Retargeting a C Compiler for a DSP Processor Master thesis performed in electronics systems by

Henrik Antelius LiTH-ISY-EX-3595-2004 Linköping 2004

Retargeting a C Compiler for a DSP Processor Master thesis in electronics systems at Linköping Institute of Technology by

Henrik Antelius LiTH-ISY-EX-3595-2004

Supervisors: Thomas Johansson Ulrik Lindblad Patrik Thalin Examiner:

Kent Palmkvist

Linköping, 2004-10-05

Avdelning, Institution Division, Department

Datum Date 2004-10-05

Institutionen för systemteknik 581 83 LINKÖPING Språk Language Svenska/Swedish X Engelska/English

Rapporttyp Report category Licentiatavhandling X Examensarbete C-uppsats D-uppsats

ISBN ISRN LITH-ISY-EX-3595-2004 Serietitel och serienummer Title of series, numbering

ISSN

Övrig rapport ____

URL för elektronisk version http://www.ep.liu.se/exjobb/isy/2004/3595/ Titel Title

Anpassning av en C-kompilator för kodgenerering till en DSP-processor Retargeting a C Compiler for a DSP Processor

Författare Author

Henrik Antelius

Sammanfattning Abstract The purpose of this thesis is to retarget a C compiler for a DSP processor. Developing a new compiler from scratch is a major task. Instead, modifying an existing compiler so that it generates code for another target is a common way to develop compilers for new processors. This is called retargeting. This thesis describes how this was done with the LCC C compiler for the Motorola DSP56002 processor.

Nyckelord Keyword retarget, compiler, LCC, DSP

Abstract The purpose of this thesis is to retarget a C compiler for a DSP processor. Developing a new compiler from scratch is a major task. Instead, modifying an existing compiler so that it generates code for another target is a common way to develop compilers for new processors. This is called retargeting. This thesis describes how this was done with the LCC C compiler for the Motorola DSP56002 processor.

Table of contents

1 Introduction 1.1 1.2 1.3 1.4

Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Purpose and goal. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The reader. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reading guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 DSP 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Motorola DSP56002. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Data buses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2 Address buses. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3 Data ALU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.4 Address generation unit . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.5 Program control unit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Instruction set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Assembly . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 Compilers

1 1 1 2 2

3 3 4 5 5 5 5 6 6 6

9

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.2 The analysis-synthesis model . . . . . . . . . . . . . . . . . . . . . . . . 9 3.3 Phases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.4 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 3.4.1 Lexical analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 3.4.2 Syntax analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 3.4.3 Semantic analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.5 Synthesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 3.5.1 Intermediate code generation . . . . . . . . . . . . . . . . . . . 15 3.5.2 Code optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 3.5.3 Code generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.6 Symbol table. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 3.7 Error handler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.8 Front and back end . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.9 Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 ix

3.9.1 Preprocessor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9.2 Assembler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9.3 Linker and loader. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.10 Compiler tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

4 LCC 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 The compiler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1 Lexical analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.2 Syntax analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.3 Semantic analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.4 Intermediate code generation . . . . . . . . . . . . . . . . . . . 4.3.5 Back end . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 Implementation 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 The compiler . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Data types and sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2 Register usage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 Memory usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.4 Frame layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.5 Calling convention. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.6 Naming convention . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Retargeting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.1 Configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.2 Declarations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.3 Rules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3.4 C code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Special features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5 Other changes to LCC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6 The environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.7 crt0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.8 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.8.1 Register targeting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.8.2 48-bit registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.8.3 Address registers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.9 Improvements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6 Conclusions

20 20 21 21

23 23 24 24 24 26 29 29 30

33 33 33 34 35 36 37 38 39 39 40 40 40 40 44 45 45 46 46 46 48 48 49

51

6.1 Retargeting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 6.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

References x

53

Table of contents

Appendix A: Instructions A.1 A.2 A.3 A.4 A.5 A.6

Arithmetic instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . Logical instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bit manipulation instructions . . . . . . . . . . . . . . . . . . . . . . . Loop instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Move instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Program control instructions . . . . . . . . . . . . . . . . . . . . . . .

Appendix B: Sample code

55 55 56 57 57 57 58

59

B.1 sample.c . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 B.2 sample.asm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

Appendix C: dsp56k.md

61

Index

79

xi

xii

1 Introduction 1.1 Background The division of Electronics Systems (ES) at the department of Electrical Engineering (ISY) at Linköping University (LiU) is currently running a project aiming at developing a DSP processor. The goal of this project is to make a DSP with a scalable structure that is instruction level compatible with the Motorola DSP56002 processor. The scalability refers to variable data word length and addition or removal of memories and instructions. The goal with scalability is to reduce the power consumption. Currently this project is nearly finished. In order to increase the usability of the DSP a C compiler is needed. It was decided that the best way to create a C compiler was to retarget an existing C compiler. Creating a compiler from scratch is a big undertaking that requires a lot of work. Retargeting a compiler is a relatively easy task compared to developing an entire compiler.

1.2 Purpose and goal The purpose of this thesis is to retarget a C compiler to the Motorola DSP56002 processor. The resulting compiler should from one or more C source files produce an executable file that can execute on the DSP. The only requirement on the compiler is that it should generate code

1

1.3 – The reader

that works correctly and functions as intended. There are no requirements on the performance or the size of the generated code. The compiler should also be compatible with Motorola’s C compiler and tools for the DSP56002. This makes it possible to mix generated code from the two compilers. It also means that the tools from Motorola can be used for the new compiler.

1.3 The reader It is assumed that the reader of this thesis has basic knowledge of the C programming language and some knowledge of assembly language. It is also assumed that the reader has a general knowledge of how processors work and what function a compiler has.

1.4 Reading guidelines This is a brief description of the chapters: • Chapter 1 contains an introduction and states the purpose of the thesis. • Chapter 2 describes how the DSP56002 processor works and how it can be used. • Chapter 3 contains general compiler theory that is needed to understand how a compiler works. • Chapter 4 describes the compiler LCC that was used in this thesis. • Chapter 5 describes the implementation and modifications that were done to LCC. • Chapter 6 lists the conclusions that were made and suggests further work.

2

2 DSP This chapter contains a description of how the Motorola DSP56002 processor works. This information is collected from [4].

2.1 Introduction Digital signal processing is, as the term suggests, the processing of signals by digital means. The signal is normally an electrical signal carried on a wire, but it can represent almost any kind of information and it can be processed in a wide variety of ways. Examples of digital signal processing include the following: • • • •

Filtering of signals. Convolution, which is the mixing of two signals. Correlation, which is the comparison of two signals. Rectification, amplification and transformation of a signal.

All of these tasks have earlier been performed by using analog circuits. Nowadays integrated circuits have enough processing power to perform these and many other functions. The devices performing these tasks are called digital signal processors, or DSPs. They are specialised microprocessors with architectures designed specifically for the types of operations required in digital signal processing. Like general-purpose microprocessors, the DSPs are programmable devices with its own native instruction set.

3

2.2 – Motorola DSP56002

DSPs can today be found in almost all electronic areas, such as mobile phones, personal computers, digital television decoders, surround receivers, and so on. The advantages of using a DSP instead of analog circuits are many. Generally, fewer components are needed, DSPs have higher noise immunity, it is easy to change the behaviour of a filter, filters with closer tolerances can be built, and so on. Also, since the DSP is a microcomputer, the same hardware design can be used in many different areas by simply changing the software for the DSP.

2.2 Motorola DSP56002 The Motorola DSP56002 is a general purpose DSP processor with a triple-bus Harvard architecture. This architecture can access multiple memories at the same time. It uses fixed-point arithmetic and has three function units; data arithmetic and logic unit (Data ALU), address generation unit (AGU) and program control unit (PCU). It does also have three memories, two for data (X and Y) and one for the program (P). A block diagram of the DSP56002 can bee seen in Figure 2.1.

Figure 2.1: Block diagram of the Motorola DSP56002 4

Chapter 2 – DSP

This architecture with multiple memories and buses makes it possible to, during one instruction cycle, make one computation in the data ALU while accessing the X and Y memories at the same time.

2.2.1 Data buses The data buses consists of four 24-bit wide buses called the Y data bus (y_dbus), the X data bus (x_dbus), the program data bus (p_dbus) and the global data bus (g_dbus). They are used for moving data between the function units and the memories. Data transfers between the data ALU and the X and Y memories occur over the X and Y data buses, respectively. All other data movements occur over the global data bus and instruction fetches occurs over the program data bus.

2.2.2 Address buses Addresses for the X data memory and the Y data memory are specified over the X address bus (x_abus) and the Y address bus (y_abus). Addresses for the program memory are specified over the P address bus (p_abus). All address buses are 16-bit wide.

2.2.3 Data ALU The data ALU performs all of the arithmetic and logical operations on the data. It uses a register set that consists of four 24-bit input registers, two 48-bit accumulator registers and two 8-bit accumulator extension registers. The input registers are called X0, X1, Y0 and Y1. They can also be combined into two 48-bit registers called X and Y. The two accumulators are called A and B and are 56 bits wide. Each consists of three concatenated registers, A2:A1:A0 and B2:B1:B0. The A2 and B2 are the 8-bit accumulator extension registers and they are used when more than 48-bit accuracy is needed. The input registers are used for operands to the instructions and the accumulator registers are used for both operands and the result from instructions.

2.2.4 Address generation unit The AGU performs all of the address storage and address calculations necessary to access the data in the memories. The AGU is divided into two identical halves, each of which has an address arithmetic unit that can generate one address each instruction cycle. The AGU has three sets of eight registers. They are the address registers R0 – R7, the offset 5

2.3 – Instruction set

registers N0 – N7 and the modifier registers M0 – M7. The R-registers are used for storing addresses that are used to address the memories. The N- and M-registers are used to update the R-registers in various ways. The registers are connected. So, for example, only N1 and M1 can be used to update R1.

2.2.5 Program control unit The PCU performs instruction prefetch, instruction decoding, hardware loop control and interrupt processing. It contains a 15-level system stack that is 32 bits wide and the following six registers: program counter (PC), loop address (LA), loop counter (LC), status register (SR), operating mode register (OMR) and stack pointer (SP).

2.3 Instruction set The instruction set can be seen in Appendix A on page 55. About half of the available instructions allow the use of parallel data moves.

2.4 Assembly The instruction syntax is organized in four columns; opcode, operands and two parallel move fields. An example of a typical assembly instruction can be seen here: Opcode MAC

Operands X0,Y0,A

XDB YDB X:(R0)+,X0 Y:(R4)+,Y0

The opcode column specifies the operation that should be performed. The operands column specifies which operands the opcode should use. The XDB and YDB columns specify optional data transfers over the X data bus and the Y data bus. The address space qualifiers X: and Y: indicate which memory is being referenced. This is an example of a small assembly program: var_a var_b

6

ORG dc dc

Y: 42 48

ORG MOVE MOVE

P:$40 Y:var_a,X0 Y:var_b,A

Chapter 2 – DSP

ADD MOVE

X0,A A,Y:var_a

This program simply adds the variables var_a and var_b and stores the result in var_a. This is a list of some of the features of the assembler that is used in this thesis: • Labels: If the first character on a line is not a space or a tab it is a label. Labels are used for variables and jump destinations. A colon is often used to end the label to increase readability of the assembly. • ORG: The ORG directive is used to indicate which memory the following statements belong to. It is also used for a lot of other memory related things. • OPT: The OPT directive is used to assign options to the assembler. • Variables: Variables are declared with a label and the DC directive to define a constant. • GLOBAL: The GLOBAL keyword is used to instruct the assembler that a variable is global. • Comments: Semicolon is used as a comment specifier. All characters to the right of the semicolon are ignored.

7

2.4 – Assembly

8

3 Compilers This chapter contains general compiler theory. Most of the information is collected from [1].

3.1 Introduction A compiler is a program that reads a program written in one language and translates it into an equivalent program in another language. An important part of this process is to report the presence of errors in the source program to the user. There exists thousands of different compilers for different source languages and target languages, and there also exists many different types of compilers. However, the basic principles of how the compilers work are the same. This chapter will discuss these basic principles.

3.2 The analysis-synthesis model There are two parts to compilation: analysis and synthesis. The analysis part breaks up the source program into consecutive pieces and creates an intermediate representation of the source program. The synthesis part constructs the desired target program from the intermediate representation.

9

3.3 – Phases

During analysis the operations stated in the source program are determined and recorded in a hierarchical structure called a tree. Often a special kind of tree called a syntax tree is used. In the synthesis part of the compilation the output is generated from the contents of the syntax tree. There is often also some sort of optimization of the generated source in this part.

3.3 Phases A compiler operates in phases, each of which transforms the source program from one representation to another. A typical decomposition of a compiler is shown in Figure 3.1. The following sections will discuss the different phases and how they are connected.

Figure 3.1: Phases of a compiler 10

Chapter 3 – Compilers

3.4 Analysis The analysis consists of three phases: lexical analysis, syntax analysis and semantic analysis.

3.4.1 Lexical analysis Lexical analysis, sometimes called scanning, is where the stream of characters that make up the source program is scanned left-to-right and transformed into groups of characters called tokens. For example, the characters in the statement result = start + rate * 60 would be transformed into the following tokens: 1. 2. 3. 4. 5. 6. 7.

The identifier result. The assignment symbol =. The identifier start. The plus sign. The identifier rate. The multiplication sign. The number 60.

The white space is normally eliminated during lexical analysis.

3.4.2 Syntax analysis Syntax analysis, or parsing, is where the tokens of the source program is grouped into grammatical phrases. Usually the phrases of the source program is represented by a parse tree. An example of a parse tree can be seen in Figure 3.2.

11

3.4 – Analysis

Figure 3.2: Parse tree for the statement result=start+rate*60

The phrase rate*60 will be grouped together because the rules of arithmetic expressions state that multiplication is performed before addition. Context free grammars The rules for the syntax analysis is often expressed by context free grammars. The grammar gives a precise and easy to understand specification of the syntax of the programming language. It is also possible to construct a parser from a grammar by using automated tools. For example, an if-else statement in C has the form: if ( expression ) statement else statement The statement is the concatenation of the keyword if, an opening parenthesis, an expression, a closing parenthesis, a statement, the keyword else, and another statement. Using the variable expr for expression and stmt for statement, this rule can be expressed as: stmt → if ( expr ) stmt else stmt The arrow may be read as “can have the form”. This kind of rule is called a production. In a production lexical elements like the keyword if and the parenthesis are called tokens. Variables like expr and stmt represent sequences of tokens and are called nonterminals. A context free grammar has four components: 1. A set of tokens, known as terminal symbols. 2. A set of nonterminals.

12

Chapter 3 – Compilers

3. A set of productions where each production consists of a nonterminal, an arrow, and a sequence of tokens and/or nonterminals. 4. A designation of one of the nonterminals as the start symbol. The following is an example of a simple grammar that can parse the right hand side of the assignment statement in Figure 3.2: expr → identifier expr → number expr → expr + expr | expr * expr The symbol | is used to separate multiple productions on one line and can be read as “or”. By using expr as the start symbol the derivation of the right hand side of the assignment statement could look like this: expr → expr + expr → identifier + expr → identifier + expr * expr → identifier + identifier * expr → identifier + identifier * number A grammar derives strings by beginning with the start symbol and repeatedly replacing a nonterminal by the right side of the production for that nonterminal. The set of token strings that can be derived from the start symbol form the language defined by the grammar. Syntax tree A more common internal representation of the syntactic structure is the syntax tree. It is a compressed representation of the parse tree where the operators appear as the nodes, and the operands of an operator are the children for that node. An example of a syntax tree is seen in Figure 3.3.

Figure 3.3: Syntax tree for the statement result=start+rate*60

13

3.4 – Analysis

3.4.3 Semantic analysis The semantic analysis phase checks the source program for semantic errors and gather type information for the code generation phase. It uses the hierarchical structure generated in the syntax analysis phase to identify the operators and operands of expressions and statements. This checking ensures that certain kinds of programming errors will be detected and reported. Examples of semantic checks can be: • Type checks: The compiler should report an error if an operator is applied to an incompatible operand. For example, if an integer variable is added to a function. It can also check that parameters to functions are correct in type and number. • Flow-of-control checks: Statements that causes the flow of control to leave a construct must have some place to which to transfer the flow of control. For example, a break statement in C causes the flow of control to leave the enclosing while, for or switch statement. If break is used outside of one of those an error is generated. • Uniqueness checks: Sometimes an object can only be defined once. For example, the case labels in a switch statement in C must be unique, and variables with the same name in the same scope is not permitted. • Name related checks: Sometimes the same name must appear at multiple locations. For example, in ADA a loop or block may have a name that appears at the beginning and at the end of the construct. There are many more different types of checks that can be needed to be performed depending on the language. In C for example, functions and variables must be declared before they are used, something that is not necessary in some languages. The type checking does not always have to result in an error. For example, a type mismatch can sometimes be resolved by converting the operand. If a in the statement a = a * 2; is a floating point number, the integer 2 must be converted to a floating point number before the multiplication can take place. This is accomplished by inserting a new node that explicitly converts an integer to a floating point number in the syntax tree.

