| Recent refactoring in changelist 66570 broke the heap mismatch detection feature. This resulted in printing an extra stack frame at the top of the stack that was internal to VLD, and not part of the users code. Please run all unit tests before checking in, otherwise quality will degrade. |
|
More...
|
about 13 years ago
|
| Minor code refactoring |
|
More...
|
about 13 years ago
|
| Fixing bug when dumping out the resolved callstack. It was cutting short a character every so often. |
|
More...
|
about 13 years ago
|
| _aligned... functions and _recalloc support added |
|
More...
|
about 13 years ago
|
| Adding Heap Validation to VLD. |
|
More...
|
about 13 years ago
|
| Infinite loop at exit fixed with some app's |
|
More...
|
about 13 years ago
|
| TestSuite bug fixed Basic test enhanced |
|
More...
|
about 13 years ago
|
| Minor fixes |
|
More...
|
about 13 years ago
|
| A little clean up of the unit tests. Added the already existing vldmfc application to the test folder in visual studio. Ran all the tests before submitting. testsuite.exe is still broken, and still throwing an assert during execution. |
|
More...
|
about 13 years ago
|
| Fixing up solution file to include the newly renamed mfc_dll project. |
|
More...
|
about 13 years ago
|
| Moving the last native console app test to the tests folder. |
|
More...
|
about 13 years ago
|
| Moving the two MFC test apps to the tests folder. |
|
More...
|
about 13 years ago
|
| Renaming the project to make it more explicit that this is a DLL project, not a console application. |
|
More...
|
about 13 years ago
|
| 45 seconds is too short a time to wait for 64 threads to finish in the unit test. Increasing the timeout value to Infinite for the multi-threaded test. |
|
More...
|
about 13 years ago
|
| Rare crash at exit on some platforms fixed |
|
More...
|
about 13 years ago
|
| assert(tls->blockprocessed) fixed when VLD disabled |
|
More...
|
about 13 years ago
|
| Modified project settings for consistency in testsuite.vcxproj. Also made the application return the difference of the expected and actual memory leaks. This should help in automated testing. Modified basics.cpp to return -1 if it had invalid arguments. This should make it consistent with the other case where it returns early. |
|
More...
|
about 13 years ago
|
| So the problem was that running Resolve twice on a CallStack pointer would erase the correct results, and provide the wrong module for the stack frame. Fixed by checking if the callstack has already been resolving before attempting to do it. If it has, then simply return early. This test in dynamic_app.exe that calls Resolve twice has been invaluable in shaking out bugs. |
|
More...
|
about 13 years ago
|
| Leaks count fixed with aggregate duplicates and resolve callstack |
|
More...
|
about 13 years ago
|
| Fixed an internal memory leak with VLD and resolved strings. Calling VisualLeakDetector::ResolveCallstacks multiple times would leak memory because CallStack::m_Resolved would just get overwritten. So the fix is to simply delete any existing memory that is there, before allocating the memory. |
|
More...
|
about 13 years ago
|
| Took out a redudant call to ReportLeaks. Also reset the leak count every time we call ReportLeaks. Also fixed some conversion compiler warnings about converting between an Unsigned Int and a size_t. Also fixed some other minor compiler warnings. |
|
More...
|
about 13 years ago
|
| Minor fix |
|
More...
|
about 13 years ago
|
| Restore import fixed Realloc callstack fixed |
|
More...
|
about 13 years ago
|
| CoTaskMemAlloc memory leak detection fixed VLDGetLeaksCount function added |
|
More...
|
about 13 years ago
|
| Problem: Uncomment out the code in RunMFCLoaderTests. Rerun the test dynamic_app.exe |
|
More...
|
about 13 years ago
|
| One more allocaiton mismatch, this time in the resolved strings for callstacks. replacing delete with delete[] |
|
More...
|
about 13 years ago
|
| Changing the memory free calls from delete to delete [] for the strings that each moduleinfo_t owns. This should eliminate some definite undefined behavior, as threading issues can arise when allocation mismatches occur. |
|
More...
|
about 13 years ago
|
| Resolved the internal memory leaks in LoadTests.cpp. What the problem was that the test unloaded the MFC Dll by calling FreeLibrary. The problem is that VLD hooks load library, but doesn't hook unload library. I don't know exactly what is going on here, but until I debug it, I"m turning off the code to unload the mfc dll. Now running dynamic_app in 64 bit mode, will report the correct 31 memory leaks with no extra warnings about internal memory leaks in VLD. |
|
More...
|
about 13 years ago
|
| Found a line of code that had been switched in VisualLeakDetector::IsModuleExcluded where a bad copy paste left address high and address low switched around. |
|
More...
|
about 13 years ago
|
| Reverting my last changes to VisualLeakDetector::IsModuleExcluded since the performance was abismal. |
|
More...
|
about 13 years ago
|
