CA850 Ver. 3.20 C Compiler Package C Language UM

CA850 Ver. 3.20 C Compiler Package C Language UM
To our customers,
Old Company Name in Catalogs and Other Documents
On April 1st, 2010, NEC Electronics Corporation merged with Renesas Technology
Corporation, and Renesas Electronics Corporation took over all the business of both
companies. Therefore, although the old company name remains in this document, it is a valid
Renesas Electronics document. We appreciate your understanding.
Renesas Electronics website: http://www.renesas.com
April 1st, 2010
Renesas Electronics Corporation
Issued by: Renesas Electronics Corporation (http://www.renesas.com)
Send any inquiries to http://www.renesas.com/inquiry.
Note that the following URLs in this document are not available:
http://www.necel.com/
http://www2.renesas.com/
Please refer to the following instead:
Development Tools | http://www.renesas.com/tools
Download | http://www.renesas.com/tool_download
For any inquiries or feedback, please contact your region.
http://www.renesas.com/inquiry
Notice
1.
2.
3.
4.
5.
6.
7.
All information included in this document is current as of the date this document is issued. Such information, however, is
subject to change without any prior notice. Before purchasing or using any Renesas Electronics products listed herein, please
confirm the latest product information with a Renesas Electronics sales office. Also, please pay regular and careful attention to
additional and different information to be disclosed by Renesas Electronics such as that disclosed through our website.
Renesas Electronics does not assume any liability for infringement of patents, copyrights, or other intellectual property rights
of third parties by or arising from the use of Renesas Electronics products or technical information described in this document.
No license, express, implied or otherwise, is granted hereby under any patents, copyrights or other intellectual property rights
of Renesas Electronics or others.
You should not alter, modify, copy, or otherwise misappropriate any Renesas Electronics product, whether in whole or in part.
Descriptions of circuits, software and other related information in this document are provided only to illustrate the operation of
semiconductor products and application examples. You are fully responsible for the incorporation of these circuits, software,
and information in the design of your equipment. Renesas Electronics assumes no responsibility for any losses incurred by
you or third parties arising from the use of these circuits, software, or information.
When exporting the products or technology described in this document, you should comply with the applicable export control
laws and regulations and follow the procedures required by such laws and regulations. You should not use Renesas
Electronics products or the technology described in this document for any purpose relating to military applications or use by
the military, including but not limited to the development of weapons of mass destruction. Renesas Electronics products and
technology may not be used for or incorporated into any products or systems whose manufacture, use, or sale is prohibited
under any applicable domestic or foreign laws or regulations.
Renesas Electronics has used reasonable care in preparing the information included in this document, but Renesas Electronics
does not warrant that such information is error free. Renesas Electronics assumes no liability whatsoever for any damages
incurred by you resulting from errors in or omissions from the information included herein.
Renesas Electronics products are classified according to the following three quality grades: “Standard”, “High Quality”, and
“Specific”. The recommended applications for each Renesas Electronics product depends on the product’s quality grade, as
indicated below. You must check the quality grade of each Renesas Electronics product before using it in a particular
application. You may not use any Renesas Electronics product for any application categorized as “Specific” without the prior
written consent of Renesas Electronics. Further, you may not use any Renesas Electronics product for any application for
which it is not intended without the prior written consent of Renesas Electronics. Renesas Electronics shall not be in any way
liable for any damages or losses incurred by you or third parties arising from the use of any Renesas Electronics product for an
application categorized as “Specific” or for which the product is not intended where you have failed to obtain the prior written
consent of Renesas Electronics. The quality grade of each Renesas Electronics product is “Standard” unless otherwise
expressly specified in a Renesas Electronics data sheets or data books, etc.
“Standard”:
8.
9.
10.
11.
12.
Computers; office equipment; communications equipment; test and measurement equipment; audio and visual
equipment; home electronic appliances; machine tools; personal electronic equipment; and industrial robots.
“High Quality”: Transportation equipment (automobiles, trains, ships, etc.); traffic control systems; anti-disaster systems; anticrime systems; safety equipment; and medical equipment not specifically designed for life support.
“Specific”:
Aircraft; aerospace equipment; submersible repeaters; nuclear reactor control systems; medical equipment or
systems for life support (e.g. artificial life support devices or systems), surgical implantations, or healthcare
intervention (e.g. excision, etc.), and any other applications or purposes that pose a direct threat to human life.
You should use the Renesas Electronics products described in this document within the range specified by Renesas Electronics,
especially with respect to the maximum rating, operating supply voltage range, movement power voltage range, heat radiation
characteristics, installation and other product characteristics. Renesas Electronics shall have no liability for malfunctions or
damages arising out of the use of Renesas Electronics products beyond such specified ranges.
Although Renesas Electronics endeavors to improve the quality and reliability of its products, semiconductor products have
specific characteristics such as the occurrence of failure at a certain rate and malfunctions under certain use conditions. Further,
Renesas Electronics products are not subject to radiation resistance design. Please be sure to implement safety measures to
guard them against the possibility of physical injury, and injury or damage caused by fire in the event of the failure of a
Renesas Electronics product, such as safety design for hardware and software including but not limited to redundancy, fire
control and malfunction prevention, appropriate treatment for aging degradation or any other appropriate measures. Because
the evaluation of microcomputer software alone is very difficult, please evaluate the safety of the final products or system
manufactured by you.
Please contact a Renesas Electronics sales office for details as to environmental matters such as the environmental
compatibility of each Renesas Electronics product. Please use Renesas Electronics products in compliance with all applicable
laws and regulations that regulate the inclusion or use of controlled substances, including without limitation, the EU RoHS
Directive. Renesas Electronics assumes no liability for damages or losses occurring as a result of your noncompliance with
applicable laws and regulations.
This document may not be reproduced or duplicated, in any form, in whole or in part, without prior written consent of Renesas
Electronics.
Please contact a Renesas Electronics sales office if you have any questions regarding the information contained in this
document or Renesas Electronics products, or if you have any other inquiries.
(Note 1) “Renesas Electronics” as used in this document means Renesas Electronics Corporation and also includes its majorityowned subsidiaries.
(Note 2) “Renesas Electronics product(s)” means any product developed or manufactured by or for Renesas Electronics.
User’s Manual
CA850 Ver. 3.20
C Compiler Package
C Language
Target Device
V850 Series
Document No. U18513EJ1V0UM00 (1st edition)
Date Published May 2007 CP(K)
© NEC Electronics Corporation 2007
Printed in Japan
[MEMO]
2
User’s Manual U18513EJ1V0UM
Windows is either a registered trademark or a trademark of Microsoft Corporation in the United States
and/or other countries.
User’s Manual U18513EJ1V0UM
3
• The information in this document is current as of May, 2007. The information is subject to change
without notice. For actual design-in, refer to the latest publications of NEC Electronics data sheets or
data books, etc., for the most up-to-date specifications of NEC Electronics products. Not all
products and/or types are available in every country. Please check with an NEC Electronics sales
representative for availability and additional information.
• No part of this document may be copied or reproduced in any form or by any means without the prior
written consent of NEC Electronics. NEC Electronics assumes no responsibility for any errors that may
appear in this document.
• NEC Electronics does not assume any liability for infringement of patents, copyrights or other intellectual
property rights of third parties by or arising from the use of NEC Electronics products listed in this document
or any other liability arising from the use of such products. No license, express, implied or otherwise, is
granted under any patents, copyrights or other intellectual property rights of NEC Electronics or others.
• Descriptions of circuits, software and other related information in this document are provided for illustrative
purposes in semiconductor product operation and application examples. The incorporation of these
circuits, software and information in the design of a customer's equipment shall be done under the full
responsibility of the customer. NEC Electronics assumes no responsibility for any losses incurred by
customers or third parties arising from the use of these circuits, software and information.
• While NEC Electronics endeavors to enhance the quality, reliability and safety of NEC Electronics products,
customers agree and acknowledge that the possibility of defects thereof cannot be eliminated entirely. To
minimize risks of damage to property or injury (including death) to persons arising from defects in NEC
Electronics products, customers must incorporate sufficient safety measures in their design, such as
redundancy, fire-containment and anti-failure features.
• NEC Electronics products are classified into the following three quality grades: "Standard", "Special" and
"Specific".
The "Specific" quality grade applies only to NEC Electronics products developed based on a customerdesignated "quality assurance program" for a specific application. The recommended applications of an NEC
Electronics product depend on its quality grade, as indicated below. Customers must check the quality grade of
each NEC Electronics product before using it in a particular application.
"Standard": Computers, office equipment, communications equipment, test and measurement equipment, audio
and visual equipment, home electronic appliances, machine tools, personal electronic equipment
and industrial robots.
"Special": Transportation equipment (automobiles, trains, ships, etc.), traffic control systems, anti-disaster
systems, anti-crime systems, safety equipment and medical equipment (not specifically designed
for life support).
"Specific": Aircraft, aerospace equipment, submersible repeaters, nuclear reactor control systems, life
support systems and medical equipment for life support, etc.
The quality grade of NEC Electronics products is "Standard" unless otherwise expressly specified in NEC
Electronics data sheets or data books, etc. If customers wish to use NEC Electronics products in applications
not intended by NEC Electronics, they must contact an NEC Electronics sales representative in advance to
determine NEC Electronics' willingness to support a given application.
(Note)
(1) "NEC Electronics" as used in this statement means NEC Electronics Corporation and also includes its
majority-owned subsidiaries.
(2) "NEC Electronics products" means any product developed or manufactured by or for NEC Electronics (as
defined above).
M8E 02. 11-1
4
User’s Manual U18513EJ1V0UM
[MEMO]
User’s Manual U18513EJ1V0UM
5
INTRODUCTION
Target Devices
The V850 Series C compiler packages create the object codes for NEC
Electronics’s V850 Series RISC microcontrollers.
Readers
This manual is intended for user engineers who wish to develop application
systems using the V850 Series C compiler package.
Purpose
This manual explains the C language specifications supported by the C
compiler (ca850) included in the package.
Organization
This manual contains the following information:
• OVERVIEW
• BASIC LANGUAGE SPECIFICATIONS
• COMPILATION ENVIRONMENT
• C LANGUAGE EXPANSION
• CALLING PROGRAM
• STARTUP ROUTINE
• LIBRARY FUNCTION
• FOR EFFICIENT USE
Notes on reading this manual
• In this manual, sections on the V850 Series peculiar to “V850E” are
specified by the title name or the mark “ [V850E] ”. Sections peculiar to
other than “V850E” are specified by the title or the mark “ [V850] ”, etc..
6
User’s Manual U18513EJ1V0UM
Related Documents
Read this manual together with the following documents.
The related documents indicated in this publication may include preliminary
versions. However, preliminary versions are not marked as such.
Documents related to development tools (user’s manuals)
Document Name
CA850 Ver. 3.20 C Compiler Package
Document No.
Operation
U18512E
C Language
This manual
Assembly Language
U18514E
Link Directives
U18515E
U18416E
PM+ Ver. 6.30 Project Manager
ID850 Ver. 3.00 Integrated Debugger
Operation
U17358E
ID850NW Ver. 3.10 Integrated Debugger
Operation
U17369E
ID850QB Ver. 3.20 Integrated Debugger
Operation
U17964E
SM+ System Simulator
Operation
U17246E
User Open Interface
U18212E
SM850 Ver. 2.50 System Simulator
Operation
U16218E
SM850 Ver. 2.00 or Later System Simulator
External Part User Open Interface Specifications
U14873E
RX850 Ver. 3.20 or Later Real-Time OS
Basics
U13430E
Installation
U17419E
Technical
U13431E
Task Debugger
U17420E
Basics
U18165E
Internal Structure
U18164E
Task Debugger
U17422E
Functionalities
U16643E
Internal Structure
U16644E
Task Debugger
U16811E
RX850 Pro Ver. 3.21 Real-Time OS
RX850V4 Ver. 4.22 Real-Time OS
AZ850 Ver. 3.30 System Performance Analyzer
U17423E
AZ850V4 Ver. 4.10 System Performance Analyzer
U17093E
TW850 Ver. 2.00 Performance Analysis Tuning Tool
U17241E
User’s Manual U18513EJ1V0UM
7
[MEMO]
8
User’s Manual U18513EJ1V0UM
CONTENTS
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS ... 17
1.1 Dependent on Processing System Stipulated ... 18
1.1.1 Data type and size ... 18
1.1.2 Translation stages ... 18
1.1.3 Diagnosis message ... 18
1.1.4 Free-standing environment ... 18
1.1.5 Executing program ... 19
1.1.6 Character set ... 19
1.1.7 Multi-byte characters ... 19
1.1.8 Meaning of character indication ... 19
1.1.9 Translation limit ... 20
1.1.10 Quantitative limit ... 21
1.1.11 Identifier ... 23
1.1.12 char type ... 23
1.1.13 Floating-point constants ... 23
1.1.14 Character constants ... 23
1.1.15 Character string ... 24
1.1.16 Header file name ... 24
1.1.17 Comment ... 24
1.1.18 Signed constants and unsigned constants ... 25
1.1.19 Floating-points and general integers ... 25
1.1.20 double type and float type ... 25
1.1.21 Signed type in operator in bit units ... 25
1.1.22 Members of structures and unions ... 25
1.1.23 sizeof operator ... 25
1.1.24 Cast operator ... 26
1.1.25 Division/remainder operator ... 26
1.1.26 Addition and subtraction operators ... 26
1.1.27 Shift operator in bit units ... 26
1.1.28 Storage area class specifier ... 26
1.1.29 Structure and union specifier ... 27
1.1.30 Enumerate type specifier ... 27
1.1.31 Type qualifier ... 27
1.1.32 Condition embedding ... 27
1.1.33 Loading header file ... 28
1.1.34 #pragma directives ... 29
1.1.35 Predefined macro names ... 31
1.1.36 Definition of special data type ... 32
1.2 ANSI Option ... 33
CHAPTER 2 COMPILATION ENVIRONMENT ... 34
2.1 Internal Representation and Value Area of Data ... 34
2.1.1 Integer type ... 34
2.1.2 Floating-point type ... 35
2.1.3 Pointer type ... 36
2.1.4 Enumerate type ... 36
2.1.5 Array type ... 36
2.1.6 Structure type ... 37
2.1.7 Union type ... 37
2.1.8 Bit field ... 38
2.1.9 Alignment conditions ... 39
2.2 General-Purpose Registers ... 41
2.3 Referencing Data ... 42
2.4 Software Register Bank ... 43
User’s Manual U18513EJ1V0UM
9
2.4.1 Register modes ... 43
2.4.2 Register mode and library ... 44
2.5 Mask Register ... 45
2.5.1 Setting mask values ... 46
2.5.2 Using mask register function ... 47
2.6 Device File ... 48
2.6.1 Specifying device file ... 48
2.6.2 Notes on specifying device file ... 49
CHAPTER 3 C LANGUAGE EXPANSION ... 50
3.1 Allocation of Data to Section ... 51
3.1.1 #pragma section directive ... 58
3.1.2 Specifying link directive of specific data section ... 60
3.1.3 Notes on section allocation ... 61
3.1.4 Example of #pragma section directive ... 65
3.2 Allocating Functions to Sections ... 69
3.2.1 #pragma text directive ... 69
3.2.2 Specifying link directive of specific text section ... 71
3.2.3 Notes on #pragma text directive ... 72
3.3 Peripheral I/O Register ... 73
3.3.1 Accessing ... 73
3.3.2 Bit access ... 74
3.4 Describing Assembler Instruction ... 75
3.5 Controlling Interrupt Level ... 78
3.5.1 __set_il function ... 78
3.5.2 __set_il function and interrupt control register ... 79
3.6 Disabling Interrupts ... 81
3.6.1 Locally disabling interrupt in function ... 81
3.6.2 Disabling interrupts in entire function ... 81
3.6.3 Notes on disabling interrupts in entire function ... 83
3.7 Interrupt/Exception Processing Handler ... 84
3.7.1 Occurrence of interrupt/exception ... 84
