函数 /* Include the sbrk function */ #include void malloc_init() { /* grab the last valid address from the OS */ last_valid_address = sbrk(0); /* we don't have any memory to manage yet, so *just set the beginning to be last_valid_address */ managed_memory_start = last_valid_address; /* Okay, we're initialized and ready to go */ has_initialized = 1; } 现在，为了完全地管理内存，我们需要能够追踪要分配和回收哪些内存。在对内存块进行了 free 调用之后，我们需要做的是诸如将它们标记为未被使用的等事情，并且，在调用 malloc 时，我们要能够定位未被使用的内存块。因此， malloc 返回的每块内存的起始处首先要有这个结构：清单 3. 内存控制块结构定义 struct mem_control_block { int is_available; int size; }; 现在，您可能会认为当程序调用 malloc 时这会引发问题 —— 它们如何知道这个结构？答案是它们不必知道；在返回指针之前，我们会将其移动到这个结构之后，把它隐藏起来。这使得返回的指针指向没有用于任何其他用途的内存。那样，从调用程序的角度来看，它们所得到的全部是空闲的、开放的内存。然后，当通过 free() 将该指针传递回来时，我们只需要倒退几个内存字节就可以再次找到这个结构。在讨论分配内存之前，我们将先讨论释放，因为它更简单。为了释放内存，我们必须要做的惟一一件事情就是，获得我们给出的指针，回退 sizeof(struct mem_control_block) 个字节，并将其标记为可用的。这里是对应的代码：清单 4. 解除分配函数 void free(void *firstbyte) { struct mem_control_block *mcb; /* Backup from the given pointer to find the * mem_control_block */ mcb = firstbyte - sizeof(struct mem_control_block); /* Mark the block as being available */ mcb->is_available = 1; /* That's It! We're done. */ return; } 如您所见，在这个分配程序中，内存的释放使用了一个非常简单的机制，在固定时间内完成内存释放。分配内存稍微困难一些。以下是该算法的略述：清单 5. 主分配程序的伪代码 1. If our allocator has not been initialized, initialize it. 2. Add sizeof(struct mem_control_block) to the size requested. 3. start at managed_memory_start. 4. Are we at last_valid address? 5. If we are: A. We didn't find any existing space that was large enough -- ask the operating system for more and return that. 6. Otherwise: A. Is the current space available (check is_available from the mem_control_block)? B. If it is: i) Is it large enough (check "size" from the mem_control_block)? ii) If so: a. Mark it as unavailable b. Move past mem_control_block and return the pointer iii) Otherwise: a. Move forward "size" bytes b. Go back go step 4 C. Otherwise: i) Move forward "size" bytes ii) Go back to step 4 我们主要使用连接的指针遍历内存来寻找开放的内存块。这里是代码：清单 6. 主分配程序 void *malloc(long numbytes) { /* Holds where we are looking in memory */ void *current_location; /* This is the same as current_location, but cast to a * memory_control_block */ struct mem_control_block *current_location_mcb; /* This is the memory location we will return. It will * be set to 0 until we find something suitable */ void *memory_location; /* Initialize if we haven't already done so */ if(! has_initialized) { malloc_init(); } /* The memory we search for has to include the memory * control block, but the users of malloc don't need * to know this, so we'll just add it in for them. */ numbytes = numbytes + sizeof(struct mem_control_block); /* Set memory_location to 0 until we find a suitable * location */ memory_location = 0; /* Begin searching at the start of managed memory */ current_location = managed_memory_start; /* Keep going until we have searched all allocated space */ while(current_location != last_valid_address) { /* current_location and current_location_mcb point * to the same address. However, current_location_mcb * is of the correct type, so we can use it as a struct. * current_location is a void pointer so we can use it * to calculate addresses. */ current_location_mcb = (struct mem_control_block *)current_location; if(current_location_mcb->is_available) { if(current_location_mcb->size >= numbytes) { /* Woohoo! We've found an open, * appropriately-size location. */ /* It is no longer available */ current_location_mcb->is_available = 0; /* We own it */ memory_location = current_location; /* Leave the loop */ break; } } /* If we made it here, it's because the Current memory * block not suitable; move to the next one */ current_location = current_location + current_location_mcb->size; } /* If we still don't have a valid location, we'll * have to ask the operating system for more memory */ if(! memory_location) { /* Move the program break numbytes further */ sbrk(numbytes); /* The new memory will be where the last valid * address left off */ memory_location = last_valid_address; /* We'll move the last valid address forward * numbytes */ last_valid_address = last_valid_address + numbytes; /* We need to initialize the mem_control_block */ current_location_mcb = memory_location; current_location_mcb->is_available = 0; current_location_mcb->size = numbytes; } /* Now, no matter what (well, except for error conditions), * memory_location has the address of the memory, including * the mem_control_block */ /* Move the pointer past the mem_control_block */ memory_location = memory_location + sizeof(struct mem_control_block); /* Return the pointer */ return memory_location; } 这就是我们的内存管理器。现在，我们只需要构建它，并在程序中使用它即可。运行下面的命令来构建 malloc 兼容的分配程序（实际上，我们忽略了 realloc() 等一些函数，不过， malloc() 和 free() 才是最主要的函数）：清单 7. 编译分配程序 gcc -shared -fpic malloc.c -o malloc.so 该程序将生成一个名为 malloc.so 的文件，它是一个包含有我们的代码的共享库。在 UNIX 系统中，现在您可以用您的分配程序来取代系统的 malloc()，做法如下：清单 8. 替换您的标准的 malloc LD_PRELOAD=/path/to/malloc.so export LD_PRELOAD LD_PRELOAD 环境变量使动态链接器在加载任何可执行程序之前，先加载给定的共享库的符号。它还为特定库中的符号赋予优先权。因此，从现在起，该会话中的任何应用程序都将使用我们的 malloc()，而不是只有系统的应用程序能够使用。有一些应用程序不使用 malloc()，不过它们是例外。其他使用 realloc() 等其他内存管理函数的应用程序，或者错误地假定 malloc() 内部行为的那些应用程序，很可能会崩溃。ash shell 似乎可以使用我们的新 malloc() 很好地工作。如果您想确保 malloc() 正在被使用，那么您应该通过向函数的入口点添加 write() 调用来进行测试。我们的内存管理器在很多方面都还存在欠缺，但它可以有效地展示内存管理需要做什么事情。它的某些缺点包括：由于它对系统中断点（一个全局变量）进行操作，所以它不能与其他分配程序或者 mmap 一起使用。当分配内存时，在最坏的情形下，它将不得不遍历全部进程内存；其中可能包括位于硬盘上的很多内存，这意味着操作系统将不得不花时间去向硬盘移入数据和从硬盘中移出数据。没有很好的内存不足处理方案（ malloc 只假定内存分配是成功的）。它没有实现很多其他的内存函数，比如 realloc()。由于 sbrk() 可能会交回比我们请求的更多的内存，所以在堆（heap）的末端会遗漏一些内存。虽然 is_available 标记只包含一位信息，但它要使用完整的 4-字节的字。分配程序不是线程安全的。分配程序不能将空闲空间拼合为更大的内存块。分配程序的过于简单的匹配算法会导致产生很多潜在的内存碎片。我确信还有很多其他问题。这就是为什么它只是一个例子! 其他 malloc 实现 malloc() 的实现有很多，这些实现各有优点与缺点。在设计一个分配程序时，要面临许多需要折衷的选择，其中包括：分配的速度。回收的速度。有线程的环境的行为。内存将要被用光时的行为。局部缓存。簿记（Bookkeeping）内存开销。虚拟内存环境中的行为。小的或者大的对象。实时保证。每一个实现都有其自身的优缺点集合。在我们的简单的分配程序中，分配非常慢，而回收非常快。另外，由于它在使用虚拟内存系统方面较差，所以它最适于处理大的对象。还有其他许多分配程序可以使用。其中包括： Doug Lea Malloc：Doug Lea Malloc 实际上是完整的一组分配程序，其中包括 Doug Lea 的原始分配程序，GNU libc 分配程序和 ptmalloc。 Doug Lea 的分配程序有着与我们的版本非常类似的基本结构，但是它加入了索引，这使得搜索速度更快，并且可以将多个没有被使用的块组合为一个大的块。它还支持缓存，以便更快地再次使用最近释放的内存。 ptmalloc 是 Doug Lea Malloc 的一个扩展版本，支持多线程。在本文后面的参考资料部分中，有一篇描述 Doug Lea 的 Malloc 实现的文章。 BSD Malloc：BSD Malloc 是随 4.2 BSD 发行的实现，包含在 FreeBSD 之中，这个分配程序可以从预先确实大小的对象构成的池中分配对象。它有一些用于对象大小的 size 类，这些对象的大小为 2 的若干次幂减去某一常数。所以，如果您请求给定大小的一个对象，它就简单地分配一个与之匹配的 size 类。这样就提供了一个快速的实现，但是可能会浪费内存。