search for: cgsccpassmanager

...> inserting functions into the call graph using 'LazyCallGraph::get', > and then adding call edges with > 'LazyCallGraph::RefSCC::insertTrivialRefEdge' and > 'LazyCallGraph::RefSCC::switchInternalEdgeToCall', I'm met with an > assertion: llvm/lib/Analysis/CGSCCPassManager.cpp:463: [...]: > Assertion `E && "No function transformations should introduce *new* " > "call edges! Any new calls should be modeled as " "promoted existing > ref edges!"' failed. > > Does anyone have any suggestions on how I can better wor...

Hi Chandler, I am now working on a bisecting tool to find mis-optimization on LLVM. I found OptBisect a very useful option and hope to make it work on the new pass manager. I have several questions about it. 1. Any plans to apply codegen stage with new pass manager? IIUC, new pass manager only works for opt stage. However the OptBisect option tries to accumulate pass counts through opt stage

...and merging SCC's in the CG and > ref graph (which the LazyCallGraph API makes one to think about > since it hides the underlying Tarjan's algorithm). > The LazyCallGraph API makes the current loop in > http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100 > very clean, but at least for my thinking about the problem, it seems > like the wrong abstraction (and most of the LazyCallGraph API seems > to be unused, so it seems like it may be overly heavyweight). > E.g. I think that maybe the easiest thing to do is to turn the >...

...ms of splitting and merging SCC's in the CG and ref graph (which the LazyCallGraph API makes one to think about since it hides the underlying Tarjan's algorithm). > The LazyCallGraph API makes the current loop in http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100 <http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100> very clean, but at least for my thinking about the problem, it seems like the wrong abstraction (and most of the LazyCallGraph API seems to be unused, so it seems like it m...

...rge modules)? 5. In the old infrastructure, `CGPassManager::runOnModule()` (in `CallGraphSCCPass.cpp`) has a do-while loop that re-runs all passes on a given SCC as long something new gets devirtualized, up to a maximum number of iterations. I don't see the equivalent loop in CGSCCPassManager, and I don't see an obvious way to accomplish the same optimization cycle. - How will we re-implement this loop? (Or did I miss it?) - There's a potential infinite loop between devirtualization and inlining on recursive virtual functions. How will we avoid it? 6...

...n terms of splitting and merging SCC's in the CG and ref graph (which the LazyCallGraph API makes one to think about since it hides the underlying Tarjan's algorithm). The LazyCallGraph API makes the current loop in http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100 very clean, but at least for my thinking about the problem, it seems like the wrong abstraction (and most of the LazyCallGraph API seems to be unused, so it seems like it may be overly heavyweight). E.g. I think that maybe the easiest thing to do is to turn the current approach inside...

...litting and merging SCC's in the CG and ref graph (which the LazyCallGraph API makes one to think about since it hides the underlying Tarjan's algorithm). >>> The LazyCallGraph API makes the current loop in http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100 very clean, but at least for my thinking about the problem, it seems like the wrong abstraction (and most of the LazyCallGraph API seems to be unused, so it seems like it may be overly heavyweight). >>> E.g. I think that maybe the easiest thing to do is to turn the current app...

...'s in the CG and ref graph (which the >> LazyCallGraph API makes one to think about since it hides the underlying >> Tarjan's algorithm). >> The LazyCallGraph API makes the current loop in >> http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100 >> very clean, but at least for my thinking about the problem, it seems like >> the wrong abstraction (and most of the LazyCallGraph API seems to be >> unused, so it seems like it may be overly heavyweight). >> E.g. I think that maybe the easiest thing to do is...

...which the >>>> LazyCallGraph API makes one to think about since it hides the underlying >>>> Tarjan's algorithm). >>>> The LazyCallGraph API makes the current loop in >>>> http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100 >>>> very clean, but at least for my thinking about the problem, it seems like >>>> the wrong abstraction (and most of the LazyCallGraph API seems to be >>>> unused, so it seems like it may be overly heavyweight). >>>> E.g. I think that m...

...; LazyCallGraph API makes one to think about since it hides the > underlying > >>>> Tarjan's algorithm). > >>>> The LazyCallGraph API makes the current loop in > >>>> > http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100 > >>>> very clean, but at least for my thinking about the problem, it seems > like > >>>> the wrong abstraction (and most of the LazyCallGraph API seems to be > unused, > >>>> so it seems like it may be overly heavyweight). > >&g...

...he LazyCallGraph API makes one to think about since it hides the >>>>>> underlying Tarjan's algorithm). >>>>>> The LazyCallGraph API makes the current loop in >>>>>> http://reviews.llvm.org/diffusion/L/browse/llvm/trunk/include/llvm/Analysis/CGSCCPassManager.h;272124$100 >>>>>> very clean, but at least for my thinking about the problem, it seems like >>>>>> the wrong abstraction (and most of the LazyCallGraph API seems to be >>>>>> unused, so it seems like it may be overly heavyweight). >>>...

...elease+Asserts build llvm[3]: Compiling BypassSlowDivision.cpp for Release+Asserts build llvm[2]: Compiling CFLAliasAnalysis.cpp for Release+Asserts build llvm[3]: Compiling CloneFunction.cpp for Release+Asserts build llvm[3]: Compiling GCOVProfiling.cpp for Release+Asserts build llvm[2]: Compiling CGSCCPassManager.cpp for Release+Asserts build llvm[3]: Compiling CloneModule.cpp for Release+Asserts build llvm[3]: Compiling FastISel.cpp for Release+Asserts build llvm[2]: Compiling CaptureTracking.cpp for Release+Asserts build llvm[3]: Compiling CmpInstAnalysis.cpp for Release+Asserts build llvm[3]: Compiling C...