3.7.2 Processing necessary in case of interrupt/exception ... 86
3.7.3 Describing interrupt/exception handler ... 89
3.7.4 Notes on describing interrupt/exception handler ... 93
3.7.5 Description example of interrupt/exception handler ... 95
3.8 Inline Expansion ... 96
3.8.1 Inline expansion ... 96
3.8.2 Conditions of inline expansion ... 97
3.8.3 Controlling inline expansion via options ... 99
3.8.4 Execution speed priority optimization and inline expansion ... 100
3.8.5 Examples of differences in inline expansion operation depending on option specification ... 101
3.9 Real-Time OS Support Function ... 102
3.9.1 Description of task ... 102
3.10 Embedded Functions ... 104
3.10.1 Interrupt control (DI/EI) ... 105
3.10.2 nop ... 105
3.10.3 halt ... 106
3.10.4 Saturated addition (satadd) ... 106
3.10.5 Saturated subtraction (satsub) ... 107
3.10.6 Halfword data byte swap (bsh) [V850E] ... 107
3.10.7 Word data byte swap (bsw) [V850E] ... 108
3.10.8 Word data halfword swap (hsw) [V850E] ... 108
3.10.9 Byte data sign extension (sxb) [V850E] ... 109
3.10.10 Halfword data sign extension (sxh) [V850E] ... 109
3.10.11 Instruction that assigns higher 32 bits of multiplication result to variable using mul instruction
[V850E] ... 110
10
User’s Manual U18513EJ1V0UM
3.10.12 Instruction that assigns higher 32 bits of unsigned multiplication result to variable using mulu
instruction [V850E] ... 111
3.10.13 Flag condition setting with logical left shift (sasf) [V850E] ... 112
3.11 Structure Packing Function ... 114
3.11.1 Structure packing specified ... 114
3.11.2 Rules of structure packing ... 115
3.11.3 Union ... 116
3.11.4 Bit field ... 117
3.11.5 Alignment condition of top structure object ... 118
3.11.6 Size of structure objects ... 118
3.11.7 Size of structure array ... 120
3.11.8 Area between objects ... 121
3.11.9 Notes concerning structure packing function ... 121
3.12 Binary Constants ... 123
CHAPTER 4 CALLING PROGRAM ... 124
4.1 Calling Between C Functions ... 124
4.1.1 Stack frame/function call ... 125
4.2 Calling Between C Function and Assembler Function ... 136
4.2.1 Calling assembler function from C function ... 136
4.2.2 Calling C function from assembler function ... 138
4.3 Prologue/Epilogue Processing of Function ... 140
4.3.1 Specifying use of runtime library function for prologue/epilogue of function ... 141
4.3.2 Calling runtime library for prologue/epilogue of function in V850Ex ... 142
4.3.3 Notes on calling runtime library for prologue/epilogue of function ... 143
4.4 Far Jump Function ... 144
4.4.1 Specifying far jump ... 144
4.4.2 File listing functions to be called by far jump function ... 145
4.4.3 Examples of using far jump function ... 146
CHAPTER 5 STARTUP ROUTINE ... 150
5.1 Operation of Startup Routine ... 150
5.1.1 Setting RESET handler when reset is input ... 152
5.1.2 Setting register mode of startup routine ... 153
5.1.3 Securing stack area and setting stack pointer (sp) ... 154
5.1.4 Securing argument area for main function ... 155
5.1.5 Setting text pointer (tp) ... 156
5.1.6 Setting global pointer (gp) ... 157
5.1.7 Setting element pointer (ep) ... 158
5.1.8 Setting mask value to mask registers (r20 and r21) ... 159
5.1.9 Initializing peripheral I/O registers that must be initialized before execution of main function ... 160
5.1.10 Initializing user target that must be initialized before execution of main function ... 162
5.1.11 Clearing sbss area to 0 ... 163
5.1.12 Clearing bss area to 0 ... 164
5.1.13 Clearing sebss area to 0 ... 165
5.1.14 Clearing tibss.byte area to 0 ... 166
5.1.15 Clearing tibss.word area to 0 ... 167
5.1.16 Clearing sibss area to 0 ... 168
5.1.17 Setting CTBP value for prologue/epilogue runtime library of functions ... 169
5.1.18 Setting BPC value of programmable peripheral I/O register ... 170
5.1.19 Setting r6 and r7 as argument of main function ... 171
5.1.20 Branching to main function ... 172
5.1.21 Branching to initialization routine of real-time OS ... 173
5.2 Example of Startup Routine ... 174
CHAPTER 6 LIBRARY FUNCTION ... 180
6.1 Supplied Libraries ... 180
6.1.1 Standard library ... 181
6.1.2 Mathematical library ... 187
User’s Manual U18513EJ1V0UM
11
6.1.3 Runtime library ... 190
6.1.4 ROMization library ... 192
6.1.5 Prologue/epilogue runtime library of functions ... 193
6.2 Header Files ... 196
6.3 Object Names Linked ... 197
6.4 Explanation of Format ... 198
6.5 Definition of Function with Variable Number of Arguments ... 199
STDARG ... 200
6.6 Management of Character String and Memory ... 202
STRING ... 203
MEMORY ... 207
6.7 Character Type Macros and Functions ... 209
CONV ... 210
CTYPE ... 212
6.8 Standard Input/Output ... 215
ERROR ... 216
FILEIO ... 217
GETS ... 219
PUTS ... 221
SPRINTF ... 223
PRINTF ... 227
SSCANF ... 230
SCANF ... 234
6.9 Standard Utility Functions ... 236
ABS ... 237
BSEARCH ... 238
DIV ... 240
ECVTF ... 242
ITOA ... 244
MALLOC ... 246
RAND ... 249
STRTODF ... 250
STRTOL ... 252
6.10 Non-Local Jump Functions ... 255
SETJMP ... 256
6.11 Mathematical Functions ... 258
BESSEL ... 260
ERFF ... 262
EXPF ... 263
FLOORF ... 265
FREXPF ... 267
GAMMAF ... 269
HYPOTF ... 270
MATHERR ... 271
SINHF ... 273
TRIG ... 275
6.12 Runtime Library ... 277
ADDF.S ... 279
CMPF.S ... 280
CVT.WS ... 282
DIV ... 283
DIVF.S ... 285
MOD ... 286
MUL ... 288
MULF.S ... 289
SUBF.S ... 290
TRNC.SW ... 291
12
User’s Manual U18513EJ1V0UM
CHAPTER 7 FOR EFFICIENT USE ... 292
7.1 volatile Qualifier ... 292
7.2 Declaration of Function Without Return Value ... 293
7.3 Pointers and Optimization ... 294
7.4 Assembler Code and Optimization ... 296
7.5 Registers ... 297
7.5.1 Register specifier ... 297
7.5.2 Static variables and external variables ... 297
7.5.3 Argument of function in K&R format ... 298
7.5.4 Optimum number of local variables to be assigned ... 298
7.5.5 Optimum number of arguments to be used for function ... 298
7.5.6 Other ... 299
7.6 Stack Size ... 300
7.7 Aligning Data ... 301
7.8 Data Type ... 302
APPENDIX A EXPANDED FUNCTIONS OF CC78Kx ... 305
A.1 #pragma Directive ... 305
A.2 Assembler Control Instructions ... 309
A.3 Specifying Interrupt/Exception Handler ... 309
A.4 Expanded Functions Not Supported ... 309
APPENDIX B CAUTIONS ... 310
APPENDIX C INDEX ... 317
User’s Manual U18513EJ1V0UM
13
LIST OF FIGURES
Figure No. Title Page
2-1
2-2
2-3
2-4
2-5
2-6
2-7
2-8
2-9
3-1
3-2
3-3
3-4
3-5
3-6
4-1
4-2
4-3
4-4
4-5
4-6
5-1
6-1
14
Internal Representation of Integer Type ... 34
Internal Representation of Floating-Point Type ... 35
Internal Representation of Pointer Type ... 36
Internal Representation of Enumerate Type ... 36
Internal Representation of Array Type ... 36
Internal Representation of Structure Type ... 37
Internal Representation of Union ... 38
Internal Representation of Bit Field ... 38
Register Modes and Usable Registers ... 44
sdata and sbss Attribute Sections ... 51
sidata and sibss Sections ... 53
sedata and sebss Sections ... 54
tidata and tibss Sections ... 55
Image of Memory Allocation of Each Section ... 57
Image of Interrupt Handler Address ... 85
Stack Frame (When Argument Register Area Is Located at Center of Stack) ... 125
Stack Frame (When Argument Register Area Is Located at Beginning of Stack) ... 126
Generation/Disappearance of Stack Frame ... 128
Stack Growth Direction of Each Area of Stack Frame ... 130
Generation/Disappearance of Stack Frame ... 132
Stack Growth Direction of Each Area of Stack Frame ... 134
Example of Startup Routine ... 174
Image of Using Runtime Library ... 277
User’s Manual U18513EJ1V0UM
LIST OF TABLES
Table No. Title Page
1-1
1-2
1-3
1-4
1-5
1-6
1-7
2-1
2-2
2-3
2-4
2-5
2-6
2-7
2-8
3-1
3-2
3-3
3-4
3-5
3-6
3-7
3-8
3-9
3 - 10
3 - 11
3 - 12
4-1
4-2
4-3
4-4
4-5
4-6
4-7
5-1
5-2
5-3
5-4
5-5
5-6
5-7
5-8
6-1
6-2
6-3
6-4
6-5
6-6
6-7
6-8
6-9
6 - 10
6 - 11
6 - 12
6 - 13
6 - 14
Expanded Notation and Meaning ... 19
Translation Limit ... 20
Limit Values Defined by limits.h File ... 21
Limit Values Defined by float.h File ... 22
A List of Supported Macros ... 31
Definition of NULL, size_t, ptrdiff_t(stddef.h File) ... 32
Processing When -ansi Option Strictly Conforming to Language Specifications Is Specified ... 33
Value Area of Integer Type ... 34
Value Area of Floating-Point Type ... 35
Alignment Condition for Basic Type ... 39
Alignment Condition for Union Type ... 39
Alignment Condition of Structure Type ... 40
Using General-Purpose Registers ... 41
Referencing Data ... 42
Register Modes Supplied by CA850 ... 43
Section Names Specified by User and Generated Section Names ... 59
Section Names Specified by User and Generated Section Names (text) ... 70
Enabling or Disabling Maskable Interrupt ... 78
Interrupt Control Functions ... 81
Interrupt/Exception Table (V850ES/SG2) ... 84
Registers for Register Variables ... 86
Stack Frame for Interrupt/Exception Handler ... 86
Stack Frame for Multiple Interrupt/Exception Handler ... 87
Usage of Registers ... 87
Processing for Saving/Restoring Registers During Interrupt ... 88
Trap Instructions and Software Exception Codes ... 91
Embedded Functions ... 104
Meanings of Macros for Functions ... 126
Method of Accessing Stack Area ... 127
Identifier ... 136
Registers for Register Variables ... 136
Registers for Register Variables ... 138
Work Registers ... 138
List of Prologue/Epilogue Runtime Functions ... 149
Startup Routine Samples ... 151
Symbols of sbss Area ... 163
Symbols of bss Area ... 164
Symbols of sebss Area ... 165
Symbols of tibss.byte Area ... 166
Symbols of tibss.word Area ... 167
Symbols of sibss Area ... 168
BPC Register ... 170
Supplied Libraries ... 180
Definition of Functions with Variable Number of Arguments ... 181
Character String Functions ... 182
Memory Management Functions ... 182
Conversion of Character ... 183
Classification of Characters ... 183
Standard I/O Functions ... 184
Standard Utility Functions ... 185
Non-Local Jump Functions ... 186
Mathematical Functions ... 188
Runtime Library ... 191
ROMization Copy Functions ... 192
List of Prologue Runtime Library Functions ... 193
List of Prologue Runtime Library Functions [V850E] ... 194
User’s Manual U18513EJ1V0UM
15
6 - 15
6 - 16
6 - 17
6 - 18
6 - 19
6 - 20
6 - 21
6 - 22
6 - 23
6 - 24
6 - 25
16
List of Epilogue Runtime Library Functions ... 194
List of Epilogue Runtime Library Functions [V850E] ... 195
Header Files ... 196
Definition of Function with Variable Number of Arguments ... 199
Functions for Character String/Memory Management ... 202
Character Type Macros ... 209
Standard Input/Output ... 215
Standard Utility Functions ... 236
Non-Local Jump Functions ... 255
Mathematical Functions ... 258
Runtime Library ... 278
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
This chapter explains the basic language specifications supported by the CA850.
The CA850 supports the language specifications stipulated by the ANSI standards. These specifications
include items that are stipulated as processing definitions. This chapter explains the language specifications of
the items dependent on the processing system of the V850 microcontrollers.
The differences between when options strictly conforming to the ANSI standards are used and when those
options are not used are also explained.
For the expanded specifications of the CA850, refer to "CHAPTER 3 C LANGUAGE EXPANSION".
User’s Manual U18513EJ1V0UM
17
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1
Dependent on Processing System Stipulated
This section explains items dependent on processing system stipulated by ANSI standards.
1.1.1
Data type and size
-
The number of bits of 1 byte is 8.
-
The number of bytes, byte order, and coding in an object are stipulated as follows.
char type
1 byte
short type
2 bytes
int, long, float, double type
4 bytes
Pointer
Same as unsigned int type
The byte order in a word is "from lower to higher". Signed integers are expressed as 2's complements. The
most significant bit indicates a sign (0 if positive or 0, and 1 if negative).
1.1.2
Translation stages
The ANSI standards specify eight translation stages (known an "phases") of priorities among syntax rules for
translation. In the third translation phase "decomposition of source file into preprocessing tokens and sequences
of white-space characters", whether each nonempty sequence of white-space characters other than new-line is
retained or replaced by one space character is implementation-defined. In this implementation, each nonempty
sequence of white-space characters other than new-line is retained, not replaced by one space character.
1.1.3
Diagnosis message
Error messages including the source file name and line number (only when the line number can be specified)
are output in response to a translation unit that violates a syntax rule or limit. These error messages are
classified into three types: "alarm", "fatal error", and "other error" messages.
1.1.4
(1)
Free-standing environment
The name and type of a function called on starting program processing are not stipulated in a free-standing
environmentNote. Therefore, they are dependent on the user-own coding or target system.
(2)
The effect of terminating a program in a free-standing environment is not stipulated. Therefore, it is
dependent on the user-own coding or target system.
Note
Environment in which a C program is executed without using the functions of the operating system.
The ANSI Standards specify two environments: a free-standing environment and a host environment.
The CA850 does not supply a host environment at present.
18
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.5
Executing program
The configuration of the interactive unit is not stipulated. Therefore, it is dependent on the uuser-own coding
and target system.
1.1.6
Character set
The values of elements of the execution environment character set are ASCII codes.
1.1.7
Multi-byte characters
Multibyte characters are not supported by character constants. However, comments and character strings in
Japanese are supported.
1.1.8
Meaning of character indication
The values of expanded notation are stipulated as follows:
Table 1 - 1 Expanded Notation and Meaning
Expanded Notation
\a
\b
\f
\n
\r
\t
\v
Value (ASCII)
07
08
0C
0A
0D
09
0B
Meaning
Alert (alarm sound)
Back space
Form feed (new page)
New line (carriage return)
Carriage return (return)
Horizontal tab
Vertical tab
User’s Manual U18513EJ1V0UM
19
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.9
Translation limit
The limit values of translation are explained below.
The values marked * are guaranteed values. These values may be exceeded in some cases, but the operation
is not guaranteed.
Table 1 - 2 Translation Limit
Contents
Number of nesting levels of compound statements, repetitive control structures, and
selective control structures (however, dependent on the number of "case" labels)
127
Number of nesting levels of condition embedding
255
Number of pointers, arrays, and function declarators (in any combination) qualifying one
arithmetic type, structure type, union type, or incomplete type in one declaration
16
Number of nesting levels enclosed by parentheses in a complete declarator
255(*)
Number of nesting levels of an expression enclosed by parentheses in a complete
expression
255(*)
Valid number of first characters in a macro name
1023(*)
Valid number of first characters of an external identifier
1022
Valid number of first characters in an internal identifier
1023
Number of identifiers having the valid block range declared by an external identifier in one
translation unit and in one basic block
4095(*)
Number of macro identifiers simultaneously defined in one translation unitNote
2047
Number of dummy arguments in one function definition and number of actual arguments in
one function call
255
Number of dummy arguments in one macro definition
127
Number of actual arguments in one macro call
127
Number of characters in one logical source line
32768
One character string constant after concatenation, or number of characters in a wide
character string constant
32768
Number of nesting levels for include (#include) files
50
Number of "case" labels for one "switch" statement (including those nested, if any)
1025
Number of members of a single structure or single union
1023(*)
Number of enumerate constants in a single enumerate type
1023(*)
Number of nesting levels of a structure or union definition in the arrangement of a single
structure declaration
63(*)
Note
20
limit Value
The upper limit of the macro identifier can be changed by a C compiler option (-Xm).
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.10
Quantitative limit
(1) The limit values of the general integer types (limits.h file)
The limits.h file stipulates the limit values of the values that can be expressed as general integer types (char
type, signed/unsigned integer type, and enumerate type).
Because multibyte characters are not supported, MB_LEN_MAX does not have a corresponding limit, and is
only defined with MB_LEN_MAX as 1.
If a -Xchar=unsigned optionNote of the CA850 is specified, CHAR_MIN is 0, and CHAR_MAX takes the same
value as UCHAR_MAX.
The limit values defined by the limits.h file are as follows.
Table 1 - 3 Limit Values Defined by limits.h File
Name
Value
Meaning
CHAR_BIT
8
The number of bits (= 1 byte) of the minimum object not
in bit field
SCHAR_MIN
- 128
Minimum value of signed char type
SCHAR_MAX
+ 127
Maximum value of signed char type
UCHAR_MAX
+ 255
Maximum value of unsigned char type
CHAR_MIN
- 128
Minimum value of char type
CHAR_MAX
+ 127
Maximum value of char type
SHRT_MIN
- 32768
Minimum value of short int type
SHRT_MAX
+ 32767
Maximum value of short int type
USHRT_MAX
+ 65535
Maximum value of unsigned short int type
INT_MIN
- 2147483648
Minimum value of int type
INT_MAX
+ 2147483647
Maximum value of int type
UINT_MAX
+ 4294967295
Maximum value of unsigned int type
LONG_MIN
- 2147483648
Minimum value of long int type
LONG_MAX
+ 2147483647
Maximum value of long int type
ULONG_MAX
+ 4294967295
Maximum value of unsigned long int type
Note
Specify a simple char type code.