在参考资料部分中，有一篇描述该实现的文章。 Hoard：编写 Hoard 的目标是使内存分配在多线程环境中进行得非常快。因此，它的构造以锁的使用为中心，从而使所有进程不必等待分配内存。它可以显著地加快那些进行很多分配和回收的多线程进程的速度。在参考资料部分中，有一篇描述该实现的文章。众多可用的分配程序中最有名的就是上述这些分配程序。如果您的程序有特别的分配需求，那么您可能更愿意编写一个定制的能匹配您的程序内存分配方式的分配程序。不过，如果不熟悉分配程序的设计，那么定制分配程序通常会带来比它们解决的问题更多的问题。要获得关于该主题的适当的介绍，请参阅 Donald Knuth 撰写的 The Art of Computer Programming Volume 1: Fundamental Algorithms 中的第 2.5 节“Dynamic Storage Allocation”（请参阅参考资料中的链接）。它有点过时，因为它没有考虑虚拟内存环境，不过大部分算法都是基于前面给出的函数。在 C++ 中，通过重载 operator new()，您可以以每个类或者每个模板为单位实现自己的分配程序。在 Andrei Alexandrescu 撰写的 Modern C++ Design 的第 4 章（“Small Object Allocation”）中，描述了一个小对象分配程序（请参阅参考资料中的链接）。基于 malloc() 的内存管理的缺点不只是我们的内存管理器有缺点，基于 malloc() 的内存管理器仍然也有很多缺点，不管您使用的是哪个分配程序。对于那些需要保持长期存储的程序使用 malloc() 来管理内存可能会非常令人失望。如果您有大量的不固定的内存引用，经常难以知道它们何时被释放。生存期局限于当前函数的内存非常容易管理，但是对于生存期超出该范围的内存来说，管理内存则困难得多。而且，关于内存管理是由进行调用的程序还是由被调用的函数来负责这一问题，很多 API 都不是很明确。因为管理内存的问题，很多程序倾向于使用它们自己的内存管理规则。C++ 的异常处理使得这项任务更成问题。有时好像致力于管理内存分配和清理的代码比实际完成计算任务的代码还要多！因此，我们将研究内存管理的其他选择。回页首半自动内存管理策略引用计数引用计数是一种半自动（semi-automated）的内存管理技术，这表示它需要一些编程支持，但是它不需要您确切知道某一对象何时不再被使用。引用计数机制为您完成内存管理任务。在引用计数中，所有共享的数据结构都有一个域来包含当前活动“引用”结构的次数。当向一个程序传递一个指向某个数据结构指针时，该程序会将引用计数增加 1。实质上，您是在告诉数据结构，它正在被存储在多少个位置上。然后，当您的进程完成对它的使用后，该程序就会将引用计数减少 1。结束这个动作之后，它还会检查计数是否已经减到零。如果是，那么它将释放内存。这样做的好处是，您不必追踪程序中某个给定的数据结构可能会遵循的每一条路径。每次对其局部的引用，都将导致计数的适当增加或减少。这样可以防止在使用数据结构时释放该结构。不过，当您使用某个采用引用计数的数据结构时，您必须记得运行引用计数函数。另外，内置函数和第三方的库不会知道或者可以使用您的引用计数机制。引用计数也难以处理发生循环引用的数据结构。要实现引用计数，您只需要两个函数 —— 一个增加引用计数，一个减少引用计数并当计数减少到零时释放内存。一个示例引用计数函数集可能看起来如下所示：清单 9. 基本的引用计数函数 /* Structure Definitions*/ /* Base structure that holds a refcount */ struct refcountedstruct { int refcount; } /* All refcounted structures must mirror struct * refcountedstruct for their first variables */ /* Refcount maintenance functions */ /* Increase reference count */ void REF(void *data) { struct refcountedstruct *rstruct; rstruct = (struct refcountedstruct *) data; rstruct->refcount++; } /* Decrease reference count */ void UNREF(void *data) { struct refcountedstruct *rstruct; rstruct = (struct refcountedstruct *) data; rstruct->refcount--; /* Free the structure if there are no more users */ if(rstruct->refcount == 0) { free(rstruct); } } REF 和 UNREF 可能会更复杂，这取决于您想要做的事情。例如，您可能想要为多线程程序增加锁，那么您可能想扩展 refcountedstruct，使它同样包含一个指向某个在释放内存之前要调用的函数的指针（类似于面向对象语言中的析构函数 —— 如果您的结构中包含这些指针，那么这是必需的）。当使用 REF 和 UNREF 时，您需要遵守这些指针的分配规则： UNREF 分配前左端指针（left-hand-side pointer）指向的值。 REF 分配后左端指针（left-hand-side pointer）指向的值。在传递使用引用计数的结构的函数中，函数需要遵循以下这些规则：在函数的起始处 REF 每一个指针。在函数的结束处 UNREF 第一个指针。以下是一个使用引用计数的生动的代码示例：清单 10. 使用引用计数的示例 /* EXAMPLES OF USAGE */ /* Data type to be refcounted */ struct mydata { int refcount; /* same as refcountedstruct */ int datafield1; /* Fields specific to this struct */ int datafield2; /* other declarations would go here as appropriate */ }; /* Use the functions in code */ void dosomething(struct mydata *data) { REF(data); /* Process data */ /* when we are through */ UNREF(data); } struct mydata *globalvar1; /* Note that in this one, we don't decrease the * refcount since we are maintaining the reference * past the end of the function call through the * global variable */ void storesomething(struct mydata *data) { REF(data); /* passed as a parameter */ globalvar1 = data; REF(data); /* ref because of Assignment */ UNREF(data); /* Function finished */ } 由于引用计数是如此简单，大部分程序员都自已去实现它，而不是使用库。