[llvm-dev] LLVM v6.0 Internalize and GlobalDCE PASS can not work together? (original) (raw)
Leslie Zhai via llvm-dev llvm-dev at lists.llvm.org
Wed Oct 25 19:45:17 PDT 2017
- Previous message: [llvm-dev] LLVM v6.0 Internalize and GlobalDCE PASS can not work together?
- Next message: [llvm-dev] Profiling data structure
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
在 2017年10月26日 10:20, Hal Finkel 写道:
On 10/25/2017 09:07 PM, Leslie Zhai wrote: Hi Hal,
Thanks for your hint! $ /opt/llvm-svn/bin/opt -S -internalize -internalize-public-api-list=main -globaldce hello3.ll -o hello3.dce.ll it works :) But I argue that
mainFunction should be inserted into ExternalNames by default: But "main" is neither necessary nor sufficient depending on the target/configuration. Internalize is really intended to be used by a linker plugin (or in some other context where a callback can be provided that has the right information). Using the pass as you're doing is really a development aid - it's not a user-facing tool by itself.
Thanks for your teaching! yes, I just use it to remove DeadFunctions for ScaffCC https://github.com/ScaffCC/ScaffCC/blob/master/scaffold/Scaffold.makefile#L122
Ali JavadiAbhari, Shruti Patil, Daniel Kudrow, Jeff Heckey, Alexey Lvov, Frederic Chong and Margaret Martonosi, ScaffCC: A Framework for Compilation and Analysis of Quantum Computing Programs, ACM International Conference on Computing Frontiers (CF 2014), Cagliari, Italy, May 2014
-Hal >>> Index: lib/Transforms/IPO/Internalize.cpp =================================================================== --- lib/Transforms/IPO/Internalize.cpp (revision 316540) +++ lib/Transforms/IPO/Internalize.cpp (working copy) @@ -61,6 +61,7 @@ if (!APIFile.empty()) LoadFile(APIFile); ExternalNames.insert(APIList.begin(), APIList.end()); + ExternalNames.insert("main"); } >> bool operator()(const GlobalValue &GV) { >>>> 在 2017年10月25日 21:20, Hal Finkel 写道: > Hi, Leslie, >>>> When you use internalize, you need to provide it with a list of > symbols to preserve as external (otherwise everything will be > internalized, including main, and then DCE will remove everything). > You can use -internalize-public-api-list=main (a comma-separated > list) or -internalize-public-api-file=somefilename where > somefilename has the list of symbols. See the comments in > lib/Transforms/IPO/Internalize.cpp. >>>> -Hal >>>> On 10/24/2017 11:22 PM, Leslie Zhai via llvm-dev wrote: >> Hi LLVM developers, >>>>>> $ cat hello.c >> #include <stdio.h> >>>>>> void foo() { >> } >>>>>> int main(int argc, char *argv[]) { >> _for (int i = 0; i < 10; i++) {_ >> printf("%d\n", i); >> } >> return 0; >> } >>>>>> $ /opt/llvm-svn/bin/clang --version >> Fedora clang version 6.0.0 (trunk 316308) (based on LLVM 6.0.0svn) >> Target: x8664-redhat-linux >> Thread model: posix >> InstalledDir: /opt/llvm-svn/bin >>>>>> $ /opt/llvm-svn/bin/clang -Xclang -disable-O0-optnone -S -emit-llvm >> hello.c -o hello2.ll >>>>>> $ cat hello2.ll >> ; ModuleID = 'hello.c' >> sourcefilename = "hello.c" >> target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" >> target triple = "x8664-unknown-linux-gnu" >>>>>> @.str = private unnamedaddr constant [4 x i8] c"%d\0A\00", align 1 >>>>>> ; Function Attrs: noinline nounwind uwtable >> define void @foo() #0 { >> entry: >> ret void >> } >>>>>> ; Function Attrs: noinline nounwind uwtable >> define i32 @main(i32 %argc, i8** %argv) #0 { >> entry: >> %retval = alloca i32, align 4 >> %argc.addr = alloca i32, align 4 >> %argv.addr = alloca i8**, align 8 >> %i = alloca i32, align 4 >> store i32 0, i32* %retval, align 4 >> store i32 %argc, i32* %argc.addr, align 4 >> store i8** %argv, i8*** %argv.addr, align 8 >> store i32 0, i32* %i, align 4 >> br label %for.cond >>>>>> for.cond: ; preds = >> %for.inc, %entry >> %0 = load i32, i32* %i, align 4 >> %cmp = icmp slt i32 %0, 10 >> br i1 %cmp, label %for.body, label %for.end >>>>>> for.body: ; preds = %for.cond >> %1 = load i32, i32* %i, align 4 >> %call = call i32 (i8*, ...) @printf(i8* getelementptr inbounds >> ([4 x i8], [4 x i8]* @.str, i32 0, i32 0), i32 %1) >> br label %for.inc >>>>>> for.inc: ; preds = %for.body >> %2 = load i32, i32* %i, align 4 >> %inc = add nsw i32 %2, 1 >> store i32 %inc, i32* %i, align 4 >> br label %for.cond >>>>>> for.end: ; preds = %for.cond >> ret i32 0 >> } >>>>>> declare i32 @printf(i8*, ...) #1 >>>>>> attributes #0 = { noinline nounwind uwtable >> "correctly-rounded-divide-sqrt-fp-math"="false" >> "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" >> "no-trapping-math"="false" "stack-protector-buffer-size"="8" >> "target-cpu"="x86-64" >> "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" >> "unsafe-fp-math"="false" "use-soft-float"="false" } >> attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" >> "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-nans-fp-math"="false" >> "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" >> "stack-protector-buffer-size"="8" "target-cpu"="x86-64" >> "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" >> "unsafe-fp-math"="false" "use-soft-float"="false" } >>>>>> !llvm.module.flags = !{!0} >> !llvm.ident = !{!1} >>>>>> !0 = !{i32 1, !"wcharsize", i32 4} >> !1 = !{!"clang version 6.0.0 (git at github.com:llvm-mirror/clang.git >> 0aed123216ad4a38a9c2b16f1783895fd5cb1a04) >> (git at github.com:llvm-mirror/llvm.git >> d209b37aec1e392dabbf9b5324ea4a60c36fbc55)"} >>>>>> $ /opt/llvm-svn/bin/opt -S -internalize -globaldce hello2.ll -o >> hello2.dce.ll >>>>>> $ cat hello2.dce.ll >> ; ModuleID = 'hello2.ll' >> sourcefilename = "hello.c" >> target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" >> target triple = "x8664-unknown-linux-gnu" >>>>>> !llvm.module.flags = !{!0} >> !llvm.ident = !{!1} >>>>>> !0 = !{i32 1, !"wcharsize", i32 4} >> !1 = !{!"clang version 6.0.0 (git at github.com:llvm-mirror/clang.git >> 0aed123216ad4a38a9c2b16f1783895fd5cb1a04) >> (git at github.com:llvm-mirror/llvm.git >> d209b37aec1e392dabbf9b5324ea4a60c36fbc55)"} >>>>>> $ /opt/llvm-svn/bin/lli hello2.dce.ll >> 'main' function not found in module. >>>>>>>>>> No more Alive Functions, so LLVM v6.0 Internalize and GlobalDCE >> PASS failed to MarkLive Function together? >>>>>>>>>> But LLVM 3.1 is able to work together: >>>>>> $ ./build/Release+Asserts/bin/opt --version >> LLVM (http://llvm.org/): >> LLVM version 3.1 >> Optimized build with assertions. >> Built Oct 23 2017 (16:22:51). >> Default target: x8664-unknown-linux-gnu >> Host CPU: corei7-avx >>>>>> $ ./build/Release+Asserts/bin/clang -S -emit-llvm hello.c -o hello0.ll >>>>>> $ cat hello0.ll >> ; ModuleID = 'hello.c' >> target datalayout = >> "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" >> target triple = "x8664-unknown-linux-gnu" >>>>>> @.str = private unnamedaddr constant [4 x i8] c"%d\0A\00", align 1 >>>>>> define void @foo() nounwind uwtable { >> entry: >> ret void >> } >>>>>> define i32 @main(i32 %argc, i8** %argv) nounwind uwtable { >> entry: >> %retval = alloca i32, align 4 >> %argc.addr = alloca i32, align 4 >> %argv.addr = alloca i8**, align 8 >> %i = alloca i32, align 4 >> store i32 0, i32* %retval >> store i32 %argc, i32* %argc.addr, align 4 >> store i8** %argv, i8*** %argv.addr, align 8 >> store i32 0, i32* %i, align 4 >> br label %for.cond >>>>>> for.cond: ; preds = >> %for.inc, %entry >> %0 = load i32* %i, align 4 >> %cmp = icmp slt i32 %0, 10 >> br i1 %cmp, label %for.body, label %for.end >>>>>> for.body: ; preds = %for.cond >> %1 = load i32* %i, align 4 >> %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds >> ([4 x i8]* @.str, i32 0, i32 0), i32 %1) >> br label %for.inc >>>>>> for.inc: ; preds = %for.body >> %2 = load i32* %i, align 4 >> %inc = add nsw i32 %2, 1 >> store i32 %inc, i32* %i, align 4 >> br label %for.cond >>>>>> for.end: ; preds = %for.cond >> ret i32 0 >> } >>>>>> declare i32 @printf(i8*, ...) >>>>>> $ ./build/Release+Asserts/bin/opt -S -internalize -globaldce >> hello0.ll -o hello0.dce.ll >>>>>> $ cat hello0.dce.ll >> ; ModuleID = 'hello0.ll' >> target datalayout = >> "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128" >> target triple = "x8664-unknown-linux-gnu" >>>>>> @.str = private unnamedaddr constant [4 x i8] c"%d\0A\00", align 1 >>>>>> define i32 @main(i32 %argc, i8** %argv) nounwind uwtable { >> entry: >> %retval = alloca i32, align 4 >> %argc.addr = alloca i32, align 4 >> %argv.addr = alloca i8**, align 8 >> %i = alloca i32, align 4 >> store i32 0, i32* %retval >> store i32 %argc, i32* %argc.addr, align 4 >> store i8** %argv, i8*** %argv.addr, align 8 >> store i32 0, i32* %i, align 4 >> br label %for.cond >>>>>> for.cond: ; preds = >> %for.inc, %entry >> %0 = load i32* %i, align 4 >> %cmp = icmp slt i32 %0, 10 >> br i1 %cmp, label %for.body, label %for.end >>>>>> for.body: ; preds = %for.cond >> %1 = load i32* %i, align 4 >> %call = call i32 (i8*, ...)* @printf(i8* getelementptr inbounds >> ([4 x i8]* @.str, i32 0, i32 0), i32 %1) >> br label %for.inc >>>>>> for.inc: ; preds = %for.body >> %2 = load i32* %i, align 4 >> %inc = add nsw i32 %2, 1 >> store i32 %inc, i32* %i, align 4 >> br label %for.cond >>>>>> for.end: ; preds = %for.cond >> ret i32 0 >> } >>>>>> declare i32 @printf(i8*, ...) >>>>>>>>>> Succeeded drop DeadFunction foo. >>>>>> Maybe I wrongly use the Internalize and GlobalDCE PASS together? >> please give me some hints, thanks a lot! >>>>>>>>>> PS: LLVM 6.0 Internalize PASS is able to work, internal foo: >>>>>> $ /opt/llvm-svn/bin/opt -S -internalize hello2.ll -o >> hello2.internal.ll >>>>>> $ cat hello2.internal.ll >> ; ModuleID = 'hello2.ll' >> sourcefilename = "hello.c" >> target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128" >> target triple = "x8664-unknown-linux-gnu" >>>>>> @.str = private unnamedaddr constant [4 x i8] c"%d\0A\00", align 1 >>>>>> ; Function Attrs: noinline nounwind uwtable >> define internal void @foo() #0 { >> entry: >> ret void >> } >>>>>> ; Function Attrs: noinline nounwind uwtable >> define internal i32 @main(i32 %argc, i8** %argv) #0 { >> entry: >> %retval = alloca i32, align 4 >> %argc.addr = alloca i32, align 4 >> %argv.addr = alloca i8**, align 8 >> %i = alloca i32, align 4 >> store i32 0, i32* %retval, align 4 >> store i32 %argc, i32* %argc.addr, align 4 >> store i8** %argv, i8*** %argv.addr, align 8 >> store i32 0, i32* %i, align 4 >> br label %for.cond >>>>>> for.cond: ; preds = >> %for.inc, %entry >> %0 = load i32, i32* %i, align 4 >> %cmp = icmp slt i32 %0, 10 >> br i1 %cmp, label %for.body, label %for.end >>>>>> for.body: ; preds = %for.cond >> %1 = load i32, i32* %i, align 4 >> %call = call i32 (i8*, ...) @printf(i8* getelementptr inbounds >> ([4 x i8], [4 x i8]* @.str, i32 0, i32 0), i32 %1) >> br label %for.inc >>>>>> for.inc: ; preds = %for.body >> %2 = load i32, i32* %i, align 4 >> %inc = add nsw i32 %2, 1 >> store i32 %inc, i32* %i, align 4 >> br label %for.cond >>>>>> for.end: ; preds = %for.cond >> ret i32 0 >> } >>>>>> declare i32 @printf(i8*, ...) #1 >>>>>> attributes #0 = { noinline nounwind uwtable >> "correctly-rounded-divide-sqrt-fp-math"="false" >> "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" >> "no-trapping-math"="false" "stack-protector-buffer-size"="8" >> "target-cpu"="x86-64" >> "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" >> "unsafe-fp-math"="false" "use-soft-float"="false" } >> attributes #1 = { "correctly-rounded-divide-sqrt-fp-math"="false" >> "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-nans-fp-math"="false" >> "no-signed-zeros-fp-math"="false" "no-trapping-math"="false" >> "stack-protector-buffer-size"="8" "target-cpu"="x86-64" >> "target-features"="+fxsr,+mmx,+sse,+sse2,+x87" >> "unsafe-fp-math"="false" "use-soft-float"="false" } >>>>>> !llvm.module.flags = !{!0} >> !llvm.ident = !{!1} >>>>>> !0 = !{i32 1, !"wcharsize", i32 4} >> !1 = !{!"clang version 6.0.0 (git at github.com:llvm-mirror/clang.git >> 0aed123216ad4a38a9c2b16f1783895fd5cb1a04) >> (git at github.com:llvm-mirror/llvm.git >> d209b37aec1e392dabbf9b5324ea4a60c36fbc55)"} >>>>>>>>
Regards, Leslie Zhai - https://reviews.llvm.org/p/xiangzhai/
- Previous message: [llvm-dev] LLVM v6.0 Internalize and GlobalDCE PASS can not work together?
- Next message: [llvm-dev] Profiling data structure
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]