User’s Manual U18513EJ1V0UM
21
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
(2) The limit values of the floating-point type (float.h file)
The limit values related to the characteristics of the floating-point type are defined by the float.h file.
The limit values defined by the float.h file are as follows.
Table 1 - 4 Limit Values Defined by float.h File
Name
22
Value
Meaning
FLT_ROUNDS
+1
Rounding mode for floating-point addition.
1 for the V850 microcontrollers (rounding in the nearest
direction).
FLT_RADIX
+2
Radix of exponent (b)
FLT_MANT_DIG
+ 24
DBL_MANT_DIG
+ 24
Number of numerals (p) with FLT_RADIX of floatingpoint mantissa as base
LDBL_MANT_DIG
+ 24
FLT_DIG
+6
DBL_DIG
+6
LDBL_DIG
+6
FLT_MIN_EXP
- 125
DBL_MIN_EXP
- 125
LDBL_MIN_EXP
- 125
FLT_MIN_10_EXP
- 37
DBL_MIN_10_EXP
- 37
LDBL_MIN_10_EXP
- 37
FLT_MAX_EXP
+ 128
DBL_MAX_EXP
+ 128
LDBL_MAX_EXP
+ 128
FLT_MAX_10_EXP
+ 38
DBL_MAX_10_EXP
+ 38
LDBL_MAX_10_EXP
+ 38
FLT_MAX
3.40282347E + 38F
DBL_MAX
3.40282347E + 38F
LDBL_MAX
3.40282347E + 38F
FLT_EPSILON
1.19209290E - 07F
DBL_EPSILON
1.19209290E - 07F
LDBL_EPSILON
1.19209290E - 07F
FLT_MIN
1.17549435E - 38F
DBL_MIN
1.17549435E - 38F
LDBL_MIN
1.17549435E - 38F
Number of digits of a decimal numberNote 1 (q) that can
round a decimal number of q digits to a floating-point
number of p digits of the radix b and then restore the
decimal number of q
Minimum negative integer (emin) that is a normalized
floating-point number when FLT_RADIX is raised to the
power of the value of FLT_RADIX minus 1.
Minimum negative integer log10bemin- 1 that falls in the
range of a normalized floating-point number when 10 is
raised to the power of its value.
Maximum integer (emax) that is a finite floating-point
number that can be expressed when FLT_RADIX is
raised to the power of its value minus 1.
Maximum integer log10((1- b- p) * bemax) that falls in the
range of a finite floating-point number when 10 is raised
to the power of its value.
Maximum number of finite floating-point numbers that
can be expressed (1 - b - p) * bemax
DifferenceNote 2 between 1.0 that can be expressed by
specified floating-point number type and the lowest
value which is greater than 1.0, b1 - p
Minimum value of normalized positive floating-point
number bemin - 1
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
Notes 1 DBL_DIG and LDBL_DIG are 10 or more in the ANSI standards but 6 in the V850 microcontrollers
because both the double and long double types are 32 bits.
2 DBL_EPSILON and LDBL_EPSILON are 1E-9 or less in the ANSI standards, but 1.19209290E-07F in
the V850 microcontrollers.
1.1.11
Identifier
An external name must consist of up to 1022 characters and must be able to be identified uniformly.
Uppercase and lowercase characters are distinguished.
1.1.12
char type
A char type with no type specifier (signed, unsigned) specified is treated as a signed integer as the default
assumption. However, a simple char type can be treated as an unsigned integer by specifying the Xchar=unsigned optionNote of the CA850.
The types of those that are not included in the character set of the source program required by the ANSI
standards (escape sequence) is converted for storage, in the same manner as when types other than char type
are substituted for a char type.
Note
Specify a simple char type code.
Example
char c = ’\777’ /* Value of c is -1. */
1.1.13
Floating-point constants
The floating-point constants conform to IEEE754Note.
Note
IEEE: Institute of Electrical and Electronics Engineers
IEEE754 is a standard to unify specifications such as the data format and numeric range in systems
that handle floating-point operations.
1.1.14
(1)
Character constants
Both the character set of the source program and the character set in the execution environment are
basically ASCII codes, and correspond to members having the same value.
For the character set of the source program, however, character codes in Japanese can be used (refer to
"1.1.15 Character string").
(2)
The last character of the value of an integer character constant including two or more characters is valid.
User’s Manual U18513EJ1V0UM
23
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
(3)
A character that cannot be expressed by the basic execution environment character set or escape
sequence is expressed as follows.
(a)
An octal or hexadecimal escape sequence takes the value indicated by the octal or hexadecimal
notation.
\777
(b)
511
The simple escape sequence is expressed as follows.
\'
'
\"
"
\?
?
\\
\
(c)
1.1.15
Character constants of multibyte characters are not supported.
Character string
The default character code is Shift JIS.
A character code can be changed by using the -Xk option of the CA850.
Option specification
-Xk=[e | euc | n | none | s | sjis]
The character codes in the output object file can be converted by the -Xkt option of the CA850.
Option specification
-Xkt=[e | euc | n | none | s | sjis]
If n or none is specified, the character code is not converted.
1.1.16
Header file name
The method to reflect the string in the two formats (< > and " ") of a header file name on the header file or an
external source file name is stipulated in "1.1.33 Loading header file".
1.1.17
Comment
A comment can be described in Japanese. The character code is the same as the character string in "1.1.15
Character string".
24
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.18
Signed constants and unsigned constants
If the value of a general integer type is converted into a signed integer of a smaller size, the higher bits are
truncated and a bit string image is copied.
If an unsigned integer is converted into the corresponding signed integer, the internal representation is not
changed.
1.1.19
Floating-points and general integers
If the value of a general integer type is converted into the value of a floating-point type, and if the value to be
converted is within a range that can be expressed but not accurately, the result is rounded to the closest
expressible valueNote.
Note
If the value is precisely in the middle, it is rounded to an even number (with the least significant bit of
the mantissa being 0).
1.1.20
double type and float type
In the processing system of the V850 microcontrollers, a double type is expressed as a floating-point number
in the same manner as a float type, and is treated as 32-bit (single-precision) data.
1.1.21
Signed type in operator in bit units
The characteristics of the shift operator conform to the stipulation in "1.1.27 Shift operator in bit units". The
other operators in bit units for signed type are calculated as unsigned values (as in the bit image).
1.1.22
Members of structures and unions
If the value of a member of a union is stored in a different member, it is stored according to an alignment
condition. Therefore, the members of that union are accessed according to the alignment condition (refer to
"2.1.6 Structure type" and "2.1.7 Union type").
In the case of a union that includes a structure sharing the arrangement of the common first members as a
member, the internal representation is the same, and the result is the same even if the first member common to
any structure is referenced.
1.1.23
sizeof operator
The value resulting from the "sizeof" operator conforms to the stipulation related to the bytes in an object in
"1.1.1 Data type and size". The number of bytes in a structure and union includes padding.
User’s Manual U18513EJ1V0UM
25
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.24
Cast operator
When a pointer is converted into a general integer type, the required size of the variable is the same as the
size of the int type. The bit string is saved as is as the conversion result.
Any integer can be converted by a pointer. However, the result of converting an integer smaller than an int type
is expanded according to the type.
1.1.25
Division/remainder operator
The result of the division operator ("/") when the operands are negative and do not divide perfectly with integer
division, is as follows:
If either the divisor or the dividend is negative, the result is the smallest integer greater than the algebraic
quotient. If both the divisor and the dividend are negative, the result is the largest integer less than the algebraic
quotient.
If the operand is negative, the result of the "%" operator takes the sign of the first operand in the expression.
1.1.26
Addition and subtraction operators
If two pointers indicating the elements of the same array are subtracted, the type of the result is int type, and
the size is 4 bytes.
1.1.27
Shift operator in bit units
If E1 of "E1 >> E2" is of signed type and takes a negative value, an arithmetic shift is executed.
1.1.28
Storage area class specifier
The storage area class specifier "register" is declared to increase the access speed as much as possible, but
this is not always effectiveNote.
Note
26
For the registers to be allocated, refer to "7.5.1 Register specifier".
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.29
(1)
Structure and union specifier
A simple int type bit field without signed or unsigned appended is treated as a signed field, and the most
significant bit is treated as the sign bit. However, the simple int type bit field can be treated as an unsigned
field by specifying the -Xbitfield optionNote of the CA850.
Note
(2)
Specify a simple int type bit field code.
To retain a bit field, a storage area unit to which any address with sufficient size can be assigned can be
allocated. If there is insufficient area, however, the bit field that does not match is packed into to the next
unit according to the alignment condition of the type of the field.
(3)
The allocation sequence of the bit field in unit is from lower to higher.
(4)
Each member of the non-bit field of one structure or union is aligned at a boundary as follows.
char, unsigned char type, and its array
Byte boundary
short, unsigned short type, and its array
Halfword boundary
Others (including pointer)
Word boundary
1.1.30
Enumerate type specifier
The type of an enumeration specifier is signed int.
When the -Xenum_type=string option is specified, however, it is as follows.
char
Treated as char
uchar
Treated as unsigned char
short
Treated as short
ushort
Treated as unsigned short
1.1.31
Type qualifier
The configuration of access to data having a type qualified to be "volatile" is dependent upon the address (I/O
port, etc.) to which the data is mapped.
1.1.32
(1)
Condition embedding
The value for the constant specified for condition embedding and the value of the character constant
appearing in the other expressions are equal.
(2)
The character constant of a single character must not have a negative value.
User’s Manual U18513EJ1V0UM
27
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.33
Loading header file
(1) A preprocessing directive in the form of "#include <character string>"
A preprocessing directive in the form of "#include <character string>" searches for a header file from the folder
specified by the -I option if "character string" does not begin with "\"Note 2, and then searches the ..\inc850 folder
with a relative path from the bin folder where the ca850 is placed.
If a uniformly identified header file is searched with a character string specified between delimiters "<" and ">",
the whole contents of the header file are replaced.
Notes
Both "\" and "/" are regarded as the delimiters of a folder.
Example
#include <header.h>
The search order is as follows:
(1)
Folder specified by -I
(2)
Standard folder
(2) A preprocessing directive in the form of "#include "character string""
A preprocessing directive in the form of "#include "character string"" searches for a header file from the folder
where the source file exists, and then searches the ..\inc850 folder via a relative path from the bin folder where
the ca850 is placed.
If a header file uniformly identified is searched with a character string specified between delimiters (") and ("),
the whole contents of the header file are replaced.
Example
#include "header.h"
The search order is as follows:
(1)
Folder where source file exists
(2)
Folder specified by -I
(3)
Standard folder
(3) The format of "#include preprocessing character phrase string"
The format of "#include preprocessing character phrase string" is treated as the preprocessing character
phrase only if the preprocessing character phrase string is a macro that is replaced to the form of <character
string> or "character string".
(4) Between a string delimited (finally) and a header file name
Between a string delimited (finally) and a header file name, the length of the alphabetic characters in the string
is identified, and the file name length valid in the compiler operating environment is valid. The folder that
searches a file conforms to the above stipulation.
28
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.34
#pragma directives
The CA850 can specify the following #pragma directives.
(1) Description with assembler instruction
#pragma
asm
#pragma
assembler instruction
endasm
Assembler directives can be described in a C language source program.
For the details of description, refer to "3.4 Describing Assembler Instruction".
(2) Inline expansion specification
#pragma inline function-name [, function-name ...]
A function that is expanded inline can be specified.
For the details of expansion specification, refer to "3.8 Inline Expansion".
(3) Device type specification
#pragma cpu device-name
Specify so that a device file defining the machine-dependent information of the device used is referenced.This
function is the same as the device specification option (-cpu) of the CA850.
For the device file, refer to "2.6 Device File".
(4) Data or program memory allocation
#pragma section section-type ["section-name"] [begin | end]
#pragma text ["section-name"] function-name
(a)
section
Allocates variables to an arbitrary section. For details about the allocation method, refer to "3.1 Allocation of
Data to Section".
(b)
text
A function to be allocated in a text section with an arbitrary name can be specified. For details about the
allocation specification, refer to "3.2 Allocating Functions to Sections" by Specifying Section Name.
(5) Peripheral I/O register name validation specification
#pragma ioreg
The peripheral I/O registers of a device are accessed by using peripheral function register names. For the
details of access, refer to "3.3 Peripheral I/O Register" by Using Register Name.
(6) Interrupt/exception handler specification
#pragma interrupt interrupt-request-name function-name [allocation-method]
User’s Manual U18513EJ1V0UM
29
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
Interrupt exception handlers are described in C language.
For details, refer to "3.7 Interrupt/Exception Processing Handler".
(7) Interrupt disable function specification
#pragma block_interrupt function-name
Interrupts are disabled for the entire function.
For description, refer to "3.6 Disabling Interrupts".
(8) Task specification
#pragma rtos_task [function-name]
A task that runs on an RTOS is described in C language.
For details, refer to "3.9.1 Description of task".
(9) Structure type packing specification
#pragma pack([1248])
Specifies the packing of a structure type. The packing value, which is an alignment value of the member, is
specified as the numeric value. A value of 1, 2, 4, or 8 can be specified. When the numeric value is not specified,
the setting is the default assumption.
For details, refer to "3.11 Structure Packing Function".
30
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.35
Predefined macro names
All the following macro names are supported.
Macros not ending with "__" are supplied for the sake of former C language specifications (K&R
specifications). To perform processing strictly conforming to the ANSI standards, use macros with "_ _" before
and after.
Table 1 - 5 A List of Supported Macros
Macro Name
Definition
__LINE_ _
Line number of source line at that point (decimal).
__FILE__
Name of assumed source file (character string constant).
__DATE__
Date of translating source file (character string constant in the form of
"Mmm dd yyyy". The name of the month is the same as that created by
the asctime function stipulated by the ANSI standards (three alphabetic
characters with only the first character being uppercase) and the first
character of dd is blank if its value is less than 10).
__TIME_ _
Translation time of source file (character string constant having format
"hh:mm:ss" similar to the time created by the asctime function).
__STDC_ _
Decimal constant 1 (defined when -ansi option is specified)Note
__v800
__v800_ _
Decimal constant 1
__v850
__v850_ _
Decimal constant 1
__v850e
__v850e__
Decimal constant 1 (defined by CA850, if V850Ex is specified as a target
device)
__v850e2
__v850e2__
Decimal constant 1 (defined by CA850, if V850E2/xxx is specified as a
target device)
__CA850
__CA850__
Decimal constant 1
__CHAR_SIGNED_ _
Decimal constant 1 (defined if signed is specified by -Xchar option or
when -Xchar option is not specified)
__CHAR_UNSIGNED__
Decimal constant 1 (defined when unsigned is specified by -Xchar option)
__DOUBLE_IS_32BITS__
Decimal constant 1
_DOUBLE_IS_32BITS
Decimal constant 1
CPU macro
Decimal constant 1 of a macro indicating the target CPU. A character
string indicated by "product type specification" in the device file with "__"
prefixed and suffixed is defined.
Register mode macro
Decimal constant 1 of a macro indicating the target CPU.
Macros defined as a register mode are as follows.
32 register mode : _ _reg32__
26 register mode : _ _reg26__
22 register mode : _ _reg22__
Note
For the processing to be performed when the -ansi option is specified, refer to "1.2 ANSI Option".
User’s Manual U18513EJ1V0UM
31
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.1.36
Definition of special data type
NULL, size_t, and ptrdiff_t defined by stddef.h file are as follows.
Table 1 - 6 Definition of NULL, size_t, ptrdiff_t(stddef.h File)
NULL / size_t / ptrdiff_t
32
Definition
NULL
((void *)0)
size_t
unsigned int
ptrdiff_t
int
User’s Manual U18513EJ1V0UM
CHAPTER 1 BASIC LANGUAGE SPECIFICATIONS
1.2
ANSI Option
If the -ansi option is specified by the CA850, processing strictly conforming to the ANSI standards is
performed. The difference between when the -ansi option is specified and when it is not specified are as follows.
Table 1 - 7 Processing When -ansi Option Strictly Conforming to Language Specifications Is Specified
Item
-ansi Specified
-ansi Not Specified
Trigraph series
Replaces trigraph series.
Does not replace.
Bit field
ErrorNote 1 occurs if type other than int is
specified for bit field.
Outputs alarm message and
permits.
Scope of argument
Multiple defined error occurs if automatic
variable having same name as argument of
function is declared.
Outputs alarm message and
validates automatic variable.
Substitution of pointer
1
Error occurs if numeric value of pointer type is
substituted into general integer typeNote 2
variable.
Outputs alarm message,
casts, and substitutes.
Substitution of pointer
2
Error occurs if pointers indicating different types
are substituted for each other.
Outputs alarm message and
permits.
Type conversion
Error occurs if conversion into pointer of array
that is not left-member value is performed.
Outputs alarm message and
permits.
Comparison operator
Error occurs if comparison is made between
arithmetic type variable and pointer.
Outputs alarm message and
permits.
Conditional operator
Error occurs if both second and third
expressions are not general integer type, same
structure, same union, or numeric value of
pointer type to type same as substitution
destination.
Outputs alarm message,
casts, and substitutes.
# line number
Error occurs.
Treated in same manner as
"#line line number"Note 3.
Character # in middle
of line
Error occurs if character '#' appears in middle of
line.