14

Chapter 3 – Compilers

Since programming languages are so different and the semantic checks needed by the languages are so different there is no systematic way perform the semantic checks. It is usually done by traversing the tree and examining the nodes or during the syntax analysis phase.

3.5 Synthesis The synthesis consists of the three phases intermediate code generation, code optimizer and code generator. It is responsible for transforming the source that is now in the form of a syntax tree to the output language.

3.5.1 Intermediate code generation After the syntax and semantic analysis some compilers generate a machine independent intermediate form of the source program. Although the source program can be translated directly to the target language from the syntax tree, there are some benefits of using an intermediate form: • A machine independent optimizer can be used on the intermediate representation. • Retargeting is made easier. Creating a compiler for a different machine can be done by replacing a smaller part of the compiler than would have otherwise been necessary. The intermediate representation should have two important properties; it should be easy to generate and it should be easy to transform into the target program. A common way to solve this is to use a so called three-address code. It is very similar to an assembly language where each memory location can be used as a register. The code consists of a sequence of instructions, each of which can have at most three operands. For example, the assignment statement from Figure 3.2 might look like this: temp1 = rate * 60 temp2 = temp1 + start result = temp2 There are also statements for conditional and unconditional jumps, procedure calls, return statements, indexed assignment to be used on arrays, and address and pointer assignments. The instruction set of three-address codes must be large enough to implement the operations in the source language, but a smaller 15

3.5 – Synthesis

instruction set is easier to implement and retarget. However, if it is too small the intermediate code generator can be forced to generate long sequences of statements for some source language operations. It will then be more difficult for the optimizer and the code generator to produce good code.

3.5.2 Code optimization The code optimizer will attempt to improve the intermediate code so that faster running machine code will be generated. It can sometimes also be of interest to make the code smaller. For DSP processors code with lower power consumption is sometimes preferred. There are two types of optimizations that can be done; machine independent and machine dependent. Machine independent optimizations are typically done using the intermediate form as the base and does not consider any details of the target architecture when making optimization decisions. It is often very general in nature. Machine dependent optimizations can be done both on the intermediate form and the generated code. These optimizations consider the target architecture specifically and uses special instructions such as hardware loops and so on. There are a number of common optimization techniques. Constant propagation Constant propagation is simply the replacement of variable references with constant references when possible. For example, the statement a = 3; function_call(a + 42); becomes function_call(3 + 42); Constant folding Expressions with constant operands can be calculated at compile time. The example above would be transformed to function_call(45); Programmers usually do not write expressions such as 3+42 directly, but these expressions are quite common after macro expansion and other optimizations such as constant propagation.

16

Chapter 3 – Compilers

Common subexpression elimination A common subexpression, or CSE, is created when two or more expressions compute the same value. The expression is calculated once to a temporary variable that is used instead of the CSE. For example, the statement array1[i + 1] = array2[i + 1]; will be transformed to temp1 = i + 1; array1[temp1] = array2[temp1]; Dead code elimination Code that is never reached or that does not affect the program can be eliminated. For example, this code fragment int global; void foo(void){ int k = 1; global = 1; global = 2; } will transform into the following int global; void foo(void){ global = 2; } Expression simplification Some expressions can be simplified by replacing them with a more efficient expression. For example, i+0 will be replaced by i, i*0 and i-i by 0, and so on. Code motion Expressions in a loop that gives the same result each time the loop is iterated can be moved outside the loop and calculated only once before entering the loop.

17

3.6 – Symbol table

Strength reduction Strength reduction replaces expensive instructions with less expensive instructions. For instance, a popular strength reduction is to replace a multiplication by a constant power of two with a left shift.

3.5.3 Code generation The final phase of the compiler is the generation of target code. The target code is usually relocatable machine code or assembly code. Memory locations are selected for each of the variables used in the source program and the intermediate instructions are translated into one or more assembly level instructions that perform the same task. A vital part of code generation is the assignment of registers to variables, since that can greatly affect the performance of the generated code. Using the example from the previous sections, the generated code might look like this MOVE MUL MOV ADD MOVE

rate, R1 #60, R1 start, R2 R1, R2 R2, result

3.6 Symbol table An essential part of the compiler is to keep track of the identifiers used in the source program and to collect information about various attributes of each identifier. These attributes contains information about the name and type of the identifier, its size, scope and so on. For functions and procedures it also contains the number and types of its arguments and the return type. In a similar way it works for more complex data types like arrays and structures. The symbol table is a data structure that contains a record for each identifier and fields for the attributes of the identifier. The data structure makes it possible to search for identifiers and add or retrieve the attributes and to add new identifiers. When an identifier is found in the lexical analysis its name is added to the symbol table if its not already there. The index in the symbol table is then passed along in the token and that index is used to refer to the identifier from there on. In the later phases of compilation information

18

Chapter 3 – Compilers

about the type and other attributes are added and is used in various ways.

3.7 Error handler It is important that the compiler can detect errors and deal with them in a reasonable way. When an error is encountered the compiler emits an error message containing the location of the error in the source program and a message stating the type of error and it then tries to continue with the compilation. It can sometimes be difficult for the compiler to know what to do to continue when an error has been detected. One way is to, for example, skip all input until the next semicolon to get to the next statement. As soon as the error count is greater than zero a flag is set and the compiler will stop execution after the semantic analyzer phase. There is no point in generating the target program when there exists errors in the source program. The compiler can also detect minor errors that will not stop the compilation and emit warnings about these errors instead.

3.8 Front and back end Often the phases are collected into a front end and a back end. The front end consists of the phases that depend on source language and are largely independent of the target machine. These normally include lexical analysis, syntax analysis, semantic analysis and the generation of intermediate code. The machine independent optimizations can also be done in the front end. The creation of the symbol table and most of the error handling is also done in the front end. The back end includes the parts of the compiler that are dependant on the target machine, and these parts usually does not depend on the source language, only the intermediate code. The back end therefore consists of the code optimizer and the code generator. It also uses the symbol table and error handler. This division of the design makes it easy to take the front end of a compiler and combine it with a new back end to produce a compiler for the same source language to a different target machine. This is called retargeting.

19

3.9 – Environment

3.9 Environment In addition to the compiler, several other programs are required if an executable program is to be created. See Figure 3.4.

Figure 3.4: A compiler system

3.9.1 Preprocessor The preprocessor produce the input to the compiler. It often performs different kinds of text processing, for example macro processing and file inclusion. In C for example, all lines beginning with a # is an instruction to the preprocessor. #define FAIL -1 causes all occurrences of FAIL to be replaced by -1, and #include will include the file file.h in the source program. In C the preprocessor also removes the comments from the source program.

3.9.2 Assembler Some compilers produce assembly code, and that must be passed to an assembler for further processing. The assembler works much like a compiler and translates the assembly source to relocatable machine code. 20

Chapter 3 – Compilers

3.9.3 Linker and loader The linker makes it possible to combine several relocatable machine code files into a single program. The different machine code files can be the result from several compilations, and some may be library files. The linker resolves external references in the input files so that data and functions from the different files can be used by each other. When all the external references are resolved the loader takes the relocatable machine code and alters all relocatable addresses to real addresses and places the code and the data in its proper locations and creates the output file.

3.10 Compiler tools Since most compilers use the same structure and function in the same way, specialized tools have been developed that helps implement the various components of the compiler. These tools use specialized languages for specifying and implementing the components, and many use algorithms that are quite sophisticated. The following is a list of some compiler construction tools: • Scanner generators: These automatically generate lexical analyzers. Usually from a specification based on regular expressions. Examples include flex and lex. • Parser generators: These produce syntax analyzers from specifications that is normally based on a context free grammar. Before the parser generators appeared, the parser was the most time consuming part to implement. Now it is considered to be one of the easiest to implement thanks to the parser generators. Examples of parser generators are yacc and bison. • Syntax-directed translation engines: These produce routines that walk the parse tree and generates intermediate code. • Automatic code generators: These tools generates routines that translates the intermediate language into the machine language for the target machine by the help of a collection of rules. The basic technique is template matching. The intermediate code statements are replaced by templates that represent sequences of machine instructions.

21

3.10 – Compiler tools

22

4 LCC This chapter describes how the compiler LCC works. Most of this information is collected from [2] and [3].

4.1 Introduction LCC is a free ANSI C compiler that is designed to be retargetable. The source code is available for download from the internet [6] under a license [7] that imposes almost no restrictions at all. This compiler was chosen because it is very small and simple. It is designed in a way so that it is easy to retarget it to generate code for other processors. There is also excellent documentation of LCC in the form of a book that describes every detail of the implementation of the entire compiler. It is called “A Retargetable C Compiler: Design and Implementation” and it was used extensively during this thesis. The thesis could probably not have been completed without the book. Another compiler candidate was the GNU C Compiler, or GCC, from the GNU Compiler Collection, which is an open source C compiler. GCC would probably have generated better and faster code, but it was not chosen because it is much bigger and more complex than LCC. Also, the same kind of documentation that was available for LCC was not available for GCC.

23

4.2 – C

4.2 C C is a general purpose programming language that was developed during the 1970’s by Brian Kernighan and Dennis Ritchie and it is still widely used today. It is a relatively low level language where the basic data types in the language correspond to real data types found in the hardware. The language provides no operations to deal directly with composite data types, such as strings, arrays, lists and so on. There are no input/output facilities and no file access facilities. All these higher level operations must be provided by library functions. This, and several other limitations, has some advantages. It makes the language small and relatively easy to learn. It does also mean that compilers for the language will be smaller and easier to construct. C has become very popular and there exists compilers for many different processors and operating systems. Although it is far from an ideal language for DSP processors, it is still extensively used for them. That is probably because it is such a simple and low level language, which makes it easier to construct a compiler that generates efficient code for the DSP processors. Over the years the C programming language has evolved and been standardized a couple of times. The first version, called K&R C (from Kernighan and Ritchie), is derived from the reference manual in the first edition of the book “The C programming language” by Brian Kernighan and Dennis Ritchie. In 1989 ANSI standardized the language, and it is commonly referred to as ANSI C or C89. ISO has released two standards for C, and they are called ISO C90 and ISO C99.

4.3 The compiler The following sections will describe the different phases of the compiler and how they work.

4.3.1 Lexical analysis The lexical analyzer reads source text and produces tokens. For each token the lexical analyzer returns its token code and zero or more associated values. The token codes for single character tokens, for example = and +, are the characters themselves. For tokens that can consist of one or more characters, for example identifiers and constants, defined constants are used. For example, the expression ptr = 42 results in the following token stream 24

Chapter 4 – LCC

ID '=' ICON

"ptr"

symbol table entry for ptr

"42"

symbol table entry for 42

The token code for the operator = is the numeric value of =, and it does not have any associated values. The token code for the identifier ptr is the value of the constant ID, and the associated values are the identifier string itself and a pointer to the symbol table entry for the identifier. The integer constant 42 returns the token ICON and the associated values "42" and a pointer to the symbol table. Keywords, such as for and switch, have their own token codes to distinguish them from identifiers. The lexical analyzer also tracks the source coordinates for each token. These coordinates contains the file name, line number and position on the line of the first character of the token. The coordinates are used to locate errors when they are found. Recognizing tokens The lexical analyzer in LCC is written by hand, it is not generated by a tool. This is due to the fact that the lexical structure in C is simple and that generated analyzers tend to be large and slow. The lexical analyzer is used by calling the function gettok(), which returns the next token. The gettok() function recognizes a token by using a switch statement on the first character in the token to classify it. It then consumes the following characters that make up the token. The following is a small sample of the code ... switch (*rcp++) { ... case '<': if (*rcp == '=') return cp++, LEQ; if (*rcp == '<') return cp++, LSHIFT; return '<'; ... rcp and cp are pointers to the next character in the input file. The code for identifying most of the tokens looks very similar to the example, but identifying numbers, strings and identifiers is a bit harder. However, it works in the same way by looking ahead in the input stream.

25

4.3 – The compiler

4.3.2 Syntax analysis The syntax analyzer, or parser, uses the stream of tokens from the lexical analyzer and confirms that it follows the syntax of the language. It also builds an internal representation of the input that is used by the rest of the compiler. The parser for LCC is also written by hand. The reason for this the same as for the lexical analyzer; C is a simple language and the code generated by tools is slow and big. Grammar LCC uses a context free grammar written in EBNF form to define the rules for the parser. The parser is constructed by writing a parsing function for each nonterminal. The idea is to write a function X() for each nonterminal X, using the productions for X as a guide to writing the code for X(). For example, the parsing function for the following production expr → term { + term } will look like void expr(void){ term(); while(t == '+'){ t = gettok(); term(); } } The { and } in the production is an EBNF feature that means “zero or more”. Abstract syntax tree When parsing the program the compiler also generates an intermediate representation of the program. This is done in the form of abstract syntax trees, or simply trees. Abstract syntax trees are parse trees without the nodes for nonterminals and nodes for useless terminals. For example, the tree for the expression (a+b)*c can be seen in Figure 4.1.

26

Chapter 4 – LCC

Figure 4.1: Tree for the expression (a+b)*c

There are no nodes for the nonterminals used when parsing this expression, and there are no nodes for the tokens ( and ). The tokens + and * are contained in the nodes ADD+I and MUL+I. The nodes with the operator ADDRG+P compute the address of the operand and INDIR+I fetches integers at the address given by their operand. The name of the nodes are constructed by an operator and a type suffix that denotes the type that the operator operates on. For example, the node ADD+I states that the node uses integer addition. Table 4.1 lists the different type suffixes available. Type suffix

Meaning

F

Floating point

I

Integer

U

Unsigned

P

Pointer

V

Void

B

Structure Table 4.1: Type suffixes

The trees can contain operators that do not appear in the source program. For example, the INDIR+I node fetches integers at an address, but there is no fetch operator in C. A list of operators that can appear in the trees is seen in Table 4.2. In addition to these, there are six more operators that are used in trees listed in Table 4.3.

27

4.3 – The compiler Operator

Type suffix

Operation

ADDRF ADDRG ADDRL CNST BCOM CVF CVI CVP CVU INDIR NEG ADD BAND BOR BXOR DIV LSH MOD MUL RSH SUB ASGN EQ GE GT LE LT NE ARG CALL RET JUMP LABEL

...P.. ...P.. ...P.. FIUP.. .IU... FI.... FIU... ..U... .IUP.. FIUP.B FI.... FIUP.. .IU... .IU... .IU... FIU... .IU... .IU... FIU... .IU... FIUP.. FIUP.B FIU... FIU... FIU... FIU... FIU... FIU... FIUP.B FIUPVB FIUPV. ....V. ....V.

Address of a parameter Address of a global Address of a local Constant Bitwise complement Convert from float Convert from signed integer Convert from pointer Convert from unsigned integer Fetch Negation Addition Bitwise AND Bitwise inclusive OR Bitwise exclusive OR Division Left shift Modulus Multiplication Right shift Subtraction Assignment Jump if equal Jump if greater than or equal Jump if greater than Jump if less than or equal Jump if less than Jump if not equal Argument Function call Function return Unconditional jump Label definition

Table 4.2: Node operators

28

Chapter 4 – LCC Operator

Operation

AND

Logical AND

OR

Logical OR

NOT

Logical NOT

COND

Conditional expression

RIGHT

Composition

FIELD

Bit-field access

Table 4.3: Tree operators

4.3.3 Semantic analysis The semantic analysis of the source program is done when the parser recognizes the input, so there is therefore no explicit phase in the compilation where this is done. Each parsing function detects and handles the semantic errors according to the semantics of each construct. When for example a type conversion is needed an extra convert node is inserted in the abstract syntax tree, and the expression x = 6 generates an error if x is not defined. There are a lot of other semantic checks that are also being done.

4.3.4 Intermediate code generation During this stage the compiler produces directed acyclic graphs, or dags, from the trees. The compiler also eliminates common subexpressions. For example, in the expression (a+b)+b*(a+b) the value of a+b is computed twice. The dag for this expression can be seen in Figure 4.2. The multiplication node (MULI4) uses the already computed values for a+b and b instead of computing them again. The names of the nodes in dags are made up of a generic operator, a type suffix and a size indicator. The + is omitted to distinguish dags from trees. For example, ADDI4 denotes a 4-byte (32-bit) integer addition.

