Can you send the IR you are using? Volkan On Thu, Oct 8, 2015 at 6:28 AM Erdem Derebaşoğlu < erdemderebasoglu at hotmail.com> wrote:> Thanks. I enabled my pass. I have one resolved issue though: > MachineMemOperand::getAddrSpace() always returns zero. How can I use it to > distinguish private memory accesses? > > Erdem > > ------------------------------ > From: vlknkls at gmail.com > Date: Sun, 4 Oct 2015 01:51:29 +0000 > > Subject: Re: [llvm-dev] How to add NOP? > To: erdemderebasoglu at hotmail.com; llvm-dev at lists.llvm.org > > Hi Erdem, > > Since it is a target specific pass, you should put your pass inside the > target's folder (e.g. lib/Target/X86/) and declare your pass in Target.h. > Then you need to modify the target's TargetPassConfig ( > http://llvm.org/docs/doxygen/html/classllvm_1_1TargetPassConfig.html) to > enable your pass. > > You can check out the existing MachineFunction passes > (e.g. X86ExpandPseudo.cpp) to get an idea. > > Volkan > > On Sat, Oct 3, 2015 at 7:21 AM Erdem Derebaşoğlu < > erdemderebasoglu at hotmail.com> wrote: > > I wrote the pass but when I try to build LLVM, I get this error: > "/home/erdem/llvm/lib/CodeGen/CodeGen.cpp:80: error: undefined reference > to 'llvm::initializeNoopInserterPass(llvm::PassRegistry&). > I need to modify Passes.h, InitializePasses.h, and Codegen.cpp files to > register my pass, right? > > Erdem > ------------------------------ > From: erdemderebasoglu at hotmail.com > To: vlknkls at gmail.com; llvm-dev at lists.llvm.org > Subject: RE: [llvm-dev] How to add NOP? > Date: Thu, 17 Sep 2015 16:38:36 +0300 > > > This seems to be what I am looking for. That was very helpful. Thank you. > > Erdem > > ------------------------------ > From: vlknkls at gmail.com > Date: Wed, 16 Sep 2015 16:25:29 +0000 > Subject: Re: [llvm-dev] How to add NOP? > To: erdemderebasoglu at hotmail.com; llvm-dev at lists.llvm.org > > Use MachineInstr::memoperands() function to get memory operands then you > can get the address space by using MachineMemOperand::getAddrSpace(). > > Volkan > > On Tue, Sep 15, 2015 at 11:40 PM Erdem Derebaşoğlu < > erdemderebasoglu at hotmail.com> wrote: > > Thanks for the response. How can I find out if an instruction uses a > private memory access? > > Erdem > ------------------------------ > From: vlknkls at gmail.com > Date: Tue, 15 Sep 2015 14:38:16 +0000 > Subject: Re: [llvm-dev] How to add NOP? > To: erdemderebasoglu at hotmail.com; llvm-dev at lists.llvm.org > > > Hi Erdem, > > I would write a MachineFunction pass. You can check each load instruction > to see if it is a private memory access and insert what you want by using > BuildMI(...) function. > > Volkan > > On Tue, Sep 15, 2015 at 4:16 AM Erdem Derebaşoğlu <llvm-dev at lists.llvm.org> > wrote: > > I would like to convert a C file that uses OpenMP library to X86 code with > a condition: I want to insert a NOP before each instruction that does a > private memory access. > I have written an optimization pass that inserts an inline assembly call > (NOP) but it is applied at IR level, and I cannot determine where to insert > it. How can I use LLVM to add NOPs accordingly? > > Thanks. > _______________________________________________ > LLVM Developers mailing list > llvm-dev at lists.llvm.org > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-dev > >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20151011/d716779e/attachment.html>
