Translation of custom attribute (defined for variables) from clang to llvm

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 adding the attribute in handleMoviAttr() function in SemaDeclAttr.cpp as shown in following snippet

D->addAttr(::new (S.Context)moviAttrAttr(Attr.getRange(), S.Context, Val, Attr.getAttributeSpellingListIndex()));

It is dumped by clang with following output

VarDecl 0x440c698 <file.c:6:17, line:10:32> col:32 a 'volatile unsigned int'
`-moviAttrAttr 0x440c6d0 <line:6:32, col:42> 1

HOWEVER, the attribute doesnt appear in IR of the program. IR is shown below.

*** IR Dump Before Pre-ISel Intrinsic Lowering ***; ModuleID = 'file.bc'
source_filename = "file.c"
target datalayout = "e-m:e-p:32:32-f64:64-i64:64-v128:64-v64:64-v32:32-v16:16-n8:16:32-S64"
target triple = "shave"

; Function Attrs: nounwind
define i32 @main() local_unnamed_addr #0 {
   %1 = alloca i32, align 4
   %2 = bitcast i32* %1 to i8*
   call void @llvm.lifetime.start(i64 4, i8* %2)
   store volatile i32 5, i32* %1, align 4
   %3 = load volatile i32, i32* %1, align 4
   %4 = load volatile i32, i32* %1, align 4
   %5 = add i32 %4, %3
   call void @llvm.lifetime.end(i64 4, i8* %2)
   ret i32 %5

; Function Attrs: argmemonly nounwind
declare void @llvm.lifetime.start(i64, i8* nocapture) #1

; Function Attrs: argmemonly nounwind
declare void @llvm.lifetime.end(i64, i8* nocapture) #1

attributes #0 = { nounwind "disable-tail-calls"="false" "less-precise-fpmad"="false" "no-frame-pointer-elim"="true" "no-frame-pointer-elim-non-leaf" "no-infs-fp-math"="false" "no-jump-tables"="false" "no-nans-fp-math"="false" "no-signed-zeros-fp-math"="false" "stack-protector-buffer-size"="8" "target-cpu"="myriad2.2" "unsafe-fp-math"="false" "use-soft-float"="false" }
attributes #1 = { argmemonly nounwind "target-cpu"="myriad2.2" }

Can someone help me please to solve the problem so that custom attribute also appears in the IR.

Thanks in advance


I think, the syntax is wrong.
You can try this way…

#define NEWATTR attribute((annotate(“moviAttr”)))

int main()
NEWATTR volatile unsigned int a = 5;
volatile unsigned int *p;
p = &a;
return (a+*p);

And it generates

; ModuleID = ‘test2.cpp’
target datalayout = “e-m:e-i64:64-f80:128-n8:16:32:64-S128”
target triple = “x86_64-pc-linux-gnu”

@.str = private unnamed_addr constant [9 x i8] c"moviAttr\00", section “llvm.metadata”
@.str1 = private unnamed_addr constant [10 x i8] c"test2.cpp\00", section “llvm.metadata”

; Function Attrs: nounwind uwtable
define i32 @main() #0 {
%1 = alloca i32, align 4
%a = alloca i32, align 4
%p = alloca i32*, align 8
store i32 0, i32* %1
%2 = bitcast i32* %a to i8*
call void @llvm.var.annotation(i8* %2, i8* getelementptr inbounds ([9 x i8]* @.str, i32 0, i32 0), i8* getelementptr inbounds ([10 x i8]* @.str1, i32 0, i32 0), i32 5)
store volatile i32 5, i32* %a, align 4
store i32* %a, i32** %p, align 8
%3 = load volatile i32* %a, align 4
%4 = load i32** %p, align 8
%5 = load volatile i32* %4, align 4
%6 = add i32 %3, %5
ret i32 %6

; Function Attrs: nounwind
declare void @llvm.var.annotation(i8*, i8*, i8*, i32) #1

attributes #0 = { nounwind uwtable “less-precise-fpmad”=“false” “no-frame-pointer-elim”=“true” “no-frame-pointer-elim-non-leaf” “no-infs-fp-math”=“false” “no-nans-fp-math”=“false” “stack-protector-buffer-size”=“8” “unsafe-fp-math”=“false” “use-soft-float”=“false” }
attributes #1 = { nounwind }

!llvm.ident = !{!0}

!0 = !{!“Ubuntu clang version 3.6.0-2ubuntu1~trusty1 (tags/RELEASE_360/final) (based on LLVM 3.6.0)”}

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 defined use; they are ignored by code generation and optimization."

Thats why I want to use my custom defined attribute (i.e. moviAttr(unsigned int))) so that i can use it later for code generation.

Syntax of adding my custom attribute in code snippet above is correct, since clang doesnt generate any error.

do you think there is anyother reason for my custom attribute to not appear in IR?

If you want the clang attribute to have an effect on the LLVM IR that
is generated then you need to modify the relevant parts of
clang/lib/CodeGen to do whatever it is you want to do. For local
variables it looks like EmitAutoVarAlloca in CGDecl.cpp is possibly
what you need to modify.

Probably the best thing to do is look around the CodeGen directory for
uses of hasAttr and try to find some other clang attribute which is
handled in a similar way to what you want to do, then do something
similar to that.


Thanks John for the hint. I will try looking CDDECL.cpp. Hope it will solve the problem.

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: %1 = alloca i32, align 4
With Custom Attribute: %1 = alloca i32, align 4, Affinity

Where affinity is an unsigned number, provided at time of variable declaration using custom attribute as shown below

unsigned int Affinity=1;
__attribute__((myCustomAttr(Affinity))) volatile unsigned int a = 5;

Can you please guide me?

Thanks in advance

Can anyone please guide me on solving the problem. I am stuck for last two days.

Thansk in advance for any help.

It's not possible to add an arbitrary operand to an alloca like that, but
you can add arbitrary metadata to an instruction. I'd have a look around
CodeGen for uses of setMetadata to see how it's done elsewhere. It looks
like you could do something like

if (D.hasAttr<myCustomAttribute>())
    Alloca->setMetadata("affinity", llvm::MDNode::get(getContext(), ValueAsMetadata::get(AffinityVal)));

where Alloca is the alloca instruction you want to add the metadata to and
AffinityVal is the llvm::Value* for the affinity value.