LLVM框架下的TriCore后端设计与实现

需积分: 9 195 浏览量更新于2024-07-09 收藏 752KB PDF 举报

"这篇文档是关于在LLVM编译框架中设计和实现一个针对TriCore架构后端的毕业论文。作者Christoph Erhardt在弗里德里希-亚历山大-埃尔兰根-纽伦堡大学的计算机科学系完成了这项工作，由Prof. Dr.-Ing. habil. Wolfgang Schröder-Preikschat和Dipl.-Inf. Fabian Scheler指导。论文始于2008年12月，完成于2009年9月。" 在本文中，作者探讨了如何为LLVM（Low-Level Virtual Machine）编译器框架开发一个专为TriCore架构设计的新后端。LLVM是一个开源的、模块化的编译器基础设施，它提供了编译器工具链的不同部分，如前端、优化器和后端，支持多种编程语言和处理器架构。TriCore是一种用于嵌入式系统的RISC（精简指令集计算机）架构，广泛应用于汽车电子和工业控制等领域。论文的核心内容可能包括以下几个方面： 1. **TriCore架构介绍**：首先，作者可能会详细解释TriCore架构的特点，包括其指令集体系结构（ISA）、寄存器布局、寻址模式以及处理器的工作原理。这对于理解如何为这种特定架构设计编译器后端至关重要。 2. **LLVM编译器框架概述**：论文会介绍LLVM的基本组件，如IR（Intermediate Representation，中间表示）和Pass（编译阶段）系统，以及它们如何支持代码生成和优化。作者可能会讨论LLVM的模块化设计如何使得添加新的后端变得更为便捷。 3. **设计新后端的挑战与策略**：这部分将详述在为LLVM添加TriCore支持时所遇到的技术挑战，例如如何将LLVM IR映射到TriCore的指令集，以及如何处理架构特定的特性。作者可能提出了有效的解决方案，如转换规则和编码策略。 4. **代码生成**：论文会详细介绍生成TriCore机器码的过程，这涉及到如何从LLVM IR生成目标代码，包括指令选择、寄存器分配和调度等步骤。这一部分可能包含具体的算法和技术。 5. **优化技术**：由于LLVM以其强大的优化能力而著称，作者很可能会讨论如何利用LLVM的优化工具链来提升针对TriCore的代码性能，包括指令级优化、循环展开、常量折叠等。 6. **实验与评估**：最后，论文可能包含了对新后端的实证评估，通过基准测试比较生成的代码效率，验证其性能是否达到预期，并与其他编译器进行对比。 7. **结论与未来工作**：作者会总结整个项目，讨论取得的成果、存在的问题以及可能的改进方向。这可能涉及到对其他嵌入式架构的支持扩展，或者对LLVM框架本身的贡献。这篇论文对于那些对LLVM编译器框架、嵌入式系统以及特定架构编译器设计感兴趣的人来说，是一个深入研究和学习的重要资源。它揭示了如何为非主流架构定制编译器后端，并展示了LLVM在面对多样化处理器需求时的灵活性。

2 The LLVM Compiler Infrastructure

Data type Width (in bits)

Integer 1, 8, 16, 32, 64, 128

IEEE-754 ﬂoating-point 32, 64, 80, 128

Boolean target-dependent

Pointer target-dependent

Integer vector various

Floating-point vector various

Array N/A

Structure N/A

Function N/A

Table 2.1: Data types of LLVM’s internal representation. Supported are mappings of all

primitive and composite types of low-level programming languages like C, and SIMD

(single instruction, multiple data) vector types.

must be mapped by the frontend. For instance, a C++ class with virtual methods can

be represented as a struct with a number of function pointers.

It must be noted that – depending on the source language – LLVM programs are not

guaranteed to be completely independent of the target machine if the language speciﬁca-

tion might be interpreted diﬀerently on diﬀerent platforms. For example, the C type

long

may be mapped to the LLVM type

i32

i64

by the frontend, according to whether

the code is being compiled for a 32-bit or a 64-bit target platform. Consequently, when

building an application written in such a language for a diﬀerent hardware architecture,

it is advised to rebuild it from source.

The instruction set is orthogonal and speciﬁes a virtual RISC-like register machine

with an unlimited amount of registers. Instructions are mostly in three-address form.

