similar to: thread safety ExecutionEngine::getFunctionAddress

> On Dec 20, 2016, at 9:13 AM, koffie drinker via llvm-dev <llvm-dev at lists.llvm.org> wrote: > > Hi, > > I'm trying to speed up the JIT time with llvm (3.9.1). > So far i've implemented the object cache, used FastISel and disabled optimizations. I can’t help with the rest, but just wanted to mention that totally disabling the IR optimizations is not necessarily

So I've made code to invoke the getfunctionaddress() in parallel. I did verify that the code was good, by substituting getfunctionaddress() with a bunch bogus computations. It seems that the code with getfunctionaddress() is being serialized. Is there a giant lock somewhere per executionengine? I have one execution engine that holds all the modules. Going through the llvm-dev list archives,

Hi, What other options do I have to reduce JIT time for large amount of code? - setting optimization level to none helps a lot - enabling FastISel doesn't seem to help much Thanks! -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20160812/f05d51fb/attachment.html>

Hi all, I'm trying to add pre-compiled object cache to my run-time. I've implemented the object cache as follow: class EngineObjectCache : public llvm::ObjectCache { private: std::unordered_map<std::string, std::unique_ptr<llvm::MemoryBuffer>> CachedObjs; public: virtual void notifyObjectCompiled(const llvm::Module *M, llvm::MemoryBufferRef Obj) { auto id =

Turns out llvm initializes memory before the constructor is invoked. Visual studio has /sdl by default on, and the __autoclassinit2 will zero the memory before the constructor is reached causing all the setters to be default zero. which clears the hashungoff var and causing the delete to flow in wrong part. When /sdl is enabled, the compiler generates code to perform these checks at run time: —

I found the root cause, but I don't know what's the best approach to fix it. Under windows, 64 bit, when a function is created the void *User::operator new(size_t Size) operator allocates space + Use*. In the Use* the HasHungOffUses is set to true. So the ptr to the use* is returned as new object. This ptr is NOT the ptr that was allocated by the system. For that ptr you need ptr - word

I made the llvm::Function() constructor public (for testing purpose) and used the non-overloaded new. auto func = ::new llvm::Function(...) if (func) func->eraseFromParent(); And the heap corruption is gone! Did something changed in llvm::User::new between 3.7.1 and 3.8.0 ? I found a bug in llvm ? On Thu, Feb 25, 2016 at 12:10 PM, koffie drinker <gekkekoe at gmail.com> wrote: > I

I'm on llvm 3.8.1 and was wondering if there's a memory leak in the removeModule impl of mcjit. In the tutorial http://llvm.org/releases/3.8.1/docs/tutorial/LangImpl4.html a module is removed from the Jit by invoking removeModule. According to the tutorial: "Its API is very simple:: addModule adds an LLVM IR module to the JIT, making its functions available for execution;

Hi Kevin, Koffie, We will start migrating to ORC for next release, but for now, this release > invoke delete after remove right? MCJIT's removeModule method does not delete the module. You'll need to do that manually. OrcMCJITReplacement is a bug-for-bug compatible implementation of MCJIT using ORC components, so it does not free the memory either. Does this mean MCJIT is

Hi Koffie, I'd highly recommend switching to ORC from MCJIT. It is much more flexible when it comes to memory-management. 1. I took the approach of 1 execution engine with multiple modules (I'm not > removing modules once they have been added). During profiling, I noticed > that the memory usage is high with a lot of code. How can I reduce the > memory usage? Is one execution

I recently upgraded from llvm 3.7.1 to a pre release of llvm (3.8.0rc2) in order to test some issues regarding bug 24233. After upgrading I starting to see heap corruption messages in vs 2015 sp1 when my program exits. "HEAP[ConsoleEngine.exe]: Invalid address specified to RtlValidateHeap( 0000000000290000, 0000000000318698 )" Initially I only got it in Release build. Debug build seems

I downloaded 3.8.0rc3 and I also have it in 3.8.0rc3. I did set a data access breakpoint on the first function ptr that causes the invalid heap. This would allow me to break whenever someone is touching that address. It did not show double deletes during debugging. Further more I managed to narrow it down to 2 function calls: // stupid code, but its just for triggering heap error auto func =

Hi everyone, In llvm 4.0 the MCJIT::getFunctionAddress function return 0 (a null address) when the symbol is not found : *uint64_t MCJIT::getSymbolAddress(const std::string &Name, bool CheckFunctionsOnly) { std::string MangledName; { raw_string_ostream MangledNameStream(MangledName); Mangler::getNameWithPrefix(MangledNameStream, Name, getDataLayout()); } return

I should have read this before sending my previous reply. :-) I'm not a big fan of default parameters, but some form of what you are suggesting may be useful. See my other comments on this topic in the other reply. -Andy From: Yaron Keren [mailto:yaron.keren at gmail.com] Sent: Saturday, October 19, 2013 3:17 PM To: Kaylor, Andrew; <llvmdev at cs.uiuc.edu> Subject: An enhancement

In MCJIT, the old JIT functions are deprecated in favor of getFunctionAddress. Code like: llvm::Function *F = M->getFunction(FuncName); void *FN = EE->getPointerToFunction(F); should be rewritten as uint64_t FN = EE->getFunctionAddress(FuncName); While functionally identical, in case the correct module is known the new version will be much slower, linear with the number of added

There's probably a lot that we could do, but I can't think of anything easy. Basically every time we need to look up a symbol by name we're going to each module and saying "Do you have this symbol?" It would likely be much better if we grabbed the function names from the module and did the search ourselves so that we could keep some information about the things that

The search is linear? If that’s really true, we should fix that. On Oct 21, 2013, at 10:14 AM, Kaylor, Andrew <andrew.kaylor at intel.com> wrote: > I should have read this before sending my previous reply. :-) > > I’m not a big fan of default parameters, but some form of what you are suggesting may be useful. See my other comments on this topic in the other reply. > >

Hi, I am wondering what the status of Clang++ and LLVM on the Windows platform ? When I last saw what the state was there was not proper linking and more recently heard that the Structured Exception Handling was not working. The status page seems somewhat out of date ? Many thanks in advance, Aaron

I don’t follow. Why are we looking at the module at all? That query should work even (especially) after the Module is deleted. We should be able to have a local symbol table that’s a DenseMap or something similar to resolve from names to target addresses. That map would be updated as part of the compilation when the object’s symbol table gets read. -Jim On Oct 21, 2013, at 4:55 PM, Kaylor,

Hi Jim, Clearly searching for name should not be linear in modules but with a map of some kind. After compilation to IR each module has a StringMap symbol table. After compiling to MC and loading the object file, the dynamic linker has a StringMap symbol table for all loaded modules. In the usual use case you'll load module(s) into MCJIT and then compile / link them and all is well, no