similar to: [LLVMdev] statically link pass to clang

2011/9/27 Zeng Bin <ezengbin at gmail.com>: > > > On Sun, Sep 25, 2011 at 8:45 PM, NAKAMURA Takumi <geek4civic at gmail.com> > wrote: >> >> 2011/9/24 Zeng Bin <ezengbin at gmail.com>: >> > A question about statically linking a pass to clang instead of >> > dynamically >> > loading it at runtime. I am stumped on this bug for a

I have seen INITIALIZE_PASS(LiveVariables, "livevars", "Live Variable Analysis", false, false); in the llvm/lib/codegen/LiveVariables.cpp, where LiveVariables is a subclass of MachineFunctionPass, and #define INITIALIZE_PASS(passName, arg, name, cfg, analysis) \ static RegisterPass<passName> passName ## _info(arg, name, cfg, analysis) in PassSupport.h Is this code used

Thanks for your last reply. Could I understand the way to adding a pass (built into the llvm rather than dynamic loadable) includes: 1. Declaring a creator function for this pass 2. Implementing the creator function for this pass 3. Instantiating this pass and get a object of it 3. Register this pass into the PassRegistry Then, for a built-into bytecode pass, task 1(declaration of the

Hi, I am trying to create a pass that should get statically linked into opt (since I'm on both mac and native win32 and win32 does not support LOADED_MODULEs, at least I get errors while trying to create one). So I have it working with CMake on windows but when I try to run the same code base on mac I get unresolved externals. Here is what I have done so far in the build system: Within

On Thu, Oct 6, 2011 at 2:20 PM, Eli Friedman <eli.friedman at gmail.com> wrote: > On Thu, Oct 6, 2011 at 2:12 PM, Zeng Bin <ezengbin at gmail.com> wrote: >> Hi all, >> >> There might be a bug in DeadStoreElimination.cpp. This pass eliminates >> stores backwards aggressively in an end BB. It does not check dependencies >> on stores in an end BB though.

If int_guard_load returns a pointer based on the passed-in pointer, it captures it (at least according to the definition of "capture" which NoCapture uses). -Eli On Thu, Oct 6, 2011 at 2:26 PM, Zeng Bin <ezengbin at gmail.com> wrote: > It does not do anything. It is an abstract function which transforms a > pointer and returns another pointer of the same type. It does not

Hi all, I am really stumped on this problem. I googled and could not find any solution. Here is the problem. I compiled and installed glibc-2.5-20090518 in a local directory with gcc version 4.1.2. I compiled glibc-2.5 because I wanted to test glibc-2.5 on gcc first llvm-gcc. However, after I installed glibc, llvm-gcc does not work any more. It reported this error message:

It does not do anything. It is an abstract function which transforms a pointer and returns another pointer of the same type. It does not visit memory or capture the pointer parameter. On Thu, Oct 6, 2011 at 2:22 PM, Eli Friedman <eli.friedman at gmail.com> wrote: > On Thu, Oct 6, 2011 at 2:20 PM, Eli Friedman <eli.friedman at gmail.com> > wrote: > > On Thu, Oct 6, 2011 at

Thanks John! My pass is registered only in 1 cpp file. It registers and executes fine without the EE. I only get the Pass registered multiple times error when I try to link to the EE by using LINK_COMPONENTS=engine in the Makefile. Without that line in the Makefile, my pass gets registered fine but it is not able to create EE(Error reading Symbol). So, how do I link LLVMExecutionEngine.o without

Hi all, I am repeating my question from yesterday coz I need to find a solution to this ASAP. How do I link the executionengine to an optimization pass. If I use LINK_COMPONENTS=engine in the Makefile, I get the Pass registered multiple times error. And if i dont use anything, it is not able to load the EE. It gives error loading symbol error. So, 1. Is it possible to use the EE within an

I've been puzzling over this for a day: Unable to schedule 'Unnamed pass: implement Pass::getPassName()' required by '<some pass>' Unable to schedule pass UNREACHABLE executed at llvm/lib/VMCore/PassManager.cpp:1213! First, why can't getPassName find the name? I have provided one in the usual way. Second, why is it unable to schedule the pass? -debug-pass=Details

bhavani krishnan wrote: > Thanks John! My pass is registered only in 1 cpp file. It registers and executes fine without the EE. > > I only get the Pass registered multiple times error when I try to link to the EE by using LINK_COMPONENTS=engine in the Makefile. Without that line in the Makefile, my pass gets registered fine but it is not able to create EE(Error reading Symbol). So, how do

I realize my questions may not have been so clear. To give more details. I am writing a fuction pass. I want to interpret some of the instructions within the function pass. In order to do so, I need to create an EE object within the function pass. Now, If I compile and execute using opt. I get an error(Error loading symbol) at the line where I am creating EE. opt: symbol lookup error:

Hi all, There might be a bug in DeadStoreElimination.cpp. This pass eliminates stores backwards aggressively in an end BB. It does not check dependencies on stores in an end BB though. For example, in this code snippet: ... 1. %sum.safe_r47.pre-phi = phi i64* [ %sum.safe_r47.pre, %entry.for.end_crit_edge ], [ %sum.safe_r42, %for.body ] 2. %call9 = call i32 @gettimeofday(%struct.timeval* %end,

bhavani krishnan wrote: > Hi all, > > I am repeating my question from yesterday coz I need to find a solution to this ASAP. > > How do I link the executionengine to an optimization pass. If I use LINK_COMPONENTS=engine in the Makefile, I get the Pass registered multiple times error. And if i dont use anything, it is not able to load the EE. It gives error loading symbol error. So,

bhavani krishnan wrote: > I realize my questions may not have been so clear. To give more details. I am writing a fuction pass. I want to interpret some of the instructions within the function pass. In order to do so, I need to create an EE object within the function pass. Now, If I compile and execute using opt. I get an error(Error loading symbol) at the line where I am creating EE. >

Hello folks, I wrote a transforming looppass for my project, opt --help can see it. I want clang to use my pass too, so that I don't have to compile to IR, opt, and then llc to binary manually. Now I am trying to add my pass to O2 by mimicking LICM pass. I add a line "MPM.add(createMyPass());" in PassManagerBuilder::populateModulePassManager, and got assertion fail "void

On Thu, Oct 6, 2011 at 2:12 PM, Zeng Bin <ezengbin at gmail.com> wrote: > Hi all, > > There might be a bug in DeadStoreElimination.cpp. This pass eliminates > stores backwards aggressively in an end BB. It does not check dependencies > on stores in an end BB though. For example, in this code snippet: >   ... > 1.  %sum.safe_r47.pre-phi = phi i64* [ %sum.safe_r47.pre,

Hi all, I am really stumped on this problem. I googled and could not find any solution. Here is the problem. I compiled and installed glibc-2.5-20090518 in a local directory with gcc version 4.1.2. I compiled glibc-2.5 because I wanted to test glibc-2.5 on gcc first llvm-gcc. However, after I installed glibc, llvm-gcc does not work any more. It reported this error message:

Hi all, Just a quick question about X86 instructions inside LLVM. I found it hard to do one-to-one mapping from the instructions in X86GenInstInfo.inc to the instructions in the intel Manual. For example, what instructions are DIVR_F32m and DIVR_F64m mapped to in the intel manual? It would be great if there is some docs about the X86 instructions. Thanks tons! Bin