similar to: [LLVMdev] Hello example plugin has unresolved symbol in llvm 2.7

Hello, I just built a virgin ToT (r98634) for release on vanilla Snow Leopard. It seems that the Hello pass doesn't want to load because of undefined symbols: builddir% ../llvm/configure --prefix=$(PWD)/../installdir --enable-optimized builddir% make builddir% make install && cd ../installdir installdir% bin/opt -load lib/libLLVMHello.dylib Error opening

Hi all, I recently experienced the same issue as below with LLVM 2.8 on Mac OS 10.5.8. I can load the pass with the debug version of opt, but not the optimized version. Does anyone know what the problem is or have any suggestions for debugging this? My install went fine except for some failures during make check (Unexpected Failures: 92). All failures were in one of the following:

While learning to write LLVM passes and following the precise instructions under http://llvm.org/docs/WritingAnLLVMPass.html, <http://llvm.org/docs/WritingAnLLVMPass.html> I got this error when loading the hello pass to run the test program: opt -load ./Release/lib/Hello.so -hello < test/test.bc > /dev/null Error opening './Release/lib/Hello.so': ./Release/lib/Hello.so:

Hey Chuck, I'm afraid I can't reproduce your error but...a problem you may run into later is that opt will complain with opt: llvm/lib/VMCore/Pass.cpp:149: void<unnamed>::PassRegistrar::RegisterPass(const llvm::PassInfo&): Assertion `Inserted && "Pass registered multiple times!"' failed. Aborted I "fixed" this by replacing the LLVMLIBS line in

Thanks, Shu, I guess I haven't updated since my post went out. There are actually 2 problems: 1. mis-compilation: My LLVM-2.5 turned out to be mis-compiled using gcc-4.4.0 (surprise to me) on Debian4-32b. I tried a few different compilers, and gcc-4.0.4 (a relatively old one, again surprised me) seems to work out fine. Question: is there a good/quick/reliable way to figure out whether a

Thanks for your response. In your experience, was it worth the conversion hassle and code expansion to have the code in a platform-independent SSA form and be able to apply LLVM-based analysis tools? Or is it better to convert to another IR (such as BAP[1]) and work in that form? Bob [1] Binary Analysis Platform, http://bap.ece.cmu.edu/ --- On Thu, 6/10/10, Nathan Jeffords <blunted2night at

I'm wondering if anyone is working on a machine code -> LLVM bitcode disassembler? Obviously, there won't be a one-to-one correspondence but it seems like you should be able to get close. There's always inline asm for the remaining fragments. So is there such a thing? Thank you, Bob

Hi!! It's me again. I guess that there is a problem in the Makefile of my source code. But I don't know where. Could you please verify it? The Makefile is written like this: # Makefile for Genetic Algorithm Pass CXXFLAGS = -Wall -I/iss/fpga3/nicole/galib247/ LDFLAGS= -L/iss/fpga3/nicole/galib247/ga/ -lga -Wl,-E # Path to top level of LLVM heirarchy LEVEL=../../../ # Name of the

Hi I have written a pass which creates some scheduling information. Its called MParSchedule. This pass works with opt. But when feeding it my experimental backend it seems to loose instructions and fails to work. I am loading this stuff as .so libraries under linux and i am using the amd64 version with a self compiled llvm-gcc. With this call it works: opt

Hi all I'm writing a pass for the llc tool. If I try to load my pass(llc -load ../mypass.so), this error message appears : Error opening '/nfs/wsi/ti/graumann/llvm-2.5/Release/lib/LLVMMicha.so': /nfs/wsi/ti/graumann/llvm-2.5/Release/lib/LLVMMicha.so: undefined symbol: _ZTVN12_GLOBAL__N_119MyRegisterAllocatorE -load request ignored. Is this code wrong for the pass