Here it is:
; ModuleID = 'omp_hello.c'target datalayout =
"e-m:e-i64:64-f80:128-n8:16:32:64-S128"target triple =
"x86_64-unknown-linux-gnu"
%ident_t = type { i32, i32, i32, i32, i8* }%struct.anon = type { i32* }
@.str = private unnamed_addr constant [30 x i8] c"Hello World from thread =
%d\0A\00", align 1 at .str.1 = private unnamed_addr constant [24 x i8]
c"Number of threads = %d\0A\00", align 1 at .str.2 = private
unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1
at 0 = private unnamed_addr constant %ident_t { i32 0, i32 66, i32 0, i32 0, i8*
getelementptr inbounds ([23 x i8], [23 x i8]* @.str.2, i32 0, i32 0) }@1 =
private unnamed_addr constant %ident_t { i32 0, i32 2, i32 0, i32 0, i8*
getelementptr inbounds ([23 x i8], [23 x i8]* @.str.2, i32 0, i32 0) }
; Function Attrs: nounwind optsize uwtabledefine i32 @main(i32 %argc, i8**
nocapture readnone %argv) #0 {entry: %a = alloca i32, align 4 %agg.captured =
alloca %struct.anon, align 8 tail call void @omp_set_num_threads(i32 4) #4 %0
= bitcast i32* %a to i8* call void @llvm.lifetime.start(i64 4, i8* %0) #1
store i32 5, i32* %a, align 4, !tbaa !1 %1 = getelementptr inbounds
%struct.anon, %struct.anon* %agg.captured, i64 0, i32 0 store i32* %a, i32**
%1, align 8, !tbaa !5 call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...)
@__kmpc_fork_call(%ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* bitcast
(void (i32*, i32*, %struct.anon*)* @.omp_outlined. to void (i32*, i32*, ...)*),
%struct.anon* nonnull %agg.captured) #1 call void @llvm.lifetime.end(i64 4, i8*
%0) #1 ret i32 0}
; Function Attrs: nounwinddeclare void @llvm.lifetime.start(i64, i8* nocapture)
#1
; Function Attrs: optsizedeclare void @omp_set_num_threads(i32) #2
; Function Attrs: nounwind optsize uwtabledefine internal void
@.omp_outlined.(i32* nocapture readonly %.global_tid., i32* nocapture readnone
%.bound_tid., %struct.anon* nocapture readonly %__context) #0 {entry: %call =
tail call i32 @omp_get_thread_num() #4 %0 = getelementptr inbounds
%struct.anon, %struct.anon* %__context, i64 0, i32 0 %ref = load i32*, i32**
%0, align 8, !tbaa !5 %1 = load i32, i32* %ref, align 4, !tbaa !1 %call1 =
tail call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([30 x i8], [30 x
i8]* @.str, i64 0, i64 0), i32 %1) #4 %cmp = icmp eq i32 %call, 0 br i1 %cmp,
label %if.then, label %if.end
if.then: ; preds = %entry call void asm
"negl %eax;", ""() %call2 = tail call i32
@omp_get_num_threads() #4 %call3 = tail call i32 (i8*, ...) @printf(i8*
getelementptr inbounds ([24 x i8], [24 x i8]* @.str.1, i64 0, i64 0), i32
%call2) #4 br label %if.end
if.end: ; preds = %if.then, %entry %2
= load i32, i32* %.global_tid., align 4, !tbaa !1 %3 = tail call i32
@__kmpc_cancel_barrier(%ident_t* nonnull @0, i32 %2) #1 ret void}
; Function Attrs: optsizedeclare i32 @omp_get_thread_num() #2
; Function Attrs: nounwind optsizedeclare i32 @printf(i8* nocapture readonly,
...) #3
; Function Attrs: optsizedeclare i32 @omp_get_num_threads() #2
declare i32 @__kmpc_cancel_barrier(%ident_t*, i32)
; Function Attrs: nounwinddeclare void @llvm.lifetime.end(i64, i8* nocapture) #1
declare void @__kmpc_fork_call(%ident_t*, i32, void (i32*, i32*, ...)*, ...)