Figure 4.2: The dag for (a+b)+b*(a+b) 29

4.3 – The compiler

Trees contain operators that are not allowed for dags. The available operators for dags are seen in Table 4.2. When the dags are constructed the operators that are not allowed are replaced by other operators instead. For example, the operator AND is replaced by a comparison and jumps and labels. Before the dags are passed to the back end they may be converted to trees again. Some back ends wants trees and some wants dags. All back ends that are included in the LCC distribution wants trees. When the conversion is done nodes that are referenced multiple times because of the common sub expression optimization are changed. The result of the common subexpression is stored in a temporary variable that is used instead. The resulting tree is still using the same data structures and representation as the dags though.

4.3.5 Back end LCC's back end is divided in a machine independent part and in a machine dependent part. The front end communicates with the back end by calling a number of interface functions. In a C program, all program code is contained in functions. To generate code for a function the front end calls the interface function function(). function() uses two functions to generate code, gencode() and emitcode(). gencode() selects and orders instructions and allocates registers. emitcode() emits the assembler code for the function and also removes unnecessary register to register copies. These register to register copies are left over from earlier optimizations and it is easier to remove them here. Selecting instructions The instruction selection is done in the function gencode(). The instruction selectors used by LCC are generated automatically from a specification by a program called lburg. lburg is a code generator generator and it emits a tree parser written in C. The core of an lburg specification is a tree grammar, which is a list of rules where each rule has a nonterminal on the left and a pattern of terminals and nonterminals on the right. For example, the rule addr: ADDI4(reg, con) matches a tree at an ADDI4 node if the node’s first child recursively matches the nonterminal reg and the second child recursively matches

30

Chapter 4 – LCC

the nonterminal con. In Figure 4.3 the tree with the selected rules for the statement i = c + 2 can be seen.

Figure 4.3: Tree with rules

Tree grammars are usually ambiguous, which means that there can be more than one selection of instructions that do the same thing. For example, increasing a register by one can be done by adding one to the register directly or by loading one into another register and adding the two registers. The cheapest implementation is preferred, so a cost is assigned to each rule and the parse tree with the lowest total cost is selected. Specifications lburg specifications uses the following format %{ configuration %} declarations %% rules %% C code The configuration part is C code and is optional. It is copied directly into the generated file. The same applies to the C code part. The declarations part contains the start symbol and a list of all the terminals. The rules part contains tree patterns. Each rule has an assembler code template, which is a quoted string that specifies what to emit when the rule is used. Rules end with an optional cost. The following is an example of a simple specification

31

4.3 – The compiler

%start stmt %term ADDI4=309 ADDRLP1=295 ASGNI4=53 %term CNSTI4=21 INDIRI4=67 %% con: CNSTI4 "1" addr: ADDRLP1 "2" addr: ADDI4(reg, con) "3" rc: con "4" rc: reg "5" reg: ADDI4(reg, rc) "6" 1 reg: addr "7" 1 stmt: ASGNI4(addr, reg) "8" 1 In this example the assembler code templates are simply rule numbers. Rule 1 states that con matches constants. Rule 2 and 3 states that addr matches trees that can be computed by address calculations, like an ADDRLP1 or the sum of a register and a constant. rc matches a constant or a reg, and reg matches any tree that can be computed into a register. Rule 6 describes an add instruction. The first operand must be in a register and the second operand can be a register or a constant. The result is stored in a register. Rule 7 describes an instruction that loads an address into a register. Rule 8 describes an instruction that stores a register at an address. The emitter The emitter in the function emitcode() is what outputs the assembler code from the assembler templates. Each rule has one assembler template. If the template ends with a newline character, lburg assumes that it is an instruction, otherwise it is assumed to be a piece of an instruction. When the emitter emits the template it treats some characters differently. %digit tells the emitter to emit the digit-th nonterminal from the pattern. %c emits the nonterminal on the left side of the production. For example, the rule areg: ADDI4(reg, rc) "add %c,%0,%1" might be emitted as add a1,r1,#60 If the template begins with #, emit2() is called to emit the instruction. This is needed to deal with tricky features in some assemblers.

32

5 Implementation 5.1 Introduction The main goal of this thesis was the design and implementation of a new back end to the LCC compiler for the DSP56002 processor. One other goal was to maintain compatibility with Motorola’s C compiler, so that the generated code would behave in the same way. This means that the two compilers use the registers in the same way, uses the same memory layout, uses the same calling convention, and so on. By doing this, code generated by Motorola’s compiler can use code compiled by this compiler, libraries for example, and vice versa. LCC is designed so that retargeting should be as easy as possible, and the included backs ends only consist of about 1000 lines of code each. This chapter will describe how the back end was constructed and why it looks and behaves as it does.

5.2 The compiler The DSP56002 digital signal processor is designed to execute DSP oriented calculations as fast as possible. As a consequence, it has an architecture that is somewhat unconventional for the C language. Because of this there are characteristics of the compiler and the generated code that are a bit unusual and will be documented here. Since this compiler should be compatible with Motorola’s compiler this section is based on information from [5]. 33

5.2 – The compiler

5.2.1 Data types and sizes Because of the word orientation of the DSP56002, all data types are aligned on word boundaries. One word is 24-bit wide. Integer data types The sizes and ranges of the integer data types is defined in Table 5.1. Data type

Size (words)

Min value

Max value

char

1

–8388608

8388607

unsigned char

1

0

0xFFFFFF

short

1

–8388608

8388607

unsigned short

1

0

0xFFFFFF

int

1

–8388608

8388607

unsigned int

1

0

0xFFFFFF

long

2

–140737488355328

140737488355327

unsigned long

2

0

0xFFFFFFFFFFFF

long long

2

–140737488355328

140737488355327

0

0xFFFFFFFFFFFF

unsigned long long 2

Table 5.1: Integer data type sizes and ranges

Floating point types The C data types float and double are implemented as fractional numbers as used by the DSP56002 processor. The precision and range can be seen in Table 5.2. This is not consistent with the Motorola compiler. It uses single precision floating point arithmetic for both float and double. However, since the DSP56002 can not do floating point arithmetic in hardware, all operations are performed by calls to an external library. The choice to implement floating point numbers as fractional numbers was made because that is what the hardware supports. Emulating “real” floating point numbers defeats the purpose of using a DSP processor in the first place since it will then be slower than a normal processor. It was also easy to implement since the only thing that is different between integers and fractional numbers is the way multiplication and division is handled. There is one problem though, since the hardware does not support 48-bit multiplication and division the compiler uses the integer version of those operations for the double

34

Chapter 5 – Implementation

data type. This pretty much makes this data type useless, but it is implemented anyway. Data type

Precision (bits)

Range

float

24

– 1.0 ≤ x < 1.0

double

48

– 1.0 ≤ x < 1.0

Table 5.2: Floating point data type precision and ranges

Pointer types All pointers are 16-bit. When computing addresses with integer arithmetic only the least significant 16 bits are relevant. See Table 5.3. Data type

Size (words)

Min value

Max value

pointers

1

0

0xFFFF

Table 5.3: Pointer size and range

5.2.2 Register usage The compiler uses all of the registers in the DSP56000 processor except the M-registers. The register usage can be seen in Table 5.4. Register

Usage

R0

Frame pointer (16-bit)

R6

Stack pointer (16-bit)

R1 – R5, R7

Address registers used for pointers and structures. (16bit)

N0 – N7

Compiler temporary. Used when updating the R-registers. (16-bit)

M0 – M7

Unused. Must be kept as 0xFFFF. (16-bit)

A

48-bit general purpose register and 48-bit function return value.

A1

24-bit general purpose register and 24-bit and 16-bit function return value.

B

48-bit general purpose register.

B1

24-bit general purpose register. Table 5.4: Register usage

35

5.2 – The compiler Register

Usage

X, Y

48-bit general purpose registers.

X0, X1, Y0, Y1

24-bit general purpose registers.

Table 5.4: Register usage

5.2.3 Memory usage Due to the architecture of the DSP56002 program and data memory are separate. The program resides in the P-memory and all data are stored in the Y-memory. Figure 5.1 illustrates the default program and data memory layout.

Figure 5.1: Program and data memory layout

The bottom of the program memory contains the interrupt table and it is filled with jumps to the subroutine Fabort, except at the first position that contains a jump to the subroutine F__start. This is because the DSP processor starts execution here. The F__start function takes care of initialization and calls the Fmain subroutine, which is the compiled main() function. The rest of the program memory is used to store the compiled functions. The data memory is split in three parts. The bottom part contains data defined in the crt0 file. The next part is used for global and static data, and the final part is used for the runt-time stack and the heap. 36

Chapter 5 – Implementation

5.2.4 Frame layout An activation record, or frame, holds all the state information needed for one invocation of a function. This includes local and temporary variables, saved registers and return address. The stack stores one frame for each active function. When a function is called it puts a new copy of the frame on the stack, and when the function returns it removes the frame. The frame pointer points into the currently active frame. It is used to access data inside the frame. The stack pointer points to the first free space on the stack and it grows upwards. Figure 5.2 shows the layout of a frame.

Figure 5.2: Frame layout

An example of what the stack looks like during a function call can be seen in Figure 5.3. The code that is executed looks like this: void main(void) { func(); } When the execution point reaches the call to func() there is a frame on the stack for the function main(). A new frame is built and 37

5.2 – The compiler

pushed onto the stack and the frame pointer is updated. When the function returns, it’s frame is removed from the stack and the frame pointer is restored.

Figure 5.3: Frame example

5.2.5 Calling convention Whenever a function is called, a strict calling convention is followed. The calling sequence is divided in three parts, the caller sequence, the callee sequence and the return sequence. Caller sequence The caller part of the calling sequence consists of: 1. Pushing the arguments onto the stack in reverse order. 2. Calling the function. 3. Adjust the stack pointer. Callee sequence During the initial part of the calling sequence, the called function is responsible for: 1. Saving the return address and the old frame pointer. 2. Updating the frame and stack pointers. 3. Saving the following registers if they are used by the function: B0, B1, X0, X1, Y0, Y1, R1 – R5 and R7. Return sequence During the final part of the calling sequence, the called function is responsible for: 1. Placing the return value in register A.

38

Chapter 5 – Implementation

2. Testing the return value. This feature is not used by this compiler, but it is needed to maintain compatibility with the calling convention used by Motorola.

5.2.6 Naming convention The compiler uses a special naming format when generating assembly code. This can be seen in Table 5.5. Label

Purpose

L#

Local labels. Used for targets of jumps. # is a unique number.

F

Global variables and functions. is the variable or function name.

F__#

Variables static to a function.



Section names. The contents of each assembly file generated by the compiler are contained in a unique section. is the file currently being compiled where the '.' is replaced by '_'.

Table 5.5: Naming convention

5.3 Retargeting The back end in LCC is split in two parts, a machine independent and a machine dependant part. All back ends use the same machine independent part. This simplifies the construction of new back ends since it does not require as much new code. The different back ends in LCC are stored in .md files (md stands for machine description). They contain everything that is needed for the back ends. To create a new back end a new file is created. When LCC is being compiled the files for the back ends are fed through the program lburg. lburg generates C code from the .md files, and that is then compiled together with the rest of the source code for LCC to create the compiler. The format of the .md file is the same as of the specification mentioned in section 4.3.5 on page 30. The complete file for the new back end is called dsp56k.md and can be seen in Appendix C on page 61. The following sections will give a more detailed description of the file. The section names correspond to the names used in section 4.3.5. 39

5.3 – Retargeting

5.3.1 Configuration This part contains declarations and function definitions. They are copied to the top of the generated C file and is almost identical for all back ends. It is only the global variables that differ somewhat. The arrays ireg[32], ireg2[32] etc. are used to hold information about the registers. The variables iregw, iregw2 etc. are used to hold an entire register set, also called a wildcard. cseg is used to hold the current segment and retstruct is a flag that is set if the function being compiled returns a structure.

5.3.2 Declarations This part contains all the terminals that are used for the rules. They are generated by a program called ops that is included in the LCC distribution. The command line given to the program was the following: ops c=1 s=1 i=1 l=2 h=2 f=1 d=2 x=2 p=1 The letter c stands for char, s for short, and so on, covering all the data types in C. x means long double and p means pointers. The number indicates how many bytes (or words) each data type use. The output from the program is a list with all the terminals that can appear with the given data type sizes and it is copied directly into the .md file.

5.3.3 Rules This is the core of the back end, and the rules were the most time consuming part to write. They required a lot of testing and fine tuning. The rules were constructed incrementally. It started with a few basic rules that was only able to compile empty programs. Then new rules were gradually added to handle more operations, but only for one data type (1-word integers). When all operators were covered additional rules to take care of all data types were added. While writing the rules, the function emit2() was also constructed to take care of the cases where the instruction templates were not enough. This function is documented further down. The other back ends were used as an inspiration when writing this, but a lot is different because of the nature of the DSP56002.

5.3.4 C code This is the actual interface to the back end that the front end uses. It is made up of a number of interface functions and a structure called the 40

Chapter 5 – Implementation

interface record that contains configuration data. The front end calls the functions to inform the back end of things during the compilation. It also calls the functions to let the back end emit the actual assembly code. This part was created in the same way and at the same time as the rules were written. At first only a few functions are needed to compile an empty program. As more and more rules were created the functions needed to have more and more features. The following is a description of the functions and the interface record in the .md file. They are listed in the same order as they appear in the file. progbeg() The front end calls progbeg() during initialization to set up variables and initialize data and to emit the boilerplate at the beginning of the generated assembly file. It is also responsible for checking and taking care of the command line arguments passed to the back end. progend() When the compilation ends the front end calls progend to give the back end a chance to clean up and finalize its output. rmap() This function is used to tell the front end which register class the operators should use. P and B is pointers and structures, they use the Rregisters. I, U and F are integers, unsigned integers and floating point numbers. They use the same registers; X and Y if they are 48-bit (2 words) and X0, X1, Y0 and Y1 if they are 24-bit (1 word). segment() The front end tells the back end which segment it should use, CODE, BSS, DATA or LIT. CODE is used for code, BSS for uninitialized variables, DATA for initialized variables and LIT for constants. For this implementation CODE uses the P memory and all the data uses the Y memory. target() This function is used because some instructions must have their operands in certain registers or they leave the result in a specific register. In this implementation there are a lot of instructions that needs to be tar41

5.3 – Retargeting

geted because of the hardware. Most of instructions in the DSP56002 use the X and Y registers as input registers and the accumulator registers A and B as output registers. Therefore almost all instructions must be forced to use specific registers. The function rtarget() is used to specify a register for the operand and setreg() is used to set the register that will contain the result from an instruction. clobber() Some instructions destroy the value in a register. This function tells the front end to insert instructions to save and restore the register before and after an instruction that destroys it. emit2() Some instructions are too complicated to be emitted using the instruction templates. They are emitted by this function instead. There are a lot of different reasons to why an instruction must be emitted by emit2() instead. For example, reg: CNSTI2 loads a constant into a register, but the hardware can not move a 48-bit constant to a register using one instruction, so the constant must be split in two parts and moved to the register using two move instructions. This can not be done with the templates so emit2() emits it instead. doarg() doarg() is called for each argument before a function call. It is used to compute the register or stack cell assigned to the next argument. For this implementation all function arguments are put on the stack. blkfetch(), blkstore() and blkloop() These functions can be used to emit code that copies blocks of data. This can, for example, be used by structure assignments. If they are empty the compiler will generate other code instead. local() This function is used by the front end to announce local variables to the back end. It is used to set the stack offset for the variables. function() function() is used by the front end to generate and emit code for a function. It is usually divided in three parts. In the first part initialization is being done. Offsets for the argument variables are calculated. 42

Chapter 5 – Implementation

After that gencode() is called to generate the code for the function. In the second part the size of the frame and the registers that need saving are known. The function prologue is emitted to save the old frame pointer and to set up a new stack pointer and save the necessary registers. After that emitcode() is called to emit the actual assembly code for the function. In the third part the function epilogue is emitted to restore registers and the frame and stack pointers. defsymbol() defsymbol() is called by the front end whenever a new symbol is defined. A symbol is an internal data type in the compiler that represents variables, constants, labels and types. This function sets up the name that the back end uses for the symbols. address() This function is used to initialize a symbol that represents an address on the form x+n, where x is a symbol name and n is a number. This is used so that addresses can be calculated by the assembler instead of at run time. defconst() This function is used to emit assembly code for constants. defaddress() This function emits assembly for pointer constants. defstring() This function is used to emit assembly code to initialize a string. There is a special case that needs to be taken care of here. Internally the compiler treats all arrays with the data type “1-byte integer” as strings. This causes a problem since char, short and int are all of this data type in this implementation. Therefore the compiler believes that all arrays of these data types are strings and tries to emit them as strings. A change in the front end was made so that if the variable n (the length of the string) is -1, the string pointer *str contains the actual value instead of pointing to a string that should be emitted. export() and import() These functions are used to emit assembly directives to import and export symbols so that variables and functions can be reached across different source files. 43

5.4 – Special features

global() The front end calls this function to emit assembly to make a variable global. space() This function is used to emit code that creates a block of words set to zero. Interface record The interface record is used to configure the back end. It consists of a structure that is assigned to the global variable IR. It is done when the compiler is initialized and it is set to the structure of the back end that the compiler chooses. The front end can then call the interface functions in the back end on the form (*IR->progbeg)(arg1,arg2) and access the interface variables on the form IR->wants_dag. The first part of the interface record sets up the size and alignment of the data types in C. After that are some flags to set up specific features. The rest is used to assign the functions documented above.

