similar to: [LLVMdev] Enabling x86-64 JIT under Visual Studio compiler

Here we go, updated patch following your advice checking max leaf and porting cpuidex for subleaf (ECX) 0. NOTE: I’ve set Haswell to be not only 60, but also 63, 69 & 70 model, following changes in Linux kernel & Xen. Also set 62 as Ivy Bridge EP aka E5 v3 (which I has in my workstation). Cheers, -- Adam Detects x86 family 6 model 60, 63, 69, 70 CPU that has AVX2 CPUID leaf 7 subleaf

Hello, Tim > Would adding cases for these assembly blocks that are in Visual Studio > syntax be enough to enable a working x86-64 JIT target under Windows > builds? I can't pretend to understand this codebase, so if there are > more roadblocks stopping 64-bit JIT on Windows targets I'd be > interested to hear... The following things are needed for enabling win64 JIT in 2.2

Hello, Everyone. Now I have some spare time and I've decided to build LLVM on Mingw. I've grab the latest 1.7 release (not CVS snapshot). Here are some issues fixed during the build. Now I'm preparing gcc build. So, I think, there will some other "parts" 1. Prerequisites 1.1 GCC 3.4.5 from mingw.org site. $ gcc --version gcc.exe (GCC) 3.4.5 (mingw special) Copyright (C)

Paolo Invernizzi wrote: > There was a similar problem some time ago, and was resolved with alloca. > I think it's a better solution to use the stack instead of the heap... I tend to agree, but the constructors won't get called if it's an object array -- anyway, this particular case there was no objects, just pointers and bools so alloca should be fine. I'll leave it to

These two subclasses of X86TargetMachine are basically identical. The *only* thing that's different is the constructor. And that *only* differs in the is64Bit argument that it passes to the X86TargetMachine constructor. Replacing the hard-coded 'true' or 'false' with 'Triple(TT).getArch()==Triple::x86_64' makes them *actually* identical. Can we just ditch the

Hey Evan, At the point of the instructions you suggested I step through, X86ISelLowering has this state: - this 0x00000000005fe728 {VarArgsFrameIndex=-842150451 RegSaveFrameIndex=-842150451 VarArgsGPOffset=3452816845 ...} llvm::X86TargetLowering * const + llvm::TargetLowering {TM={...} TD=0x00000000008edac0

Hi Morten, If you can make the VS2005 project files availiable on the net then I can test them as I have VS2005 now, so then with Chris'es okay then they could be distributed with LLVM. Thanks, Aaron ----- Original Message ----- From: "Morten Ofstad" <morten at hue.no> To: "LLVM Developers Mailing List" <llvmdev at cs.uiuc.edu> Sent: Thursday, January 26,

Thanks, now it links. If nobody objects, I will commit the following patch: diff --git a/lib/Target/X86/X86JITInfo.cpp b/lib/Target/X86/X86JITInfo.cpp index 44d8cce..8acc220 100644 --- a/lib/Target/X86/X86JITInfo.cpp +++ b/lib/Target/X86/X86JITInfo.cpp @@ -339,6 +339,8 @@ extern "C" { /// must locate the start of the stub or call site and pass it into the JIT /// compiler function.

Hi Patrik, On 12/06/13 09:48, Patrik Hägglund H wrote: > Thanks, now it links. > > If nobody objects, I will commit the following patch: > > diff --git a/lib/Target/X86/X86JITInfo.cpp b/lib/Target/X86/X86JITInfo.cpp > index 44d8cce..8acc220 100644 > --- a/lib/Target/X86/X86JITInfo.cpp > +++ b/lib/Target/X86/X86JITInfo.cpp > @@ -339,6 +339,8 @@ extern "C" {

The project files need frequent updating. I cannot maintain VS2005 project files, so while they could be distributed with LLVM, they will become broken fast. Also, VS2003 and VS2005 project and solution files cannot coexist in the same directories, further complicating matters. Aaron Gray wrote: > Hi Morten, > > If you can make the VS2005 project files availiable on the net then I

