similar to: Miscompilation while switching from clang-4 to clang-5

On 31 May 2018 at 08:41, Alex Denisov via llvm-dev <llvm-dev at lists.llvm.org> wrote: > I understand the linking problem. What I do not understand is how to debug the difference between two versions of compilers. > This what I do (briefly): > > clang++-4 foobar.cpp -lLLVM -l/opt/llvm-3.9/lib > > clang++-5 foobar.cpp -lLLVM -l/opt/llvm-3.9/lib If you add -v then Clang will

On 05/30/2018 02:39 PM, Alex Denisov via llvm-dev wrote: > Hello everyone, > > I observe a weird behavior switching from clang-4 to clang-5 (and any higher version). > I compile an executable that depends on LLVM. Everything works fine with clang-4, but when I run the executable compiled with clang-5 I see the following error: > : CommandLine Error: Option

Hi Tom, hi Michael, Thank you for your help. I understand the linking problem. What I do not understand is how to debug the difference between two versions of compilers. This what I do (briefly): clang++-4 foobar.cpp -lLLVM -l/opt/llvm-3.9/lib clang++-5 foobar.cpp -lLLVM -l/opt/llvm-3.9/lib The first command produces a working executable, while the other one does not. I'm trying to

2018-06-01 2:06 GMT-05:00 Alex Denisov <1101.debian at gmail.com>: > If there are any other hints: send them my way :) Debug with gdb, but a breakpoint at https://github.com/llvm-mirror/llvm/blob/master/lib/Transforms/Utils/SymbolRewriter.cpp#L91. It should hit twice (which is the error reported). At each break you can look at where the machine code is coming from (e.g. a shared object),

Hi there, I have a question regarding this documentation page: http://llvm.org/docs/Projects.html IIRC LLVM has moved away from Makefile's and uses CMake only. The question is: Is the page outdated or is it a recommended way for creating LLVM based project? P.S. I’m trying to create a project using CMake, though I’m seeing some issues (didn’t manage to include headers so far). --

Hi everyone, I’m working with the LLVM C API now. I see that several functions are deprecated, however the only notion is in comments around the function. Is there any specific reason why __attribute__((deprecated)) is omitted? Will it make sense to send a patch with such additions? -- AlexDenisov Software Engineer, http://lowlevelbits.org -------------- next part -------------- A non-text

Hi there, I’m curious is there any (official or not) communication channel for upcoming EuroLLVM? If there is none besides mailing lists and IRC, shall we then setup something like Slack? -- AlexDenisov Software Engineer, https://lowlevelbits.org

There seems to be an issue with DNS: > nslookup llvm.org ;; Got SERVFAIL reply from 192.168.178.1, trying next server Server: 192.168.178.1 Address: 192.168.178.1#53 ** server can't find llvm.org: NXDOMAIN -- AlexDenisov Software Engineer, https://lowlevelbits.org -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type:

Hi everyone, I run some functions using ORC JIT, now I need to add custom instrumentation. I want to add two callbacks to each function: ‘enterFunction' at the beginning and ‘leaveFunction' at the end. Intuition says that I could ‘just' insert CallInst's to the first and the last basic blocks in the function. Am I correct? Are there any other/better way to do this? Is there

Hello everyone, My name is Alex, I'm hobbyist compiler hacker who lives in Berlin, Germany now. I'm thinking about organising "LLVM Social" meetups in Berlin, however I don't really know how many people are here, and how many of them are interested in getting in touch. If you want to participate or have any questions - feel free to answer to this thread, or to send me a

Thanks for the hint, I didn’t know about this option. That’s a great reference! However, I am trying to be a compiler/language agnostic. Also (for whatever reasons) I need a numeric ID of a function rather then its address. So the question is still opened. May I assume that the following always holds: The first basic block in a function is an entry point and the last basic block in a function is

Appreciations for the initiative! There are some people in the Berlin C++ user group [1] who may be interested. Also there was an active compiler meetup group [3] a few years ago. As far as I know these guys mostly joined the Strange Group Berlin [2], but may still be interested in a more compiler-focused meetup. For the people cc'd: please let us know who'd like to participate in a LLVM

Hello, everybody. We are working on a tool for mutation testing. The work is still in progress and far away from being done. However, we have got some results already. And we would like to share them with you. But, let me give you a brief introduction first. ### Mutation Testing In a nutshell, Mutation Testing is a way to evaluate a quality of a test suite. The approach suggests introducing a

Hi Tim, The verbose mode is a good idea, thank you. I've gathered all compilation commands with -v flag for clang-4 and clang-5 and they are identical (excerpt of compiler version). I will try to find what has changed in clang/llvm between the versions. If there are any other hints: send them my way :) > On 31. May 2018, at 09:54, Tim Northover <t.p.northover at gmail.com> wrote:

Hi devs, hi Lang, I use ORC’s SimpleCompiler to compile modules into object files and it works just great, however I’m having an issue: a module gets changed a bit when I compile it (some transformations/optimizations applied). I tried to set optimization level of a TargetMachine to None, but the module gets changed anyway. The question is: how can I prevent a module from being modified during

Hi there, I try to JIT compile some bitcode and seeing the following error: LLVM ERROR: Cannot select: 0x28ec830: ch,glue = X86ISD::CALL 0x28ec7c0, 0x28ef900, Register:i32 %EDI, Register:i8 %AL, RegisterMask:Untyped, 0x28ec7c0:1 0x28ef900: i32 = X86ISD::Wrapper TargetGlobalAddress:i32<void (i8*, ...)* @_ZN5FooBr7xprintfEPKcz> 0 0x28ec520: i32 = TargetGlobalAddress<void (i8*, ...)*

Hello! I am looking for a handful of people to help write blog posts for the LLVM blog (blog.llvm.org <http://blog.llvm.org/>) or to help recruit volunteers to write blog posts. The LLVM Project blog is a great place to share details about recent changes to LLVM, Clang, and related sub-projects. It also is a great place to highlight users of LLVM. You can write about your work or the work

Hi everybody, I do write some Clang/LLVM related articles on my blog[1][2], and I will be happy to write for LLVM’s blog. However, I can’t omit bike-shedding :) Forgive me my directness, but current blog doesn’t look like something close to 2016. The blog already has lots of great articles. But it’s so hard to grasp valuable information when you have to read non-highlighted C++ code. I think I

Hi -- I've recently added some documentation to the LLVM docs subdirectory, and was wondering about a couple of things: - How should I best debug whether the docs (in RST) are syntactically correct? Are there recommended practices for doing a sanity check on the docs? - How often do the documents get built and uploaded to the website? Cheers -- Dean

Hello everybody, I am trying to do some static analysis, e.g. find which other functions accessible from a function. Current naive implementation goes over each instruction and whether it is a call site or not. It works great so far, but there are some cases where it doesn’t work. For example: declare no_source(function: f) // uses f internally define foo() { ... } define bar() { ... }