phasar LLVM静态分析框架介绍

一、简介

简介:PhASAR是一个基于LLVM的静态分析框架,用户可以进行数据流分析,框架会帮助进行指向分析、调用图构建。

安装:ubuntu下运行如下脚本。

# 安装
$ git clone https://github.com/secure-software-engineering/phasar.git
$ sudo ./bootstrap.sh
# 测试:若能获取正常输出,则表示安装正确。
$ phasar-llvm --module test/build_systems_tests/installation_tests/module.ll -D ifds-solvertest

架构:PhASAR调用LLVM API实现,可扩展性强,可以编写插件来增加功能。提供了命令行接口,也提供了封装的API来开发新工具。

[外链图片转存失败,源站可能有防盗链机制,建议将图片保存下来直接上传(img-WNYmbFvn-1595854253757)(./picture/1-phasar-architecture.png)]

二、实现

(1)IFDS分析

要解决IFDS(Inter-procedural Finite Distributive Subset)问题,开发者需要定义一个类型,实现FlowFunctions接口。

template <typename N, typename D, typename M> struct FlowFunctions{
    virtual ~FlowFunctions() = default;
    // 用于处理所有的过程间流
    virtual FlowFunction<D> *getnormalFlowFunction(N curr, N succ) = 0;
    // 用于处理调用站点(call-site)上的过程间流,通常,这个流函数工厂的任务是将位于给定调用站点的数据流映射到调用方法的作用域
    virtual FlowFunction<D> *getCallFlowFunction (N callStmt, M destMthd) = 0;
    // 处理函数退出时候的过程间流(如Return)。这将被调用方的返回值以及可能通过引用或指针参数离开函数的数据流事实映射回调用方的上下文/范围。
    virtual FlowFunction<D> *getRetFlowFunction (N callSite, M calleeMthd, N exitStmt, N retSite) = 0;
    // 沿调用站点传播调用中未涉及的所有数据流事实,通常是堆栈未被参数引用的本地数据。
    virtual FlowFunction<D> *getCallToRetFlowFunction (N callSite, N retSite, set<M> callees) = 0;
}

这些函数会由解决器(Solver)自动调用,根据过程间控制流图。

上面展示是函数工厂,他们会返回FlowFunction类型的小函数对象,而FlowFunction本身就是一个接口,开发人员必须提供一个适当的实现,其成员函数computeTargets接受类型为D的数据流值,并计算一组相同类型的新数据流事实。它指定了表示控制流的语句的二部图是如何构造的,并且回答了图中哪些边必须绘制这一问题。

template <typename D> struct FlowFunction{    virtual ~FlowFunction() = default;    virtual set<D> computeTargets(D source) = 0;}

很多常用的流函数,像GenGenIfKillKillAllIdentity已经被实现了,可以直接使用。

IDE分析

(过段时间更新

Monotone分析

(过段时间更新

内部函数和Libc函数调用的处理

LLVM目前有大约130个内部函数。这些函数用于描述分析和优化阶段的语义,没有实际的实现。稍后在编译器管道中,后端可以自由地用软件或硬件实现替换对内部函数的调用(如果目标体系结构存在这样的实现)。在LLVM中引入新的内部函数比引入新的指令更可取,因为在引入一条新指令时,必须重新审视构建在LLVM之上的所有优化、分析和工具,以使它们了解新指令。对内部函数的调用可以作为普通函数调用来处理。

LIBC标准库中包含的函数代表特殊的目标,这些函数实际上是所有实际的C和C++程序使用的。此外,标准库中包含的函数不能自己分析,因为它们大多是围绕系统调用的非常薄的包装器,通常无法用于分析。然而,在许多情况下,在执行数据流分析时不必分析这些函数。PhASAR将所有这些都建模为同一函数。分析开发人员可以通过使用特殊的摘要函数来更改默认行为并为不同的效果建模。SpecialSummaries类可用于注册除标识外的流和边缘函数。这个类知道所有的内在函数和libc函数。