> I promise I'll do the review of your code after that. Tim, I don’t want to push too much. But since there’s 3.4 release on the horizon, maybe you could find a moment review this patch. Especially Haswell is all there since few months. Cheers, -- Adam --- lib/Support/Host.cpp | 8 ++++++++ lib/Target/X86/X86Subtarget.cpp | 3 ++- 2 files changed, 10 insertions(+), 1

On 02/03/2010 01:08 PM, Kristaps Straupe wrote: > Hello again. > > I still think that you are wrong. Realignement with and esp,-16 not > always changes stack poiner. If esp is already aligned to 16 byte > boundary, it will not change! Take a look at following example. > Assume esp has value 0x000001000 at start of X86CompilationCallback > function. Then execution of it will

_WIN32, WIN32, WIN64, and _WIN64 are the wrong definitions to be gating this on in the first place. They aren't at all meant to be x86/x86-64 specific. At best, they're 32-bit/64-bit specific, but that's not the intended use in the code. The correct definitions are _M_IX86 and _M_X64, as Martin said. I sent a patch to this ML that fixed these a few months ago but it was

The target description .td files are allowed to change the default allocation order on a register class by overriding the allocation_order_begin() and allocation_order_end() methods on TargetRegisterClass. Previously, this was used all the time to filter out stack and frame pointers and other reserved registers. I was able to remove most of these custom allocation orders in the tree because the

I first sent this to the Clang dev list, but they told me to come here: ---------- Forwarded message ---------- From: Ruben Van Boxem <vanboxem.ruben at gmail.com> Date: 2010/12/31 Subject: LLVM on Windows MSVC 10 To: cfe-dev at cs.uiuc.edu Hi, I'm trying to build svn LLVM with Visual Studio 2010: cd build cmake .. -G"NMake Makefiles" nmake and several link steps fail

2010/12/31 Francois Pichet <pichet2000 at gmail.com>: > I don't normally build using nmake.. but I just tried and it worked 100% > here. > Are you sure you are using the trunk? > > On Fri, Dec 31, 2010 at 7:32 AM, Ruben Van Boxem <vanboxem.ruben at gmail.com> > wrote: >> >> I first sent this to the Clang dev list, but they told me to come here:

Yeah that's the one. On Nov 6, 2014 1:23 PM, "Tristan Matthews" <le.businessman at gmail.com> wrote: > On Thu, Nov 6, 2014 at 4:20 PM, Cameron Gutman <aicommander at gmail.com> > wrote: > > _WIN32, WIN32, WIN64, and _WIN64 are the wrong definitions to be gating > this > > on in the first place. They aren't at all meant to be x86/x86-64 >

Ruben Van Boxem <vanboxem.ruben at gmail.com> writes: >>> I'm trying to build svn LLVM with Visual Studio 2010: >>> >>> cd build >>> cmake .. -G"NMake Makefiles" >>> nmake >>> >>> and several link steps fail due to a missing symbol: >>> >>> > LLVMX86CodeGen.lib(X86JITInfo.cpp.obj) : error

I don't normally build using nmake.. but I just tried and it worked 100% here. Are you sure you are using the trunk? On Fri, Dec 31, 2010 at 7:32 AM, Ruben Van Boxem <vanboxem.ruben at gmail.com>wrote: > I first sent this to the Clang dev list, but they told me to come here: > > > ---------- Forwarded message ---------- > From: Ruben Van Boxem <vanboxem.ruben at

Compiling llvm on a linux/amd64 box produces: home/rafael/dev/llvm/build/Debug/lib/LLVMX86.o: In function `_X86CompilationCallback': (.text+0x316fe): undefined reference to `_X86CompilationCallback2' /home/rafael/dev/llvm/build/Debug/lib/LLVMX86.o: In function `llvm::X86JITInfo::getLazyResolverFunction(void* (*)(void*))': /home/rafael/dev/llvm/cvs/lib/Target/X86/X86JITInfo.cpp:219: