similar to: [LLVMdev] opt and llc use datalayout differently - bug?

hey, so I'm writing in cpp. the documentation says that TheModule -> getDataLayout should return const DataLayout, but instead it is returning std::string. I require it to return DataLayout, as I generalize my function pass manager to accept the DataLayout constant as an argument, it being the only thing in common amongst both the ExecutionEngine and the Module class. Any pointers to

I’m debugging SAFECode source code files, and in AllocatorInfo.cpp, in the function getObjectSize(Value* V), the function getAnalysis<DataLayout>() is called. I’ve run into this problem before, and the first time I saw it, my employer recommended I use M.getDataLayout() instead, where M was a reference to a Module. However, in getObjectSize(Value* V), there is no Module reference in the

Hi everyone. I am using LLVM 2.4 release to practise the tutorial at Kaleidoscope Chapter 4. I built the LLVM library, and created my own project, and then paste the code from the tutorial into my own project. Then after compiled and built it, I passed the built stage. (I am using visual studio 2005 ) However, when I tried to input some code in the command line window, the sample program crashed.

On Sun, 21 Jan 2007, Marcel Weiher wrote: > just installed the new llvm 1.9 build and noticed that my code no > longer worked. It seems something has changed with > addPassesToEmitFile(). First, the arguments to that method changed so > that it no longer takes a PassManager, but only a > FunctionPassManager. Instead there is a addPassesToEmitWholeFile() > method, but that is

I may have found a bug in the andersends pass, but before I try to strip the code down to something reasonable I'd like to make sure I'm not misunderstanding anything. This is against the current SVN head, on my Mac OS 10.5 box. 1) I'm compiling a bunch of C++ code with llvm-g++ (4.0.1). The compiled code includes the gcc 4.0.1 implementation for dynamic_cast (so it can be

Jianzhou Zhao <jianzhou at seas.upenn.edu> writes: [snip] > The other question is about > target datalayout = > "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" > > If I use this string to creat a TargetData object directly, I got an > assertion error: > >

For those targets supported by LLVM, you can get their TargetData by creating TargetMachine first (take X86 as example): ==== BEGIN CODE SNIPPET ==== const std::string TripleStr = "i686-unknown-linux"; // hard coded for example const std::string FeatureStr = ""; // hard coded for example std::string Err; const Target* T; TargetMachine* TM = NULL; const

Hi Matthijs, Thanks for giving some code so we can discuss this in more concrete detail. In terms of the information we need, I think you have it right. We just need a description of how the different address spaces relate and I don't see much of an issue with how you implemented to InstructionCombining. As you also mentioned, I don't like that we pass a reference to

Hi All, I have a C code: ////////////////////////////// #include "stdio.h" int main () { putchar('a'); return 0; } llvm-gcc -emit-llvm, I got //////////////////////////////////////// ; ModuleID = 't1.bc' target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:32:32" target triple

Hello, So the current JIT will be superseded by the MCJIT completely and no further development should be expected for the current JIT? In any case, I am definitely interested in submitting a patch if you could help me by sending me in the right direction, since I really want this working. I managed to reproduce this behavior in LLVM 3.0 by modifying llc to read my .bc file and try to JIT the

I have a problem where I add an Andersens AA pass to the pass manager, but it appears to get invalidated by another pass, and never rerun. My understanding from reading the documentation is that when a pass gets invalidated, it should be rerun before any other passes that requires it. Here is a simple example of the problem I am seeing: PassManager passManager; passManager.add(new

Hi all, > If I read the standard correctly, the properties of these address spaces can > be fully captured by defining the relationship between every pair of address > spaces (disjoint, identical, subset/superset). > > I think it would make sense to make these relationships backend/platform > specific, but for clang and the optimization passes to properly work with > address

On Wednesday 18 February 2009 21:43, Dan Gohman wrote: > I agree, that doesn't look right. It looks like this > is what was intended: > > Index: lib/CodeGen/SelectionDAG/DAGCombiner.cpp > =================================================================== > --- lib/CodeGen/SelectionDAG/DAGCombiner.cpp (revision 65000) > +++ lib/CodeGen/SelectionDAG/DAGCombiner.cpp

On Feb 20, 2009, at 3:05 PM, David Greene wrote: > On Wednesday 18 February 2009 21:43, Dan Gohman wrote: >> I agree, that doesn't look right. It looks like this >> is what was intended: >> >> Index: lib/CodeGen/SelectionDAG/DAGCombiner.cpp >> =================================================================== >> ---

I just tried cmake for the first time and it mostly worked. I had been using autotools to configure. With cmake configure environment when I try to do: make TESTSUITE=CodeGen/mblaze check it runs the whole test suite. Under autotools it would just run the mblaze tests. Is there an equivalent command in a camke configured environment or is this broken? I know nothing about cmake so I'm

Hi folks, just installed the new llvm 1.9 build and noticed that my code no longer worked. It seems something has changed with addPassesToEmitFile(). First, the arguments to that method changed so that it no longer takes a PassManager, but only a FunctionPassManager. Instead there is a addPassesToEmitWholeFile() method, but that is marked as optional, and when I change my code to

I got bit by this in LLVM 2.4 DagCombiner.cpp and it's still in trunk: SDValue DAGCombiner::visitSTORE(SDNode *N) { [...] // If this is a store of a bit convert, store the input value if the // resultant store does not need a higher alignment than the original. if (Value.getOpcode() == ISD::BIT_CONVERT && !ST->isTruncatingStore() && ST->isUnindexed()) {

Hello, We are happy to release version 0.10.0 of llvmlite. llvmlite is a light-weight Python binding for LLVM (compatible with Python 2.7 as well as Python 3.4 and later). It stems from the needs of the Numba community and is geared towards creating JIT compilers. llvmlite is available both as source code and as binaries for a number of platforms. Source code is available on PyPI and Github:

To figure out the size of data varaibles for my backend, I tries this code stolen from X86 backend: const TargetData& TD = m_tm.getTargetData(); unsigned size = TD.getTypeSize(initializer->getType()); However, the getTypeInfo function in TargetData.cpp contains this: case Type::VoidTyID: case Type::BoolTyID: case Type::UByteTyID: case Type::SByteTyID: Size = 1;

Hi folks, just installed the new llvm 1.9 build and noticed that my code no longer worked. It seems something has changed with addPassesToEmitFile(). First, the arguments to that method changed so that it no longer takes a PassManager, but only a FunctionPassManager. Instead there is a addPassesToEmitWholeFile() method, but that is marked as optional, and when I change my code to