5.4 Special features There are some special features in the compiler that needs to be mentioned. The C standard specifies that floating point constants default to double if they are not ended with an f. For example, 1.0 is a double and 1.0f is a float. But since the double data type is not fully implemented this was changed so that floating point numbers default to float and double is used if the number ends with a d. This violates the C standard, but it makes the compiler more usable. 1.0 is not a valid number for float and double since the ranges for these data types are – 1.0 ≤ x < 1.0 . But it is still possible to use it and it will be converted to the largest number allowed for that data type. This avoids statements like float x = 0.999999999999; to get the largest possible number.

44

Chapter 5 – Implementation

5.5 Other changes to LCC One “internal” goal when implementing the compiler was to avoid any changes to the front end. This could however not be accomplished. The biggest change that needed to be done involved the byte size. LCC assumes that a byte is 8 bits wide, and the DSP56002 processor uses a byte size (or word size) of 24 bits. LCC is hard coded with this, so the expression 8*someting appeared in a lot of places in the source code. This was changed by replacing the 8 with a global variable with the byte size that is set during initialization when the back end is chosen. Other things that needed changes in the front end was the change of the default floating point type from double to float and the stringarray-problem mentioned for defstring(). A new command line option was added that turns on debug messages in the back end (-x), and some minor fixes to deal with 48-bit constants (LCC’s maximum is 32-bit) and the fractional numbers. LCC uses strength reduction for optimization and replaces multiplications and divisions with a power of two with left and right shifts. This was disabled because the DSP56002 can only shift one step at a time. Also, the multiplication and the shift operation have the same execution time. All of these changes were made so that they only affected this back end, and not the other back ends, so that they would still work. One reason for this was that the other back ends were used to test and compare this back end so it was crucial that they still worked.

5.6 The environment The actual compiler executable, called rcc, is used together with a number of other programs to form a complete compiler environment. rcc only compiles pre processed C code to assembly code. The main executable for LCC is called lcc and is a so called driver. It drives the compilation by taking care of file names and starting the different programs with the correct command line parameters. The following is a list of the programs used: • Preprocessor: The C preprocessor from the GNU Compiler Collection is used. The actual executable is called cpp.

45

5.7 – crt0

• Assembler: The assembler from Motorola is used. It is called asm56000 and it is run with the following command line options: -B -OIL,NOW. This makes the assembler output a relocatable object file and disables both output to the terminal and warning messages. • Linker: Also the linker from Motorola is used. It is called dsplnk and is run with the command line options -d and -B. -B causes the linker to output an absolute executable file. -d is an undocumented switch that is also used by the Motorola compiler when it invokes the linker. It instructs the linker to define the variable DSIZE to the amount of used memory in the Y memory. It is used to initialize the stack pointer. The switch may have other effects, but it is unknown.

5.7 crt0 The crt0 file is linked with all programs when they are compiled. It contains the C bootstrap code and provides the environment to execute a C program. The bootstrap code is put first in the resulting executable and is always the code that gets executed first when a program runs. It defines some global variables, initializes the memories, stack and frame pointers and registers. It then jumps to the main() function in the C program.

5.8 Problems There have been some problems during the development of the back end. Most of them originates from the fact that LCC is intended for normal processors and not DSP processors.

5.8.1 Register targeting The most severe problem has to do with register targeting. In LCC registers can be grouped together in something that is called wildcards. Each data type in LCC (I1, I2, U1 and so on) is assigned to use registers from one wildcard. This is done in the function rmap(). When registers are selected for the instructions they are all assigned from the same wildcard since they are the same data type. This is not suitable for the DSP52002 since it uses input registers and accumulator registers. Most of the instructions must use the input registers for input and produces the output in the accumulator registers. 46

Chapter 5 – Implementation

The solution to this problem is to use the target() function which forces the register allocator to select a specific register for an instruction. The input registers are allocated from the wildcard and target() sets the output register to one of the accumulator registers. This results in another problem that is caused by the optimization being done in the register allocator. When compiling expressions like a*b*c the compiler generates temporary variables to store the intermediate result between the two multiplications. This intermediate result would be calculated to the accumulator register, and the register allocator uses this as an input register to the next multiplication. This is not allowed on the DSP56002 processor. For example, the expression a*b*c would compile to mpy mpy

y0,y1,b b,x1,b

The solution to this is to target the input registers for some instructions as well. The result of this is that the register allocator can not choose between the free registers when targeting these instructions, but has to use a specific register that may already be used. This causes the compiler to emit extra instructions to save that register to memory and also a lot of register to register moves. There is still a problem with the register allocator. This could possibly be caused by a bug in the front end. When compiling the expression a*a the register allocator will use the same input register for both operands. This is not allowed on the DSP56002. The following code is generated mpy

y0,y0,b

even though the input registers are targeted to Y0 and Y1. This problem is handled in the function emit2(). There is one final problem with the register allocator that could not be solved. In the function progbeg() the variables tmask[] and vmask[] are used to configure which registers can be used for temporary values and which registers can be used for variables. When compiling larger programs the register selector will crash and complain that it can not find any free registers unless vmask[] is set to zero (meaning no registers). This means that variables will always be fetched and stored directly to and from memory when they are used, instead of being stored in a register between calculations. For example, the statements

47

5.8 – Problems

x = a + b; y = a + c; will cause the value of a to be fetched two times instead of being fetched and then reused. The reason to why the register allocator crashes is unknown, but if the problem could be fixed the performance of the generated code would increase a lot.

5.8.2 48-bit registers The emit2() function is quite big as a result of another problem. The DSP52002 does not handle 48-bit numbers very well. The and, or, eor and not instructions can only operate on 24-bit numbers, so that needs to be done in two steps. The mpy, div and mod instructions only supports 24-bit input, so special algorithms are emitted for them to handle 48-bit input. Moving 48-bit registers to and from memory needs special treatment. The DSP56002 can only transfer 24 bits at a time over the data bus to the memory, so that is done in two steps.

5.8.3 Address registers The DSP56002 can only address the memory by using the R-registers together with the N- and M-registers. This causes some problems because LCC often accesses the memory by adding a constant offset to an address in a register. The local variables are accessed by adding a constant to the stack pointer and the function arguments are accessed by adding a constant to the frame pointer. To do this on the DSP56002 the constant must be moved to one of the N-registers before the memory can be accessed. For example, something that could have looked like this move

Y:(r0)-3,a

will instead look like this move lua move

#-3,n0 (r0)+n0,r7 Y:(r7),a

The assembler will also insert two nop instructions because the new values in the R- and N-registers are not available until one instruction has passed. This is due to pipeline delays in the processor. So, a total of five instructions are required to access any local or argument varia48

Chapter 5 – Implementation

ble. This is very costly since it is used for both local variables and function arguments.

5.9 Improvements There are some improvements that could be made to the compiler. One obvious thing to improve is the register allocator. The first thing to do is to fix the crashing so it can promote variables to registers. One other thing that can be done is to introduce new register classes for the input and accumulator registers so it can take full advantage of all available registers. The DSP56002 has some hardware features that could be used. Some of them may be a bit tricky to implement: • The DSP can do hardware looping, but that would be difficult to use since it has a lot of restrictions. • The instruction MAC, which means multiply and accumulate, could be used to speed up the generated code. That would relatively easy to implement since it could be realised with a rule that looks like this: reg: ADDI4(MULI4(reg, reg), reg) It would match two nodes in the tree and use fewer instructions. There are other instructions that could be used in the same way, for example INC, which increases a register by one. • The DSP can execute one instruction and move data to and from the memories at the same time. This can be utilized to speed up the code, but it would be difficult to implement. The Motorola compiler uses this by running the generated assembly code trough an external program called alo. It analyses the code and concatenates MOVE instructions with other instructions. For example, the code move add

x0,Y:(r1) x1,a

would be transformed to add

x1,a

x0,Y:(r1)

This kind of optimization would increase the performance of the code a lot.

49

5.9 – Improvements

50

6 Conclusions 6.1 Retargeting The goal of this thesis was to retarget a C compiler for the Motorola DSP56002 processor. The resulting compiler should generate working code that functions as intended. That goal was fulfilled. A working compiler was constructed. The performance of the code that the compiler generates is far from optimal, but since that was not a requirement, only a limited amount of time was spent to improve it. The second goal was to make the compiler compatible with the compiler from Motorola. This was also accomplished and the compiler can be used with the assembler and linker from Motorola to form a complete environment that compiles the C code. The code that the compiler generates was tested in a simulator for the DSP56002 processor to verify that it works. This was very useful and a lot of bugs were found using the simulator. There is probably still a few bugs left in the compiler since there was not enough time to test everything. An example of a compilation of a simple program can be seen in Appendix B on page 59.

6.2 Future work The compiler works, but it generates code that does not perform very well. Much of this has to do with the way the compiler selects and 51

6.2 – Future work

uses registers. As mentioned in section 5.9, this could be fixed by changing how the register allocator works. Hardware features can be used to speed up the code. Some of the hardware features may be very difficult to use, such as hardware looping, but others may be implemented with very little difficulty. And, of course, more testing to find and fix bugs can always be done.

52

References Books and papers [1]

A. V. Aho, R. Sethi and J. D. Ullman, Compilers - Principles, Techniques, and Tools, Addison-Wesley, 1985

[2]

C. W. Fraser and D. R. Hanson, A Retargetable C Compiler: Design and Implementation, Addison-Wesley, 1995

[3]

C. W. Fraser and D. R. Hanson, The lcc 4.x Code-Generation Interface, Microsoft Research, 2003

[4]

Motorola, DSP56000 24-bit Digital Signal Processor Family Manual, Motorola Inc., 1994

[5]

Motorola, Motorola DSP56000 Family Optimizing C Compiler User’s Manual, Release 6.3, Motorola Inc., 1999

Internet [6]

LCC home page: http://www.cs.princeton.edu/software/lcc/

[7]

LCC license: http://www.cs.princeton.edu/software/lcc/4.1/CPYRIGHT

53

54

A Instructions This appendix lists all instructions available for the Motorola DSP56002. The instructions are arranged in groups according to their functionality. The * indicates that a parallel data move is allowed for the instruction.

A.1 Arithmetic instructions Instruction

Description

ABS

Absolute value *

ADC

Add long with carry *

ADD

Add *

ADDL

Shift left then add *

ADDR

Shift right then add *

ASL

Arithmetic shift accumulator left *

ASR

Arithmetic shift accumulator right *

CLR

Clear accumulator *

CMP

Compare *

CMPM

Compare magnitude *

DEC

Decrement accumulator Table A.1: Arithmetic instructions

55

Instruction

Description

DIV

Divide iteration

INC

Increment accumulator

MAC

Signed multiply-accumulate *

MACR

Signed multiply-accumulate and round *

MPY

Signed multiply *

MPYR

Signed multiply and round *

NEG

Negate accumulator *

NORM

Normalize accumulator iteration

RND

Round accumulator *

SBC

Subtract long with carry *

SUB

Subtract *

SUBL

Shift left then subtract *

SUBR

Shift right then subtract *

Tcc

Transfer Conditionally

TFR

Transfer data ALU register *

TST

Test accumulator * Table A.1: Arithmetic instructions

A.2 Logical instructions Instruction

Description

AND

Logical AND *

ANDI

AND immediate with control register

EOR

Logical exclusive OR *

LSL

Logical shift accumulator left *

LSR

Logical shift accumulator right *

NOT

Logical complement on accumulator *

OR

Logical inclusive OR *

ORI

OR immediate with control register

ROL

Rotate accumulator left *

ROR

Rotate accumulator right * Table A.2: Logical instructions

56

Appendix A – Instructions

A.3 Bit manipulation instructions Instruction

Description

BCHG

Bit test and change

BCLR

Bit test and clear

BSET

Bit test and set

BTST

Bit test on memory Table A.3: Bit manipulation instructions

A.4 Loop instructions Instruction

Description

DO

Start hardware loop

ENDDO

Exit from hardware loop

Table A.4: Loop instructions

A.5 Move instructions Instruction

Description

LUA

Load updated address

MOVE

Move data

MOVEC

Move control register

MOVEM

Move program memory

MOVEP

Move peripheral data

Table A.5: Move instructions

57

A.6 Program control instructions Instruction

Description

DEBUG

Enter debug mode

DEBUGcc

Enter debug mode conditionally

ILLEGAL

Illegal instruction interrupt

Jcc

Jump conditionally

JCLR

Jump if bit clear

JMP

Jump

JScc

Jump to subroutine conditionally

JSCLR

Jump to subroutine if bit clear

JSET

Jump if bit set

JSSET

Jump to subroutine if bit set

JSR

Jump to subroutine

NOP

No operation

REP

Repeat next instruction

RESET

Reset on-chip peripheral devices

RTI

Return from interrupt

RTS

Return from subroutine

STOP

Stop processing (low power standby)

SWI

Software interrupt

WAIT

Wait for interrupt (low power standby)

Table A.6: Program control instructions

58

B Sample code This appendix contains the resulting assembly file of a compilation of a small test program called sample.c.

B.1 sample.c int res = 1; int x = 10; void main(void) { if(res == 1) res = res + x; else res = x; }

59

B.2 sample.asm section sample_c opt so,nomd,rp org y: global Fres

boilerplate

Fres dc global

$1 Fx

dc org p:

$a

initialized data

Fx

;;;; Function main starts global Fmain Fmain: move r0,Y:(r6)+ lua (r6)+,r0 move ssh,Y:(r6)+ move y1,Y:(r6)+ move Y:Fres,a move #>1,y1 cmp y1,a jne L2 move Y:Fres,a move Y:Fx,y1 add y1,a move a,Y:Fres jmp L3 L2: move Y:Fx,y1 move y1,Y:Fres L3: L1: move Y:-(r6),y1 move Y:-(r6),ssh tst a Y:-(r6),r0 rts ;;;; Function main ends endsec end

60

entry sequence/ function prologue

body of main()

exit sequence/ function epilogue

boilerplate

C dsp56k.md This appendix contains the entire contents of the file dsp56k.md. It is used to create the back end for the Motorola DSP56002. %{ #include “c.h” #include “time.h” #define debug2(x) (void)(xflag&&((x),0)) #define #define #define #define #define static static static static static static static static static static static static static static static static static static static static static

NODEPTR_TYPE Node OP_LABEL(p) ((p)->op) LEFT_CHILD(p) ((p)->kids[0]) RIGHT_CHILD(p) ((p)->kids[1]) STATE_LABEL(p) ((p)->x.state)

void void void void void void void void void void void void void void void void void void void void void

address(Symbol, Symbol, long); blkfetch(int, int, int, int); blkloop(int, int, int, int, int, int[]); blkstore(int, int, int, int); defaddress(Symbol); defconst(int, int, Value); defstring(int, char *); defsymbol(Symbol); doarg(Node); emit2(Node); export(Symbol); clobber(Node); function(Symbol, Symbol [], Symbol [], int); global(Symbol); import(Symbol); local(Symbol); progbeg(int, char **); progend(void); segment(int); space(int); target(Node);

static int equal(Node p, int a); static static static static

Symbol Symbol Symbol Symbol

ireg[32], ireg2[32], areg[32], areg2[32]; rreg[32]; iregw, ireg2w, aregw, areg2w; rregw;

static int cseg; static int retstruct; %} %start stmt %term CNSTF1=1041 CNSTF2=2065 %term CNSTI1=1045 CNSTI2=2069 %term CNSTP1=1047

61

%term CNSTU1=1046 CNSTU2=2070 %term %term %term %term %term

ARGB=41 ARGF1=1057 ARGF2=2081 ARGI1=1061 ARGI2=2085 ARGP1=1063 ARGU1=1062 ARGU2=2086

%term %term %term %term %term

ASGNB=57 ASGNF1=1073 ASGNF2=2097 ASGNI1=1077 ASGNI2=2101 ASGNP1=1079 ASGNU1=1078 ASGNU2=2102

%term %term %term %term %term

INDIRB=73 INDIRF1=1089 INDIRF2=2113 INDIRI1=1093 INDIRI2=2117 INDIRP1=1095 INDIRU1=1094 INDIRU2=2118

%term CVFF1=1137 CVFF2=2161 %term CVFI1=1141 CVFI2=2165 %term CVIF1=1153 CVIF2=2177 %term CVII1=1157 CVII2=2181 %term CVIU1=1158 CVIU2=2182 %term CVPU1=1174 %term CVUI1=1205 CVUI2=2229 %term CVUP1=1207 %term CVUU1=1206 CVUU2=2230 %term NEGF1=1217 NEGF2=2241 %term NEGI1=1221 NEGI2=2245 %term %term %term %term %term %term