> Question: is there a good/quick/reliable way to figure out whether a > certain gcc compiler will mis-compile? http://llvm.org/docs/GettingStarted.html#brokengcc lists documented problems with different versions of gcc. I use gcc-4.3.2 with ubuntu 8.10 and haven't had any problems > 2. I ran into exactly the problem you pointed out. (Thank you) > The Makefile needs some update

Hi, My pass compiles fine with Gmake,but when i run it with 'opt' it was initially breaking with the following error message : "Error opening '../../../Release/lib/ReusePass.so': ../../../Release/lib/ReusePass.so: undefined symbol: _ZNK4llvm4Pass5printERSoPKNS_6ModuleE -load request ignored. opt: Unknown command line argument '-ReusePass'." Then i

Hi, I'm writing my first hello world Pass with the class name First, but when I tried to load it using opt, I got the following error: /var/soft/llvm-2.2-build/lib/Transforms/Hello$ opt -load ../../../Debug/lib/First.so --help opt: /var/soft/llvm-2.2/lib/VMCore/Pass.cpp:157: void<unnamed>::PassRegistrar::RegisterPass(llvm::PassInfo&): Assertion `Inserted && "Pass

It turned out to be the problem of this line in Makefile LLVMLIBS = LLVMCore.a LLVMSupport.a LLVMSystem.a The error is gone when I remove the above line. However, the pass output only contains one line of the main function. First: main It does not have puts and __main as shown in this page: http://llvm.org/docs/WritingAnLLVMPass.html#running So I guess the drived FunctionPass only works on

Hello LLVM, I am following the document "Writing an LLVM Pass". When I ran "opt -load ../../../Debug/lib/MyPass.so -mypass < hello.bc > /dev/null" I got the next error: ***@ubuntu:~/test$ opt -load ../../llvm/Debug/lib/MyPass.so -mypass < hello.bc > /dev/null opt: Pass.cpp:159: void<unnamed>::PassRegistrar::RegisterPass(const llvm::PassInfo&): Assertion

Juan Carlos Martinez Santos wrote: > Hello LLVM, > > I am following the document "Writing an LLVM Pass". When I ran "opt > -load ../../../Debug/lib/MyPass.so -mypass < hello.bc > /dev/null" I > got the next error: > > ***@ubuntu:~/test$ opt -load ../../llvm/Debug/lib/MyPass.so -mypass < > hello.bc > /dev/null > opt: Pass.cpp:159:

I am using llvm 1.8 and have patched ltdl.c and ltdl.h to the latest version. My Makefile has LLVMLIBS = LLVMCore.a LLVMSupport.a LLVMSystem.a included. > Try using 'nm -m' on the 'opt' executable itself. Using plain nm on my > local one yields: > 00121a4e T > __ZN4llvm12FunctionPass16addToPassManagerEPNS_20FunctionPassManagerTERNS_13AnalysisUsageE > > and nm

And this brings back CommandLine Error: Arugment blah defined more than once! And without this, opt -load .../LLVMHello.dylib -hello does not work on Darwin because dyld is not able to find SlowerOperationInformer ;) thoughts ? - Devang On Apr 13, 2007, at 11:28 AM, Devang Patel wrote: > > > Changes in directory llvm/lib/Transforms/Hello: > > Makefile updated: 1.7 ->

Hi Tim, Dear List, Am 18.04.2014 um 10:08 schrieb Tim Northover <t.p.northover at gmail.com>: > One common problem (the only one I really know about, unfortunately) > is trying to load a module into Xcode's copy of "opt" rather than one > from the same tree as your module. I just double-checked that I use the opt executable from the same tree I use to build the

On Fri, 13 Apr 2007, Devang Patel wrote: > And this brings back CommandLine Error: Arugment blah defined more > than once! > > And without this, opt -load .../LLVMHello.dylib -hello does not work > on Darwin because dyld is not able to find SlowerOperationInformer ;) I think libhello should drop its use of SlowOperationInformer. -Chris > On Apr 13, 2007, at 11:28 AM, Devang