Outputs warning message
and enables the character
_asm
Outputs warning message and treats the
character as function call.
However, __ asm is valid.
Treated as assembler
insertionNote 4
__STDC_ _
Defines as macro with value of 1.
Does not define.
B ina r y c on stan t
Error occurs if "0b" or "0B" is followed by one or
more "0" or "1".
Treats "0b" or "0B" followed
by one or more "0" or "1" as
a binary constant.
Notes 1 Normal error beginning with "E". The same applies hereafter.
2 char type, signed/unsigned integer type, and enumerate type
3 Refer to the ANSI standards.
4 Refer to "3.4 Describing Assembler Instruction".
User’s Manual U18513EJ1V0UM
33
CHAPTER 2 COMPILATION ENVIRONMENT
CHAPTER 2 COMPILATION ENVIRONMENT
This chapter explains how the CA850 handles data, registers, and the environment during execution.
2.1
Internal Representation and Value Area of Data
This section explains the internal representation and value area of each type for the data handled by the
CA850.
2.1.1
Integer type
(1) Internal representation
The leftmost bit in an area is a sign bit with a signed type (type declared without "unsigned"). The value of a
signed type is expressed as 2's complement.
If -Xchar=unsigned is specified, however, a char type specified without "signed" or "unsigned" is assumed to
be unsigned.
Figure 2 - 1 Internal Representation of Integer Type
char (no sign bit for unsigned)
7
0
short (no sign bit for unsigned)
15
0
int,long (no sign bit for unsigned)
31
0
(2) Value area
Table 2 - 1 Value Area of Integer Type
Type
34
Value Area
charNote
-128 to +127
short
-32768 to +32767
int
-2147483648 to +2147483647
long
-2147483648 to +2147483647
unsigned char
0 to 255
User’s Manual U18513EJ1V0UM
CHAPTER 2 COMPILATION ENVIRONMENT
Table 2 - 1 Value Area of Integer Type
Type
Value Area
unsigned short
0 to 65535
unsigned int
0 to 4294967295
unsigned long
0 to 4294967295
Note
The value area is 0 to 255 if "-Xchar=unsigned" is specified by the CA850.
Caution 64-bit operation cannot do the CA850.
2.1.2
Floating-point type
(1) Internal representation
The internal representation of floating-point type data conforms to IEEE754Note.
The leftmost bit in the area is the sign bit. If the value of this sign bit is 0, the data is a positive value; if it is 1,
the data is a negative value.
A double type is a floating-point representation the same as a float type, and is handled as 32-bit (singleprecision) data.
Note
IEEE: Institute of Electrical and Electronics Engineers
IEEE754 is a standard to unify specifications such as the data format and numeric range in systems
that handle floating-point operations.
Figure 2 - 2 Internal Representation of Floating-Point Type
float,double
S
E
M
31
23 22
0
S : Sign bit of mantissa
E : Exponent (8 bits)
M : Mantissa (23 bits)
(2) Value area
Table 2 - 2 Value Area of Floating-Point Type
Type
float,double
Value Area of Absolute Value
1.18 x 10-38 to 3.40 x 1038
User’s Manual U18513EJ1V0UM
35
CHAPTER 2 COMPILATION ENVIRONMENT
2.1.3
Pointer type
(1) Internal representation
The internal representation of a pointer type is the same as that of an unsigned int type.
Figure 2 - 3 Internal Representation of Pointer Type
31
2.1.4
0
Enumerate type
(1) Internal representation
The internal representation of an enumerate type is the same as that of a signed int type. The leftmost bit of
the area is the sign bit.
Figure 2 - 4 Internal Representation of Enumerate Type
31
2.1.5
0
Array type
(1) Internal representation
The internal representation of an array type arranges the elements of an array in the form that satisfies the
array condition(alignment) of the elements.
char a[8] = { 1, 2, 3, 4, 5, 6, 7, 8 }
The internal representation of the array shown above is as follows.
Figure 2 - 5 Internal Representation of Array Type
7
36
0 7
0 7
0 7
0 7
0 7
User’s Manual U18513EJ1V0UM
0 7
0 7
0
CHAPTER 2 COMPILATION ENVIRONMENT
2.1.6
Structure type
(1) Internal representation
The internal representation of a structure type arranges the elements of a structure in a form that satisfies the
alignment condition of the elements.
Example
struct {
short
int
char
long
} tag;
s1;
s2;
s3;
s4;
The internal representation of the structure shown above is as follows.
Figure 2 - 6 Internal Representation of Structure Type
s4
s3
31
0 31
8 7
s2
0 31
s1
0 31
16 15
0
For the internal representation when the structure type packing function is used, refer to "3.11 Structure
Packing Function".
2.1.7
Union type
(1) Internal representation
A union is considered as a structure whose members all start with offset 0 and that has sufficient size to
accommodate any of its members. The internal representation of a union type is like each element of the union is
placed separately at the same address.
Example
union {
int
short
char
long
} tag;
u1;
u2;
u3;
u4;
The internal representation of the union shown in the above example is as follows.
User’s Manual U18513EJ1V0UM
37
CHAPTER 2 COMPILATION ENVIRONMENT
Figure 2 - 7 Internal Representation of Union
31
0
tag
tag.u3(1byte)
tag.u2(2bytes)
tag.u1,tag.u4(4bytes)
2.1.8
Bit field
(1) Internal representation
An area including the declared number of bits is reserved for a bit field. The most significant bit of the area for
a bit field declared to be of signed type is a sign bit.
The bit field declared first is allocated from the least significant bit of a word area. If the alignment condition of
the type specified in the declaration of a bit field is exceeded as a result of allocating an area that immediately
follows the area of the preceding bit field to the bit field, the area is allocated starting from a boundary that
satisfies the alignment condition.
Example
struct {
unsigned int
int
unsigned int
} flag;
f1 : 30;
f2 : 14;
f3 : 6;
The internal representation for the bit field in the above example is as follows.
Figure 2 - 8 Internal Representation of Bit Field
f3
31
20 19
f2
14 13
f1
0 31 29
0
The ANSI standards do not allow char and short types to be specified for a bit field, but CA850 allows this. In
this case, a warning message is output, and padding is performed according to the alignment condition of the
specified typeNote.
For the internal representation of bit field when the structure type packing function is used, refer to "3.11
Structure Packing Function".
Note
38
An error occurs if -ansi is specified as an option of the CA850.
User’s Manual U18513EJ1V0UM
CHAPTER 2 COMPILATION ENVIRONMENT
2.1.9
Alignment conditions
(1) Alignment conditions for basic type
Table 3-3 shows the alignment conditions for basic types. If -Xi of the CA850 is specified, however, all the
alignment types are word boundaries.
Table 2 - 3 Alignment Condition for Basic Type
Basic Type
Alignment Condition
(unsigned) char and its array type
Byte boundary
(unsigned) short and its array type
Halfword boundary
Other basic types (including pointer)
Word boundary
(2) Alignment condition for union type
The alignment condition for the union type varies as shown in Table 3-4, depending on the maximum member
size.
Table 2 - 4 Alignment Condition for Union Type
Maximum Member Size
Alignment Condition
2 bytes < size
Word boundary
Size <= 2 bytes
Maximum member size boundary
Here are examples of the respective cases:
Example 1
union tug1 {
unsigned short i;
unsigned char c;
};
/* 2-byte member */
/* 1-byte member */
/* The union is aligned with 2 bytes. */
Example 2
union tug2 {
unsigned int i;
unsigned char c;
};
/* 4-byte member */
/* 1-byte member */
/* The union is aligned with 4 bytes. */
User’s Manual U18513EJ1V0UM
39
CHAPTER 2 COMPILATION ENVIRONMENT
(3) Alignment condition for structure type
The alignment condition for the structure type differs as shown in Table 3-5, depending on the size of the
structure (excluding the size of the integer). If -Xi of the CA850 is specified, however, all the structure types are
word boundaries.
Table 2 - 5 Alignment Condition of Structure Type
Structure Size
Alignment Condition
2 bytes < size
Word boundary
Size <= 2 bytes
If member of type more than int type exists
Word boundary
If there is no member of type more than int type,
and 1 byte < size <= 2 bytes
Halfword boundary
If there is no member of type more than int type,
and size <= 1 byte
Byte boundary
Here are examples of the respective cases:
Example 1
struct SS {
int
i;
/* 4-byte member */
char
c;
/* 1-byte member */
};
/* Structure is aligned with 4 bytes. */
Example 2
struct BIT_I {
int
};
i1 : 5;
/* 4-byte member (size is 1 byte or less) */
/* Structure is aligned with 4 bytes because member type is int. */
Example 3
struct BIT_C {
char
c1 : 5;
/* 1-byte member */
};
/* Structure is aligned with 1 byte. */
Example 4
struct BIT_CC {
char
c1 : 5;
/* 1-byte member */
char
c2 : 5;
/* 1-byte member */
};
/* Structure is aligned with 2 bytes because size is 2 bytes. */
(4) Alignment condition for function argument
The alignment condition for a function argument is a word boundary.
(5) Alignment condition for executable program
The alignment condition when an executable object file is created by linking object files is a halfword boundary.
40
User’s Manual U18513EJ1V0UM
CHAPTER 2 COMPILATION ENVIRONMENT
2.2
General-Purpose Registers
Table 3-6 shows how the CA850 uses the general-purpose registers.
The general-purpose registers includes the following functions.
(1) Software register bank
The number of the work registers (r10 through r19) and register variable registers (r20 through r29) used can
be reduced by the -reg option of CA850 (refer to "2.4 Software Register Bank").
(2) Mask register function
In the 32-register mode and 22-register mode, registers r20 and r21 can be used to set a mask value (refer to
"2.5 Mask Register").
Table 2 - 6 Using General-Purpose Registers
Register
Usage
r0
Zero register
Used for operation as value of 0.
Also used to reference data located at const section
(read-only section placed in ROM area)Note.
r1
Assembler-reserved register
Used for instruction expansion by assembler.
r2(hp)
Handler stack pointer
Reserved for system.
r3(sp)
Stack pointer
Used to indicate beginning of stack frame.
r4(gp)
Global pointer
Used to reference external variable.
r5(tp)
Text pointer
Used to indicate beginning of text section (.text section)
r6 - r9
Argument registers
Used to pass argument.
r10 - r19
Work registers
Used as work register during operation (r10 is also
used to pass return value of function).
r20 - r29
Register variable registers
Used as an area for register variables.
r30(ep)
Element pointer
Used to reference external variable specified to be
located in internal RAM or external RAM sectionNote.
r31(lp)
Link pointer
Used to pass return address of function.
Note
For the allocation of data to a section, refer to "3.1 Allocation of Data to Section".
User’s Manual U18513EJ1V0UM
41
CHAPTER 2 COMPILATION ENVIRONMENT
2.3
Referencing Data
How the CA850 references data are as follows.
Table 2 - 7 Referencing Data
Type
42
Referencing Method
Numeric constant
Immediate
Character constant
Offset from global pointer (gp)
Offset from r0 register
Automatic variable,
argument
Offset from stack pointer (sp)
External variable,
static variable in function
Offset from global pointer (gp)
Offset from element pointer (ep)
Offset from r0 register
Function address
Operated during execution by using offset from text pointer (tp)
User’s Manual U18513EJ1V0UM
CHAPTER 2 COMPILATION ENVIRONMENT
2.4
Software Register Bank
Because the CA850 implements a register bank function by software, three register modes are provided. By
specifying these register modes efficiently, the contents of some registers do not need to be saved or restored
when an interrupt occurs or the task is switched. As a result, the processing speed can be improved.
The register modes are specified by using the register mode specification option (-reg) of CA850.
This function reduces the number of registers internally used by the CA850 on a step-by-step basis.
As a result, the following effects can be expected:
(1)
The registers not used can be used for the application program (that is, a source program in assembly
language).
(2)
The overhead required for saving and restoring registers can be reduced.
Note
In an application program that has many variables to be allocated to registers by the CA850, the
variables so far allocated to a register are accessed from memory when a register mode has been
specified. As a result, the processing speed may drop.
2.4.1
Register modes
Next table and next Figure show the three register modes supplied by the CA850.
Table 2 - 8 Register Modes Supplied by CA850
Mode
Work Registers
Register Variable Registers
32-register mode (default)
r10 - r19
r20 - r29
26-register mode
r10 - r16
r23 - r29
22-register mode
r10 - r14
r25 - r29
User’s Manual U18513EJ1V0UM
43
CHAPTER 2 COMPILATION ENVIRONMENT
Figure 2 - 9 Register Modes and Usable Registers
32-register mode
26-register mode
31
0
31
22-register mode
0
31
0
r0
r0
r0
r10
r10
r10
Other Registers
r14
r15
Work Registers
r16
r17
r19
r20
r22
r23
r24
r25
Register Variable
Registers
Other Registers
r29
r29
r29
r31
r31
r31
Register that can be used
freely for application
Specification example on command line
> ca850 -cpu 3201 -reg26 file.c -- compiled in 26-register mode
2.4.2
Register mode and library
A library supplied by the CA850 (refer to "CHAPTER 6 LIBRARY FUNCTION") is provided for each register
mode.
The standard folders that search a library are "Install Folder\lib850" and "Install Folder\lib850\r32" as the
default assumption. If the 22- or 26-register mode is specified by the CA850, however, "Install Folder \lib850\r22"
or "Install Folder\lib850\r26" is used as the standard folder for the library, in the place of "Install
Folder\lib850\r32".
If ld850 is not started from the CA850 but object files are linked by directly starting ld850 from the command
line, however, a library suitable for each register mode must be specified by specifying the -reg option of ld850.
44
User’s Manual U18513EJ1V0UM
CHAPTER 2 COMPILATION ENVIRONMENT
2.5
Mask Register
When byte data or halfword data is loaded from the memory to a register, the V850 microcontrollers signextends the data to a word length according to the value of the most significant bit of the data. Therefore, mask
codes of the higher bits may be generated during an unsigned char or unsigned short type data (refer to "7.8
Data Type") operation. When storing the result of an operation to a register variable, mask codes are generated
to clear the higher bits if the result of the operation is unsigned byte data or unsigned halfword data.
Generation of mask codes can be prevented if word data is used. If word data cannot be used and the mask
codes are generated, the code size can be reduced by using the mask register function.
However, to decide whether the mask register function is to be used or not, the following points must be
carefully considered for the code where the mask register function may be used.
(1)
Whether the program outputs many mask codes
(2)
Two register variable registers will not be able to be used because they will be used as mask registers.Will
this cause any difficulties
The CA850 uses r20 and r21 as mask registers, as shown in the example below, when the mask register
function is used. Note that mask values must be set to the mask registers by program.
Mask code generation example
unsigned char UC;
unsigned short US;
void func(void)
{
register unsigned char ruc;
register unsigned short rus;
:
UC *= UC;
:
ruc = UC;
rus = US;
:
}
- - Normal code
ld.b
$UC, r11
andi
0xff, r11, r11
mulh
r11, r11
st.b
r11, $UC
:
ld.b
$UC, r29
andi
0xff, r29, r29
ld.h
$US, r28
andi
0xffff, r28, r28
-- Code when mask register is used
ld.b
$UC, r11
and
r20, r11
mulh
r11, r11
st.b
r11, $UC
:
ld.b
$UC, r29
and
r20, r29
ld.h
$US, r28
and
r21, r28
User’s Manual U18513EJ1V0UM
45
CHAPTER 2 COMPILATION ENVIRONMENT
An instruction that executes "an operation on unsigned data" has been added to the V850Ex and the CA850
outputs a code that uses this instruction. When the V850Ex is used, therefore, setting to use the mask register
may not have as much effect as expected.
2.5.1
Setting mask values
Mask values (0xff and 0xffff) must be set to r20 and r21, which are used as mask registers, via the program.
The CA850 generates mask codes using the mask registers, assuming that the mask values have been set.
Example of setting of mask value
_ _ start :
mov
mov
:
mov
mov
:
jarl
#_ _ tp_TEXT, tp
#_ _ gp_DATA, gp
0xff, r20
0xffff, r21
-- Sets mask value to r20
-- Sets mask value to r21
_main, lp
If the program uses an RTOS, however, the mask values are automatically set according to the RTOS type.
(1) When RX850 is used
Because the mask values are automatically set by the initialization routine of the RX850, they do not have to
be set by program.
(2) When RX850 Pro is used
The mask values must be set in advance by using the startup module.
(3) When real-time OS is not used
Set the mask values in advance by using the startup moduleNote.
Note
The startup module crtN.s (for 32-register mode) supplied with the package sets the mask values
(refer to "CHAPTER 5 STARTUP ROUTINE").
46
User’s Manual U18513EJ1V0UM
CHAPTER 2 COMPILATION ENVIRONMENT
2.5.2
Using mask register function
This section describes the specifications for using the mask register functions and points to be noted.
(1) To newly compile C language source file
By specifying the mask register function option (-Xmask_reg) of the CA850, an assembly language source
program including the mask codes that use the mask registers and information indicating that the mask register
function is used (".option mask_reg" directive) is output.
(2) Checking during linking
Once the link editor has been started by specifying the mask register function option (-Xmask_reg) of the
compiler, the object file with the file name information (information specified by the ".file" directive) that indicates
that the object file has been created from the .c file is checked while the object file is linked. If an object using the
mask register function and an object that does not use the function exist together at this time, an error occurs.