CALLB=217 CALLF1=1233 CALLF2=2257 CALLI1=1237 CALLI2=2261 CALLP1=1239 CALLU1=1238 CALLU2=2262 CALLV=216

%term %term %term %term %term

RETF1=1265 RETF2=2289 RETI1=1269 RETI2=2293 RETP1=1271 RETU1=1270 RETU2=2294 RETV=248

%term ADDRGP1=1287 %term ADDRFP1=1303 %term ADDRLP1=1319 %term %term %term %term

ADDF1=1329 ADDF2=2353 ADDI1=1333 ADDI2=2357 ADDP1=1335 ADDU1=1334 ADDU2=2358

%term %term %term %term

SUBF1=1345 SUBF2=2369 SUBI1=1349 SUBI2=2373 SUBP1=1351 SUBU1=1350 SUBU2=2374

%term LSHI1=1365 LSHI2=2389 %term LSHU1=1366 LSHU2=2390 %term MODI1=1381 MODI2=2405 %term MODU1=1382 MODU2=2406 %term RSHI1=1397 RSHI2=2421 %term RSHU1=1398 RSHU2=2422 %term BANDI1=1413 BANDI2=2437 %term BANDU1=1414 BANDU2=2438 %term BCOMI1=1429 BCOMI2=2453 %term BCOMU1=1430 BCOMU2=2454 %term BORI1=1445 BORI2=2469 %term BORU1=1446 BORU2=2470 %term BXORI1=1461 BXORI2=2485 %term BXORU1=1462 BXORU2=2486 %term DIVF1=1473 DIVF2=2497 %term DIVI1=1477 DIVI2=2501 %term DIVU1=1478 DIVU2=2502 %term MULF1=1489 MULF2=2513 %term MULI1=1493 MULI2=2517 %term MULU1=1494 MULU2=2518

62

Appendix C – dsp56k.md %term EQF1=1505 EQF2=2529 %term EQI1=1509 EQI2=2533 %term EQU1=1510 EQU2=2534 %term GEF1=1521 GEF2=2545 %term GEI1=1525 GEI2=2549 %term GEU1=1526 GEU2=2550 %term GTF1=1537 GTF2=2561 %term GTI1=1541 GTI2=2565 %term GTU1=1542 GTU2=2566 %term LEF1=1553 LEF2=2577 %term LEI1=1557 LEI2=2581 %term LEU1=1558 LEU2=2582 %term LTF1=1569 LTF2=2593 %term LTI1=1573 LTI2=2597 %term LTU1=1574 LTU2=2598 %term NEF1=1585 NEF2=2609 %term NEI1=1589 NEI2=2613 %term NEU1=1590 NEU2=2614 %term JUMPV=584 %term LABELV=600 %term VREGP=711 %term %term %term %term %term %term %term %term

LOADI1=1253 LOADI2=2277 LOADU1=1254 LOADU2=2278 LOADB=233 LOADF1=1249 LOADF2=2273 LOADP1=1255

%% reg: INDIRI1(VREGP) reg: INDIRI2(VREGP) reg: INDIRU1(VREGP) reg: INDIRU2(VREGP) reg: INDIRF1(VREGP) reg: INDIRF2(VREGP) rreg: INDIRP1(VREGP) stmt: stmt: stmt: stmt: stmt: stmt: stmt: con: reg: con: reg: con: reg: reg:

ASGNI1(VREGP, ASGNI2(VREGP, ASGNU1(VREGP, ASGNU2(VREGP, ASGNF1(VREGP, ASGNF2(VREGP, ASGNP1(VREGP,

“# “# “# “# “# “# “#

read read read read read read read

reg) reg) reg) reg) reg) reg) rreg)

CNSTI1 CNSTI2 CNSTU1 CNSTU2 CNSTP1 CNSTF1 CNSTF2

“# “# “# “# “# “# “#

register\n” register\n” register\n” register\n” register\n” register\n” register\n”

write write write write write write write

“%a” “# \tmove “%a” “# \tmove “%a” “# \tmove “# \tmove

register\n” register\n” register\n” register\n” register\n” register\n” register\n”

\t#>%a,%c\n” 2+2 \t#>%a,%c\n” 2+2 \t#>%a,%c\n” 2 \t#>%a,%c\n” 2+2

reg: LOADI1(reg) reg: LOADU1(reg) reg: LOADU1(rreg) reg: LOADI2(reg) reg: LOADU2(reg) reg: LOADF1(reg) reg: LOADF2(reg) rreg: LOADP1(rreg) rreg: LOADP1(reg)

“\tmove \t%0,%c\n” 2 “\tmove \t%0,%c\n” 2 “\tmove \t%0,%c\n” 2 “# \tmove \t%0,%c\n” 2 “# \tmove \t%0,%c\n” 2 “\tmove \t%0,%c\n” 2 “# \tmove \t%0,%c\n” 2 “\tmove \t%0,%c\n” 2 “\tmove \t%0,%c\n” 2

reg: con rreg: con

“\tmove \t#>%0,%c\n” 2 “\tmove \t#%0,%c\n” 2

rreg: ADDRGP1 rreg: ADDRFP1 rreg: ADDRLP1

“\tmove \t#%a,%c\n” 2 “\tmove \t#%a,n0\n\tlua \t(r0)+n0,%c\n” 2+4 “\tmove \t#%a,n6\n\tlua \t(r6)+n6,%c\n” 2+4

stmt: reg stmt: LABELV

““ “%a:\n”

addr: rreg addr: ADDRGP1

“Y:(%0)” “Y:%a”

stmt: stmt: stmt: stmt:

ASGNI1(addr, ASGNI2(addr, ASGNU1(addr, ASGNU2(addr,

reg) reg) reg) reg)

“\tmove \t%1,%0\n” 2 “# \tmove \t%1,%0\n” 2 “\tmove \t%1,%0\n” 2 “# \tmove \t%1,%0\n” 2

63

stmt: stmt: stmt: stmt:

ASGNF1(addr, reg) ASGNF2(addr, reg) ASGNP1(addr, rreg) ASGNB(rreg, INDIRB(rreg))

“\tmove \t%1,%0\n” 2 “# \tmove \t%1,%0\n” 2 “\tmove \t%1,%0\n” 2 “# ASGNB\n”

reg: INDIRI1(addr) reg: INDIRI2(addr) reg: INDIRU1(addr) reg: INDIRU2(addr) reg: INDIRF1(addr) reg: INDIRF2(addr) rreg: INDIRP1(addr)

“\tmove \t%0,%c\n” 2 “# \tmove \t%0,%c\n” 2 “\tmove \t%0,%c\n” 2 “# \tmove \t%0,%c\n” 2 “\tmove \t%0,%c\n” 2 “# \tmove \t%0,%c\n” 2 “\tmove \t%0,%c\n” 2

reg: reg: reg: reg: reg: reg:

ADDI1(reg, ADDI2(reg, ADDU1(reg, ADDU2(reg, ADDF1(reg, ADDF2(reg,

reg) reg) reg) reg) reg) reg)

“\tadd “\tadd “\tadd “\tadd “\tadd “\tadd

\t%1,%0\n” \t%1,%0\n” \t%1,%0\n” \t%1,%0\n” \t%1,%0\n” \t%1,%0\n”

2 2 2 2 2 2

reg: reg: reg: reg: reg: reg:

SUBI1(reg, SUBI2(reg, SUBU1(reg, SUBU2(reg, SUBF1(reg, SUBF2(reg,

reg) reg) reg) reg) reg) reg)

“\tsub “\tsub “\tsub “\tsub “\tsub “\tsub

\t%1,%0\n” \t%1,%0\n” \t%1,%0\n” \t%1,%0\n” \t%1,%0\n” \t%1,%0\n”

2 2 2 2 2 2

rc: reg rc: con rreg: ADDP1(rc, rreg) rreg: ADDP1(rreg, rc)

“%0” “%#%0” “# \tmove \t%0,n8\n\tlua \t(%1)+n8,%c\n” 2+4 “# \tmove \t%1,n8\n\tlua \t(%0)+n8,%c\n” 2+4

rreg: SUBP1(rc, rreg) rreg: SUBP1(rreg, rc)

“# \tmove \t%0,n8\n\tlua \t(%1)-n8,%c\n” 2+4 “# \tmove \t%1,n8\n\tlua \t(%0)-n8,%c\n” 2+4

reg: reg: reg: reg: reg: reg:

MULI1(reg, MULI2(reg, MULU1(reg, MULU2(reg, MULF1(reg, MULF2(reg,

reg) reg) reg) reg) reg) reg)

“# “# “# “# “# “#

\tmpy \tmpy \tmpy \tmpy \tmpy \tmpy

\t%0,%1,%c\n\tasr \t%c\n\tmove \t%c0,%c\n” 2+2+2 \t%0,%1,%c\n” 2 \t%0,%1,%c\n\tasr \t%c\n\tmove \t%c0,%c\n” 2+2+2 \t%0,%1,%c\n” 2 \t%0,%1,%c\n\tmove \t%c0,%c\n” 2+2 \t%0,%1,%c\n” 2

reg: reg: reg: reg: reg: reg:

DIVI1(reg, DIVI2(reg, DIVU1(reg, DIVU2(reg, DIVF1(reg, DIVF2(reg,

reg) reg) reg) reg) reg) reg)

“# “# “# “# “# “#

\tdiv \tdiv \tdiv \tdiv \tdiv \tdiv

\t%0,%1\n” \t%0,%1\n” \t%0,%1\n” \t%0,%1\n” \t%0,%1\n” \t%0,%1\n”

2 2 2 2 2 2

reg: reg: reg: reg:

MODI1(reg, MODI2(reg, MODU1(reg, MODU2(reg,

reg) reg) reg) reg)

“# “# “# “#

\tmod \tmod \tmod \tmod

\t%0,%1\n” \t%0,%1\n” \t%0,%1\n” \t%0,%1\n”

2 2 2 2

reg: reg: reg: reg: reg: reg:

CVII1(reg) CVII2(reg) CVIU1(reg) CVIU2(reg) CVIF1(reg) CVIF2(reg)

“# “# “# “# “# “#

\tmove \tmove \tmove \tmove \tmove \tmove

\t%0,%c\n” \t%0,%c\n” \t%0,%c\n” \t%0,%c\n” \t%0,%c\n” \t%0,%c\n”

move(a) move(a) move(a) move(a) move(a) move(a)

reg: CVUI1(reg) reg: CVUI2(reg) reg: CVUU1(reg) reg: CVUU2(reg) rreg: CVUP1(reg)

“# \tmove \t%0,%c\n” move(a) “# \tmove \t%0,%c\n” move(a) “# \tmove \t%0,%c\n” move(a) “# \tmove \t%0,%c\n” move(a) “\tmove \t%0,%c\n” move(a)

reg: CVPU1(rreg)

“\tmove \t%0,%c\n” move(a)

reg: reg: reg: reg:

“# “# “# “#

CVFF1(reg) CVFF2(reg) CVFI1(reg) CVFI2(reg)

stmt: JUMPV(jaddr)

\tmove \tmove \tmove \tmove

\t%0,%c\n” \t%0,%c\n” \t%0,%c\n” \t%0,%c\n”

move(a) move(a) move(a) move(a)

“\tjmp \t%0\n” 4

reg: CALLB(jaddr, rreg)

“\tmove \t%1,%c\n\tjsr \t%0\n” 2+4

reg: CALLI1(jaddr) reg: CALLI1(jaddr) reg: CALLI1(jaddr)

“\tjsr \t%0\n\tmove \t#%a,n6\n\tmove \t(r6)-n6\n” 4+2 “\tjsr \t%0\n\tmove \t(r6)-\n” equal(a, 1) “\tjsr \t%0\n” equal(a, 0)

reg: CALLI2(jaddr) reg: CALLI2(jaddr) reg: CALLI2(jaddr)

“\tjsr \t%0\n\tmove \t#%a,n6\n\tmove \t(r6)-n6\n” 4+2 “\tjsr \t%0\n\tmove \t(r6)-\n” equal(a, 1) “\tjsr \t%0\n” equal(a, 0)

reg: CALLP1(jaddr) reg: CALLP1(jaddr) reg: CALLP1(jaddr)

“\tjsr \t%0\n\tmove \t#%a,n6\n\tmove \t(r6)-n6\n” 4+2 “\tjsr \t%0\n\tmove \t(r6)-\n” equal(a, 1) “\tjsr \t%0\n” equal(a, 0)

reg: CALLU1(jaddr) reg: CALLU1(jaddr)

“\tjsr \t%0\n\tmove \t#%a,n6\n\tmove \t(r6)-n6\n” 4+2 “\tjsr \t%0\n\tmove \t(r6)-\n” equal(a, 1)

64

Appendix C – dsp56k.md reg: CALLU1(jaddr)

“\tjsr \t%0\n” equal(a, 0)

reg: CALLU2(jaddr) reg: CALLU2(jaddr) reg: CALLU2(jaddr)

“\tjsr \t%0\n\tmove \t#%a,n6\n\tmove \t(r6)-n6\n” 4+2 “\tjsr \t%0\n\tmove \t(r6)-\n” equal(a, 1) “\tjsr \t%0\n” equal(a, 0)

reg: CALLF1(jaddr) reg: CALLF1(jaddr) reg: CALLF1(jaddr)

“\tjsr \t%0\n\tmove \t#%a,n6\n\tmove \t(r6)-n6\n” 4+2 “\tjsr \t%0\n\tmove \t(r6)-\n” equal(a, 1) “\tjsr \t%0\n” equal(a, 0)

reg: CALLF2(jaddr) reg: CALLF2(jaddr) reg: CALLF2(jaddr)

“\tjsr \t%0\n\tmove \t#%a,n6\n\tmove \t(r6)-n6\n” 4+2 “\tjsr \t%0\n\tmove \t(r6)-\n” equal(a, 1) “\tjsr \t%0\n” equal(a, 0)

stmt: CALLV(jaddr) stmt: CALLV(jaddr) stmt: CALLV(jaddr)

“\tjsr \t%0\n\tmove \t#%a,n6\n\tmove \t(r6)-n6\n” 4+2 “\tjsr \t%0\n\tmove \t(r6)-\n” equal(a, 1) “\tjsr \t%0\n” equal(a, 0)

jaddr: ADDRGP1 jaddr: CNSTP1 jaddr: rreg

“%a” “%a” “(%0)”

stmt: stmt: stmt: stmt: stmt: stmt: stmt: stmt:

ARGI1(reg) “\tmove \t%0,Y:(r6)+\n” 2 ARGI2(reg) “# \tmove \t%0,Y:(r6)+\n” 2 ARGU1(reg) “\tmove \t%0,Y:(r6)+\n” 2 ARGU2(reg) “# \tmove \t%0,Y:(r6)+\n” 2 ARGF1(reg) “\tmove \t%0,Y:(r6)+\n” 2 ARGF2(reg) “# \tmove \t%0,Y:(r6)+\n” 2 ARGP1(rreg) “\tmove \t%0,Y:(r6)+\n” 2 ARGB(INDIRB(rreg)) “# ARGB\n” 8

stmt: stmt: stmt: stmt: stmt: stmt: stmt:

RETI1(reg) RETI2(reg) RETU1(reg) RETU2(reg) RETF1(reg) RETF2(reg) RETP1(rreg)

“# “# “# “# “# “# “#

stmt: stmt: stmt: stmt: stmt: stmt:

EQI1(reg, EQI2(reg, EQU1(reg, EQU2(reg, EQF1(reg, EQF2(reg,

reg) reg) reg) reg) reg) reg)

“\tcmp “\tcmp “\tcmp “\tcmp “\tcmp “\tcmp

\t%1,%0\n\tjeq \t%1,%0\n\tjeq \t%1,%0\n\tjeq \t%1,%0\n\tjeq \t%1,%0\n\tjeq \t%1,%0\n\tjeq

\t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n”

2+4 2+4 2+4 2+4 2+4 2+4

stmt: stmt: stmt: stmt: stmt: stmt:

GEI1(reg, GEI2(reg, GEU1(reg, GEU2(reg, GEF1(reg, GEF2(reg,

reg) reg) reg) reg) reg) reg)

“\tcmp “\tcmp “\tcmp “\tcmp “\tcmp “\tcmp

\t%1,%0\n\tjge \t%1,%0\n\tjge \t%1,%0\n\tjge \t%1,%0\n\tjge \t%1,%0\n\tjge \t%1,%0\n\tjge

\t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n”

2+4 2+4 2+4 2+4 2+4 2+4

stmt: stmt: stmt: stmt: stmt: stmt:

GTI1(reg, GTI2(reg, GTU1(reg, GTU2(reg, GTF1(reg, GTF2(reg,

reg) reg) reg) reg) reg) reg)