不过，它们依赖于 malloc 和 free 等低层的分配程序来实际地分配和释放它们的内存。在 Perl 等高级语言中，进行内存管理时使用引用计数非常广泛。在这些语言中，引用计数由语言自动地处理，所以您根本不必担心它，除非要编写扩展模块。由于所有内容都必须进行引用计数，所以这会对速度产生一些影响，但它极大地提高了编程的安全性和方便性。以下是引用计数的益处：实现简单。易于使用。由于引用是数据结构的一部分，所以它有一个好的缓存位置。不过，它也有其不足之处：要求您永远不要忘记调用引用计数函数。无法释放作为循环数据结构的一部分的结构。减缓几乎每一个指针的分配。尽管所使用的对象采用了引用计数，但是当使用异常处理（比如 try 或 setjmp()/ longjmp()）时，您必须采取其他方法。需要额外的内存来处理引用。引用计数占用了结构中的第一个位置，在大部分机器中最快可以访问到的就是这个位置。在多线程环境中更慢也更难以使用。 C++ 可以通过使用智能指针（smart pointers）来容忍程序员所犯的一些错误，智能指针可以为您处理引用计数等指针处理细节。不过，如果不得不使用任何先前的不能处理智能指针的代码（比如对 C 库的联接），实际上，使用它们的后果通实比不使用它们更为困难和复杂。因此，它通常只是有益于纯 C++ 项目。如果您想使用智能指针，那么您实在应该去阅读 Alexandrescu 撰写的 Modern C++ Design 一书中的“Smart Pointers”那一章。内存池内存池是另一种半自动内存管理方法。内存池帮助某些程序进行自动内存管理，这些程序会经历一些特定的阶段，而且每个阶段中都有分配给进程的特定阶段的内存。例如，很多网络服务器进程都会分配很多针对每个连接的内存 —— 内存的最大生存期限为当前连接的存在期。Apache 使用了池式内存（pooled memory），将其连接拆分为各个阶段，每个阶段都有自己的内存池。在结束每个阶段时，会一次释放所有内存。在池式内存管理中，每次内存分配都会指定内存池，从中分配内存。每个内存池都有不同的生存期限。在 Apache 中，有一个持续时间为服务器存在期的内存池，还有一个持续时间为连接的存在期的内存池，以及一个持续时间为请求的存在期的池，另外还有其他一些内存池。因此，如果我的一系列函数不会生成比连接持续时间更长的数据，那么我就可以完全从连接池中分配内存，并知道在连接结束时，这些内存会被自动释放。另外，有一些实现允许注册清除函数（cleanup functions），在清除内存池之前，恰好可以调用它，来完成在内存被清理前需要完成的其他所有任务（类似于面向对象中的析构函数）。要在自己的程序中使用池，您既可以使用 GNU libc 的 obstack 实现，也可以使用 Apache 的 Apache Portable Runtime。GNU obstack 的好处在于，基于 GNU 的 Linux 发行版本中默认会包括它们。Apache Portable Runtime 的好处在于它有很多其他工具，可以处理编写多平台服务器软件所有方面的事情。要深入了解 GNU obstack 和 Apache 的池式内存实现，请参阅参考资料部分中指向这些实现的文档的链接。下面的假想代码列表展示了如何使用 obstack：清单 11. obstack 的示例代码 #include #include /* Example code listing for using obstacks */ /* Used for obstack macros (xmalloc is a malloc function that exits if memory is exhausted */ #define obstack_chunk_alloc xmalloc #define obstack_chunk_free free /* Pools */ /* Only permanent allocations should go in this pool */ struct obstack *global_pool; /* This pool is for per-connection data */ struct obstack *connection_pool; /* This pool is for per-request data */ struct obstack *request_pool; void allocation_failed() { exit(1); } int main() { /* Initialize Pools */ global_pool = (struct obstack *) xmalloc (sizeof (struct obstack)); obstack_init(global_pool); connection_pool = (struct obstack *) xmalloc (sizeof (struct obstack)); obstack_init(connection_pool); request_pool = (struct obstack *) xmalloc (sizeof (struct obstack)); obstack_init(request_pool); /* Set the error handling function */ obstack_alloc_failed_handler = &allocation_failed; /* Server main loop */ while(1) { wait_for_connection(); /* We are in a connection */ while(more_requests_available()) { /* Handle request */ handle_request(); /* Free all of the memory allocated * in the request pool */ obstack_free(request_pool, NULL); } /* We're finished with the connection, time * to free that pool */ obstack_free(connection_pool, NULL); } } int handle_request() { /* Be sure that all object allocations are allocated * from the request pool */ int bytes_i_need = 400; void *data1 = obstack_alloc(request_pool, bytes_i_need); /* Do stuff to process the request */ /* return */ return 0; } 基本上，在操作的每一个主要阶段结束之后，这个阶段的 obstack 会被释放。不过，要注意的是，如果一个过程需要分配持续时间比当前阶段更长的内存，那么它也可以使用更长期限的 obstack，比如连接或者全局内存。