Notes 1 Objects included in an archive file (.a file) are not checked. To use an .a file created by the user,
confirm that the mask registers are not used.
2 To start ld850 alone from the command line of the DOS window, an option that performs checking
during linking (-mc) must be specified.
(3) When using created assembly language source file
If the program is described in an assembly language from the beginning, check that the contents of the mask
registers are not lost. The mask registers are not checked during linking because the file name information is not
".c".
Whether or not the contents of the mask registers are lost can be confirmed by a warning message that is
output when the assembler is executed, if the -m option that specifies the use of the mask registers is specified
in the assembler.
(4) Supplied library restrictions
Although the object files in the archive file are not checked during linking, almost all the libraries in the package
do not destroy the contents of the mask registers. The bsearch function in the standard library, however, may
destroy the contents of the mask registers because it calls an application function. Therefore, do not use the
bsearch function when the mask register function is used (the compiler does not output an error even if the
bsearch function is used)Note.
Note
The bsearch function in the standard library, however, may destroy the contents of the mask registers
because it calls an application function. Therefore, do not use the bsearch function when the mask
register function is used (The CA850 does not output an error even if the bsearch function is used).
User’s Manual U18513EJ1V0UM
47
CHAPTER 2 COMPILATION ENVIRONMENT
2.6
Device File
A device file is a binary file that contains information dependent upon the device type. One device file is
available for each device or group of target devices as a package. The compiler references a device file to
generate object codes corresponding to the target system that is used in the application system. Therefore,
place the device file to be used under the standard folder for the device file. If the device is placed under any
other folder, specify the folder using a compiler option; otherwise an error occurs during compilation because the
device file is not found.
2.6.1
Specifying device file
A device file that is referenced by a program in C language can be specified in the following two ways.
(1)
Specifying device name using compiler option (-cpu device-name)
Example
> ca850 -cpu 3201 file.c
When building a program with PM+, specifying a device has an effect equivalent to specifying this option.
(2)
Specifying device name using #pragma directive (#pragma cpu device-name) in C language source file
Example
#pragma cpu 3201
In this example, the device name is "3201" (V850ES/SA2).
The character strings that can be specified as "device name" are common to option specification and the
#pragma directive. Uppercase and lowercase characters are not distinguished.
For the character strings that can be specified as a device name, refer to the Architecture User's Manual of
each device.
Notes 1 When specifying a device name using the #pragma directive, device specification must be
described in all source files.
2 Specify a device name at the beginning of a source file when using the #pragma directive. Only
preprocessing that has nothing to do with C language syntax and comments can be described
before specification of the device name. If a device name is specified in C language syntax, the
compiler outputs the following error message and stops processing.
F2625: illegal placement ' #pragma cpu '
Example of incorrect specification
#include <stdio.h>
int
i;
#pragma cpu 3201
:
48
User’s Manual U18513EJ1V0UM
CHAPTER 2 COMPILATION ENVIRONMENT
2.6.2
Notes on specifying device file
(1) If no device name is specified
If a device file is specified by neither the #pragma directive nor the -cpu option, and if neither the -cn option nor
the -cnv850e optionNote is specified, the compiler outputs the following error message and stops compiling.
F2620: unknown cpu type, cannot compile
Note
A device file is necessary during linking even if the -cn or -cnv850e option is specified.
(2) If device is specified by both option and #pragma directive
The compiler outputs a warning message and continues processing, giving priority to the option. If different
device names are specified by two or more options or #pragma directives, the compiler outputs the following
message and stops processing.
F2622: duplicated cpu type
(3) Program described in assembler instructions
In this case also, a device must be specified by an assembler option or the .option quasi directive when an
object file that can be linked is created.
User’s Manual U18513EJ1V0UM
49
CHAPTER 3 C LANGUAGE EXPANSION
CHAPTER 3 C LANGUAGE EXPANSION
This chapter explains the language specifications expanded by the CA850.
The expanded specifications include how to specify section location of data and access the internal peripheral
function registers of the device, how to insert assembler code in a C language source program, how to specify
inline expansion for each function, how to define a handler when an interrupt or exception occurs, how to disable
interrupts at the C language level, the valid RTOS functions when a real-time OS is used for the target
environment, and how to embed instructions in a C language source program.
50
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.1
Allocation of Data to Section
When external variables and data are defined in a source, the CA850 allocates them to memory.
The memory location to which the variables and data are allocated is, basically, an area that can be referenced
by an offset from the address pointed to by the global pointer (gp). If the variables or data are accessed in the
program, therefore, the CA850 attempts to output a code that accesses the area using gp, by default.
At this time, the CA850 attempts to output a code that allocates data to an area that can be referenced from gp
by one instruction, in order to enhance the object efficiency and execution efficiency as much as possible. Since
the range that can be referenced by one instruction from gp must be within +32 K bytes (a total of 64 K bytes)
from gp according to the V850 architecture, the CA850 has dedicated sections in the +32 K bytes area from gp.
These sections are called the sdata and sbss attribute sections.
Figure 3 - 1 sdata and sbss Attribute Sections
High Address
sdata attribute
sbss attribute
section
32K bytes
(0x8000)
gp
32K bytes
(0x8000)
In many cases, however, variables exceed in this range when using an application that uses many variables.
In this case, the variables must be allocated to other sections. The CA850 has many sections to which variables
and data can be allocated, in addition to the sdata and sbss attribute sections. Each of these sections has its
own feature and sections to which variables that must be accessed quickly can be allocated are also available,
so that the sections can be selected depending on the usage. The sections that can be used in the CA850 are
explained below.
(1) sdata and sbss attribute sections
These sections can be referenced from gp with one instruction and must be allocated within +32 K bytes from
gp.
Data with initial values is allocated to the sdata attribute section, and data without initial values is allocated to
the sbss attribute section.
The CA850 first attempts to generate a code that is to be allocated to these sections. If the code exceeds the
upper limit of the section of these attributes, the compiler generates a code that allocates data to a data or bss
attribute section.
To increase the amount of data to be allocated to the sdata or sbss attribute sections, the upper size limit for
the data to be allocated can be specified with the "-G" option of the CA850 so that data in excess of this upper
limit is not allocated to the sdata or sbss attribute sections (refer to CA850 for Operation User’s Manual for
details of this option).
User’s Manual U18513EJ1V0UM
51
CHAPTER 3 C LANGUAGE EXPANSION
Use the #pragma section directive to specify a variable to be allocated to the sdata or sbss attribute section in
the program (refer to "3.1.1 #pragma section directive" for details).
#pragma
int
int
#pragma
section sdata begin
a=1;
/* allocated to sdata attribute section */
b;
/* allocated to sbss attribute section */
section sdata end
(2) data and bss attribute sections
These sections can be referenced from gp with two instructions. Since access is performed after address
generation, the code becomes correspondingly longer and the execution speed also drops, but the entire 32-bit
space can be accessed. In other words, these sections can be allocated anywhere as long as they are in RAM.
Use the #pragma section directive to specify a variable to be allocated to the data or bss attribute section in
the program (refer to "3.1.1 #pragma section directive" for details).
#pragma section data begin
int
a=1;
/* allocated to data attribute section */
int
b;
/* allocated to bss attribute section */
#pragma section data end
(3) sconst attribute section
This is a section that can be referenced from r0, in other words from address 0, with 1 instruction, and must be
allocated within +32 bytes from address 0. Basically, data that can be fixed to ROM is allocated to this section.
In the case of V850 devices with internal ROM, in many cases the internal ROM is assigned from address 0,
and data that should be referenced with 1 instruction and that can be fixed to ROM is allocated to the sconst
attribute section. Variables/data declared by adding the const qualifier are subject to allocation to the sconst
attribute section. If the data exceeds the upper limit of this attribute section, it is allocated to the const attribute
section.
To increase the amount of data to be allocated to the sconst attribute section, the upper size limit for the data
to be allocated can be specified with the "-Xsconst" option of the CA850 so that data in excess of this upper limit
is not allocated to the sconst attribute section (refer to CA850 for Operation User’s Manual for details of this
option).
Use the #pragma section directive to specify a variable to be allocated to the sconst attribute section in the
program (refer to "3.1.1 #pragma section directive" for details).
#pragma section sconst begin
const int a=1;
/* allocated to sconst attribute section */
#pragma section sconst end
52
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
(4) const attribute section
This is a section that can be referenced from r0, in other words from address 0, with two instructions. Data that
can be fixed to ROM that exceeds the upper limit of the sconst attribute section, or data that should be allocated
to external ROM in the case of ROMless devices of the V850 microcontrollers, is allocated to the const attribute
section. Variables/data declared by adding the const qualifier are subject to allocation to the const attribute
section. The variables declared by adding the const qualifier are allocated to the const attribute section, string
literal even if allocation to the .const section is not specified by the #pragma section directive. Since access is
performed after address generation, the code becomes correspondingly longer and the execution speed also
drops, but the entire 32-bit space can be accessed. In other words, the const attribute section can be allocated
anywhere as long as it is in the 32-bit space. Use the #pragma section directive to specify a variable to be
allocated to the const attribute section in the program (refer to "3.1.1 #pragma section directive" for details).
#pragma section const begin
const int a=1;
/* allocated to const attribute section */
#pragma section const end
(5) sidata and sibss attribute sections
These sections can be referenced from ep (element pointer) with 1 instruction toward higher addresses. In
other words, these sections are allocated in the 32 K bytes space toward higher addresses from ep.
Figure 3 - 2 sidata and sibss Sections
High Address
sidata attribute
sibss attribute
32K bytes
(0x8000)
section
ep
Data with initial values is allocated to the tidata attribute section, and data without initial values is allocated to
the tibss attribute section. If variables that exceed the upper limit of the sdata and sbss attribute sections that can
be accessed from gp with 1 instruction, but which need to be accessed with 1 instruction still exist, they can be
allocated in the range that can be accessed with 1 instruction using ep.The sidata and sibss attribute sections
are sections for access toward higher addresses from ep; the sedata and sebss attribute sections are sections
for access toward lower addresses from ep.
Use the #pragma section directive to specify a variable to be allocated to the sidata or sibss attribute section in
the program (refer to "3.1.1 #pragma section directive" for details).
#pragma section sidata begin
int
a=1;
/* allocated to sidata attribute section */
int
b;
/* allocated to sibss attribute section */
#pragma section sidata end
User’s Manual U18513EJ1V0UM
53
CHAPTER 3 C LANGUAGE EXPANSION
(6) sedata and sebss attribute sections
These sections can be referenced from ep (element pointer) with 1 instruction toward lower addresses. In
other words, these sections are allocated within 64 K bytes toward lower addresses from ep.
Figure 3 - 3 sedata and sebss Sections
ep
High Address
sedata attribute
sebss attribute
32K bytes
(0x8000)
section
Data with initial values is allocated to the sedata attribute section, and data without initial values is allocated to
the sebss attribute section. If variables that exceed the upper limit of the sdata and sbss attribute sections that
can be accessed from gp with 1 instruction, but which need to be accessed with 1 instruction still exist, they can
be allocated in the range that can be accessed with 1 instruction using ep. The sidata and sibss attribute
sections are sections for access toward higher addresses from ep; the sedata and sebss attribute sections are
sections for access toward lower addresses from ep.
Use the #pragma section directive to specify a variable to be allocated to the sedata or sebss attribute section
in the program (refer to "3.1.1 #pragma section directive" for details).
#pragma
int
int
#pragma
54
section sedata begin
a=1;
/* allocated to sedata attribute section */
b;
/* allocated to sebss attribute section */
section sedata end
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
(7) tidata (tidata.byte, tidata.word) and tibss (tibss.byte, tibss.word) attribute sections
These sections can be referenced from ep (element pointer) with 1 instruction toward higher addresses. These
sections are accessed with 1 instruction in the same manner as the sidata and sibss attribute sections, but differ
in terms of the assembler instruction to be used.
The sidata and sibss attribute sections use the 4-byte st/ld instruction for store/reference, whereas the tidata
and tibss attribute sections use the 2-byte sst/sld instruction to perform access. In other words, the code
efficiency of the tidata and tibss attribute sections is better than that of the sidata and sibss attribute sections.
However, the range in which sst/sld instructions can be applied is small, so it is not possible to allocate a large
number of variables.
Figure 3 - 4 tidata and tibss Sections
High Address
tidata.byte attribute
tibss.byte attribute
tidata.word attribute
256 bytes
(0x100)
tibss.word attribute
section
ep
Data with initial values is allocated to the tidata (tidata.byte, tidata.word) attribute section, and data without
initial values is allocated to the tibss (tibss.byte, tibss.word) attribute section. Specify the tidata.byte/tibss.byte
attribute to allocate byte data, and specify the tidata.word/tibss.word attribute to allocate word data. To select
automatic byte/word judgment by the CA850, specify the tidata/tibss attribute.
The tidata and tibss attribute sections are used to allocate data that must be accessed the fastest in the
system. However, the data to be allocated to these sections must be carefully selected because the quantity of
data that can be allocated to these sections is limited. Use the #pragma section directive to specify variables to
be allocated to the tidata.byte/tibss.byte or tidata.word/tibss.word attribute section in the program (refer to "3.1.1
#pragma section directive" for details).
User’s Manual U18513EJ1V0UM
55
CHAPTER 3 C LANGUAGE EXPANSION
#pragma section
char
unsigned char
#pragma section
tidata_byte begin
a=1;
/* allocated to tidata.byte attribute section */
b;
/* allocated to tibss.byte attribute section */
tidata_byte end
#pragma
int
short
#pragma
section
a=1;
b;
section
tidata_word begin
/* allocated to tidata.word attribute section */
/* allocated to tibss.word attribute section */
tidata_word end
#pragma
int
char
#pragma
section
a=1;
b;
section
tidata begin
/* allocated to tidata.word attribute section */
/* allocated to tibss.byte attribute section */
tidata end
The efficiency can be enhanced in terms of execution speed if variables or data that are especially frequently
used in the system are selected and allocated to the tidata (tidata.byte, tidata.word) or tibss (tibss.byte or
tibss.word) attribute section.
The CA850 has a section file generator that investigates the frequency of reference.
The frequency information obtained as a result of the investigation is output as a frequency information file.
The code that allocates data to the tidata (tidata.byte, tidata.word) or tibss (tibss.byte, tibss.word) attribute
section is output based on this information. The user can edit the frequency information file to select variables
that should be allocated to the tidata (tidata.byte, tidata.word) or tibss (tibss.byte, tibss.word) attribute section by
priority. The variables can then be allocated to these sections without qualifying the source.
Refer to CA850 for Operation User’s Manual for details of the section file generator and frequency information
file.
Figure 3 - 5 shows an example of memory allocation of each section.
56
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
Figure 3 - 5 Image of Memory Allocation of Each Section
Peripheral
I/O register
.sibss section
.sidata section
Within 32 K bytes
.tibss.word section
Within
256 bytes
.tidata.word section
.tibss.byte section
Within
128 bytes
.tidata.byte section
ep
.sebss section
ep is generally set at
the beginning of internal RAM
Within 32 K bytes
.sedata section
.const section
.bss section
gp points to the start address
of the .sdata section +32 K bytes
.sbss section
gp
.sbss and .sdata are allocated within 64 K bytes
.sdata section
.data section
r0-relative access area
ep-relative access area
.text section
gp-relative access area
tp
tp-relative access area
.sconst section
Interrupt/exception table
Within 32KB
Other
Address 0
User’s Manual U18513EJ1V0UM
57
CHAPTER 3 C LANGUAGE EXPANSION
3.1.1
#pragma section directive
How to allocate data to a section using the #pragma section directive is explained below.
(1) To use default section name as is
Describe the #pragma section directive in the following format when using the section name defined by the
CA850 as is.
#pragma section section-type begin
variable-declaration/definition
#pragma section section-type end
The following can be specified as the section-type.
-
data
-
sdata
-
sedata
-
sidata
-
tidata
-
tidata.word
-
tidata.byte
-
sconst
-
const
The name of the bss attribute section must not be specified as the section type. The CA850 automatically
allocates declared or defined data with initial values to the data attribute section, and data without initial values to
the bss attribute section.
#pragma
int
int
#pragma
section sdata begin
a=1;
/* allocated to sdata attribute section */
b;
/* allocated to sbss attribute section */
section sdata end
In the above case, "variable a" is allocated to the data-attribute .sdata section because it has an initial value,
and "variable b" is allocated to the sbss-attribute sbss section because it does not have an initial value.
Two or more variable declarations or definitions can be described between "#pragma section section-type
begin" and "#pragma section section-type end". Enumerate variables to be allocated for each section type.
Use "_" (underscore) instead of "." (period) to specify tidata.word or tidata.byte as the section type, as shown
below.
58
-
tidata_word
-
tidata_byte
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
(2) To assign original section name
The user can assign a specific name to the sections with the following attributes, and can allocate variables
and data to those sections.
-
data
-
sdata
-
sconst
-
const
In this case, describe the #pragma section directive in the following format.
#pragma section section-type "created-section-name" begin
Variable declaration/definition
#pragma section section-type "created-section-name" end
However, ".section-type" is appended to a section name actually generated by this method as follows.
created-section-name.section-type
This is to prevent a section with another attribute and having the same name from being created because the
section attribute is classified into data or bss depending on whether the data has an initial value or not. Specify a
generated section name when specifying a section in a link directive file. Refer to "3.1.2 Specifying link directive
of specific data section" for an example of specification in a link directive file.