三、phasar使用

3.1 使用示例——Example uses
(1)clang编译程序
// 示例程序 main.cpp
int main() {
    int i = 1;
    int j = 2;
    int k = i + j;
    return 0;
}

编译为LLVM IR:

$ clang++ -emit-llvm -S main.cpp

生成main.ll

; ModuleID = 'main.cpp'
source_filename = "main.cpp"
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

; Function Attrs: noinline norecurse nounwind optnone uwtable
define 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
  store i32 1, i32* %2, align 4
  store i32 2, i32* %3, align 4
  %5 = load i32, i32* %2, align 4
  %6 = load i32, i32* %3, align 4
  %7 = add nsw i32 %5, %6
  store i32 %7, i32* %4, align 4
  ret i32 0
}

attributes #0 = { noinline norecurse nounwind optnone 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" }

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

!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{!"clang version 5.0.1 (tags/RELEASE_501/final 332326)"}
(2)phasar分析与mem2reg优化

使用phasar进行IFDS分析时加-m选项,就会默认从main()第一条指令开始分析。

$ phasar-llvm -m path/to/your/main.ll -D ifds-solvertest

使用mem2reg pass来优化以上中间码,目标是尽可能多使用寄存器、少使用内存,少使用alloc:

$ opt -mem2reg -S main.ll
; ModuleID = 'main-mem2reg.ll'
source_filename = "main.cpp"
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

; Function Attrs: noinline norecurse nounwind uwtable
define dso_local i32 @main() #0 {
  %1 = add nsw i32 1, 2
  ret i32 0
}

attributes #0 = { noinline norecurse nounwind 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"="+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 8.0.0 "}
(3)复杂程序(函数调用)
int function(int x) {
	return x + 1;
}

int main() {
	int i = 42;
	int j = function(i);
	return 0;
}

IR(无mem2reg优化):

; ModuleID = 'main.cpp'
source_filename = "main.cpp"
target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
target triple = "x86_64-unknown-linux-gnu"

; Function Attrs: noinline nounwind optnone uwtable
define i32 @_Z8functioni(i32) #0 {
  %2 = alloca i32, align 4
  store i32 %0, i32* %2, align 4
  %3 = load i32, i32* %2, align 4
  %4 = add nsw i32 %3, 1
  ret i32 %4
}

; Function Attrs: noinline norecurse nounwind optnone uwtable
define i32 @main() #1 {
  %1 = alloca i32, align 4
  %2 = alloca i32, align 4
  %3 = alloca i32, align 4
  store i32 0, i32* %1, align 4
  store i32 42, i32* %2, align 4
  %4 = load i32, i32* %2, align 4
  %5 = call i32 @_Z8functioni(i32 %4)
  store i32 %5, i32* %3, align 4
  ret i32 0
}

attributes #0 = { noinline nounwind optnone 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 = { noinline norecurse nounwind optnone 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" }

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

!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{!"clang version 5.0.1 (tags/RELEASE_501/final 332326)"}

运行IFDSSolverTest分析,生成IFDS/IDE结果,和intra/inter monotone framework结果不同。

IFDS/IDE结果中,每个语句N处,都保存所有的data-flow facts D。

3.2 运行分析

使用phasar进行数据流分析

(1)选择phasar自带的分析
# 使用-D选项
$ phasar-llvm -m module.ll -D IFDSSolverTest 
$ phasar-llvm -m module.ll -D IFDSUninitializedVariables

如果未指定分析,则只生成调用图和其他数据结构。目前phasar支持的分析如下:

