By the way, I'm actively working on getting the test suite completely running on a buildbot, but I need to get down to a green baseline first. Still a few weeks out, but hopefully this type of thing will be caught automatically in the future.
On Fri, Aug 21, 2015 at 5:00 PM Zachary Turner <ztur...@google.com> wrote: > Hi Stephane, I'm sorry but I had to revert this change. When I tested it > locally, I did it without running the full test suite, but when I did run > the full test suite it showed about 15-20 new test failures. I should have > caught this earlier by running the full test suite, so sorry about that. > > I'm not 100% sure what the cause of the failure is, but I suspect it could > be related to DynamicLoaderWindows adding modules to the list *and* > ProcessWindows adding modules to the list as well. > > If you need help setting up a Windows machine to test this on let me > know. Unfortunately it's not that trivial, but luckily once you get it set > up and correctly configured it becomes pretty easy to keep it going > smoothly. > > Let me know if you need more information. > > On Fri, Aug 21, 2015 at 12:29 PM Stephane Sezer <s...@cd80.net> wrote: > >> This revision was automatically updated to reflect the committed changes. >> Closed by commit rL245725: Implement basic DidAttach and DidLaunch for >> DynamicLoaderWindowsDYLD. (authored by sas). >> >> Changed prior to commit: >> http://reviews.llvm.org/D12245?vs=32842&id=32853#toc >> >> Repository: >> rL LLVM >> >> http://reviews.llvm.org/D12245 >> >> Files: >> >> lldb/trunk/source/Plugins/DynamicLoader/Windows-DYLD/DynamicLoaderWindowsDYLD.cpp >> >> Index: >> lldb/trunk/source/Plugins/DynamicLoader/Windows-DYLD/DynamicLoaderWindowsDYLD.cpp >> =================================================================== >> --- >> lldb/trunk/source/Plugins/DynamicLoader/Windows-DYLD/DynamicLoaderWindowsDYLD.cpp >> +++ >> lldb/trunk/source/Plugins/DynamicLoader/Windows-DYLD/DynamicLoaderWindowsDYLD.cpp >> @@ -9,6 +9,7 @@ >> >> #include "DynamicLoaderWindowsDYLD.h" >> >> +#include "lldb/Core/Log.h" >> #include "lldb/Core/PluginManager.h" >> #include "lldb/Target/Process.h" >> #include "lldb/Target/Target.h" >> @@ -72,11 +73,38 @@ >> void >> DynamicLoaderWindowsDYLD::DidAttach() >> { >> + Log *log(GetLogIfAnyCategoriesSet(LIBLLDB_LOG_DYNAMIC_LOADER)); >> + if (log) >> + log->Printf("DynamicLoaderWindowsDYLD::%s()", __FUNCTION__); >> + >> + DidLaunch(); >> + >> + m_process->LoadModules(); >> } >> >> void >> DynamicLoaderWindowsDYLD::DidLaunch() >> { >> + Log *log(GetLogIfAnyCategoriesSet(LIBLLDB_LOG_DYNAMIC_LOADER)); >> + if (log) >> + log->Printf("DynamicLoaderWindowsDYLD::%s()", __FUNCTION__); >> + >> + ModuleSP executable = GetTargetExecutable(); >> + >> + if (!executable.get()) >> + return; >> + >> + ModuleList module_list; >> + module_list.Append(executable); >> + // FIXME: We probably should not always use 0 as the load address >> + // here. Testing showed that when debugging a process that we start >> + // ourselves, there's no randomization of the load address of the >> + // main module, therefore an offset of 0 will be valid. >> + // If we attach to an already running process, this is probably >> + // going to be wrong and we'll have to get the load address somehow. >> + UpdateLoadedSections(executable, LLDB_INVALID_ADDRESS, 0); >> + >> + m_process->GetTarget().ModulesDidLoad(module_list); >> } >> >> Error >> >> >>
_______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