The following table shows specific examples of section names specified by the user and generated section
names.
Table 3 - 1 Section Names Specified by User and Generated Section Names
Section Name
Specified by User
Section Type
Character String
Appended
Generated Section Name
mydata
data attribute
.data/.bss
mydata.data/mydata.bss
mysdata
sdata attribute
.sdata/.sbss
mysdata.sdata/mysdata.sbss
myconst
const attribute
.const
myconst.const
mysconst
sconst attribute
.sconst
mysconst.sconst
If the name is specified as follows, "variable a" is allocated to the mysdata.sdata section because it has an
initial value, and "variable b" is allocated to the mysdata.sbss section because it does not have an initial value.
#pragma section sdata "mysdata" begin
int
a=1;
/* allocated to mysdata.sdata attribute section */
int
b;
/* allocated to mysdata.sbss attribute section */
#pragma section sdata "mysdata" end
User’s Manual U18513EJ1V0UM
59
CHAPTER 3 C LANGUAGE EXPANSION
3.1.2
Specifying link directive of specific data section
When a specific section is created using the #pragma section directive, describe that section in a link directive
file as explained below.
If "variable a" and "variable b" are specified as follows in a C language source, "variable a" is allocated to the
mysdata.sdata section because it has an initial value, and "variable b" is allocated to the mysdata.sbss section
because it does not have an initial value.
#pragma
int
int
#pragma
section sdata "mysdata" begin
a=1;
/* allocated to mysdata.sdata attribute section */
b;
/* allocated to mysdata.sbss attribute section */
section sdata "mysdata" end
At this time, the variable can be allocated to a specific section if the mapping directive in the link directive file is
described as follows.
.data = $PROGBITS ?AW .data;
.bss = $NOBITS ?AW .bss;
mysdata.data = $PROGBITS ?AW mysdata.data;
mysdata.bss = $NOBITS ?AW mysdata.bss;
Since the variables are allocated in the order in which they are described, change the description order to
change the allocation order. It is also possible to specify the start address of the section directly (generally, a
segment is created first and a mapping directive, which specifies the start address of a section in segment units,
is then described in that segment).
Because the attribute of mysdata.data is "$PROGBITS?AW" and that of mysdata.bss is "$NOBITS?AW", do
not omit the input section (".data", ".bss", "mysdata.data", and "mysdata.bss" on the rightmost side of the
mapping directive in the above example) from mapping directives that have the same attribute as these.
Example
.data = $PROGBITS ?AW;
.bss = $NOBITS ?AW;
If an input section is omitted from a mapping directive having the same "$PROGBITS?AW" or "$NOBITS?AW"
attribute, the linker links and locates all the sections having that attribute. Consequently, data is not allocated to
the specific section created by the user. This means that the data that should be allocated to the mysdata.data
section is allocated to the .data section, and the data that should be allocated to the mysdata.bss section is
allocated to the .bss section.
Refer to CA850 for Link Directive User’s Manual for details of the format of the link directive file.
60
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.1.3
Notes on section allocation
Notes below must be noted when sections are allocated by the #progma section directive, the const qualifier,
or the section file.
(1)
An error occurs during compilation if the #pragma section directive is specified as follows.
-
Section allocation is nested.
-
begin and end of #pragma section cross.
-
Either begin or end of #pragma section is missing.
Example of incorrect specification: "Nesting of sections"
#pragma section data begin
int
a=1
#pragma
short
char
#pragma
section sdata begin
b;
c=0x10
section sdata end
int
d;
#pragma section data end
Example of incorrect specification: "Crossing sections"
#pragma section data begin
int
a=1
#pragma
short
char
#pragma
section sdata begin
b;
c=0x10
section data end
int
d;
#pragma section sdata end
(2)
If a section is specified for an automatic variable, the specification is ignored. Section specification is a
function for external variables.
(3)
When specifying a specific section name, keep the length of the name to within 256 characters.
User’s Manual U18513EJ1V0UM
61
CHAPTER 3 C LANGUAGE EXPANSION
(4)
A variable declaration that is not set with an initial value is usually treated as a tentative definition. When a
section is specified, however, it is treated as a "definition". Do not allow variable declarations which do not
have their initial values set to get mixed in with definitions.
/* Variable declaration not using
#pragma section */
/* Variable declaration using
#pragma section */
int i;
int i=10;
#pragma
int
int
#pragma
/* tentative definition */
/* definition */
/* Error does not occur. */
section
i;
i=10;
section
sedata begin
/* definition */
/* definition */
sedata end
/* Duplicated definition error */
If a section is specified for the tentative definition of an ordinary external variable, it is treated as a
"definition". Be sure to make extern declaration in files that reference an external variable. In the example
below, a duplicated definition error occurs if extern is missing in the tentative definition of the variable in
file1.c.
[file1.c]
#pragma section sedata begin
extern int i;
#pragma section sedata end
[file2.c]
#pragma section sedata begin
int
i;
#pragma section sedata end
[Duplicated definition error occurs if extern is not declared]
(5)
When a variable specified by a section is referenced by another file, the section must be specified with the
same section type for the extern declaration of that variable. An error occurs if a type of section different
from that of the section specified when a variable is defined is specified.
For example, if "#pragma section data begin - #pragma section data end" is specified on the definition
side and "#pragma section data begin - #pragma section data end" is not specified on the tentative
definition side (extern declaration), it is assumed on the tentative definition side that the variable is
allocated to the sdata section. This means that a code that accesses the variable from gp with two
instructions is output on the definition side and that a code that accesses the variable from gp with one
instruction is output on the tentative definition side. In other words, a contradiction occurs. Consequently,
the following error message is output during linking.
F4163: output section ".data" overflowed or illegal label reference for symbol
"symbol" in file "file" (value: value, input section: section, offset: offset,
type:R_V850_GPHWLO_1). "symbol" is allocated in section ".data" (file: file).
62
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
Example of correct specification
[file1.c]
#pragma section sedata begin
int
i=1;
#pragma section sedata end
[file2.c]
#pragma section sedata begin
extern int i;
#pragma section sedata end
Example of incorrect specification 1
[file1.c]
int
i=1;
[file2.c]
#pragma section sedata begin
extern int i;
#pragma section sedata end
"variable i" defined by file1.c is allocated to the sbss or bss attribute section. However, file2.c outputs a
code that accesses the sebss attribute section for "variable i". As a result, the linker outputs the following
error message.
User’s Manual U18513EJ1V0UM
63
CHAPTER 3 C LANGUAGE EXPANSION
F4163: output section ".sebss" overflowed or illegal label reference forsymbol
"_i" in file "file2.o" (value: value, input section: section, offset: offset,
type: type). "_i" isallocated in section ".sbss" (file: file1.o).
Example of incorrect specification 2
[file1.c]
#pragma section sedata begin
int
i=1;
#pragma section sedata end
[file2.c]
extern int
i;
f"variable i" defined by file1.c is allocated to the sebss attribute section but file2.c outputs a code that
accesses the sbss attribute section or bss attribute section to "variable i". Consequently, the linker outputs
the following mismatch error.
F4156: can not find GP-symbol in segment "*DUMMY*" or illegal labelreference
for symbol "_i" in file "file2.o" (section: section, offset: offset,
type:R_V850_GPHWLO_1). "_i" is allocated in section ".sedata" (file: file1.o).
(6)
When defining a variable with the sconst or const attribute using the #pragma section directive, be sure to
make a const specification for the variable. A const specification is also necessary at the location of the
tentative definition made by extern declaration.
If the const declaration is missing when a variable is declared, the variable is not allocated to the sconst
section or const section (the #pragma section directive is ignored) even if "#pragma section sconst begin #pragma section sconst end" or "#pragma section const begin - #pragma section const end" is specified,
but to a gp-relative section such as the sdata section or data section. In other words, allocation is not
performed as intended.
[file1.c]
#pragma section sconst begin
const int
i=1;
#pragma section sconst end
[file2.c]
#pragma section sconst begin
int
i;
#pragma section sconst end
A code that allocates "variable i" to the sconst section is output in file1.c. In file2.c, however, the #pragma
section specification is ignored because the const specification is missing from "variable i", and therefore
the variable is treated as a gp-relative variable. In other words, a code that allocates the variable to the
sdata or data section is output. Consequently, "variable i" is not allocated to the sconst section during
linking. A const specification is also necessary at the location of the tentative definition with extern
declaration, as shown below.
[file1.c]
#pragma section sconst begin
const int
i=10;
#pragma section sconst end
64
[file2.c]
#pragma section sconst begin
extern const int
i;
#pragma section sconst end
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.1.4
Example of #pragma section directive
Here are some examples of using the #pragma section directive.
(1)
Allocating "variable a" to tidata.word section and "variable b" to tibss.word section
#pragma
int
short
#pragma
(2)
section tidata_word begin
a=1;
/* allocated to tidata.word attribute section */
b;
/* allocated to tibss.word attribute section */
section tidata_word end
Allocating "variable c" to tidata.byte section and "variable d" to tibss.byte section
#pragma
char
char
#pragma
section tidata_byte begin
c=0x10;
/* allocated to tidata.byte attribute section */
d;
/* allocated to tibss.byte attribute section */
section tidata_byte end
In the tidata section, word data or halfword data is allocated to the tidata_word or tibss_word section, and
byte data is allocated to the tidata_byte or tibss_byte section. If char-type arrays are declared in the C language source, however, they are allocated to the tidata.word section. The tidata.word section can be used up
to 256 bytes. Because the arrays are of char type, a code using sld.b or sst.b is output. However, the sld.b and
sst.b instructions cannot access more than 128 bytes. Therefore, if a char-type array is declared and if the
array itself is of more than 128 bytes or is located at a place exceeding 128 bytes relatively from ep, an error
occurs during linking. Take this point into consideration when allocating char-type arrays to the tidata section.
(3)
Allocating "variable e" specified by const to the sconst section and character string constant data indicated
by pointer p to sconst section
#pragma section sconst begin
const int
e=0*10;
const char *p="Hello World";
#pragma section sconst end
In the above description, "Hello World" indicated by pointer p is allocated to the sconst section, and pointer
variable "p" itself is allocated to the sdata section or data section. The allocation position of the pointer variable
and the contents indicated by the pointer vary depending on how const is specified.
Example 1
const char
*p="Hello World";
If this declaration is made, the pointer variable and character sting constant indicated by the pointer are
allocated as follows.
User’s Manual U18513EJ1V0UM
65
CHAPTER 3 C LANGUAGE EXPANSION
Pointer variable "p"
Can be rewritten ("p = 0" can be compiled).
Character string constant
"Hello World"
Cannot be rewritten
("*p = 0" cannot be compiled).
Describe as shown below to allocate what the pointer variable indicates to a section with the const
attribute. This description is used when the pointer itself is fixed to ROM.
#pragma section sconst begin
const char *p="Hello World";
#pragma section sconst end
With the above definition, the pointer and character string constant are allocated to the following sections.
Pointer variable "p"
sdata/data section
Character string constant
"Hello World"
sconst section
Example 2
char
*const p;
Pointer variable "p"
Cannot be rewritten
("p = 0" cannot be compiled).
Describe as shown below to allocate the pointer variable to a section with the const attribute. This is used
to fix the pointer itself to ROM.
char *const p="Hello World";
The above description allocates both the pointer variable and character string constant "Hello World" to a
section with the const attribute.
#pragma section sconst begin
char *const p="Hello World";
#pragma section sconst end
The above definition allocates the pointer variable and constant to the following sections.
Pointer variable "p"
Character
string
"Hello World"
66
sconst section
constant
sconst section
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
Example 3
const char
*const p;
Pointer variable "p"
Cannot be rewritten
("p = 0" cannot be compiled).
Describe as shown above to allocate the pointer variable and the destination it indicates to a section with
the const attribute. Both the above descriptions are used to fix the pointer to ROM.
const char *const
p="Hello World";
The above description allocates both the pointer variable and character string constant "Hello World" to a
section with the const attribute.
#pragma section sconst begin
const char *const
p="Hello World";
#pragma section sconst end
The above definition allocates the pointer variable and constant to the following sections.
Pointer variable "p"
sconst section
Character string constant
"Hello World"
sconst section
In addition to the #pragma section directive, the compiler option "-Xconst" can be used to allocate a
variable specified by const to the sconst section.
User’s Manual U18513EJ1V0UM
67
CHAPTER 3 C LANGUAGE EXPANSION
(4)
Make the extern declaration of the #pragma section directive in a commonly used header file and include it
in the C language source.
[header.h]
#pragma section sidata begin
extern int k;
#pragma section sidata end
[file1.c]
#include "header.h"
#pragma section sidata begin
int
k;
#pragma section sidata end
[file2.c]
#include "header.h"
void func1(void)
{
k = 0x10;
}
If the extern declaration of the #pragma section directive is made in a header file as shown above, the errors
decrease and the source is visually simplified.
68
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.2
Allocating Functions to Sections
The CA850 allocates the functions of a C language source program, i.e., program codes, to the .text section
by default. When the .text section allocation address is specified in the link directive file, the program is allocated
from that address. However, it may be necessary to change the allocation address for each function or distribute
the allocation address because of the layout of the memory. To satisfy these necessities, the CA850 has the
#pragma text directive. Using this directive, any name can be given to a section with the text attribute, and the
allocation address can be changed in the link directive file.
3.2.1
#pragma text directive
Using the #pragma text directive, any name can be given to a section with the text attribute. The #pragma text
directive can be used in the following two ways.
-
Specifying the function name to be allocated to a section to be created using the #pragma text directive
-
Describing the #pragma text directive before the main body of a function (function definition) but not
specifying a function name
(1)
Specifying the function name to be allocated to a section to be created using the #pragma text directive
#pragma text "created section name" function-name
Describe functions that are described in the C language. In the case of a function, "void func1() {}", specify
"func1". The created section name can be omitted. In this case, a function specified by "function name" is
allocated to the default .text section.
(2)
Describing the #pragma text directive before the main body of a function (function definition) but not
specifying a function name
#pragma text "created section name"
The created section name can be omitted. In this case, specification of the name of section to be created by
"#pragma text" specified immediately before is canceled, and the subsequent functions are allocated to the
default .text section. However, ".text" is appended to a section name actually generated by this method as
follows.
section-name.text
Specify the generated section name when specifying a section in a link directive file. Refer to "3.2.2
Specifying link directive of specific text section" for an example of specifying in a link directive file.
User’s Manual U18513EJ1V0UM
69
CHAPTER 3 C LANGUAGE EXPANSION
The following table shows specific examples of section names specified by the user and generated section
names.
Table 3 - 2 Section Names Specified by User and Generated Section Names (text)
Section Name
Specified by User
mytext
Section Type
text attribute
Character String
Appended
.text
Generated Section Name
mytext.text
If the name is specified as follows, "func1" is allocated to the mytext1.text section, and "func2" is allocated to
the .text section by default, because the #pragma text directive is not used.
#pragma text "mytext1" func1
void func1(void)
{
:
}
void func2(void)
{
:
}
If the name is specified as follows, "func1" and "func2" are allocated to the mytext2.text section, "func3" to
the "mytext3.text section", and "func4" to the default .text section because the #pragma text "mytext3"
immediately before is canceled.
#pragma text "mytext2"
void func1(void)
{
:
}
void func2(void)
{
:
}
#pragma text "mytext3"
void func3(void)
{
:
}
#pragma text
void func4(void)
{
:
}
70
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.2.2
Specifying link directive of specific text section
When a specific section is created using the #pragma section directive, describe that section in a link directive
file as explained below.
#pragma text "mytext2"
void func1(void)
{
:
}
void func2(void)
{
:
}
#pragma text "mytext3"
void func3(void)
{
:
}
#pragma text
void func4(void)
{
:
}
If the #pragma text directive is specified in a C language source as shown above, "func1" and "func2" are
allocated to the mytext2.text section, "func3" to the mytext3.text section, and "func4" to the default .text section
because the #pragma text "mytext3" immediately before is canceled.
.text
= $PROGBITS
mytext2 = $PROGBITS
mytext3 = $PROGBITS
?AX .text;
?AX mytext2.text;
?AX mytext3.text;
Since the functions are allocated in the order in which they are described, change the description order to
change the allocation order. It is also possible to specify the start address of the function directly (generally, a
segment is created first and a mapping directive, which specifies the start address of a function in segment units,
is then described in that segment).
Because the attribute of mytext2.text and mytext3.text is "$PROGBITS ?AW", do not omit the input section
(".text", "mytext2.text", and "mytext3.text" on the rightmost side of the mapping directive in the above example)
from mapping directives that have the same attribute as these.
User’s Manual U18513EJ1V0UM
71
CHAPTER 3 C LANGUAGE EXPANSION
Example
.text
= $PROGBITS ?AX;
If an input section is omitted from a mapping directive having the same "$PROGBITS ?AX" attribute, the linker
links and locates all the sections having that attribute. Consequently, data is not allocated to the specific section
created by the user. This means that the program that should be allocated to the mytext2.text or mytext3.text
section is allocated to the .text section.
Refer to CA850 for Link Directive User’s Manual for details of the format of the link directive file.