All LLVM programs are in static single assignment (SSA) form, i. e., for every use of

a virtual register there is exactly one reaching deﬁnition and that deﬁnition strictly

dominates the use. This makes data ﬂow analysis signiﬁcantly easier. The SSA-based

optimizer supports three classes of optimizations:

•

standard scalar optimizations (e. g., dead code elimination, constant propagation),

• aggressive loop transformations (e. g., unrolling), and

• advanced inter-procedural optimizations (e. g., inlining).

Figure 2.2 shows an LLVM code example for a function that computes the factorial of a

given integer. Note the obvious segmentation into basic blocks, the use of explicit

instructions instead of implicit fall-through edges, and the SSA

phi

instruction where

two reaching deﬁnitions from diﬀerent paths meet.

“LLVM’s virtual instruction set is a ﬁrst class language which has a textual, binary,

and in-memory representation” [

], so LLVM programs can be stored on disk either in a

human-readable assembly-like text form (

.ll

) or as compressed binary code (

.bc

) ﬁles.

Due to the project’s extremely modular architecture, this makes it possible to use the

middle-end as a standalone component.

2 The LLVM Compiler Infrastructure

At one stroke, LLVM suddenly supported the entire set of programming languages

oﬀered by GCC – including Ada, C, C++, Fortran, Java, Objective-C, and many

more.

LLVM-GCC can be used as a drop-in replacement for GCC without any adaptations

to the source code or build systems of existing applications

The main disadvantage is the fact that the GCC frontend is rather slow, memory-hungry,

and carries a lot of legacy code.

2.5.2 Clang

Clang is a relatively new eﬀort to create a modern frontend for C, C++ and Objective-C

from scratch. It aims to be fully language-conformant while providing compatibility to

GCC’s extensions. Clang has a range of beneﬁts over LLVM-GCC:

•

Diagnostics (errors/warnings) are much more descriptive and clear. Clang keeps

track of all column numbers and for each diagnostic outputs the respective line of

code with a marker placed below the responsible expression. Moreover, the original

information of

typedef

s is retained, eliminating the problem of literally “drowning”

in STL-related error messages.

•

Clang adds a huge performance boost to the compilation process. Measurements

have shown the frontend to be more than twice as fast as LLVM-GCC and to

consume signiﬁcantly less memory.

•

Just like the rest of LLVM, Clang is designed to be modular, extensible and easily

reusable (e. g., by source analysis tools or IDEs).

While the C and Objective-C parts are already in a reasonably mature state

, the C++

part is currently still a work in progress.

2.6 Code Generation

One of the points of criticism at GCC is that retargeting is rather unpleasant and

requires to more or less reinvent the wheel. This is where LLVM’s modular design pays

oﬀ once again. It provides a generic code generation framework that includes a set of

reusable target-independent algorithms for register allocation, instruction scheduling, etc.

This implicates a considerable simpliﬁcation and uniﬁcation of the retargeting process.

Chapter 4 discusses this process and the code generation framework in detail.

The LLVM 2.5 release bundles code generators for the Alpha, ARM, Cell SPU, Itanium,

MIPS, PIC16, PowerPC, SPARC, x86/x86-64, and XCore hardware architectures. The

This is true provided that the application in question does not make use of any builtins / pragmas /

inline assembly not (yet) implemented in LLVM.

As of July 2009, Clang is able to compile the complete FreeBSD kernel and most of the userland. The

Linux kernel however still has a few build issues.

剩余60页未读，继续阅读

jiankangshiye

粉丝: 73

LLVM框架下的TriCore后端设计与实现

英飞凌TC397数据手册详细解读与应用指南

英飞凌TC297安全管理单元代码紧急停止功能实现

TC265单片机CAN及CANFD通信编程示例

TriCore_TC162P_core_architecture_volume_1_of_2.pdf

Lcf_Tasking_Tricore_Tc.lsl

ADC_2.rar_Tricore_tricore dsp

esx-3xl_datenblatt_en.pdf

XC2268M_11.PDF

Appnote AURIX GTM Capture and Compare basics_CH.pdf

2014-05 Appnote AURIX TC27xC Startup And Initialization_CH.pdf

最新资源