attributes #0 = { nounwind optsize uwtable
"disable-tail-calls"="false"
"less-precise-fpmad"="false"
"no-frame-pointer-elim"="false"
"no-infs-fp-math"="false"
"no-nans-fp-math"="false"
"stack-protector-buffer-size"="8"
"target-cpu"="x86-64"
"target-features"="+sse,+sse2"
"unsafe-fp-math"="false"
"use-soft-float"="false" }attributes #1 = { nounwind
}attributes #2 = { optsize "disable-tail-calls"="false"
"less-precise-fpmad"="false"
"no-frame-pointer-elim"="false"
"no-infs-fp-math"="false"
"no-nans-fp-math"="false"
"stack-protector-buffer-size"="8"
"target-cpu"="x86-64"
"target-features"="+sse,+sse2"
"unsafe-fp-math"="false"
"use-soft-float"="false" }attributes #3 = { nounwind optsize
"disable-tail-calls"="false"
"less-precise-fpmad"="false"
"no-frame-pointer-elim"="false"
"no-infs-fp-math"="false"
"no-nans-fp-math"="false"
"stack-protector-buffer-size"="8"
"target-cpu"="x86-64"
"target-features"="+sse,+sse2"
"unsafe-fp-math"="false"
"use-soft-float"="false" }attributes #4 = { nounwind optsize
}
!llvm.ident = !{!0}
!0 = !{!"clang version 3.8.0 (trunk 243832)"}!1 = !{!2, !2, i64 0}!2 =
!{!"int", !3, i64 0}!3 = !{!"omnipotent char", !4, i64 0}!4
= !{!"Simple C/C++ TBAA"}!5 = !{!3, !3, i64 0}
Erdem
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20151012/e587d1e1/attachment.html>
According to LLVM Language Reference Manual ( http://llvm.org/docs/LangRef.html#global-variables), the address space is 0 for those global variables. "A global variable may be declared to reside in a target-specific numbered address space. For targets that support them, address spaces may affect how optimizations are performed and/or what target instructions are used to access the variable. The default address space is zero. The address space qualifier must precede any other attributes." Volkan On Mon, Oct 12, 2015 at 5:31 AM Erdem Derebaşoğlu < erdemderebasoglu at hotmail.com> wrote:> Here it is: > > ; ModuleID = 'omp_hello.c' > target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" > target triple = "x86_64-unknown-linux-gnu" > > %ident_t = type { i32, i32, i32, i32, i8* } > %struct.anon = type { i32* } > > @.str = private unnamed_addr constant [30 x i8] c"Hello World from thread > = %d\0A\00", align 1 > @.str.1 = private unnamed_addr constant [24 x i8] c"Number of threads > %d\0A\00", align 1 > @.str.2 = private unnamed_addr constant [23 x i8] > c";unknown;unknown;0;0;;\00", align 1 > @0 = private unnamed_addr constant %ident_t { i32 0, i32 66, i32 0, i32 0, > i8* getelementptr inbounds ([23 x i8], [23 x i8]* @.str.2, i32 0, i32 0) } > @1 = private unnamed_addr constant %ident_t { i32 0, i32 2, i32 0, i32 0, > i8* getelementptr inbounds ([23 x i8], [23 x i8]* @.str.2, i32 0, i32 0) } > > ; Function Attrs: nounwind optsize uwtable > define i32 @main(i32 %argc, i8** nocapture readnone %argv) #0 { > entry: > %a = alloca i32, align 4 > %agg.captured = alloca %struct.anon, align 8 > tail call void @omp_set_num_threads(i32 4) #4 > %0 = bitcast i32* %a to i8* > call void @llvm.lifetime.start(i64 4, i8* %0) #1 > store i32 5, i32* %a, align 4, !tbaa !1 > %1 = getelementptr inbounds %struct.anon, %struct.anon* %agg.captured, > i64 0, i32 0 > store i32* %a, i32** %1, align 8, !tbaa !5 > call void (%ident_t*, i32, void (i32*, i32*, ...)*, ...) > @__kmpc_fork_call(%ident_t* nonnull @1, i32 1, void (i32*, i32*, ...)