1
2
3
sudo apt install lld-12
sudo ln -s /lib/llvm-9/bin/llc /bin/llc
sudo ln -s /lib/llvm-9/bin/opt /bin/opt

官方文档：LLVM Language Reference Manual — LLVM 16.0.0git documentation

1
clang -Xclang -ast-dump -fsyntax-only test.c

生成 AST

  • -S Only run preprocess and compilation steps
  • -emit-llvm Use the LLVM representation for assembler and object files
  • -c Only run preprocess, compile, and assemble steps （生成字节码的 bc 文件）
1
clang -S -emit-llvm test.c

生成的 ir 中

1
2
3
4
5
define dso_local i32 @main() #0 {
  %1 = alloca i32, align 4
  store i32 0, i32* %1, align 4
  ret i32 0
}

如果开启优化

1
clang -S -emit-llvm -O3 test.c

会直接变成

1
2
3
define dso_local i32 @main() local_unnamed_addr #0 {
  ret i32 0
}

然后用 llc 生成汇编

1
llc test.ll

ll 到 bc 文件可以用 llvm-as
反过来用 llvm-dis
注意，ll 和 bc 和内存中的形式是等价的。

dso_local 是一个 Runtime Preemption 说明符，表明该函数会在同一个链接单元（即该函数所在的文件以及包含的头文件）内解析符号。

对于

1
2
3
4
5
6
7
8
9
10
11
// main.c
int foo(int first, int second) {
    return first + second;
}

int a = 5;

int main() {
    int b = 4;
    return foo(a, b);
}

生成

  • alloca 就是在栈中分配空间
  • 先把传入的值放入栈中再拿出来（嫩麻烦
  • nsw : no signed wrap
  • 所有的全局变量都以 @ 为前缀
  • 这里 #0 与之后的 attributes #0 相对应
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
; ModuleID = 'main.c'
source_filename = "main.c"
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-pc-linux-gnu"

@a = dso_local global i32 5, align 4

; Function Attrs: noinline nounwind optnone uwtable
define dso_local i32 @foo(i32, i32) #0 {
  %3 = alloca i32, align 4
  %4 = alloca i32, align 4
  store i32 %0, i32* %3, align 4
  store i32 %1, i32* %4, align 4
  %5 = load i32, i32* %3, align 4
  %6 = load i32, i32* %4, align 4
  %7 = add nsw i32 %5, %6
  ret i32 %7
}

; Function Attrs: noinline nounwind optnone uwtable
define dso_local i32 @main() #0 {
  %1 = alloca i32, align 4
  %2 = alloca i32, align 4
  store i32 0, i32* %1, align 4
  store i32 4, i32* %2, align 4
  %3 = load i32, i32* @a, align 4
  %4 = load i32, i32* %2, align 4
  %5 = call i32 @foo(i32 %3, i32 %4)
  ret i32 %5
}

attributes #0 = { noinline nounwind optnone uwtable "correctly-rounded-divide-sqrt-fp-math"="false" "disable-tail-calls"="false" "less-precise-fpmad"="false" "min-legal-vector-width"="0" "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"="+cx8,+fxsr,+mmx,+sse,+sse2,+x87" "unsafe-fp-math"="false" "use-soft-float"="false" }

!llvm.module.flags = !{!0}
!llvm.ident = !{!1}

!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{!"clang version 9.0.1-12 "}

对于库外函数 要使用就需要 declare 这个函数的签名

1
declare i32 @getint()

# 条件分支

1
2
3
4
5
6
7
8
9
10
11
12
13
//if.c
int main() {
    int a = getint();
    int b = getint();
    int c = 0;
    if (a == b) {
        c = 5;
    } else {
        c = 10;
    }
    putint(c);
    return 0;
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
define dso_local i32 @main() #0 {
  %1 = alloca i32, align 4
  %2 = alloca i32, align 4
  %3 = alloca i32, align 4
  %4 = alloca i32, align 4
  store i32 0, i32* %1, align 4
  %5 = call i32 (...) @getint()
  store i32 %5, i32* %2, align 4
  %6 = call i32 (...) @getint()
  store i32 %6, i32* %3, align 4
  store i32 0, i32* %4, align 4
  %7 = load i32, i32* %2, align 4
  %8 = load i32, i32* %3, align 4
  %9 = icmp eq i32 %7, %8
  br i1 %9, label %10, label %11

10:                                               ; preds = %0
  store i32 5, i32* %4, align 4
  br label %12

11:                                               ; preds = %0
  store i32 10, i32* %4, align 4
  br label %12

12:                                               ; preds = %11, %10
  %13 = load i32, i32* %4, align 4
  %14 = call i32 (i32, ...) bitcast (i32 (...)* @putint to i32 (i32, ...)*)(i32 %13)
  ret i32 0
}

语法 br + 标志位 + truelabel + falselabel

# cfg 图

1
2
3
4
5
6
7
8
9
10
int max(int a, int b) {
  if (a > b) {
    return a;
  } else {
    return b;
  }
}
int main(){
    return max(1,2);
}

opt -dot-cfg test.ll 生成

为了可视化

1
sudo apt-get install -y graphviz-doc libgraphviz-dev graphviz
1
dot .max.dot  -Tpng -o max.png

image-20221018103046162
如果在 clang 的时候用 O3 编译
这里就变成了选择函数

1
2
3
4
5
define dso_local i32 @max(i32, i32) local_unnamed_addr #0 {
  %3 = icmp sgt i32 %0, %1
  %4 = select i1 %3, i32 %0, i32 %1
  ret i32 %4
}

# SSA & phi node

这个概念在 ghidra，南大静态分析，各种文章里都看过了。。
SSA form enables and simplifies a vast number of compiler optimizations, and is the de-facto standard for intermediate representations in compilers of imperative programming languages.
看这个就行 SSA Explained

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
define i32 @max(i32 %a, i32 %b) {
entry:
  %0 = icmp sgt i32 %a, %b
  br i1 %0, label %btrue, label %bfalse

btrue:                                      ; preds = %2
  br label %end

bfalse:                                     ; preds = %2
  br label %end

end:                                     ; preds = %btrue, %bfalse
  %retval = phi i32 [%a, %btrue], [%b, %bfalse]
  ret i32 %retval
}

依据控制流分支选择变量。

1
llc -O0 -filetype=asm test.ll
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
# %bb.0:                                # %entry
	cmpl	%esi, %edi
	movl	%edi, -4(%rsp)          # 4-byte Spill
	movl	%esi, -8(%rsp)          # 4-byte Spill
	jle	.LBB0_2
# %bb.1:                                # %btrue
	movl	-4(%rsp), %eax          # 4-byte Reload
	movl	%eax, -12(%rsp)         # 4-byte Spill
	jmp	.LBB0_3
.LBB0_2:                                # %bfalse
	movl	-8(%rsp), %eax          # 4-byte Reload
	movl	%eax, -12(%rsp)         # 4-byte Spill
	jmp	.LBB0_3
.LBB0_3:                                # %end
	movl	-12(%rsp), %eax         # 4-byte Reload
	retq

汇编里是实现就是每一条分支都往 -12(%rsp) 上放置数据，然后 end 分支读取。