“\tcmp “\tcmp “\tcmp “\tcmp “\tcmp “\tcmp

\t%1,%0\n\tjgt \t%1,%0\n\tjgt \t%1,%0\n\tjgt \t%1,%0\n\tjgt \t%1,%0\n\tjgt \t%1,%0\n\tjgt

\t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n”

2+4 2+4 2+4 2+4 2+4 2+4

stmt: stmt: stmt: stmt: stmt: stmt:

LEI1(reg, LEI2(reg, LEU1(reg, LEU2(reg, LEF1(reg, LEF2(reg,

reg) reg) reg) reg) reg) reg)

“\tcmp “\tcmp “\tcmp “\tcmp “\tcmp “\tcmp

\t%1,%0\n\tjle \t%1,%0\n\tjle \t%1,%0\n\tjle \t%1,%0\n\tjle \t%1,%0\n\tjle \t%1,%0\n\tjle

\t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n”

2+4 2+4 2+4 2+4 2+4 2+4

stmt: stmt: stmt: stmt: stmt: stmt:

LTI1(reg, LTI2(reg, LTU1(reg, LTU2(reg, LTF1(reg, LTF2(reg,

reg) reg) reg) reg) reg) reg)

“\tcmp “\tcmp “\tcmp “\tcmp “\tcmp “\tcmp

\t%1,%0\n\tjlt \t%1,%0\n\tjlt \t%1,%0\n\tjlt \t%1,%0\n\tjlt \t%1,%0\n\tjlt \t%1,%0\n\tjlt

\t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n”

2+4 2+4 2+4 2+4 2+4 2+4

stmt: stmt: stmt: stmt: stmt: stmt:

NEI1(reg, NEI2(reg, NEU1(reg, NEU2(reg, NEF1(reg, NEF2(reg,

reg) reg) reg) reg) reg) reg)

“\tcmp “\tcmp “\tcmp “\tcmp “\tcmp “\tcmp

\t%1,%0\n\tjne \t%1,%0\n\tjne \t%1,%0\n\tjne \t%1,%0\n\tjne \t%1,%0\n\tjne \t%1,%0\n\tjne

\t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n” \t%a\n”

2+4 2+4 2+4 2+4 2+4 2+4

reg: reg: reg: reg:

NEGI1(reg) NEGI2(reg) NEGF1(reg) NEGF2(reg)

“\tneg “\tneg “\tneg “\tneg

\t%0\n” \t%0\n” \t%0\n” \t%0\n”

con1: CNSTI1 reg: reg: reg: reg:

LSHI1(reg, LSHI1(reg, LSHI2(reg, LSHI2(reg,

rts\n” rts\n” rts\n” rts\n” rts\n” rts\n” rts\n”

4 4 4 4 4 4 4

2 2 2 2

“%a” range(a,1,1) rc) con1) rc) con1)

“\trep “\tasl “\trep “\tasl

\t%1\n\tasl \t%0\n” 4+2 \t%0\n” 4 \t%1\n\tasl \t%0\n” 4+2 \t%0\n” 4

65

reg: reg: reg: reg:

LSHU1(reg, LSHU1(reg, LSHU2(reg, LSHU2(reg,

rc) con1) rc) con1)

“\trep “\tlsl “\trep “\tasl

reg: reg: reg: reg: reg: reg: reg: reg:

RSHI1(reg, RSHI1(reg, RSHI2(reg, RSHI2(reg, RSHU1(reg, RSHU1(reg, RSHU2(reg, RSHU2(reg,

rc) con1) rc) con1) rc) con1) rc) con1)

“\trep \t%1\n\tasr \t%0\n” 4+2 “\tasr \t%0\n” 4 “\trep \t%1\n\tasr \t%0\n” 4+2 “\tasr \t%0\n” 4 “\trep \t%1\n\tlsr \t%0\n” 4+2 “\tlsr \t%0\n” 4 “\tmove \t#0,%02\n\trep \t%1\n\tasr \t%0\n” 4+2 “\tmove \t#0,%02\n\tasr \t%0\n” 4

reg: reg: reg: reg:

BANDI1(reg, BANDI2(reg, BANDU1(reg, BANDU2(reg,

reg) reg) reg) reg)

reg: reg: reg: reg:

BORI1(reg, BORI2(reg, BORU1(reg, BORU2(reg,

reg: reg: reg: reg:

BXORI1(reg, BXORI2(reg, BXORU1(reg, BXORU2(reg,

reg: reg: reg: reg:

BCOMI1(reg) BCOMI2(reg) BCOMU1(reg) BCOMU2(reg)

reg) reg) reg) reg) reg) reg) reg) reg)

\t%1\n\tlsl \t%0\n” 4+2 \t%0\n” 4 \t%1\n\tasl \t%0\n” 4+2 \t%0\n” 4

“\tand \t%0,%1\n” 2 “# \tand \t%0,%1\n” 2 “\tand \t%0,%1\n” 2 “# \tand \t%0,%1\n” 2 “\tor \t%0,%1\n” 2 “# \tor \t%0,%1\n” 2 “\tor \t%0,%1\n” 2 “# \tor \t%0,%1\n” 2 “\teor \t%0,%1\n” 2 “# \teor \t%0,%1\n” 2 “\teor \t%0,%1\n” 2 “# \teor \t%0,%1\n” 2 “\tnot \t%0\n” 2 “# \tnot \t%0\n” 2 “\tnot \t%0\n” 2 “# \tnot \t%0\n” 2

%% static int equal(Node p, int a){ if(p->syms[0]->u.c.v.i == a){ return 0; } else { return LBURG_MAX; } } static void progbeg(int argc, char **argv) { int i, n; char section[256]; time_t res = time(NULL); print(“;;;; LCC-DSP56k - compiled on %s\n”, asctime(localtime(&res))); parseflags(argc, argv); ireg[0] ireg[1] ireg[2] ireg[3]

= = = =

mkreg(“x0”, mkreg(“x1”, mkreg(“y0”, mkreg(“y1”,

0, 1, 2, 3,

1, 1, 1, 1,

IREG); IREG); IREG); IREG);

ireg2[0] = mkreg(“x”, 0, 1, ireg2[1] = mkreg(“y”, 2, 1, ireg2[0]->x.regnode->mask = ireg2[1]->x.regnode->mask =

IREG); IREG); 3; // 00000011 12; // 00001100

areg[0] areg[1] areg[2] areg[3]

IREG); IREG); IREG); IREG);

= = = =

mkreg(“a0”, mkreg(“a1”, mkreg(“b0”, mkreg(“b1”,

4, 5, 6, 7,

1, 1, 1, 1,

areg2[0] = mkreg(“a”, 4, 1, areg2[1] = mkreg(“b”, 6, 1, areg2[0]->x.regnode->mask = areg2[1]->x.regnode->mask =

IREG); IREG); 48; // 00110000 192; // 11000000

for(i = 0; i < 8; i++) { rreg[i] = mkreg(stringf(“r%d”, i), i+8, 1, IREG); } iregw = mkwildcard(ireg); ireg2w = mkwildcard(ireg2); aregw = mkwildcard(areg); areg2w = mkwildcard(areg2); rregw = mkwildcard(rreg); tmask[IREG] tmask[FREG] vmask[IREG] vmask[FREG]

= = = =

0x0000BEFF; // R7,5-1 Y,X Y1,Y0,X1,X0 0x00000000; 0x00000000; 0x00000000;

// convert . to _ and remove the path in filename // (../path/to/file.c -> file_c) if(firstfile) { for(i = n = 0; firstfile[i] && i < 255; i++){

66

Appendix C – dsp56k.md char ch = firstfile[i] == ‘.’ ? ‘_’ : firstfile[i]; if(ch == ‘/’) { n = 0; } else { section[n++] = ch; } } section[i]=0; print(“\tsection %s\n”, section); } // so - write symbols // nomd - do not write macro defs // rp - generate nop insn to accomodate pipeline delay print(“\topt \tso,nomd,rp\n”); } static void progend(void) { print(“\n\tendsec\n”); print(“\tend\n”); } static Symbol rmap(int opk) { debug2(fprint(stderr, “Inside %s: %d\n”, “rmap”, opk)); switch(optype(opk)) { case P: case B: return rregw; break; case I: case U: case F: if(opsize(opk) == 1) { return iregw; } else { return ireg2w; } break; default: return 0; } } static void segment(int n) { debug2(fprint(stderr, “Inside %s: %d\n”, “segment”, n)); if(n == cseg) { return; } cseg = n; if(n == CODE){ print(“\torg p:\n”); } else { print(“\torg y:\n”); } } static void target(Node p) { debug2(fprint(stderr, “Inside %s: %x\n”, “target”, p)); switch(specific(p->op)) { case CALL+B: setreg(p, areg2[0]); //a //rtarget(p, 1, areg2[0]); //a break; case ADD+I: case SUB+I: case ADD+U: case SUB+U: case ADD+F: case SUB+F: setreg(p, areg2[0]); //a rtarget(p, 0, areg2[0]); //a break; case MUL+I: case MUL+U: case MUL+F: if(opsize(p->op) == 1) { rtarget(p, 0, ireg[2]); //y0 This is bad but prevents this rtarget(p, 1, ireg[3]); //y1 error: a*b*c -> mpy y0,y1,b setreg(p, areg2[1]); //b mpy b,x1,b } else { rtarget(p, 0, areg2[0]); //a rtarget(p, 1, areg2[1]); //b setreg(p, areg2[1]); //b } break; case DIV+I: case DIV+U: case DIV+F: if(opsize(p->op) == 1) { rtarget(p, 0, areg2[0]); //a rtarget(p, 1, ireg[0]); //x0 setreg(p, areg2[0]); //a } else { rtarget(p, 0, areg2[0]); //a rtarget(p, 1, areg2[1]); //b setreg(p, areg2[0]); //a } break; case MOD+I: case MOD+U: if(opsize(p->op) == 1) { rtarget(p, 0, areg2[0]); //a rtarget(p, 1, ireg[0]); //x0

67

setreg(p, areg2[0]); //a } else { rtarget(p, 0, areg2[0]); //a rtarget(p, 1, areg2[1]); //b setreg(p, areg2[0]); //a } break; case LSH+I: case RSH+I: case LSH+U: case RSH+U: if(opsize(p->op) == 1) { rtarget(p, 0, areg2[0]); //a setreg(p, areg[1]); //a1 } else { rtarget(p, 0, areg2[0]); //a setreg(p, areg2[0]); //a1 } break; case NEG+I: case NEG+F: rtarget(p, 0, areg2[0]); //a setreg(p, areg2[0]); //a break; case BAND+I: case BOR+I: case BAND+U: case BOR+U: case BXOR+I: case BXOR+U: if(opsize(p->op) == 1) { rtarget(p, 0, ireg[3]); //y1 rtarget(p, 1, areg[1]); //a1 setreg(p, areg[1]); //a1 } else { rtarget(p, 1, areg2[0]); //a setreg(p, areg2[0]); //a } break; case BCOM+I: case BCOM+U: if(opsize(p->op) == 1) { rtarget(p, 0, areg[1]); //a1 setreg(p, areg[1]); //a1 } else { rtarget(p, 0, areg2[0]); //a setreg(p, areg2[0]); //a } break; case EQ+I: case GE+I: case GT+I: case LE+I: case LT+I: case NE+I: case EQ+U: case GE+U: case GT+U: case LE+U: case LT+U: case NE+U: case EQ+F: case GE+F: case GT+F: case LE+F: case LT+F: case NE+F: rtarget(p, 0, areg2[0]); //a if(opsize(p->op) == 2) { rtarget(p, 1, areg2[1]); //b } break; case CALL+I: case CALL+U: case CALL+F: case CALL+P: case CALL+V: setreg(p, areg2[0]); //a break; case RET+I: case RET+U: case RET+F: rtarget(p, 0, areg2[0]); //a break; case RET+P: rtarget(p, 0, areg[1]); //a1 break; case CVI+I: case CVI+U: case CVI+F: case CVU+I: case CVU+U: case CVF+I: case CVF+F: if(opsize(p->op) == 1){ setreg(p, areg2[0]); //a } else { setreg(p, areg2[0]); //a } break; case LOAD+I: case LOAD+U: case LOAD+F: if(p->kids[0]->x.inst == _rreg_NT){ break; } if(opsize(p->op) == 1){ rtarget(p, 0, areg2[0]); //a } else { rtarget(p, 0, areg2[0]); //a } break; default: break; } } static void clobber(Node p) { debug2(fprint(stderr, “Inside %s: %x\n”, “clobber”, p)); assert(p); } static char *reg_name(int reg){ if(reg >= 0 && reg <= 3){ return ireg[reg]->x.name; } else if(reg == 4 || reg == 6){ return areg2[reg/2 - 2]->x.name; } else if(reg == 5 || reg == 7){ return areg[reg - 4]->x.name;

68

Appendix C – dsp56k.md } else { assert(0); } } static void emit2(Node p) { debug2(fprint(stderr, “Inside %s: %x\n”, “emit2”, p)); switch(specific(p->op)) { case CNST+F: assert(0); break; case CNST+I: case CNST+U: { assert(opsize(p->op) == 2); //reg: CNSTI2 “# \tmove \t%a,%c\n” 0 int reg = getregnum(p); print(“\tmove \t#>%d,%s\n”, (int)(p->syms[0]->u.c.v.i & 0xFFFFFF), reg_name(reg)); print(“\tmove \t#>%d,%s\n”, (int)(p->syms[0]->u.c.v.i >> 24), reg_name(reg + 1)); break; } case INDIR+I: case INDIR+U: case INDIR+F: { if(p->kids[0]->op != VREG+P){ if(opsize(p->op) == 2){ int dst = getregnum(p); //reg: INDIRI2(addr) “\tmove \t%0,%c\n” print(“\tmove \t”); emitasm(p->kids[0], _addr_NT); if(p->kids[0]->x.inst == _rreg_NT){ print(“+,%s\n”, reg_name(dst)); } else { print(“,%s\n”, reg_name(dst)); } print(“\tmove \t”); emitasm(p->kids[0], _addr_NT); if(p->kids[0]->x.inst == _rreg_NT){ print(“-,%s\n”, reg_name(dst + 1)); } else { print(“+1,%s\n”, reg_name(dst + 1)); } } } break; } case ASGN+I: case ASGN+U: case ASGN+F: { if(p->kids[0]->op != VREG+P){ if(opsize(p->op) == 2){ int src = getregnum(p->kids[1]); //stmt: ASGNI2(addr, reg) “\tmove \t%1,%0\n” print(“\tmove \t%s,”, reg_name(src)); emitasm(p->kids[0], _addr_NT); if(p->kids[0]->x.inst == _rreg_NT){ print(“+\n”); } else { print(“\n”); } print(“\tmove \t%s,”, reg_name(src+1)); emitasm(p->kids[0], _addr_NT); if(p->kids[0]->x.inst == _rreg_NT){ print(“-\n”); } else { print(“+1\n”); } } } break; } case ASGN+B: { int src = getregnum(p->x.kids[1]) - 8; int dst = getregnum(p->x.kids[0]) - 8; int lab = genlabel(1); print(“\tmove \tr%d,n%d x1,Y:(r6)\n”, src, src); print(“\tmove \tr%d,n%d\n”, dst, dst); print(“\tdo \t#%d,L%d\n”, (int)p->syms[0]->u.c.v.i, lab); print(“\tmove \tY:(r%d)+,x0\n”, src); print(“\tmove \tx0,Y:(r%d)+\n”, dst); print(“L%d:\n”, lab); print(“\tmove \tn%d,r%d\n”, src, src); print(“\tmove \tn%d,r%d Y:(r6),x1\n”, dst, dst); break; } case ADD+P: //rreg: ADDP1(rreg, rc) “# \tmove \t%1,n8\n\tlua \t(%0)+n8,%c\n” //rreg: ADDP1(rc, rreg) “# \tmove \t%0,n8\n\tlua \t(%1)+n8,%c\n” if(p->kids[0]->x.inst == _rreg_NT){ int reg = getregnum(p->x.kids[0]) - 8; int dst = getregnum(p) - 8; print(“\tmove \t”); emitasm(p->kids[1], _rc_NT); print(“,n%d\n\tlua \t(r%d)+n%d,r%d\n”, reg, reg, reg, dst);