* > bitcast (void (i32*, i32*, %struct.anon*)* @.omp_outlined. to void (i32*, > i32*, ...)*), %struct.anon* nonnull %agg.captured) #1 > call void @llvm.lifetime.end(i64 4, i8* %0) #1 > ret i32 0 > } > > ; Function Attrs: nounwind > declare void @llvm.lifetime.start(i64, i8* nocapture) #1 > > ; Function Attrs: optsize > declare void @omp_set_num_threads(i32) #2 > > ; Function Attrs: nounwind optsize uwtable > define internal void @.omp_outlined.(i32* nocapture readonly > %.global_tid., i32* nocapture readnone %.bound_tid., %struct.anon* > nocapture readonly %__context) #0 { > entry: > %call = tail call i32 @omp_get_thread_num() #4 > %0 = getelementptr inbounds %struct.anon, %struct.anon* %__context, i64 > 0, i32 0 > %ref = load i32*, i32** %0, align 8, !tbaa !5 > %1 = load i32, i32* %ref, align 4, !tbaa !1 > %call1 = tail call i32 (i8*, ...) @printf(i8* getelementptr inbounds > ([30 x i8], [30 x i8]* @.str, i64 0, i64 0), i32 %1) #4 > %cmp = icmp eq i32 %call, 0 > br i1 %cmp, label %if.then, label %if.end > > if.then: ; preds = %entry > call void asm "negl %eax;", ""() > %call2 = tail call i32 @omp_get_num_threads() #4 > %call3 = tail call i32 (i8*, ...) @printf(i8* getelementptr inbounds > ([24 x i8], [24 x i8]* @.str.1, i64 0, i64 0), i32 %call2) #4 > br label %if.end > > if.end: ; preds = %if.then, > %entry > %2 = load i32, i32* %.global_tid., align 4, !tbaa !1 > %3 = tail call i32 @__kmpc_cancel_barrier(%ident_t* nonnull @0, i32 %2) > #1 > ret void > } > > ; Function Attrs: optsize > declare i32 @omp_get_thread_num() #2 > > ; Function Attrs: nounwind optsize > declare i32 @printf(i8* nocapture readonly, ...) #3 > > ; Function Attrs: optsize > declare i32 @omp_get_num_threads() #2 > > declare i32 @__kmpc_cancel_barrier(%ident_t*, i32) > > ; Function Attrs: nounwind > declare void @llvm.lifetime.end(i64, i8* nocapture) #1 > > declare void @__kmpc_fork_call(%ident_t*, i32, void (i32*, i32*, ...)*, > ...) > > attributes #0 = { nounwind optsize uwtable "disable-tail-calls"="false" > "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" > "no-infs-fp-math"="false" "no-nans-fp-math"="false" > "stack-protector-buffer-size"="8" "target-cpu"="x86-64" > "target-features"="+sse,+sse2" "unsafe-fp-math"="false" > "use-soft-float"="false" } > attributes #1 = { nounwind } > attributes #2 = { optsize "disable-tail-calls"="false" > "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" > "no-infs-fp-math"="false" "no-nans-fp-math"="false" > "stack-protector-buffer-size"="8" "target-cpu"="x86-64" > "target-features"="+sse,+sse2" "unsafe-fp-math"="false" > "use-soft-float"="false" } > attributes #3 = { nounwind optsize "disable-tail-calls"="false" > "less-precise-fpmad"="false" "no-frame-pointer-elim"="false" > "no-infs-fp-math"="false" "no-nans-fp-math"="false" > "stack-protector-buffer-size"="8" "target-cpu"="x86-64" > "target-features"="+sse,+sse2" "unsafe-fp-math"="false" > "use-soft-float"="false" } > attributes #4 = { nounwind optsize } > > !llvm.ident = !{!0} > > !0 = !{!"clang version 3.8.0 (trunk 243832)"} > !1 = !{!2, !2, i64 0} > !2 = !{!"int", !3, i64 0} > !3 = !{!"omnipotent char", !4, i64 0} > !4 = !{!"Simple C/C++ TBAA"} > !5 = !{!3, !3, i64 0} > > Erdem >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20151014/95a094a3/attachment-0001.html>