3.2.3
Notes on #pragma text directive
Note the following points when using the #pragma text directive.
(1)
Describe the #pragma text directive before the function definition in the same file; otherwise a warning
message is output and the directive is ignored. However, the order of prototype declaration of a function is
not affected.
(2)
If a function specified by the #pragma text directive is an interrupt handler specified as direct allocation, a
warning message is output and the #pragma text directive is ignored. Refer to "3.7 Interrupt/Exception
Processing Handler" for details of direct allocation specification.
(3)
A function specified by #pragma text cannot be expanded inline by a #pragma inline specification or an
optimization option. Inline expansion specification is ignored.
(4)
72
When specifying a section name, keep the length of the name to within 256 characters.
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.3
Peripheral I/O Register
Peripheral I/O registers are used to control the internal peripheral functions of a device.
By using the peripheral I/O register name defined by the device, the internal I/O can be accessed at C
language level. The peripheral I/O register names can be treated in the C language source program as if they
were normal unsigned external variables.
For the register names and attributes that can be specified, refer to the Relevant Device’s Hardware User’s
Manual of each device.
3.3.1
Accessing
A peripheral function register name is validated by describing the following pragma directive.
#pragma ioreg
In a C language source file in which "#pragma ioreg" directive is described, the peripheral function register
name described after the pragma directive can be used.
If this directive is not used or if a peripheral function register name is used without specifying an applicable
device name, an "undefined variable" error occurs. An error also occurs if the access attribute peculiar to the
specified register is violated.
Of the examples as follows, Example 1 is correct, but Examples 2 and 3 cause an error.
P0, P1, P2, RXB0, and OVF0 in the following examples indicate the peripheral I/O registers of the V850
microcontrollers.In this way, symbols defined by the device file are specified as "register names".
Next shows specification examples.
Example 1
#pragma ioreg
void func1(void)
{
int
i;
P0 = 1;
i = RXB0;
}
void func2(void)
{
P1 = 0;
}
/* Writes to P0 */
/* Reads from RXB0 */
/* Writes to P1 */
Example 2
void func(void)
{
P1 = 0;
}
/* Undefined error */
User’s Manual U18513EJ1V0UM
73
CHAPTER 3 C LANGUAGE EXPANSION
Example 3
#pragma ioreg
void functorial)
{
RXB0 = 1;
/* Error that occurs if attribute of RXB0 is read-only */
}
3.3.2
Bit access
The CA850 can access each bit of a peripheral function register.
"bit number" is specified as 0 to 31 in the case of a 32-bit register.
(register name).(bit number) = ...
(1) Cautions of case of bit access
(a)
If a value other than 0 or 1 is substituted in accessing a bit, the binary least significant value of that value is
set (In this case, no message is output.).
Specification examples 1
#pragma ioreg
void func(void)
{
P0.1 = 1;
P2.3 = 0;
}
(b)
/* Sets bit 1 of P0 to 1 */
/* Resets bit 3 of P2 to 0 */
The bits of the flag of each register can be accessed by using a bit name.Specify a name defined by the
device file as the bit name
Specification examples 2
#pragma ioreg
void func(void)
{
OVF0 = 1;
}
74
/* Sets bit name OVF0 to 1 */
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.4
Describing Assembler Instruction
With the CA850, assembler instruction can be described in the functions of a C language source program in
the following format.
-
asm declaration
-
#pragma directive
To use registers with an inserted assembler, save or restore the contents of the registers in the program
because they are not saved or restored by the CA850.
It is advisable to insert assembler in a function. It the instructions are described outside a function, the
following restrictions apply and a warning message is output.
-
The output sequence of the function and code is not guaranteed.
-
The code is not output in a file where the function does not exist.
(1) asm declaration
_ _ asm(character string constant); or _asm(character string constant);
[Cautions]
(a)
The _asm format is provided to maintain compatibility with the conventional language specifications. If the
-ansi option is specified, the compiler outputs a warning message to the _asm format and treats the option
as a function call. When specifying the -ansi option, use the __asm format.
(b)
If the asm declaration is specified, the compiler suffixes a new-line character (\n) to the specified character
string constantNote and passes it to the assembler.
Note
The specified character string constant is unlike the normal character string constant, "\" followed by a
character other than a new line indicates the following character itself ("\" followed by a new line
causes an error).
Example
_ _ asm("nop ");
_ _ asm(".str \"string\\0\"");
(c)
__asm or __asm is a declaration and is not treated as a statement. Therefore, because of the syntax of
the C language source, an error occurs in a structure that does not allow the use of a declaration only, as
shown in Example 1 below. Therefor, enclose the statement in "{ }" as shown in Example 2 to make it a
compound statement.
Example 1
if(i == 0)
__ asm("mov r11, r10"); /* Error occurs because only declaration is made. */
User’s Manual U18513EJ1V0UM
75
CHAPTER 3 C LANGUAGE EXPANSION
Example 2
if(i == 0){
__asm("mov r11, r10"); /* Can be used because this is compound statement. */
}
(2) #pragma directive
In the range enclosed by the above #pragma directives, assembler instructions can be described as is. This is
useful for using two or more assembler instructions.
#pragma asm
assembler instruction
#pragma endasm
A description of example 1 to show next is same to a description of example 2.
Example 1
extern int i;
void f(void)
{
#pragma asm
mov
r0, r10
st.w
r10, $_i
:
#pragma endasm
}
Example 2
extern int
i;
void f(void)
{
_ _asm("mov r0, r10");
_ _asm("st.w r10, $_i");
:
}
The description from "#pragma asm" to "#pragma endasm" is passed to the assembler as is. In other words,
the CA850 internally creates an assembler instruction and starts the assembler. Therefore, a quasi directive of
the assembler can be used after the #pragma asm declaration. A local variable in a C language source must not
be used with the assembler. Because the local variable is allocated to the stack or a register by the CA850, it
cannot be used with an inline assembler.
76
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
A local variable in a C language source must not be used with the assembler. Because the local variable is
allocated to the stack or a register by the CA850, it cannot be used with an inline assembler. A symbol defined
using #define in the C language source file cannot be used in the description from "#pragma asm" to "#pragma
endasm", therefore expand a macro defined by #define in a file by an assembler instruction, as follows.
-
Define the macro by using the .macro instruction in the #pragma asm - #pragma endasm directives.
-
Call an assembler instruction from the C language source program by means of a function call.
Another method is to write an assembler instruction without making a macro definition.
User’s Manual U18513EJ1V0UM
77
CHAPTER 3 C LANGUAGE EXPANSION
3.5
Controlling Interrupt Level
3.5.1
__set_il function
The CA850 can manipulate the interrupts of the V850 microcontrollers as follows in a C language source.
-
By controlling interrupt level
-
By enabling or disabling acknowledgment of maskable interrupts (by masking interrupts)
In other words, the interrupt control register can be manipulated. For this purpose, the "_ _set_il" function is
used. Specify this function as follows to manipulate the interrupt priority level.
_ _ set_il(interrupt-priority-level, "interrupt-request-name");
The "interrupt request name" that can be specified is the "maskable interrupt request name" defined in the
device file. Because a request name defined in the device file is used, the #pragma ioreg directive must be
described in the C language source that uses this function. Integer values 1 to 8 can be specified as the interrupt
priority level. With the V850, eight steps, from 0 to 7, can be specified as the interrupt priority level. To set the
interrupt priority level to "5", therefore, specify the interrupt priority level as "4" by this function.
Example
_ _ set_il(2, "INTP0");
This specification sets the interrupt priority level of interrupt INTP0 to 1.
Specify the _ _set_il function as follows to enable or disable acknowledgment of a maskable interrupt.
_ _ set_il(enables/disables maskable interrupt, "interrupt request name");
"-1" or "0" can be specified to enable or disable the maskable interrupt.
Table 3 - 3 Enabling or Disabling Maskable Interrupt
Set Value
Operation
-1
Disables acknowledgment of maskable interrupt (masks interrupt).
0
Enables acknowledgement of maskable interrupt (unmasks interrupt).
Example
_ _ set_il(-1, "INTP0");
If the function is specified as shown above, acknowledging maskable interrupt INTP0 is disabled (INTP0
is masked). Note that the _ _set_il function does not manipulate the ep flag (that indicates that exception
processing is in progress) in the program status word (PSW).
78
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.5.2
__set_il function and interrupt control register
The interrupt control register of the V850 microcontrollers is configured as follows.
7
6
5
4
3
2
1
0
xxIFn
xxMKn
0
0
0
xxPRn2
xxPRn1
xxPRn0
If the __set_il function is used, either "priority level" or "interrupt mask flag" is set. This means that the
__set_il function cannot set an interrupt request flag.
To set the interrupt priority level to 6 when the interrupt request name is "INTP000" and the interrupt control
register name is "P00IC0", for example, describe the function as follows.
_ _ set_il(7, "INTP000");
The following codes will be output.
ld.b
andi
ori
st.b
P00IC0, r1
0xf8, r1, r1
0x6, r1, r1
r1, P00IC0
Therefore, codes that change only the lower 3 bits (xxxPR02 to xxxPR00) of the setting of the priority level are
output.
Describe the _ _set_il function as follows to enable a maskable interrupt when the interrupt request name is
"INTP000" and the interrupt control register name is "P00IC0".
_ _ set_il(0, "INTP000");
The following code will then be output.
clr1
6, P00IC0
A code that changes only the interrupt mask flag is output.
If a value is directly written to the interrupt control register, values are set to the priority level, interrupt mask
flag, and interrupt request flag.
Example
When the interrupt control register name is "P00IC0"
P00IC0=0x6;
User’s Manual U18513EJ1V0UM
79
CHAPTER 3 C LANGUAGE EXPANSION
The above description outputs the following codes.
mov
st.b
0x6, r29
r29, P00IC0
The meanings of these codes are as follows.
80
-
Sets the priority level to 6.
-
Enables the maskable interrupt.
-
Clears the interrupt request flag.
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.6
Disabling Interrupts
The CA850 can disable the maskable interrupts in a C language source. This can be done in the following two
ways.
-
Locally disabling interrupt in function
-
Disabling interrupts in entire function
3.6.1
Locally disabling interrupt in function
The "di instruction" and "ei instruction" of the assembler instruction can be used to disable an interrupt locally
in a function described in C language. However, the CA850 has functions that can control the interrupts in a C
language source.
Table 3 - 4 Interrupt Control Functions
Interrupt Control Function
Operation
Processing by CA850
__DI( ) ;
Disables interrupt.
Generates di instruction.
__EI( ) ;
Enables interrupt.
Generates ei instruction.
Example (How to use the __ DI() and _ _EI() functions and the codes to be output are shown below.)
[C language source]
void func1(void)
{
:
__DI( ) ;
/* describe processing to be performed with interrupt disabled */
__EI( ) ;
:
}
[output code of C language source above]
_func1:
- - prologue code
:
di
- - processing to be performed with interrupt disabled
ei
:
- - epilogue code
jmp [lp]
3.6.2
Disabling interrupts in entire function
The CA850 has a "#pragma block_interrupt" directive that disables the interrupts of an entire function.
User’s Manual U18513EJ1V0UM
81
CHAPTER 3 C LANGUAGE EXPANSION
This directive is described as follows.
#pragma block_interrupt function-name
Describe functions that are described in the C language. In the case of a function, "void func1() {}", specify
"func1".
The interrupt to the function specified by "function-name" above is disabled.
As explained in "3.6.1 Locally disabling interrupt in function", "_ _DI()" can be described at the beginning of a
function and "__EI()", at the end. In this case, however, an interrupt to the prologue code and epilogue code
output by the CA850 cannot be disabled or enabled, and therefore, interrupts in the entire function cannot be
disabled.
Using the #pragma block_interrupt directive, interrupts are disabled immediately before execution of the
prologue code, and enabled immediately after execution of the epilogue code. As a result, interrupts in the entire
function can be disabled.
Example (How to use the #pragma block_interrupt directive and the code that is output are shown below.)
[C language source]
#pragma block_interrupt func1
void func1(void)
{
:
/* describe processing to be performed with interrupt disabled */
:
}
[output code of C language source above]
_func1:
di
- - prologue code
:
- - processing to be performed with interrupt disabled
:
- - epilogue code
ei
jmp [lp]
82
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.6.3
Notes on disabling interrupts in entire function
Note the following points when disabling interrupts in an entire function.
(1)
If an interrupt handler and a #pragma block_interrupt directive are specified for the same interrupt, the
interrupt handler takes precedence, and the setting of disabling interrupts is ignored.
(2)
If the following functions are called in a function in which an interrupt is disabled, the interrupt is enabled
when execution has returned from the call.
-
Function specified by #pragma block_interrupt
-
Function that disables interrupt at the beginning and enables interrupt at the end
(3)
Describe the #pragma block_interrupt directive before the function definition in the same file; otherwise an
error occurs during compilation. However, the order of prototype declaration of a function is not affected.
(4)
Neither #pragma inline nor inline expansion can be specified by an optimization option for the function
specified by a #pragma block_interrupt directive. The inline expansion specification is ignored.
(5)
A code that manipulates the ep flag (that indicates exception processing is in progress) in the program
status word (PSW) is not output even if #pragma block_interrupt is specified.
User’s Manual U18513EJ1V0UM
83
CHAPTER 3 C LANGUAGE EXPANSION
3.7
Interrupt/Exception Processing Handler
The CA850 can describe an interrupt handler or exception handler that is called if an interrupt or exception
occurs. This section explains how to describe these handlers.
3.7.1
Occurrence of interrupt/exception
If an interrupt or exception occurs in the V850, the program jumps to a handler address corresponding to the
interrupt or exception. An interrupt source and a handler address correspond one by one. A collection of handler
addresses is called an interrupt/exception table. For example, the interrupt/exception table of the V850ES/SG2 is
as shown below (only the top part is shown).
Table 3 - 5 Interrupt/Exception Table (V850ES/SG2)
Address
Interrupt Name
Interrupt Trigger
0x00000000
RESET
RESET pin input/reset by internal source
0x00000010
NMI
Valid edge input to NMI pin
0x00000020
INTWDT2
Overflow of WDT2
0x00000040
TRAP0n
TRAP instruction
0x00000050
TRAP1n
TRAP instruction
0x00000060
LGOP/DBG0
Illegal instruction code/DBTRAP instruction
0x00000080
INTLVI
Low voltage detection
0x00000090
INTP0
Detection of input edge of external interrupt pin (INTP0)
0x000000A0
INTP1
Detection of input edge of external interrupt pin (INTP1)
0x000000B0
INTP2
Detection of input edge of external interrupt pin (INTP2)
0x000000C0
INTP3
Detection of input edge of external interrupt pin (INTP3)
:
:
The arrangement of the handler addresses and the available interrupts vary depending on the device of the
V850. Refer to the Relevant Device’s Hardware User’s Manual of each device for details.
Each handler address has a 16-byte area. If an interrupt occurs, an instruction written in that 16-byte area is
executed. This means that, if the processing code does not exceed 16 bytes, it is performed only in the handler
address. If not, an instruction that branches to a function (interrupt handler) where the processing is written is
described.
84
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
Figure 3 - 6 Image of Interrupt Handler Address
Address
0x00000090
jr _func_intp0
Interrupt handler
address of INTP0
0x00000100
jr _func_intp1
Interrupt handler
address of INTP1
If the INTP0 interrupt occurs in the V850ES/SG2, the program jumps to address
0x90 and executes the code written at that address.
In this example, the program jumps to the func_intp0 function because a code
that branches to that function is written.
This means that func_intp0 is the interrupt handler of INTP0.
The above description is at an assembly language source level. With the CA850, users do not have to pay
much attention to this when describing interrupt servicing in C language source. How to describe interrupt
servicing is explained specifically in "3.7.3 Describing interrupt/exception handler".
User’s Manual U18513EJ1V0UM
85
CHAPTER 3 C LANGUAGE EXPANSION
3.7.2
Processing necessary in case of interrupt/exception
If an interrupt/exception occurs while a function or a task is being executed, interrupt/exception processing
must be immediately executed. When the interrupt/exception processing is completed, execution must return to
the function or task that was interruptedNote. Therefore, the register information at that time must be saved
when an interrupt/exception occurs, and the register information must be restored when interrupt/exception
processing is complete.
Note
When a real-time OS is used, execution may not return to a task that is interrupted if a system call is
issued during interrupt servicing. Refer to the User's Manual of each real-time OS for details.
The prologue and epilogue codes of an ordinary function save and restore the registers for register variables.
The registers for register variables are shown below. Those that must be saved and restored are saved and
restored.
Table 3 - 6 Registers for Register Variables
Register Mode
Registers for Register Variables
22-register mode
r25, r26, r27, r28, r29
26-register mode
r23, r24, r25, r26, r27, r28, r29
32-register mode
r20, r21, r22, r23, r24, r25, r26, r27, r28, r29
When execution shifts to an interrupt/exception handler, the following registers that must be saved, in addition
to the registers shown in the above table, are also saved as a stack frame for the interrupt/exception handler.