69

} else if(p->kids[1]->x.inst == _rreg_NT){ int reg = getregnum(p->x.kids[1]) - 8; int dst = getregnum(p) - 8; print(“\tmove \t”); emitasm(p->kids[0], _rc_NT); print(“,n%d\n\tlua \t(r%d)+n%d,r%d\n”, reg, reg, reg, dst); } else assert(0); break; case SUB+P: //rreg: SUBP1(rc, rreg) “# \tmove \t%0,n8\n\tlua \t(%1)-n8,%c\n” //rreg: SUBP1(rreg, rc) “# \tmove \t%1,n8\n\tlua \t(%0)-n8,%c\n” if(p->kids[0]->x.inst == _rreg_NT){ int reg = getregnum(p->x.kids[0]) - 8; int dst = getregnum(p) - 8; print(“\tmove \t”); emitasm(p->kids[1], _rc_NT); print(“,n%d\n\tlua \t(r%d)-n%d,r%d\n”, reg, reg, reg, dst); } else if(p->kids[1]->x.inst == _rreg_NT){ int reg = getregnum(p->x.kids[1]) - 8; int dst = getregnum(p) - 8; print(“\tmove \t”); emitasm(p->kids[0], _rc_NT); print(“,n%d\n\tlua \t(r%d)-n%d,r%d\n”, reg, reg, reg, dst); } else assert(0); break; case MUL+I: case MUL+U: case MUL+F: { //reg: MULI1(reg, reg) “\tmpy \t%0,%1,%c\n\tasr \t%c\n // \tmove \t%c0,%c\n” if(opsize(p->op) == 1) { int s1 = getregnum(p->x.kids[0]); int s2 = getregnum(p->x.kids[1]); int d = getregnum(p); char *src1 = p->x.kids[0]->syms[RX]->x.name; char *src2 = p->x.kids[1]->syms[RX]->x.name; if(s1 == s2){ //temp_reg = x1, x0 if x1 is taken char *temp_reg = (s1 == 0) ? (ireg[1]->x.name) : (ireg[0]->x.name); print(“\tmove \t%s,Y:(r6)\n”, temp_reg); print(“\tmove \t%s,%s\n”, src2, temp_reg); print(“\tmpy \t%s,%s,%s\n”,src1, temp_reg, p->syms[RX]->x.name); if(optype(p->op) == I || optype(p->op) == U){ print(“\tasr \t%s\n”, p->syms[RX]->x.name); print(“\tmove \t%s0,%s Y:(r6),%s\n”, p->syms[RX]->x.name, p->syms[RX]->x.name, temp_reg); } } else { print(“\tmpy \t%s,%s,%s\n”,src1, src2, p->syms[RX]->x.name); if(optype(p->op) == I || optype(p->op) == U){ print(“\tasr \t%s\n”, p->syms[RX]->x.name); print(“\tmove \t%s0,%s\n”, p->syms[RX]->x.name, p->syms[RX]->x.name); } } } else { // long mpy - copied from motorola // a and b is input regs and b is output reg int lab1 = genlabel(1); int lab2 = genlabel(1); if(optype(p->op) == F){ warning(“double multiply is not implemented; using long multiply instead\n”); } print(“\t;; begin long multiply\n”); print(“\tmoven0,y:(r6)+\n”); print(“\tmovea0,y:(r6)+\n”); print(“\tmovea1,y:(r6)+\n”); print(“\tmovex0,y:(r6)+\n”); print(“\tmovex1,y:(r6)+\n”); print(“\tmovey0,y:(r6)+\n”); print(“\tmove#$0,n0\n”); print(“\tmovea1,x0\n”); print(“\teorx0,bb1,x1\n”); print(“\tjplL%d\n”, lab1); print(“\tmove#$1,n0\n”); print(“L%d:\n”, lab1); print(“\tabsa x1,b1\n”); print(“\tabsb y1,y:(r6)+\n”); print(“\tmoveb0,y:(r6)+\n”); print(“\tmoveb1,y:(r6)\n”); print(“\tclrb b1,x0\n”); print(“\tmovex0,b0\n”); print(“\taslb\n”); print(“\taslb\n”); print(“\tclrb b1,x0\n”); print(“\tmovea1,b0\n”); print(“\taslb\n”); print(“\taslb #$7fffff,y1\n”); print(“\tmoveb1,x1\n”); print(“\tmovey:(r6)-,b\n”); print(“\tmovey:(r6),b0\n”); print(“\tmovex0,y:(r6)+\n”); print(“\tmovex1,y:(r6)\n”);

70

Appendix C – dsp56k.md print(“\taslb\n”); print(“\tandy1,b\n”); print(“\tmoveb1,x0\n”); print(“\tasrb\n”); print(“\tmoveb0,b\n”); print(“\tandy1,b\n”); print(“\tmoveb1,x1\n”); print(“\tasla\n”); print(“\tandy1,ay1,b1\n”); print(“\tmovea1,y0\n”); print(“\tasra\n”); print(“\tmovea0,y1\n”); print(“\tandy1,b\n”); print(“\tmoveb1,y1\n”); print(“\tmpyx1,y1,b\n”); print(“\tmpyx1,y0,a\n”); print(“\tmacx0,y1,a\n”); print(“\tmovea1,a2\n”); print(“\tmovea0,a1\n”); print(“\tmove#$0,a0\n”); print(“\tasra\n”); print(“\tadda,b\n”); print(“\tmpyx0,y0,a\n”); print(“\tmovey:(r6)-,y0\n”); print(“\tmovey:(r6)-,x0\n”); print(“\tmacy0,x1,a\n”); print(“\tmacx0,y1,a\n”); print(“\tclra a0,x0\n”); print(“\tmovex0,a2\n”); print(“\tasra y:(r6)-,y1\n”); print(“\tasra y:(r6)-,y0\n”); print(“\tadda,by:(r6)-,x1\n”); print(“\tasrb y:(r6)-,x0\n”); print(“\tmoven0,a\n”); print(“\ttsta y:(r6)-,a\n”); print(“\tjeqL%d\n”, lab2); print(“\tnegb\n”); print(“L%d:\n”, lab2); print(“\ttstb y:(r6)-,a0\n”); print(“\tmovey:(r6),n0\n”); print(“\t;; end long multiply\n”); } break; } case DIV+I: case DIV+U: case DIV+F:{ if(opsize(p->op) == 1){ char *src = ireg[getregnum(p->x.kids[1])]->x.name; char *dst = p->syms[RX]->x.name; int lab = genlabel(1); print(“\tmove \tb,Y:(r6)+\n”); print(“\tabs \t%s %s,b\n”, dst, dst); //b if(optype(p->op) == I || optype(p->op) == U){ print(“\tclr \t%s %s1,Y:(r6)\n”, dst, dst); print(“\tmove \tY:(r6),%s0\n”, dst); print(“\tasl \t%s\n”, dst); } print(“\trep \t#24\n”); print(“\tdiv \t%s,%s\n”, src, dst); print(“\teor \t%s,b\n”, src); //b print(“\tjpl \tL%d\n”, lab); print(“\tneg \t%s\n”, dst); print(“L%d:\tmove \t%s0,%s\n”, lab, dst, dst); print(“\tmove \tY:-(r6),b\n”); } else { // long div - copied from motorola // a and b is input regs and a is output reg int lab[11]; int i; for(i = 0; i < 11; i++){ lab[i] = genlabel(1); } if(optype(p->op) == F){ warning(“double division is not implemented; using long division instead\n”); } print(“\t;;-- begin long division\n”); print(“move n0,y:(r6)+\n”); print(“move b0,y:(r6)+\n”); print(“move b1,y:(r6)+\n”); print(“move x0,y:(r6)+\n”); print(“move x1,y:(r6)+\n”); print(“move y0,y:(r6)+\n”); print(“move y1,y:(r6)+\n”); print(“\n”); print(“move #$0,n0\n”); print(“move b1,y1\n”); print(“eor y1,a a1,y0\n”); print(“jpl L%d\n”, lab[2]); print(“move #$1,n0\n”); print(“L%d:\n”, lab[2]); print(“abs b y0,a1\n”); print(“abs a #$0,x1\n”); print(“\n”);

71

print(“move b1,y1\n”); print(“clr b b0,y0\n”); print(“move b1,y:(r6)\n”); print(“\n”); print(“ori #$04,mr\n”); print(“asl a #>$1,x0\n”); print(“andi #$fe,ccr\n”); print(“jec L%d\n”, lab[3]); print(“ori #$01,ccr\n”); print(“L%d:\n”, lab[3]); print(“andi #$f3,mr\n”); print(“div x1,b\n”); print(“\n”); print(“do #$2f,L%d\n”, lab[4]); print(“btst #23,y:(r6)\n”); print(“jcs L%d\n”, lab[5]); print(“add x,a\n”); print(“move #$0,a2\n”); print(“ori #$04,mr\n”); print(“asl a\n”); print(“andi #$fe,ccr\n”); print(“jec L%d\n”, lab[6]); print(“ori #$01,ccr\n”); print(“L%d:\n”, lab[6]); print(“andi #$f3,mr\n”); print(“div x1,b\n”); print(“sub y,b\n”); print(“jmp L%d\n”, lab[7]); print(“L%d:\n”, lab[5]); print(“move #$0,a2\n”); print(“ori #$04,mr\n”); print(“asl a\n”); print(“andi #$fe,ccr\n”); print(“jec L%d\n”, lab[8]); print(“ori #$01,ccr\n”); print(“L%d:\n”, lab[8]); print(“andi #$f3,mr\n”); print(“div x1,b\n”); print(“add y,b\n”); print(“L%d:\n”, lab[7]); print(“move a1,x1\n”); print(“move b1,a1\n”); print(“eor y1,a\n”); print(“move b1,y:(r6)\n”); print(“move x1,a1\n”); print(“move #$0,x1\n”); print(“L%d:\n”, lab[4]); print(“btst #23,y:(r6)\n”); print(“jcs L%d\n”, lab[9]); print(“add x,a\n”); print(“L%d:\n”, lab[9]); print(“move #$80,x1\n”); print(“eor x1,a (r6)-\n”); print(“move #$0,a2\n”); print(“\n”); print(“move n0,b\n”); print(“tst b\n”); print(“jeq L%d\n”, lab[10]); print(“neg a\n”); print(“L%d:\n”, lab[10]); print(“move y:(r6)-,y1\n”); print(“move y:(r6)-,y0\n”); print(“move y:(r6)-,x1\n”); print(“move y:(r6)-,x0\n”); print(“move y:(r6)-,b\n”); print(“tst a y:(r6)-,b0\n”); print(“move y:(r6),n0\n”); print(“\t;;-- end long division\n”); } break; } case MOD+I: case MOD+U: { if(opsize(p->op) == 1){ char *src = ireg[getregnum(p->x.kids[1])]->x.name; char *dst = p->syms[RX]->x.name; int lab = genlabel(1); print(“\tmove \tb,Y:(r6)+\n”); print(“\tabs \t%s %s,b\n”, dst, dst); //b print(“\tclr \t%s %s1,Y:(r6)\n”, dst, dst); print(“\tmove \tY:(r6),%s0\n”, dst); print(“\tasl \t%s\n”, dst); print(“\trep \t#24\n”); print(“\tdiv \t%s,%s\n”, src, dst); print(“\tmove \t%s1,Y:(r6)\n”, dst); print(“\tmove \t%s,%s\n”, src, dst); print(“\tabs \t%s Y:(r6),%s\n”, dst, src);//destroy x0 (src) print(“\tadd \t%s,a\n”, src); print(“\tasr \t%s\n”, dst); print(“\ttst \tb\n”); //b print(“\tjge \tL%d\n”, lab);

72

Appendix C – dsp56k.md print(“\tneg \t%s\n”, dst); print(“L%d:\n”, lab); print(“\tmove \tY:-(r6),b\n”); } else { // long modulo - copied from motorola // a and b is input regs and b is output reg int lab[14]; int i; for(i = 0; i < 14; i++){ lab[i] = genlabel(1); } print(“\t;;-- begin long modulo\n”); print(“move n0,y:(r6)+\n”); print(“move b0,y:(r6)+\n”); print(“move b1,y:(r6)+\n”); print(“move x0,y:(r6)+\n”); print(“move x1,y:(r6)+\n”); print(“move y0,y:(r6)+\n”); print(“move y1,y:(r6)+\n”); print(“move r0,y:(r6)+\n”); print(“move r1,y:(r6)+\n”); print(“move #$0,n0\n”); print(“tst a\n”); print(“jpl L%d\n”, lab[12]); print(“abs a\n”); print(“move #$1,n0\n”); print(“L%d:\n”, lab[12]); print(“move b1,y1\n”); print(“tfr a,b b0,y0\n”); print(“\n”); print(“clr a #>$1,x0\n”); print(“tst b #$0,x1\n”); print(“jpl L%d\n”, lab[1]); print(“sub x,a\n”); print(“L%d:\n”, lab[1]); print(“move b1,x1\n”); print(“move a1,b1\n”); print(“eor y1,b r6,r0\n”); print(“move b1,y:(r6)+\n”); print(“move r6,r1\n”); print(“move b1,y:(r6)+\n”); print(“move x1,b1\n”); print(“move #$0,b2\n”); print(“ori #$04,mr\n”); print(“asl b #$0,x1\n”); print(“andi #$fe,ccr\n”); print(“jec L%d\n”, lab[2]); print(“ori #$01,ccr\n”); print(“L%d:\n”, lab[2]); print(“andi #$f3,mr\n”); print(“div x1,a\n”); print(“\n”); print(“do #$2f,L%d\n”, lab[3]); print(“btst #23,y:(r1)\n”); print(“jcs L%d\n”, lab[4]); print(“add x,b\n”); print(“move #$0,b2\n”); print(“ori #$04,mr\n”); print(“asl b\n”); print(“andi #$fe,ccr\n”); print(“jec L%d\n”, lab[5]); print(“ori #$01,ccr\n”); print(“L%d:\n”, lab[5]); print(“andi #$f3,mr\n”); print(“div x1,a\n”); print(“sub y,a\n”); print(“jmp L%d\n”, lab[6]); print(“L%d:\n”, lab[4]); print(“move #$0,b2\n”); print(“ori #$04,mr\n”); print(“asl b\n”); print(“andi #$fe,ccr\n”); print(“jec L%d\n”, lab[7]); print(“ori #$01,ccr\n”); print(“L%d:\n”, lab[7]); print(“andi #$f3,mr\n”); print(“div x1,a\n”); print(“add y,a\n”); print(“L%d:\n”, lab[6]); print(“move b1,x1\n”); print(“move a1,b1\n”); print(“eor y1,b\n”); print(“move b1,y:(r1)\n”); print(“move x1,b1\n”); print(“move #$0,x1\n”); print(“L%d:\n”, lab[3]); print(“btst #23,y:(r1)\n”); print(“jcs L%d\n”, lab[8]); print(“add x,b\n”); print(“L%d\n”, lab[8]); print(“move #$80,x1\n”); print(“eor x1,b\n”);

73

print(“move #$0,x1\n”); print(“btst #23,y:(r0)\n”); print(“jcc L%d\n”, lab[9]); print(“add x,b\n”); print(“L%d:\n”, lab[9]); print(“move b1,x1\n”); print(“move y:(r0),b1\n”); print(“move y:(r1),x0\n”); print(“eor x0,b\n”); print(“move x1,b1\n”); print(“jpl L%d\n”, lab[10]); print(“btst #23,y:(r0)\n”); print(“jcc L%d\n”, lab[11]); print(“sub y,a\n”); print(“jmp L%d\n”, lab[10]); print(“L%d:\n”, lab[11]); print(“add y,a\n”); print(“L%d:\n”, lab[10]); print(“move n0,b\n”); print(“tst b\n”); print(“jeq L%d\n”, lab[13]); print(“neg a\n”); print(“L%d:\n”, lab[13]); print(“move (r6)-\n”); print(“move (r6)-\n”); print(“move (r6)-\n”); print(“move y:(r6)-,r1\n”); print(“move y:(r6)-,r0\n”); print(“move y:(r6)-,y1\n”); print(“move y:(r6)-,y0\n”); print(“move y:(r6)-,x1\n”); print(“move y:(r6)-,x0\n”); print(“move y:(r6)-,b\n”); print(“tst a y:(r6)-,b0\n”); print(“move y:(r6),n0\n”); print(“\t;;-- end long modulo\n”); } break; } case ARG+I: case ARG+U: case ARG+F:{ if(opsize(p->op) == 2){ int src = getregnum(p->x.kids[0]); //stmt: ARGI2(reg) “# \tmove \t%0,Y:(r6)+\n” print(“\tmove \t%s,Y:(r6)+\n”, reg_name(src)); print(“\tmove \t%s,Y:(r6)+\n”, reg_name(src + 1)); } break; } case ARG+B: { int src = getregnum(p->x.kids[0]) - 8; int label = genlabel(1); int size = p->syms[0]->u.c.v.i; print(“\tmove \t#%d,n6\n”, size); print(“\tmove \tr%d,n%d\n”, src, src); print(“\tmove \tx0,Y:(r6+n6)\n”); if(size > 1) { print(“\tdo \t#%d,L%d\n”, size , label); } print(“\tmove \tY:(r%d)+,x0\n”, src); print(“\tmove \tx0,Y:(r6)+\n”); if(size > 1) { print(“L%d:\n”, label); } print(“\tmove \tn%d,r%d\n”, src, src); print(“\tmove \tY:(r6),x0\n”); break; } case BAND+I: case BAND+U: case BOR+I: case BOR+U: case BXOR+I: case BXOR+U: { //reg: BANDI2(reg, reg) “\tand \t%0,%1\n” int src = getregnum(p->x.kids[0]); int dst = getregnum(p->x.kids[1]); print(“\tand \t%s,”, reg_name(src + 1)); emitasm(p->kids[1], _reg_NT); print(“\t%s,Y:(r6)\n”, reg_name(dst)); print(“\tmove \t%s,%s\n”, reg_name(dst + 1), reg_name(dst)); print(“\tmove \tY:(r6),%s\n”, reg_name(dst + 1)); switch(generic(p->op)){ case BAND: print(“\tand \t%s,”, reg_name(src)); break; case BOR: print(“\tor \t%s,”, reg_name(src)); break; case BXOR: print(“\teor \t%s,”, reg_name(src)); break; } emitasm(p->kids[1], _reg_NT);