DataFlowAnalysisTypeParameter
DataFlowAnalysisType::IFDSConstAnalysis“IFDSConstAnalysis”
DataFlowAnalysisType::IFDSLinearConstantAnalysis“IFDSLinearConstantAnalysis”
DataFlowAnalysisType::IFDSSolverTest“IFDSSolverTest”
DataFlowAnalysisType::IFDSTaintAnalysis“IFDSTaintAnalysis”
DataFlowAnalysisType::IFDSTypeAnalysis“IFDSTypeAnalysis”
DataFlowAnalysisType::IFDSUninitializedVariables“IFDSUninitializedVariables”
DataFlowAnalysisType::IDELinearConstantAnalysis“IDELinearConstantAnalysis”
DataFlowAnalysisType::IDESolverTest“IDESolverTest”
DataFlowAnalysisType::IDETaintAnalysis“IDETaintAnalysis”
DataFlowAnalysisType::IDETypeStateAnalysis“IDETypeStateAnalysis”
DataFlowAnalysisType::IntraMonoFullConstantPropagation“IntraMonoFullConstantPropagation”
DataFlowAnalysisType::IntraMonoSolverTest“IntraMonoSolverTest”
DataFlowAnalysisType::InterMonoSolverTest“InterMonoSolverTest”
DataFlowAnalysisType::InterMonoTaintAnalysis“InterMonoTaintAnalysis”
DataFlowAnalysisType::Plugin“Plugin”
DataFlowAnalysisType::None“None”
(2)命令行接口
$ phasar-llvm --help 
PhASAR v0120
A LLVM-based static analysis framework

Allowed options:

Command-line options:
  -v [ --version ]                      Print PhASAR version
  -h [ --help ]                         Print help message
  --more-help                           Print more help
  -c [ --config ] arg                   Path to the configuration file, options
                                        can be specified as 'parameter = 
                                        option'
  -s [ --silent ]                       Suppress any non-result output

Configuration file options:
  -m [ --module ] arg                   Path to the module(s) under analysis
  -E [ --entry-points ] arg             Set the entry point(s) to be used
  -O [ --output ] arg (=results.json)   Filename for the results
  -D [ --data-flow-analysis ] arg       Set the analysis to be run
  --analysis-strategy arg (=WPA)
  --analysis-config arg                 Set the analysis's configuration (if 
                                        required)
  -P [ --pointer-analysis ] arg (=CFLAnders)
                                        Set the points-to analysis to be used 
                                        (CFLSteens, CFLAnders)
  -C [ --call-graph-analysis ] arg (=OTF)
                                        Set the call-graph algorithm to be used
                                        (NORESOLVE, CHA, RTA, DTA, VTA, OTF)
  -H [ --classhierarchy-analysis ]      Class-hierarchy analysis
  -S [ --statistical-analysis ]         Statistics
  -M [ --mwa ]                          Enable Modulewise-program analysis mode
  -R [ --printedgerec ]                 Print exploded-super-graph edge 
                                        recorder
  -L [ --log ]                          Enable logging
  --emit-ir                             Emit preprocessed and annotated IR of 
                                        analysis target
  --emit-raw-results                    Emit unprocessed/raw solver results
  --emit-text-report                    Emit textual report of solver results
  --emit-graphical-report               Emit graphical report of solver results
  --emit-esg-as-dot                     Emit the Exploded super-graph (ESG) as 
                                        DOT graph
  --emit-th-as-text                     Emit the type hierarchy as text
  --emit-th-as-dot                      Emit the type hierarchy as DOT graph
  --emit-cg-as-text                     Emit the call graph as text
  --emit-cg-as-dot                      Emit the call graph as DOT graph
  --emit-pta-as-text                    Emit the points-to information as text
  --emit-pta-as-dot                     Emit the points-to information as DOT 
                                        graph
  --right-to-ludicrous-speed            Uses ludicrous speed (shared memory 
                                        parallelism) whenever possible
  --analysis-plugin arg                 Analysis plugin(s) (absolute path to 
                                        the shared object file(s))
  --callgraph-plugin arg                ICFG plugin (absolute path to the 
                                        shared object file)
  -I [ --project-id ] arg (=default-phasar-project)
                                        Project Id used for the database
  -A [ --pamm-out ] arg (=PAMM_data.json)
                                        Filename for PAMM's gathered data
(3)运行分析
$ clang++ -emit-llvm -S main.cpp
$ phasar-llvm -m path/to/your/main.ll -D IFDSSolverTest

参考:

论文笔记:PhASAR: An Inter-procedural Static Analysis Framework for C/C++

PhASAR静态分析除零错误

Phasar-IFDS框架学习笔记

评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值