similar to: [LLVMdev] type attributes

Hi Zach, > Does the bitcode representation retain any information about gcc attributes > placed on types? yes it does, at least for some of them. For example, suppose you specify that a type should be 16 byte aligned. This will not change the LLVM type, instead every time an object of your type is created the object creation operation (eg: an alloca) will ensure that the object is 16

Hi everyone, I am using the dragonegg plugin to produce bitcode for the following small C program: int main() { 1: int x = -5; 2: x = 6; 3: return 0; } The assignments on lines 1 and 2 are removed even when no optimization flags are specified. I tried making variable "x" volatile, which prevents the assignments from being removed, however I am no longer able to recover the original

I don't know how to file a PR, but I have a patch (see below), that should work regardless of abi differences, since it relies on the compiler to do the though job. void X86CompilationCallback_SSE(void) { char * SAVEBUF= (char*) alloca(64+12); // alloca is 16byte aligned asm volatile ( "movl %%eax,(%0)\n" "movl %%edx,4(%0)\n" // Save EAX/EDX/ECX

Hello, I'm developing a llvm backend. It seems that, if i16 is not a legal type (no register can hold i16 types in RegisterInfo.td and as a RegisterClass in SelLowering.cpp), i16 should be promoted to i32. Nonotheless, this simple program: int main(){ volatile short a; a= 3; return 0; } which is trasformed in this IR: define i32 @main() nounwind readnone { entry: %a = alloca i16,

On Tue, Jun 9, 2015 at 12:32 PM, David Majnemer <david.majnemer at gmail.com> wrote: > 'load volatile i32 addrspace(1)* null' seems fine to me. However, it > looks like instcombine will turn: > define i32 @foo() { > entry: > %std_ld.i = load volatile i32, i32 addrspace(1)* null > ret i32 %std_ld.i > } > > into: > define i32 @foo() { > entry:

Hi All, I need your guidance about a custom attribute. I have defined one for variables. It is accepted in the source code (without any warnings from clang), for example in following snippet. #define NEWATTR __attribute__((moviAttr(1))) int main() { NEWATTR volatile unsigned int a = 5; volatile unsigned int *p; p = &a; return (a+*p); } and actually when I Dump the declaration, after

On Dec 16, 2011, at 12:39 PM, Kostya Serebryany wrote: > > Do we consider the above transformation legal? Yes, the transformation is perfectly legal for the normal compiler. > > I would argue that it should not be legal. We don't actually know what > > comes after the 22 byte object. Is it another memory object? A > > memory-mapped I/O device? Unmapped memory?

Hi John, I have looked into the EmitAutoVarAlloca() in CGDecl.cpp. However, I could not figure out how to employ my custom attribute for code generation. For example, my custom attribute is visible in CGDecl.cpp but how can I generate based on my custom attribute if (D.hasAttr<myCustomAttri>()) { //What to do here? } What I wan in IR is something like below. Without Custom Attribute:

Hi Asit, thanks for the reply. But I guess I was not clear in my question. Actually, i dont want to use __ATTRIBUTE__((ANNOTATE("MOVIATTR"))), since in documentation it is stated that "This intrinsic allows annotation of local variables with arbitrary strings. This can be useful for special purpose optimizations that want to look for these annotations. These have no other

Hi All, I am working to cross compile (just generate assembly code) a simple C code in ARM. First, I use CLANG to get LLVM bytecode, then I use llc to generate assembly for ARM. The problem is it never uses any other register except r0-r3 and always uses spill code even if other register are available to use. Anyone has any idea? Thanks Best Regards, A. Yazdanbakhsh

Hello all, I am implementing some simple obfuscation transformations in LLVM. One of the obfuscations involves searching for particular constants, and "unrolling" them throughout a procedure using arithmetic. In effect, certain constants are broken up into smaller constants and recombined as needed using the appropriate operators. I perform this on intermediate LLVM instructions. After

Fairly straight code motion. Split non-virtualizable pieces of processor.h into default and VMI subarches. CPUID is non-virtualizable, since it doesn't trap, and very often the hypervisor will want to hide specific feature bits from the kernel. To provide a replacement for call sites that use CPUID as a serializing instruction, the sync_core() macro is still available. Signed-off-by:

Fairly straight code motion. Split non-virtualizable pieces of processor.h into default and VMI subarches. CPUID is non-virtualizable, since it doesn't trap, and very often the hypervisor will want to hide specific feature bits from the kernel. To provide a replacement for call sites that use CPUID as a serializing instruction, the sync_core() macro is still available. Signed-off-by:

On Thu, 13 Dec 2007, Matt Fredrikson wrote: > After I run opt on an un-obfuscated bitcode file to produce an > obfuscated bitcode file, I verify that my transformations were placed ok > in the file using llvm-dis. At this point, the changes appear to have > been made. However, if I run the obfuscated bitcode file through llc > to produce x86 assembly, the obfuscations vanish. I

Hello, Corrado > Before you can correctly invoke a function via the Procedure Linkage > Table (plt), the ABI mandates that ebx is pointing to the GOT (Global > Offset Table) (see http://www.greyhat.ch/lab/downloads/pic.html) This is known issue, just nobody realized, that we have bunch of non- PIC-aware assembler code. :) Fixing would be not so trivial though, mostly due to ABI

Marking volatile accesses as !unordered seems excessively conservative. For instance, LLVM is not able to optimize declare void @escape(i32*) declare i32 @read_only(i32) readonly define i32 @f(i1* %c) { entry: %a = alloca i32 %b = alloca i32 call void @escape(i32* %a) call void @escape(i32* %b) %a0 = load i32, i32* %b, align 4 %lv = load volatile i32, i32* %b %a1 = load i32,

LLVM developers, I was wondering if the program would still be safe if I strip the byval attributes from the parameters in the entire bitcode. LLVM language reference manual states that "The attribute implies that a hidden copy of the pointee is made between the caller and the callee, so the callee is unable to modify the value in the callee. This attribute is only valid on LLVM pointer

Thanks David, It turns out, that the address space I was using was not 0, and yet the pointer was constant folded to null. Here is the sequence: Unoptimized code: define i32 @foo() #0 { entry: %address.addr.i = alloca i32, align 4 %value.i = alloca i32, align 4 store i32 0, i32* %address.addr.i, align 4 %0 = load i32* %address.addr.i, align 4 %1 = inttoptr i32 %0 to i32 addrspace(1)*

Hi, In my llvm code I want to create some function calls. The function prototype is as follows: int memoize ( char *function_name, int *int_params, unsigned num_ints, double *double_params, unsigned num_doubles) In order to create these calls I do the following for example: %88 = alloca [7 x i8] store volatile [7 x i8] c"ORACLE\00", [7 x i8]*

Hi All, Using clang+llvm at head, I noticed a weird behaviour with the following reduced testcase : $ cat test.c #include <stdint.h> struct R { uint16_t a; uint16_t b; }; volatile struct R * const addr = (volatile struct R *) 416; void test(uint16_t a) { struct R r = { a, 1 }; *addr = r; } $ clang -O2 -o - -emit-llvm -S -c test.c ; ModuleID = 'test.c' target