74

Appendix C – dsp56k.md print(“\n\tmove \t%s,Y:(r6)\n”, reg_name(dst + 1)); print(“\tmove \t%s,”, reg_name(dst)); emitasm(p->kids[1], _reg_NT); print(“\n\tmove \tY:(r6),%s\n”, reg_name(dst)); break; } case BCOM+I: case BCOM+U: { //reg: BCOMI2(reg) “\tnot \t%0\n” int src = getregnum(p->x.kids[0]); print(“\tnot \t%s\n”, reg_name(src + 1)); print(“\tmove \t%s,Y:(r6)\n”, reg_name(src + 1)); print(“\tmove \t%s,%s\n”, reg_name(src), reg_name(src + 1)); print(“\tnot \t%s\n”, reg_name(src + 1)); print(“\tmove \t%s,%s\n”, reg_name(src + 1), reg_name(src)); print(“\tmove \tY:(r6),%s\n”, reg_name(src + 1)); break; } case RET+I: case RET+U: case RET+F: break; case CVI+I: case CVI+U: case CVI+F: case CVU+I: case CVU+U: case CVF+I: case CVF+F: { char *src = p->kids[0]->syms[RX]->x.name; char *dst = p->syms[RX]->x.name; if(src == areg2[1]->x.name){ print(“\ttfr \t%s,%s\n”, src, dst); } else { print(“\tmove \t%s,%s\n”, src, dst); } break; } case LOAD+I: case LOAD+U: case LOAD+F: assert(opsize(p->op) == 2); if(opsize(p->x.kids[0]->op) == 2){ int src = getregnum(p->x.kids[0]); int dst = getregnum(p); print(“\tmove \t%s,%s\n”, reg_name(src), reg_name(dst)); print(“\tmove \t%s,%s\n”, reg_name(src+1), reg_name(dst+1)); } else { print(“\tmove \t%s,%s\n”, p->kids[0]->syms[RX]->x.name, p->syms[RX]->x.name); assert(0); } break; default: break; } } static void doarg(Node p) { debug2(fprint(stderr, “Inside %s: %x\n”, “doarg”, p)); mkactual(1, p->syms[0]->u.c.v.i); } static void blkfetch(int k, int off, int reg, int tmp) { debug2(fprint(stderr, “Inside %s\n”, “blkfetch”)); } static void blkstore(int k, int off, int reg, int tmp) { debug2(fprint(stderr, “Inside %s\n”, “blkstore”)); } static void blkloop(int dreg, int doff, int sreg, int soff, int size, int tmps[]) { debug2(fprint(stderr, “Inside %s\n”, “blkloop”)); } static void local(Symbol p) { debug2(fprint(stderr, “Inside %s: %s, retstruct: %d\n”, “local”, p->name, retstruct)); if (retstruct) { assert(p == retv); p->sclass = REGISTER; if(askregvar(p, rreg[7]) == 0) { assert(0); } retstruct = -1; return; } if(askregvar(p, rmap(ttob(p->type))) == 0) { mkauto(p); } } static void function(Symbol f, Symbol caller[], Symbol callee[], int n) { int i; debug2(fprint(stderr,”Inside %s: %s, %d\n”, “function”, f->name, n)); usedmask[0] = usedmask[1] = 0; freemask[0] = freemask[1] = ~(unsigned)0; offset = -3; for(i = 0; callee[i]; i++) {

75

Symbol p = callee[i]; Symbol q = caller[i]; assert(q); if(q->type->size == 2) { //long, double p->x.offset = q->x.offset = offset - 1; p->x.name = q->x.name = stringf(“%d”, offset - 1); } else { //char, short, int, float p->x.offset = q->x.offset = offset; p->x.name = q->x.name = stringf(“%d”, offset); } p->sclass = q->sclass = AUTO; offset -= q->type->size; } assert(caller[i] == 0); offset = maxoffset = 0; retstruct = isstruct(freturn(f->type)); debug2(fprint(stderr, “gencode(%s):\n”, f->name)); gencode(caller, callee); print(“\n ;;;; Function %s starts\n”, f->name); print(“\tglobal\t%s\n”, f->x.name); print(“%s:\n”, f->x.name); print(“\tmove \tr0,Y:(r6)+ \t;; save frame pointer\n”); print(“\tlua \t(r6)+,r0 \t;; set new frame pointer\n”); print(“\tmove \tssh,Y:(r6)+ \t;; save return address\n”); debug2(fprint(stderr, “usedmask[IREG]: %x, usedmask[FREG]: %x\n”, usedmask[IREG], usedmask[FREG])); for(i = 0; i < 4; i++) { if(usedmask[IREG] & (1 << i)) { print(“\tmove \t%s,Y:(r6)+ \t;; save ireg\n”, ireg[i]->x.name); } } for(i = 6; i < 8; i++) { if(usedmask[IREG] & (1 << i)) { print(“\tmove \t%s,Y:(r6)+ \t;; save areg\n”, areg[i-4]->x.name); } } for(i = 8; i < 16; i++) { if(usedmask[IREG] & (1 << i)) { print(“\tmove \t%s,Y:(r6)+ \t;; save rreg\n”, rreg[i-8]->x.name); } } debug2(fprint(stderr, “offset: %d, maxoffset: %d, argoffset: %d, maxargoffset: %d\n”, offset, maxoffset, argoffset, maxargoffset)); if(maxoffset > 0) { print(“\tmove \t#%d,n6 \t\t;; update stack with local and temp offset\n”, maxoffset); print(“\tmove \t(r6)+n6 \t;;\n”); } if(retstruct == -1){ print(“\tmove \ta,r7\n”); retstruct = 0; } debug2(fprint(stderr, “emitcode(%s):\n”, f->name)); emitcode(); if(maxoffset > 0) { print(“\tmove \t#%d,n6 \t\t;; restore stack\n”, maxoffset); print(“\tmove \t(r6)-n6 \t;;\n”); } for(i = 15; i >= 8; i--) { if(usedmask[IREG] & (1 << i)) { print(“\tmove \tY:-(r6),%s \t;; restore rreg\n”, rreg[i-8]->x.name); } } if(usedmask[IREG] & (1 << 7)) { //FIX b1.b0 ->b.b0 print(“\tmove \t%Y:-(r6),%s \t;; restore areg\n”, areg2[1]->x.name); } if(usedmask[IREG] & (1 << 6)) { print(“\tmove \t%Y:-(r6),%s \t;; restore areg\n”, areg[2]->x.name); } for(i = 3; i >= 0; i--) { if(usedmask[IREG] & (1 << i)) { print(“\tmove \t%Y:-(r6),%s \t;; restore ireg\n”, ireg[i]->x.name); } }

76

Appendix C – dsp56k.md print(“\tmove \tY:-(r6),ssh \t;; restore return address\n”); print(“\ttst \ta Y:-(r6),r0 \t;; restore frame pointer and test a\n”); print(“\trts \n”); print(“ ;;;; Function %s ends\n\n”, f->name); } static void defsymbol(Symbol p) { debug2(fprint(stderr, “Inside %s: %s -> “, “defsymbol”, p->name)); if(p->scope >= LOCAL && p->sclass == STATIC) { p->x.name = stringf(“F__%s%d”, cfunc->name, genlabel(1)); } else if(p->generated) { p->x.name = stringf(“L%s”, p->name); } else if(p->scope == CONSTANTS && (isint(p->type) || isptr(p->type))) { if(p->name[0] == ‘0’ && p->name[1] == ‘x’) { p->x.name = stringf(“$%s”, &p->name[2]); } else { p->x.name = p->name; } } else { p->x.name = stringf(“F%s”, p->name); } debug2(fprint(stderr, “%s\n”, p->x.name)); } static void address(Symbol q, Symbol p, long n) { debug2(fprint(stderr, “Inside %s: %s, %s, %d\n”, “address”, q->name, p->name, n)); if (p->scope == GLOBAL || p->sclass == STATIC || p->sclass == EXTERN) q->x.name = stringf(“%s%s%D”, p->x.name, n >= 0 ? “+” : ““, n); else { assert(n <= INT_MAX && n >= INT_MIN); q->x.offset = p->x.offset + n; q->x.name = stringd(q->x.offset); } } static void defconst(int suffix, int size, Value v) { debug2(fprint(stderr, “Inside %s: %d, %d, %d \n”, “defconst”, suffix, size, v.i)); if (suffix == F) { if(size == 1) { unsigned u; if(v.d == 1.0){ u = 0x7FFFFF; } else { u = (unsigned)(v.d * 8388608) & 0xFFFFFF; } print(“\tdc\t$%x\n”, u); } else { long long u; if(v.d == 1.0){ u = 0x7FFFFFFFFFFFLL; } else { u = (long long)(v.d * 16777216*8388608) & 0xFFFFFFFFFFFFLL; } print(“\tdc\t$%x\n”, (unsigned)(u & 0xFFFFFF)); print(“\tdc\t$%x\n”, (unsigned)(u >> 24)); } } else if (suffix == P) { print(“\tdc\t$%x\n”, (unsigned)v.p); } else if (size == 1) { print(“\tdc\t$%x\n”, (unsigned)(suffix == I ? v.i : v.u)); } else if (size == 2) { print(“\tdc\t$%x\n”, (unsigned long)((suffix == I ? v.i : v.u) & 0xFFFFFF)); print(“\tdc\t$%x\n”, (unsigned long)((suffix == I ? v.i : v.u) >> 24)); } } static void defaddress(Symbol p) { debug2(fprint(stderr, “Inside %s: %s\n”, “defaddress”, p->name)); print(“\tdc\t%s\n”, p->x.name); } static void defstring(int n, char *str) { int i; debug2(fprint(stderr, “Inside %s: “, “defstring”)); if(n == -1){ print(“\tdc\t%d\n”, *(int *)str); } else { for(i = 0; i < n; i++) { debug2(fprint(stderr, “%d - %c, “, str[i], str[i])); print(“\tdc\t%d\n”, str[i]); } } debug2(fprint(stderr, “, %d\n”, n)); } static void export(Symbol p) { debug2(fprint(stderr, “Inside %s: %s\n”, “export”, p->name)); } static void import(Symbol p) {

77

debug2(fprint(stderr, “Inside %s: %s\n”, “import”, p->name)); } static void global(Symbol p) { debug2(fprint(stderr, “Inside %s: %s\n”, “global”, p->name)); print(“\tglobal\t%s\n%s\n”, p->x.name, p->x.name); } static void space(int n) { debug2(fprint(stderr, “Inside %s: %d\n”, “space”, n)); print(“\tbsc\t%d\n”, n); } Interface dsp56kIR = { 1, 1, 0, /* char */ 1, 1, 0, /* short */ 1, 1, 0, /* int */ 2, 2, 0, /* long */ 2, 2, 0, /* long long */ 1, 1, 1, /* float */ 2, 2, 1, /* double */ 2, 2, 1, /* long double */ 1, 1, 0, /* T * */ 0, 1, 0, /* struct */ 1, /* little_endian */ 0, /* mulops_calls */ 1, /* wants_callb */ 1, /* wants_argb */ 0, /* left_to_right */ 0, /* wants_dag */ 0, /* unsigned_char */ address, blockbeg, blockend, defaddress, defconst, defstring, defsymbol, emit, export, function, gen, global, import, local, progbeg, progend, segment, space, 0, 0, 0, 0, 0, 0, 0, { 1, /* max_unaligned_load */ rmap, blkfetch, blkstore, blkloop, _label, _rule, _nts, _kids, _string, _templates, _isinstruction, _ntname, emit2, doarg, target, clobber, } }; static char rcsid[] = “$Id: dsp56k.md,v 1.24 2004/08/19 13:42:43 henan702 Exp $”;

78

Index A

H

abstract syntax tree . . . . . . . . accumulator register . . . . . . . activation record . . . . . . . . . address register . . . . . . . . . . ANSI C . . . . . . . . . . . . . . . .

26 .5 37 .5 24

B boilerplate bootstrap .

. . . . . . . . . . . . . . 41 . . . . . . . . . . . . . . 46

. . . . . . . .4 . . . . . . . 36

I input register

. . . . . . . . . . . . .5

K . . . . . . . . . . . . . . . . . 24

K&R C

L

C C89 C90 C99

Harvard architecture heap . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . 24 . . . . . . . . . . . . . . . . . . . 24 . . . . . . . . . . . . . . . . . . . 24

D dag . . . . . . . . . . . . . . . . . . . DC . . . . . . . . . . . . . . . . . . . derivation . . . . . . . . . . . . . . directed acyclic graph . . . . . .

29 .7 13 29

E EBNF

. . . . . . . . . . . . . . . . . 26

F fixed-point . . . . . . . . . . . . . . fractional number . . . . . . . . .

.4 34

G GCC . . . . . . . . . . . . . . . . . . 23 GLOBAL . . . . . . . . . . . . . . . . 7 global data bus . . . . . . . . . . . . 5 GNU C Compiler . . . . . . . . . 23 GNU Compiler Collection . . . 23

lburg

. . . . . . . . . . . . . . . 30, 39

M modifier register

. . . . . . . . . . .6

N nonterminal

. . . . . . . . . . . . . 12

O offset register . . . . . . . . . . . . . 5 ops . . . . . . . . . . . . . . . . . . . . 40 OPT . . . . . . . . . . . . . . . . . . . . 7 ORG . . . . . . . . . . . . . . . . . . . . 7

P P address bus . . . . . . . . . . . . . 5 parse tree . . . . . . . . . . . . 11, 26 parser . . . . . . . . . . . . . . . . . . 26 parsing . . . . . . . . . . . . . . . . . 11 production . . . . . . . . . . . . . . 12 program data bus . . . . . . . . . . 5

R retarget

. . . . . . . . . . . . . . . . . 19 79

S scanning . . . . . . . . . . . . . . . stack . . . . . . . . . . . . . . . . . . start symbol . . . . . . . . . . . . . syntax tree . . . . . . . . . . . . . .

11 36 13 13

T template . . . . . . . . . . . . 31, 32 terminal . . . . . . . . . . . . . . . . 12 three-address code . . . . . . . . 15 token . . . . . . . . . . . . . . . 11, 24 tree grammar . . . . . . . . . . . . 30 tree parser . . . . . . . . . . . . . . 30

W wildcard

. . . . . . . . . . . . 40, 46

X X address bus . . . . . . . . . . . . X data bus . . . . . . . . . . . . . . .

5 5

Y Y address bus . . . . . . . . . . . . Y data bus . . . . . . . . . . . . . . .

80

5 5

På svenska Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare – under en längre tid från publiceringsdatum under förutsättning att inga extra-ordinära omständigheter uppstår. Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner, skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för ickekommersiell forskning och för undervisning. Överföring av upphovsrätten vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av dokumentet kräver upphovsmannens medgivande. För att garantera äktheten, säkerheten och tillgängligheten finns det lösningar av teknisk och administrativ art. Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i den omfattning som god sed kräver vid användning av dokumentet på ovan beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan form eller i sådant sammanhang som är kränkande för upphovsmannens litterära eller konstnärliga anseende eller egenart. För ytterligare information om Linköping University Electronic Press se förlagets hemsida http://www.ep.liu.se/

In English The publishers will keep this document online on the Internet - or its possible replacement - for a considerable time from the date of publication barring exceptional circumstances. The online availability of the document implies a permanent permission for anyone to read, to download, to print out single copies for your own use and to use it unchanged for any non-commercial research and educational purpose. Subsequent transfers of copyright cannot revoke this permission. All other uses of the document are conditional on the consent of the copyright owner. The publisher has taken technical and administrative measures to assure authenticity, security and accessibility. According to intellectual property law the author has the right to be mentioned when his/her work is accessed as described above and to be protected against infringement. For additional information about the Linköping University Electronic Press and its procedures for publication and for assurance of document integrity, please refer to its WWW home page: http://www.ep.liu.se/ © Henrik Antelius