传递给 obstack_free() 的 NULL 指出它应该释放 obstack 的全部内容。可以用其他的值，但是它们通常不怎么实用。使用池式内存分配的益处如下所示：应用程序可以简单地管理内存。内存分配和回收更快，因为每次都是在一个池中完成的。分配可以在 O(1) 时间内完成，释放内存池所需时间也差不多（实际上是 O(n) 时间，不过在大部分情况下会除以一个大的因数，使其变成 O(1)）。可以预先分配错误处理池（Error-handling pools），以便程序在常规内存被耗尽时仍可以恢复。有非常易于使用的标准实现。池式内存的缺点是：内存池只适用于操作可以分阶段的程序。内存池通常不能与第三方库很好地合作。如果程序的结构发生变化，则不得不修改内存池，这可能会导致内存管理系统的重新设计。您必须记住需要从哪个池进行分配。另外，如果在这里出错，就很难捕获该内存池。回页首垃圾收集垃圾收集（Garbage collection）是全自动地检测并移除不再使用的数据对象。垃圾收集器通常会在当可用内存减少到少于一个具体的阈值时运行。通常，它们以程序所知的可用的一组“基本”数据 —— 栈数据、全局变量、寄存器 —— 作为出发点。然后它们尝试去追踪通过这些数据连接到每一块数据。收集器找到的都是有用的数据；它没有找到的就是垃圾，可以被销毁并重新使用这些无用的数据。为了有效地管理内存，很多类型的垃圾收集器都需要知道数据结构内部指针的规划，所以，为了正确运行垃圾收集器，它们必须是语言本身的一部分。收集器的类型复制（copying）：这些收集器将内存存储器分为两部分，只允许数据驻留在其中一部分上。它们定时地从“基本”的元素开始将数据从一部分复制到另一部分。内存新近被占用的部分现在成为活动的，另一部分上的所有内容都认为是垃圾。另外，当进行这项复制操作时，所有指针都必须被更新为指向每个内存条目的新位置。因此，为使用这种垃圾收集方法，垃圾收集器必须与编程语言集成在一起。标记并清理（Mark and sweep）：每一块数据都被加上一个标签。不定期的，所有标签都被设置为 0，收集器从“基本”的元素开始遍历数据。当它遇到内存时，就将标签标记为 1。最后没有被标记为 1 的所有内容都认为是垃圾，以后分配内存时会重新使用它们。增量的（Incremental）：增量垃圾收集器不需要遍历全部数据对象。因为在收集期间的突然等待，也因为与访问所有当前数据相关的缓存问题（所有内容都不得不被页入（page-in）），遍历所有内存会引发问题。增量收集器避免了这些问题。保守的（Conservative）：保守的垃圾收集器在管理内存时不需要知道与数据结构相关的任何信息。它们只查看所有数据类型，并假定它们可以全部都是指针。所以，如果一个字节序列可以是一个指向一块被分配的内存的指针，那么收集器就将其标记为正在被引用。有时没有被引用的内存会被收集，这样会引发问题，例如，如果一个整数域中包含一个值，该值是已分配内存的地址。不过，这种情况极少发生，而且它只会浪费少量内存。保守的收集器的优势是，它们可以与任何编程语言相集成。 Hans Boehm 的保守垃圾收集器是可用的最流行的垃圾收集器之一，因为它是免费的，而且既是保守的又是增量的，可以使用 --enable-redirect-malloc 选项来构建它，并且可以将它用作系统分配程序的简易替代者（drop-in replacement）（用 malloc/ free 代替它自己的 API）。实际上，如果这样做，您就可以使用与我们在示例分配程序中所使用的相同的 LD_PRELOAD 技巧，在系统上的几乎任何程序中启用垃圾收集。如果您怀疑某个程序正在泄漏内存，那么您可以使用这个垃圾收集器来控制进程。在早期，当 Mozilla 严重地泄漏内存时，很多人在其中使用了这项技术。这种垃圾收集器既可以在 Windows® 下运行，也可以在 UNIX 下运行。垃圾收集的一些优点：您永远不必担心内存的双重释放或者对象的生命周期。使用某些收集器，您可以使用与常规分配相同的 API。其缺点包括：使用大部分收集器时，您都无法干涉何时释放内存。在多数情况下，垃圾收集比其他形式的内存管理更慢。垃圾收集错误引发的缺陷难于调试。如果您忘记将不再使用的指针设置为 null，那么仍然会有内存泄漏。回页首结束语一切都需要折衷：性能、易用、易于实现、支持线程的能力等，这里只列出了其中的一些。为了满足项目的要求，有很多内存管理模式可以供您使用。每种模式都有大量的实现，各有其优缺点。对很多项目来说，使用编程环境默认的技术就足够了，不过，当您的项目有特殊的需要时，了解可用的选择将会有帮助。下表对比了本文中涉及的内存管理策略。表 1. 内存分配策略的对比策略分配速度回收速度局部缓存易用性通用性实时可用 SMP 线程友好定制分配程序取决于实现取决于实现取决于实现很难无取决于实现取决于实现简单分配程序内存使用少时较快很快差容易高否否 GNU malloc 中快中容易高否中 Hoard 中中中容易高否是引用计数 N/A N/A 非常好中中是（取决于 malloc 实现）取决于实现池中非常快极好中中是（取决于 malloc 实现）取决于实现垃圾收集中（进行收集时慢）中差中中否几乎不增量垃圾收集中中中中中否几乎不增量保守垃圾收集中中中容易高否几乎不参考资料您可以参阅本文在 developerWorks 全球站点上的英文原文。 Web 上的文档 GNU C Library 手册的 obstacks 部分提供了 obstacks 编程接口。 Apache Portable Runtime 文档描述了它们的池式分配程序的接口。基本的分配程序 Doug Lea 的 Malloc 是最流行的内存分配程序之一。 BSD Malloc 用于大部分基于 BSD 的系统中。 ptmalloc 起源于 Doug Lea 的 malloc，用于 GLIBC 之中。 Hoard 是一个为多线程应用程序优化的 malloc 实现。 GNU Memory-Mapped Malloc（GDB 的组成部分）是一个基于 mmap() 的 malloc 实现。池式分配程序 GNU Obstacks（GNU Libc 的组成部分）是安装最多的池式分配程序，因为在每一个基于 glibc 的系统中都有它。 Apache 的池式分配程序（Apache Portable Runtime 中）是应用最为广泛的池式分配程序。 Squid 有其自己的池式分配程序。 NetBSD 也有其自己的池式分配程序。 talloc 是一个池式分配程序，是 Samba 的组成部分。智能指针和定制分配程序 Loki C++ Library 有很多为 C++ 实现的通用模式，包括智能指针和一个定制的小对象分配程序。垃圾收集器 Hahns Boehm Conservative Garbage Collector 是最流行的开源垃圾收集器，它可以用于常规的 C/C++ 程序。关于现代操作系统中的虚拟内存的文章 Marshall Kirk McKusick 和 Michael J. Karels 合著的 A New Virtual Memory Implementation for Berkeley UNIX 讨论了 BSD 的 VM 系统。 Mel Gorman's Linux VM Documentation 讨论了 Linux VM 系统。关于 malloc 的文章 Poul-Henning Kamp 撰写的 Malloc in Modern Virtual Memory Environments 讨论的是 malloc 以及它如何与 BSD 虚拟内存交互。 Berger、McKinley、Blumofe 和 Wilson 合著的 Hoard -- a Scalable Memory Allocator for Multithreaded Environments 讨论了 Hoard 分配程序的实现。 Marshall Kirk McKusick 和 Michael J. Karels 合著的 Design of a General Purpose Memory Allocator for the 4.3BSD UNIX Kernel 讨论了内核级的分配程序。 Doug Lea 撰写的 A Memory Allocator 给出了一个关于设计和实现分配程序的概述，其中包括设计选择与折衷。 Emery D. Berger 撰写的 Memory Management for High-Performance Applications 讨论的是定制内存管理以及它如何影响高性能应用程序。关于定制分配程序的文章 Doug Lea 撰写的 Some Storage Management Techniques for Container Classes 描述的是为 C++ 类编写定制分配程序。 Berger、Zorn 和 McKinley 合著的 Composing High-Performance Memory Allocators 讨论了如何编写定制分配程序来加快具体工作的速度。 Berger、Zorn 和 McKinley 合著的 Reconsidering Custom Memory Allocation 再次提及了定制分配的主题，看是否真正值得为其费心。关于垃圾收集的文章 Paul R. Wilson 撰写的 Uniprocessor Garbage Collection Techniques 给出了垃圾收集的一个基本概述。 Benjamin Zorn 撰写的 The Measured Cost of Garbage Collection 给出了关于垃圾收集和性能的硬数据（hard data）。 Hans-Juergen Boehm 撰写的 Memory Allocation Myths and Half-Truths 给出了关于垃圾收集的神话（myths）。 Hans-Juergen Boehm 撰写的 Space Efficient Conservative Garbage Collection 是一篇描述他的用于 C/C++ 的垃圾收集器的文章。 Web 上的通用参考资料内存管理参考中有很多关于内存管理参考资料和技术文章的链接。关于内存管理和内存层级的 OOPS Group Papers 是非常好的一组关于此主题的技术文章。 C++ 中的内存管理讨论的是为 C++ 编写定制的分配程序。 Programming Alternatives: Memory Management 讨论了程序员进行内存管理时的一些选择。垃圾收集 FAQ 讨论了关于垃圾收集您需要了解的所有内容。 Richard Jones 的 Garbage Collection Bibliography 有指向任何您想要的关于垃圾收集的文章的链接。书籍 Michael Daconta 撰写的 C++ Pointers and Dynamic Memory Management 介绍了关于内存管理的很多技术。 Frantisek Franek 撰写的 Memory as a Programming Concept in C and C++ 讨论了有效使用内存的技术与工具，并给出了在计算机编程中应当引起注意的内存相关错误的角色。 Richard Jones 和 Rafael Lins 合著的 Garbage Collection: Algorithms for Automatic Dynamic Memory Management 描述了当前使用的最常见的垃圾收集算法。在 Donald Knuth 撰写的 The Art of Computer Programming 第 1 卷 Fundamental Algorithms 的第 2.5 节“Dynamic Storage Allocation”中，描述了实现基本的分配程序的一些技术。在 Donald Knuth 撰写的 The Art of Computer Programming 第 1 卷 Fundamental Algorithms 的第 2.3.5 节“Lists and Garbage Collection”中，讨论了用于列表的垃圾收集算法。 Andrei Alexandrescu 撰写的 Modern C++ Design 第 4 章“Small Object Allocation”描述了一个比 C++ 标准分配程序效率高得多的一个高速小对象分配程序。 Andrei Alexandrescu 撰写的 Modern C++ Design 第 7 章“Smart Pointers”描述了在 C++ 中智能指针的实现。 Jonathan 撰写的 Programming from the Ground Up 第 8 章“Intermediate Memory Topics”中有本文使用的简单分配程序的一个汇编语言版本。来自 developerWorks 自我管理数据缓冲区内存（developerWorks，2004 年 1 月）略述了一个用于管理内存的自管理的抽象数据缓存器的伪 C （pseudo-C）实现。 A framework for the user defined malloc replacement feature （developerWorks，2002 年 2 月）展示了如何利用 AIX 中的一个工具，使用自己设计的内存子系统取代原有的内存子系统。掌握 Linux 调试技术（developerWorks，2002 年 8 月）描述了可以使用调试方法的 4 种不同情形：段错误、内存溢出、内存泄漏和挂起。在处理 Java 程序中的内存漏洞（developerWorks，2001 年 2 月）中，了解导致 Java 内存泄漏的原因，以及何时需要考虑它们。在 developerWorks Linux 专区中，可以找到更多为 Linux 开发人员准备的参考资料。从 developerWorks 的 Speed-start your Linux app 专区中，可以下载运行于 Linux 之上的 IBM 中间件产品的免费测试版本，其中包括 WebSphere® Studio Application Developer、WebSphere Application Server、DB2® Universal Database、Tivoli® Access Manager 和 Tivoli Directory Server，查找 how-to 文章和技术支持。通过参与 developerWorks blogs 加入到 developerWorks 社区。可以在 Developer Bookstore Linux 专栏中定购打折出售的 Linux 书籍。关于作者 Jonathan Bartlett 是 Programming from the Ground Up 一书的作者，这本书介绍的是 Linux 汇编语言编程。Jonathan Bartlett 是 New Media Worx 的总开发师，负责为客户开发 Web、视频、kiosk 和桌面应用程序。您可以通过 johnnyb@eskimo.com 与 Jonathan 联系。

