similar to: [LLVMdev] cross-compiling current trunk fails

Dear LLVM contributors, I am Naoki Shibata, an associate professor at Nara Institute of Science and Technology. I and Hal Finkel would like to jointly propose to add my vectorized math library to LLVM. The library has been available as public domain software for years, I am going to double-license the library if necessary. ******** Below is a proposal to add my vectorized math library,

Hi, I'm trying to add function calls in the LLVM IR using the IRBuilder class. I've read both tutorials about functions but I still get assertion errors. I allocate memory for all function arguments and pass them as explained in the tutorial. My code is (this function is supposed to add a call to f in bb at pos): void addCallSite(Function *f, BasicBlock *bb, BasicBlock::iterator pos)

Hi Rantao, perhaps you should use $llvmpath/utils/profile.pl for profiling? It works well. If you don't want to use it, there is at least some code in the script that should help you. BTW: I'm not sure if it is possible to instrument the bitcode with function and basic block code at the same time. But it should work sequential (data in llvmprof.out are accumulated after each run

CreateAlloca(Type) returns an object of type Type*, the memory that can hold an object of type Type. You probably don't want to be creating allocas just before calling the function since 1) if that call winds up in a loop they'll grow your stack frame without bound, and 2) the memory they point to is initially uninitialized. Where did the tutorial tell you to do that? In general, when I

Hello, My name is Rantao Chen and i am a student at the University of Maryland. I am currently doing some research for a professor to do Basic Block profiling using LLVM. I have set up the environment correctly and built the profile_rt.so shared library, but i am having some problems using the LLVM profiling tools. I run "opt -insert-function-profiling -insert-block-profiling input.bc

Hello LLVMdev!! Yesterday I posted a patch request to the llvm-commits list requesting that someone could apply a patch to Triple.h and Triple.cpp for me. I didn't get any response so I wondered whether I should have posted to this list instead. My story is as follows: we are trying to get lldb/llvm support for CSRs range of Kalimba DSPs. Eventually we are planning to hire someone to

Hi! When compiling the attached module with the JIT engine on an Intel KNL I see wrong code getting emitted. I attach a complete exploit program which shows the bug in LLVM 3.8. It loads and JIT compiles the module and prints the assembler. I stumbled on this since the result of an actual calculation was wrong. So, it's not only the text version of the assembler also the machine

Hi Frank, > We are looking at a variety of target architectures. Ultimately we aim to run on BG/Q and Intel Xeon Phi (native). However, running on those architectures with the LLVM technology is planned in some future. As a first step we would target vanilla x86 with SSE/AVX 128/256 as a proof-of-concept. Great! It should be easy to support these targets. When you said wide-vectors I assumed

Dear Customer! This is an automatic response from our support system, because you have sent a message to our support system e-mail address, which does not refer to an existing support case tracking number. We regret to inform you that we can no longer process such messages and your e-mail has, therefore, NOT been delivered. We have recently implemented a new support case management system in

What would be needed in order to make MCJIT work on a new architecture? I am thinking BG/Q and Xeon Phi (native). Let's assume the components required for JIT (core, mcjit, native, etc.) can be cross-compiled and linked with an Intel or IBM compiler for such an architecture, and somehow one manages to execute the application. (I didn't try all that yet.) Also, let's assume there is

first: thank you for helping out >I might try compiling your source from the VS's Developer Command Prompt: >cmake --build . --target Fibonacci >Does this work? call "C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\VC\Auxiliary\Build\vcvars64.bat" cd llvm-build cmake --build . --config Debug --target Fibonacci building Fibonacci example works BUT

Hi, I'd like to experiment with the MIPS and PowerPC backends, but, considering that they aren't widely used processors, I'd like to start with the same environment (OS/ABI/linker) used by the people who work with these backends. So, what OS/ABI/linker use the people who use these backends for production work? Thanks!!

Hi! Sorry for delay, just returned from vacation. On 12/11/15 23:44, Kostya Serebryany via llvm-dev wrote: > +Maxim and Yuri, as I think this is their code. > > On Thu, Nov 12, 2015 at 3:02 AM, Jay Foad <jay.foad at gmail.com > <mailto:jay.foad at gmail.com>> wrote: > > (Resending with the correct mailing list address.) > > Hi, > > Currently

(Resending with the correct mailing list address.) Hi, Currently test/asan/TestCases/alloca_vla_interact.cc is XFAILed for powerpc64. I've had a look at why it doesn't work. I think the only problem is in the call to __asan_allocas_unpoison that is inserted at the end of the "for" loop (just before a stackrestore instruction). The call function is created something like this

Jay, do you have a PowerPC64 target? If so, could you please check attached patch on PPC box? This is a draft patch, but it would be nice to make sure that we are moving to right direction here. Thanks, -Maxim On 18/11/15 00:12, Jay Foad wrote: >>> Currently test/asan/TestCases/alloca_vla_interact.cc is XFAILed for >>> powerpc64. I've had a look at why it

Hi there! The attached patch fixes a few places where 32bit register classes or instructions are used on PowerPC64. This lets me compile a working "vim" at -O0 on PowerPC64/FreeBSD. Previously it would crash the compiler with llvm_unreachable("Impossible reg-to-reg copy"); As it was mixing 64 bit registers with 32bit ones. Vim compiled with -O2 still crashes. OK

In LowerGET_DYNAMIC_AREA_OFFSET() you're calling MFI->getMaxCallFrameSize(), but it looks like that doesn't return useful information until after the PrologEpilogInserter's PEI::calculateCallsInformation() has run. So maybe the lowering has to be done as part of frame index elimination? (I'm not too familiar with this code.) Jay. On 23 November 2015 at 13:07, Jay Foad

I updated one of the powerpc64 lnt buildbot machines to a new OS release and the buildbot started failing every run when it tries to check out the test suite. The checkout will run for a while then always fails like this: /usr/bin/svn checkout --non-interactive --no-auth-cache --revision 323463 http://llvm.org/svn/llvm-project/test-suite/trunk at 323463 test/test-suite in dir

Hi! I always have to specify the flag -mminimize-toc when compiling on a PowerPC64 architecture (Linux, AIX). Is this the preferred way or are there other/better solutions? If this flag is always required then I would create a patch to automatically set this flag. Regards Kai

Hi, Some days ago, I have used the maildir-size-fix.pl script (cited in this mail thread: http://www.dovecot.org/list/dovecot/2013-January/087942.html, in available here: http://dovecot.org/tools/maildir-size-fix.pl). This is a very good script and helped me a lot! However, some compressed mails where misdetected as non-compressed mail. So I discovered that some versions of the file command