similar to: LLVM module attributes question

Oops, missed initializing some stuff. Added: LLVMLinkInMCJIT(); LLVMInitializeNativeTarget(); LLVMInitializeNativeAsmPrinter(); LLVMInitializeNativeAsmParser(); Now it crashes in LLVMGetFunctionAddress(). Hmm. On Wed, Mar 8, 2017 at 5:14 PM, Toshiyasu Morita <toshi at tensyr.com> wrote: > Made it a bit further. Here's the current code: > >

Or do you mean I need to load the module into memory before calling LLVMGetBitcodeModuleInContext2? > Yes, you need to load the module into memory first. > LLVMCreateMemoryBufferWithContentsOfFile will do that for you. Thanks! On Wed, Mar 8, 2017 at 3:48 PM, Friedman, Eli <efriedma at codeaurora.org> wrote: > On 3/8/2017 3:44 PM, Toshiyasu Morita wrote: > > >>

Thanks for the tip - getting closer: $ ./capi_test 5 6 args[0]: 5 args[1]: 6 result: 4294959200 Here's the code I changed: printf("args[0]: %d\n", (int)LLVMGenericValueToInt(args[0], 0)); printf("args[1]: %d\n", (int)LLVMGenericValueToInt(args[1], 0)); uint64_t (*func)(); func = (uint64_t (*)())LLVMGetFunctionAddress(engine, "sum");

Hi, I have a problem - looking for advice. I have a source code file with two functions which are compiled into a .bc file. When the bitcode file is loaded, I can dump the module and see the two functions: ... ; Materializable ; Function Attrs: noinline nounwind uwtable define void @matmul(double*, double*, double*, i32, i32, i32) #0 {} ; Materializable ; Function Attrs: noinline nounwind

I'm trying to use LLVMGetBitcodeModuleInContext2 to load a .bc file. However, it's not working. The code looks something like this: void llvm_load_IR_library(char *path) { LLVMContextRef global_context; LLVMMemoryBufferRef module_path; LLVMModuleRef ir_lib_module; bool flag; module_path = LLVMCreateMemoryBufferWithMemoryRange(path, strlen(path),

Hi, I'm attempting to use MCJIT with the C interface with llvm-3.9.0, and it doesn't seem to work. This code is derived from Paul Smith's example code: int main(int argc, char const* argv[]) { LLVMModuleRef mod = LLVMModuleCreateWithName("my_module"); LLVMTypeRef param_types[] = {LLVMInt32Type(), LLVMInt32Type()}; LLVMTypeRef ret_type =

> > > module_path = LLVMCreateMemoryBufferWithMemoryRange(path, > strlen(path), "path", 1); > LLVMCreateMemoryBufferWithContentsOfFile takes a path. Erm...no...the code is calling LLVMCreateMemoryBufferWithMemoryRange, not LLVMCreateMemoryBufferWithContentsOfFile... Or do you mean I need to load the module into memory before calling

Hi, We couldn't delete any files and directories when connecting glusterfs to windows via NFS. We could make/change files and directories. Although it has written that glusterfs can be connected with windows NFS client in the page shown below. Any ideas to workaround/fix this ?

Adding the flags in an opt pass seems to do the trick. I added the flags as follows M.addModuleFlag(llvm::Module::Warning, "Dwarf Version", 3); M.addModuleFlag(llvm::Module::Error, "Debug Info Version", llvm::DEBUG_METADATA_VERSION); llvm-dwarfdump and addr2line now point at a line in .ll files, not the source files. Thanks! /Muneeb On 15 Oct

On Wed, Oct 15, 2014 at 9:04 AM, Adrian Prantl <aprantl at apple.com> wrote: > I have no experience with debug-ir, but is it possible that the debug-ir > pass never sets the debug info version? There should be a named metadata > node > !llvm.module.flags = !{..., !123} > !123 = metadata !{i32 1, metadata !"Debug Info Version", i32 2} > in the .ll output. If

Hi, I wanted to know if it was possible to force ARM backend to compile a function in ARM while the rest is in Thumb mode. I tried the attributes which is used in GCC but it doesn't work. Here is what I tried: https://pastebin.com/jCr5LPUY Thanks in advance, Uvekilledkenny -------------- next part -------------- An HTML attachment was scrubbed... URL:

Hi Anton > > > The interesting for me part of the CallInst is printf(i8* noalias %0, i32 > > 123). > > I was diging in doxygen documentation but I really can't see the easy way > to > > compare those instructions and again finish with reinvented (but working) > > wheel ;). > Ah, sorry. I missed that you're doing variadic calls, not casting >

Hi, It looks like SimplifyLibCalls has a tendency to emit calls to libm functions without checking with TLI whether these calls are available. For example, PowOpt has this code: struct PowOpt : public UnsafeFPLibCallOptimization { PowOpt(bool UnsafeFPShrink) : UnsafeFPLibCallOptimization(UnsafeFPShrink) {} virtual Value *callOptimizer(Function *Callee, CallInst *CI, IRBuilder<> &B)

As we encode more CodeGen and target-specific options in bitcode to support LTO, we risk crippling `llc` as a debugging tool. In particular, `llc` command-line options are generally ignored when a function has an attribute set explicitly, but the plan of record is for `clang` to explicitly encode all (or most) CodeGen options -- even the target defaults. Changing `clang` to store target defaults

Hello All, I am in need of dragonegg support for the latest version of llvm. I tried to make dragonegg with the latest versions but it fails. It is able to be maked with the llvm version 3.3 only with GCC 4.6. I am trying this on my ubuntu system. I tried to make dragonegg with various combinations of llvm and GCC versions but the only combination which seems to be working is LLVM-3.3, GCC-4.6

Hi All, While reading LLVM source code, sometimes I am wondering when should we use auto instead of iterator/const_iterator. I want to use the patch [1] I sent before as an example. Could someone give me advice/guideline here? Also, I have another question. Sometimes the for-loop uses const_iterator, say for (SUnit::const_succ_iterator I = SU->Succs.begin(), E = SU->Succs.end();

Hi, Artur > // CallSites have equal signatures > bool MSILWriter::cmpCallSite(CallSite A, CallSite B) { >   return (getCallSiteFType(A)==getCallSiteFType(B) && >     A.getAttributes()==B.getAttributes()); > } As it is impossible to honour argument attributes in MSIL I don't see why you should compare attributes. You seems to have the same MSIL call signature for calls

Hi Duncan, Been thinking about this a bit and a few comments/questions. I may have misunderstood some things in your mail though so please feel free to explain at me :) > Changing `clang` to store target defaults on the module will allow us to > continue to override them when running `llc`. The right precedence > would be: > > 1. Explicit attributes set on the function. >

Hi, We have been investigating an issue when running LTO with our proprietary linker, which links against libLTO dynamically. The issue is that when we pass -time-passes via the lto_codegen_debug_options function in the LTO C API, no time information is produced during compilation. The reason for this is that time information is stored in state owned by a ManagedStatic instance, and is only

Is it possible to use different port numbers for UDP and TCP? I'd like to open the TCP connection to one port on the remote server and stream the UDP packets to a different port. I've tried specifying both as BindToAddress and Address lines but it always just uses TCP. On 2 May 2017 17:36, "Guus Sliepen" <guus at tinc-vpn.org> wrote: On Tue, May 02, 2017 at 09:53:15PM