高校技术转移办公室人员如何提升成果转化项目的市场匹配效率？.docx

科易网基于AI模型+行业大数据，在技术转移、成果转化、技术经纪、知识产权、产业创新、科技招商等垂直领域的多样化应用场景，推出科技创新领域的AI+工具与数智化解决方案，推动科技创新与产业创新智能化发展。

产业园区如何高效识别并匹配优质技术成果与产业需求？.docx

科易网基于AI模型+行业大数据，在技术转移、成果转化、技术经纪、知识产权、产业创新、科技招商等垂直领域的多样化应用场景，推出科技创新领域的AI+工具与数智化解决方案，推动科技创新与产业创新智能化发展。

华为鲲鹏920s serdestool工具

华为鲲鹏920s serdestool工具，用来进行眼图测试

【软件授权管理】VC5.0-OLP版本及插件授权信息：包含主程序与扩展模块的永久许可配置方案

内容概要：本文档为VC5.0 OLP版本的软件授权文件，主要包含VC5.0主程序及插件“addon_vc_solver”的授权信息。其中，主程序“prod_vc_prem_olp”有两个授权条目：一个有效期至2050年4月3日午夜，另一个为永久授权（无过期日期），均标明为独占许可。插件“addon_vc_solver”版本5.0同样拥有永久授权，且为独占性质。所有授权均以加密字符串形式呈现，附带版本号、许可类型和有效期说明。; 适合人群：企业IT管理员、软件授权管理人员、需要使用VC5.0软件及其求解器插件的技术人员。; 使用场景及目标：①确认VC5.0软件及插件的合法授权状态和有效期；②管理企业内部软件许可部署，确保合规使用；③识别永久授权与限时授权的区别，便于长期规划软件使用。; 阅读建议：此授权文件包含加密内容，普通用户无需解析密文，应重点关注明文部分的有效期和版本信息，必要时可联系供应商验证授权真实性。

两相交错并联Boost电路仿真模型（Simulink仿真实现）

内容概要：本文系统介绍了“两相交错并联Boost电路仿真模型”的Simulink仿真实现，属于电力电子变换器领域的高阶仿真设计内容。该模型采用交错并联结构，有效提升系统效率、降低输入电流纹波、增强带载能力，适用于中高功率DC-DC升压场合，尤其在新能源、电动汽车、微电网及储能系统中有广泛应用前景。文档还汇总了丰富的电力电子系统仿真资源，涵盖LLC谐振变换器、PFC功率因数校正电路、双向Buck-Boost、三相PWM整流器等多种典型拓扑的闭环控制、软启动、变频控制及多相交错技术的Simulink/Psim仿真模型，并配套提供详细的计算书、参考文献与说明文档，构建了一个完整的科研与工程实践支持体系，具有较强的学术研究与工程复现价值。; 适合人群：电气工程、自动化、电力电子及相关专业的研究生、科研人员及具备一定Matlab/Simulink仿真基础的工程技术人员；特别适用于从事新能源变换器设计、电源系统开发、电力电子装置研发或承担SCI/EI论文复现与课题攻关的专业人员。; 使用场景及目标：① 深入学习两相交错并联Boost电路的工作机理与建模仿真方法，掌握其在抑制纹波与提升效率方面的优势；② 借助所提供的多样化仿真案例，开展不同拓扑结构的性能对比、控制策略研究（如双闭环控制、变频控制、软开关技术）及系统级优化设计；③ 支持科研项目中的模型验证、实验复现、创新拓扑设计与高水平论文撰写。; 阅读建议：建议结合文中列举的LLC、PFC、Buck等其他相关仿真模型进行横向对比学习，充分利用配套的计算书与参考文献深化理论理解；推荐按照资源目录顺序循序渐进地实践操作，重点关注控制策略与系统参数之间的耦合关系，全面提升电力电子系统的仿真建模能力与工程应用水平。

产业园区运营负责人如何精准识别优质科技成果以提升招商成功率？.docx

科易网基于AI模型+行业大数据，在技术转移、成果转化、技术经纪、知识产权、产业创新、科技招商等垂直领域的多样化应用场景，推出科技创新领域的AI+工具与数智化解决方案，推动科技创新与产业创新智能化发展。

VCSEL阵列行业分析：如何在AI光互连与固态激光雷达之间重新定价.pdf

DuCut_1793670348591_917.mp4

YOLO26算法光伏太阳能板识别+训练好的模型+123张数据集+pyqt可视化界面.zip

附训练好的 YOLO 权重与 PyQt 可视化检测界面，页面底部含可视化效果预览，毕设/课设开箱即用。【数据集概况】 · 检测类别（中文）：[太阳能电池板(Solar panels)] 【训练曲线与评估图】【模型训练配置】参数 | 值模型 | yolo26n 训练轮数 | 100 epochs 输入尺寸 | 640x640 批次大小 | 24 优化器 | auto 初始学习率 | 0.01 训练设备【关键指标汇总】训练了 95 个 epoch，最终轮指标：指标 | 数值 mAP50 | **0.9287** mAP50-95 | 0.7723 Precision | 0.8463 Recall | 0.8504 train/box_loss | 0.6116 train/cls_loss | 0.6324 val/box_loss | 0.6968 val/cls_loss | 0.7057 【训练过程分析】我开始着手训练 yolo26n 算法，准备了123张数据集。训练初期，曲线走势让人既兴奋又焦虑。刚开始的20轮训练进展非常快，box_loss 从初始的0.75迅速下降至0.61，train/box_loss 值从0.81降至0.61，期间模型明显学习到了一些关键特征，但同时又有些不稳定。我注意到，虽然整体趋势向上，但在第15轮时出现了个小平台期，持续了大约5轮，让人感觉训练进展似乎停滞了。好在，经过20轮左右，box_loss 和 cls_loss 都稳定下来，进入了一个相对平缓的下降通道，整体收敛性良好。在第60轮后，loss 逐渐稳定在较低水平，说明模型逐渐掌握了更多细节特征。随着训练深入，我开始关注 mAP 的变化。mAP50 从第30轮开始逐渐提升，经过60轮后，终于在第8... 使用场景：毕业设计、课程设计、深度学习入门、工业场景落地验证。

V2G电动汽车ACDC整流+双向CLLC谐振变换器充放电仿真（Simulink仿真实现）

内容概要：本文介绍了一个基于Simulink的V2G（Vehicle-to-Grid）电动汽车充放电系统仿真模型，重点实现了AC-DC整流与双向CLLC谐振变换器的联合仿真。该系统涵盖交直流转换、谐振软开关技术、双向能量流动控制等核心环节，支持充电与放电模式的灵活切换，并引入闭环控制策略以提升系统的动态响应速度与运行稳定性。文档详细阐述了关键参数设计方法、仿真模型构建流程，并提供了参考文献与配套资料，适用于电力电子、新能源并网及智能电网领域的科研与教学实践。; 适合人群：电气工程、自动化、新能源等相关专业的研究生、科研人员及具备一定电力电子理论基础与Simulink仿真经验的工程技术人员。; 使用场景及目标：①开展V2G系统中双向能量变换拓扑结构的研究与性能验证；②掌握CLLC谐振变换器在高效率、软开关条件下的工作原理与控制策略；③辅助完成课程设计、科研项目申报或高水平论文复现；④深化对电动汽车与电网双向互动机制的理解，提升系统建模与仿真分析能力。; 阅读建议：建议结合文中提供的仿真模型与参考资料，逐步搭建系统并进行参数调试，重点关注控制环路设计、谐振元件匹配及闭环稳定性分析，同时利用Matlab/Simulink工具进行波形观测与性能评估，以实现理论与实践的深度融合，提升科研与工程应用能力。

Chrome WebDriver Chromedriver（谷歌浏览器Selenium驱动包118版本！！！）

微软常用运行库合集 2026.11.10.exe.zip

全桥LLC谐振变化器参数设计资料(闭环仿真+参数设计报告）

内容概要：本文档系统性地介绍了全桥LLC谐振变换器的参数设计与闭环控制技术，涵盖隔离型全桥LLC谐振变换器的建模、变频与移相混合控制策略、软启动过程、恒压恒流双闭环控制等核心技术环节，并提供了完整的仿真模型、设计报告与计算书。资料基于Simulink/Plecs平台实现系统级仿真，包含2.5kW双向全桥LLC调频闭环设计实例及配套PPT说明，具备高度的工程实践价值。此外，文档融合了状态估计、扩展卡尔曼滤波（EKF）与神经网络（如BP、EKF+BP、PF）等先进算法在轨迹估计与参数优化中的应用，体现了电力电子与智能控制多学科交叉的技术趋势。; 适合人群：从事电力电子变换器、新能源电源系统、车载充电机、数据中心供电等领域研发的工程师，以及高校电气工程、自动化等相关专业的研究生，需具备扎实的电路理论基础和一定的仿真软件操作能力。; 使用场景及目标：① 掌握全桥LLC谐振变换器的关键参数设计方法与闭环控制系统构建流程；② 复现并优化典型拓扑的Simulink/Plecs仿真模型，提升实际工程项目开发效率；③ 结合状态估计算法与智能控制策略，提升系统动态响应与稳定性，拓展在高精度电源管理与智能电网中的研究与应用。; 阅读建议：建议按照文档目录结构循序渐进学习，优先研读核心设计报告与仿真模型，结合Mathcad计算书进行参数推导，并动手搭建与调试Simulink模型以深化理解；同时可参考所提供的MATLAB代码与参考文献，进一步开展算法融合与创新性研究。

产业园区运营负责人需要哪些数智工具支撑企业技术对接？.docx

科易网基于AI模型+行业大数据，在技术转移、成果转化、技术经纪、知识产权、产业创新、科技招商等垂直领域的多样化应用场景，推出科技创新领域的AI+工具与数智化解决方案，推动科技创新与产业创新智能化发展。

科技中介服务机构如何借助平台降低技术交易成本提高服务质量？.docx

科易网基于AI模型+行业大数据，在技术转移、成果转化、技术经纪、知识产权、产业创新、科技招商等垂直领域的多样化应用场景，推出科技创新领域的AI+工具与数智化解决方案，推动科技创新与产业创新智能化发展。

目录