Table 3 - 7 Stack Frame for Interrupt/Exception Handler
Register Mode
86
Registers Saved/Restored in Case of Interrupt/Exception
22-register mode
r1, r6, r7, r8, r9,
r10, r11, r12, r13, r14,
r31(lp), CTPC [V850E], CTPSW [V850E]
26-register mode
r1, r6, r7, r8, r9,
r10, r11, r12, r13, r14, r15, r16,
r31(lp), CTPC [V850E], CTPSW [V850E]
32-register mode
r1, r6, r7, r8, r9,
r10, r11, r12, r13, r14, r15, r16, r17, r18, r19,
r31(lp), CTPC [V850E], CTPSW [V850E]
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
When execution shifts to an interrupt/exception handler, the following registers that must be saved, in addition
to the registers shown in the above table, are also saved as a stack frame for the interrupt/exception handler.
Table 3 - 8 Stack Frame for Multiple Interrupt/Exception Handler
Register Mode
Registers Saved/Restored in Case of Multiple Interrupts/Exceptions
22-register mode
r1, r6, r7, r8, r9,
r10, r11, r12, r13, r14,
r31(lp), EIPC, EIPSW, CTPC [V850E], CTPSW [V850E]
26-register mode
r1, r6, r7, r8, r9,
r10, r11, r12, r13, r14, r15, r16,
r31(lp), EIPC, EIPSW, CTPC [V850E], CTPSW [V850E]
32-register mode
r1, r6, r7, r8, r9,
r10, r11, r12, r13, r14, r15, r16, r17, r18, r19,
r31(lp), EIPC, EIPSW, CTPC [V850E], CTPSW [V850E]
The usage of the above registers is as follows.
Table 3 - 9 Usage of Registers
Register
Usage
r1
Assembler-reserved register
r6 - r9
Registers for arguments (registers to set arguments of function)
r10 - r19
Work registers (registers used by CA850 to generate codes)
r31
Link pointer
CTPC [V850E]
Program counter (PC) when CALLT instruction is executed
CTPSW [V850E]
Program status word (PSW) when CALLT instruction is executed
EIPC
Program counter (PC) during interrupt/exception processing
EIPSW
Program status word (PSW) during EIPSW interrupt/exception processing
When interrupt/exception processing is completed, the code which restores saved registers is output, the reti
instruction is output. This instruction notifies the V850 that the interrupt servicing is completed.
If codes for saving/restoring registers or outputting the reti instruction are described as explained in "3.7.3
Describing interrupt/exception handler", the CA850 automatically outputs the relevant code. The code for saving/
restoring registers is output as explained in Table 3 - 10. The user therefore does not have to pay much
attention to this and can concentrate on describing the processing of the main body of the interrupt handler.
User’s Manual U18513EJ1V0UM
87
CHAPTER 3 C LANGUAGE EXPANSION
Table 3 - 10 Processing for Saving/Restoring Registers During Interrupt
Register Name
Explanation
r1 register
r1
Always saved/restored at interrupt.
Argument registers
r6 - r9
r6 is always saved/restored when the interrupt source is TRAP0/
TRAP1.
Saved/restored when a function call (including runtime functions)
exists.
Saved/restored if a function call does not exist but is used with an
interrupt function.
Work
registers
22-register
mode
r10 - r14
26-register
mode
r10 - r16
Saved/restored when a function call exists.
Saved/restored if a function call does not exist but is used with an
interrupt function.
32-register
mode
r10 - r19
22-register
mode
r25 - r29
26-register
mode
r23 - r29
32-register
mode
r20 - r29
Register
variable
registers
88
Register
Saved/restored as necessary, as with ordinary functions.
Link pointer
r31(lp)
Saved/restored when a function call (including runtime functions)
exists.
Saved/restored if a function call does not exist.
Interrupt-related system
registers
EIPCE,
EIPSW
Saved/restored with functions using the multiple interrupt qualifier
__multi_interrupt.
Not saved/restored with the __interrupt qualifier.
callt instruction-related
system registers [V850E]
CTPC,
CTPSW
Always saved/restored with interrupt functions being compiled
with a V850E/V850ES/V850E2 core device specified.
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.7.3
Describing interrupt/exception handler
The format in which an interrupt/exception handler is described does not differ from ordinary C functions, but
the functions described in C must be recognized as an interrupt/exception handler by the CA850. With the
CA850, an interrupt/exception handler is specified using the #pragma interrupt directive and _ _interrupt qualifier,
or #pragma interrupt directive and _ _multi_interrupt qualifier.
(1)
When specifying interrupt handler
#pragma interrupt Interrupt-request-name Function-name Allocation-method
_ _ interrupt Function-definition, or Function-declaration
(2)
When specifying multiple-interrupt handler
#pragma interrupt Interrupt-request-name Function-name Allocation-method
_ _ multi_interrupt Function-definition, or Function-declaration
Describe functions that are described in the C language. In the case of a function, "void func1() {}", specify
"func1".
"Specifying multiple-interrupt handler" means to "specify a function that can be interrupted more than once"
and does not mean "to specify a function that interrupts more than once".
(a)
Interrupt request name
Interrupt request names registered in the device file can be specified. Refer to the interrupt request
names described in the Relevant Device’s Architecture User’s Manual of each device; they are the interrupt
request names registered in the device file.
A non-maskable interrupt (NMI) can also be specified in this way, but a reset interrupt (RESET) cannot be
specified. Processing after reset must be described with assembler instructions. Processing after reset is
generally described in the startup routine, so refer to "CHAPTER 5 STARTUP ROUTINE" for details.
(b)
Function name
Specify the names of functions that are used as an interrupt handler. Describe the function name in C lan-
guage source. When specifying the function "void func1(void)", specify the function name as "func1".
(c)
Allocation method
Specify whether the main body of the function is directly allocated to the handler address, or only the
instruction that branches to the interrupt handler function is allocated. Specify "direct" when the main body of
the function is directly allocated; otherwise describe nothing as "allocation method".
By specifying "direct", all functions are allocated from the handler address of the specified interrupt
source. Note, however, that the areas for the subsequent handler address are also used.
When specifying "direct", be sure to describe the #pragma interrupt directive before the function definition;
otherwise an error occurs during compilation.
User’s Manual U18513EJ1V0UM
89
CHAPTER 3 C LANGUAGE EXPANSION
Next, the roles of the #pragma interrupt directive, __interrupt qualifier, and __multi_interrupt qualifier are
explained.
(1)
#pragma interrupt directive
Allocates an instruction (jr) that branches to the specified function to a handler address corresponding to the
interrupt request name specified by the #pragma interrupt directive. When the -Xj option is specified, this
directive allocates an instruction that saves the r1 register contents to the stack and an instruction (jmp) that
branches to the specified function.
(2)
__interrupt qualifier
Adds processing to save/restore the register contents by an interrupt/exception handler to a function with
the __interrupt qualifier and adds the reti instruction at the end. When the -Xj option is specified, processing to
save the r1 register contents is output to the handler address, so only restore processing is output for the
function.
(3)
multi_interrupt qualifier
Adds processing to save/restore the register contents by an interrupt handler and processing to save/
restore the contents of the EIPC and EIPSW registers to a function with the __multi_interrupt qualifier. This
directive also adds the reti instruction at the end. When the -Xj option is specified, processing to save the r1
register contents is output to the handler address, so only restore processing is output for the function.
When the #pragma interrupt directive, __interrupt qualifier, and __multi_interrupt qualifier are specified at the
same time, the following codes are output and the handler completes the interrupt servicing routine.
-
Allocation of an instruction branching to the specified interrupt handler to the handler address
-
Addition of processing to save/restore the register contents by an interrupt handler (and processing to
save/restore the contents of EIPC and EIPSW if the __ multi_interrupt qualifier is specified)
-
Addition of the reti instruction at the end of the handler
In this case, function definition and the #pragma interrupt directive can be described in separate files in any
order. If "direct" is specified for the allocation method, however, they cannot be described in separate files.The
following codes are output if only the __ interrupt qualifier or __ multi_interrupt qualifier is specified.
-
Addition of processing to save/restore the register contents by an interrupt handler (and processing to
save/restore the contents of EIPC and EIPSW if the __ multi_interrupt qualifier is specified)
-
Addition of the reti instruction at the end of the handler
Therefore, the function can be started as an interrupt handler but the processing to allocate "an instruction to
branch to the interrupt handler to the handler address" output by the #pragma interrupt directive is not
performed.
90
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
Example
The #pragma interrupt is specified as follows when the interrupt handler "void intp0_func(void)" is used for
the interrupt request name "INTP0" without "direct" being specified and multiple interrupts being enabled.
#pragma interrupt INTP0 intp0_func
_ _ interrupt
void intp0_func(void)
{
:
/* main body of interrupt servicing */
:
}
Next, the function type that can be specified as an interrupt handler is explained.
(1)
Function type
The type of a handler that handles a maskable interrupt or NMI is as follows.
void func(void) type
The argument and return value of this function are void type.
The type of a software exception processing (trap) handler is as follows.
void func(unsigned int) type
EICC (exception code) of the interrupt source register (ECR) is set as the argument. Unless the function is
specified by this type, an error occurs during compilation. Refer to the next paragraph for the software
exception processing function.
(2)
Software exception processing (trap processing) handler
When software exception processing (trap processing) is used, two entry points, TRAP0 (address 0x40) and
TRAP1 (address 0x50), are used according to the specifications of the V850 microcontrollers.
When the software exception "trap 0x00 to trap 0x0f" occurs, execution branches to TRAP0 (address 0x40);
if "trap 0x10 to trap0x1f" occurs, it branches to TRAP1 (address 0x50). At this time, the value "0x40 to 0x4f" is
set to the interrupt source register (ECR) as a software exception code in the case of TRAP0. In the case of
TRAP1, the value "0x50 to 0x5f" is set to the ECR.
Table 3 - 11 Trap Instructions and Software Exception Codes
Trap Instruction
Software Exception Code
trap 0x00
0x40
trap 0x01
0x41
trap 0x02
0x42
:
:
:
:
trap 0x0a
0x4a
trap 0x0b
0x4b
User’s Manual U18513EJ1V0UM
91
CHAPTER 3 C LANGUAGE EXPANSION
Table 3 - 11 Trap Instructions and Software Exception Codes
Trap Instruction
Software Exception Code
:
:
:
:
trap 0x10
0x50
trap 0x11
0x51
trap 0x12
0x52
:
:
:
:
trap 0x1e
0x5e
trap 0x1f
0x5f
When software exception processing for TRAP0 or TRAP1 is described, that function has one argument and
the type of the variable is "unsigned int". The software exception code set to the interrupt source register (ECR)
is set as the argument. In the case of TRAP0, the value is "0x40 to 0x4f". In the case of TRAP1, it is "0x50 to
0x5f". Processing must be described in the handler depending on these values.
#pragma interrupt TRAP0 trap0_func
_ _ interrupt
void trap0_func(unsigned int codenum)
{
:
/* describe processing of each exception code. */
:
}
92
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.7.4
(1)
Notes on describing interrupt/exception handler
"Specifying multiple-interrupt handler" with the __multi_interrupt qualifier means to "specify a function that
can be interrupted more than once" and does not mean "to specify a function that interrupts more than
once".
(2)
Even if a handler that enables multiple interrupts is specified by _ _multi_interrupt, interrupts are not
enabled when the interrupt handler is activated. Therefore, be sure to issue an interrupt enabling
instruction (such as __EI()) in the interrupt handler, and issue an interrupt disabling instruction (such as
__DI()) at the end of the handler. If the interrupt disabling instruction is not issued at the end of the handler,
an interrupt may be acknowledged while the contents of a register are being restored, which may cause a
hang-up.
(3)
The reset interrupt cannot be specified by the #pragma interrupt directive.
#pragma interrupt RESET reset_func /* error */
If the above description is made, an error occurs during compilation. Processing after reset must be
described with assembler instructions. Processing after reset is generally described in the startup routine,
so refer to "CHAPTER 5 STARTUP ROUTINE" for details.
(4)
The #pragma interrupt directive and _ _multi_interrupt qualifier do not support multiple exceptions and
multiple NMIs. To use multiple exceptions or multiple NMI, add a code that saves or restores the necessary
system registers (such as FEPC and FEPSW). Refer to the Relevant Device’s Hardware User’s Manualt of
each device for the necessary system registers.
(5)
The user is not required to additionally describe an interrupt handler address in the link directive file; it is
output internally by the CA850.
(6)
The same interrupt request name must not be specified for two or more functions.
(7)
Both the __ interrupt qualifier and __multi_interrupt qualifier must not be specified for the same function.
(8)
An error occurs during compilation if a function is declared with the __interrupt qualifier or __multi_interrupt
qualifier after the function is defined without the __interrupt qualifier or __multi_interrupt qualifier being
specified.
(9)
A function specified as an interrupt/exception handler cannot be expanded inline. The #pragma inline
directive is ignored even if specified.
(10) An interrupt to a function specified as an interrupt/exception handler is disabled. Therefore, the #pragma
block_interrupt directive is ignored even if specified.
(11) A function specified as an interrupt/exception handler cannot be called by an ordinary function call. If it is
called from another file, the compiler cannot check it.
User’s Manual U18513EJ1V0UM
93
CHAPTER 3 C LANGUAGE EXPANSION
(12) When an assembler program is called from an interrupt/exception handler and the registers for register
variables and the registers shown in Table 3 - 6 and Table 3 - 7 are used, processing to save/restore the
register contents must be described. Processing to save/restore the register contents must also be
described when sp (r3), gp (r4), tp (r5), and ep (r30) are rewritten.
(13) The #pragma interrupt directive, __interrupt qualifier, and __multi_interrupt qualifier do not issue a
processing end report (EOI command) to the external interrupt controller. The user should therefore
execute this directive, if necessary.
(14) Disable interrupts at the end of multiple interrupts because a code that restores EIPC and EIPSW must be
described.
(15) If "direct" is not specified, an instruction to branch to the interrupt/exception handler is allocated to the
handler address. In this case, the CA850 outputs the jr instruction to enhance the code efficiency.
However, the range in which the jr instruction can branch execution is limited to +21 bits from the jr
instruction. If the main body of the interrupt handler is not within the range in which the jr instruction can
branch execution, an error occurs during linking. In this case, specify the compilation option "-Xj" to replace
the jr instruction with the jmp instruction.
94
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.7.5
Description example of interrupt/exception handler
Examples of describing interrupt/exception handlers are shown below. Note that the interrupt request name
differs depending on the device. Refer to the Relevant Device’s Architecture User’s Manual of each device.
Example 1 (Non-maskable interrupt)
#pragma interrupt NMI func1
/* non-maskable interrupt */
_ _ interrupt
void func1(void)
{
:
}
Example 2 (Trap)
#pragma interrupt TRAP0 func2
_ _ interrupt
void func2(unsigned int num)
{
switch(num){
:
}
/* trap 0 */
/* for every exception code */
}
Example 3 (#pragma interrupt and __ interrupt qualifier in separate files)
[a. c]
_ _ interrupt
void func1(void)
{
:
}
[b. c]
/* _ _ interrupt specification */
#pragma interrupt NMI func1 /* can be described after definition or in separate file */
Example 4 (Specification of multiple interrupts)
#pragma interrupt INTP0 func1
_ _ multi_interrupt
/* multiple-interrupt function specified */
void func1(void)
{
:
}
User’s Manual U18513EJ1V0UM
95
CHAPTER 3 C LANGUAGE EXPANSION
3.8
Inline Expansion
The CA850 allows inline expansion of each function. This section explains how to specify inline expansion.
3.8.1
Inline expansion
Inline expansion is used to expand the main body of a function at a location where the function is called. This
decreases the overhead of function call and increases the possibility of optimization. As a result, the execution
speed can be increased. If inline expansion is executed, however, the object size increases.
Specify the function to be expanded inline using the #pragma inline directive.
#pragma inline function-name [,function-name...]
Describe functions that are described in the C language. In the case of a function, "void func1() {}", specify
"func1". Two or more function names can be specified with each delimited by "," (comma).
#pragma inline func1,func2
void func1(){ ... }
void func2(){ ... }
void func(void)
{
func1(); /* function subject to inline expansion */
func2(); /* function subject to inline expansion */
}
96
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.8.2
Conditions of inline expansion
At least the following conditions must be satisfied for inline expansion of a function specified using the
#pragma inline directive. If optimization other than "size priority optimization (-Os)" and "execution speed priority
optimization (-Ot)" is specified, however, inline expansion may not be executed even if the following conditions
are satisfied, because of the internal processing of the CA850.
(1)
A function that expands inline and a function that is expanded inline are described in the same file.
A function that expands inline and a function that is expanded inline, i.e., a function call and a function
definition must be in the same file. This means that a function described in another C language source cannot
be expanded inline. If it is specified that a function described in another C language source is expanded inline,
the CA850 does not output a warning message and ignores the specification.
(2)
The #pragma inline directive is described before function definition.
If the #pragma inline directive is described after function definition, the CA850 outputs a warning message
and ignores the specification. However, prototype declaration of the function may be described in any order.
Here is an example.
Example
Valid Inline Expansion Specification
#pragma inline func1,func2
/* prototype declaration */
void func1();
void func2();
/* function definition */
void func1() { /* ... */ }
void func2() { /* ... */ }
(3)
Invalid Inline Expansion Specification
/* prototype declaration */
void func1();
void func2();
/* function definition */
void func1() { /* ... */ }
void func2() { /* ... */ }
#pragma inline func1,func2
The number of arguments is the same between "call" and "definition" of the function to be expanded inline.
If the number of arguments is different between "call" and "definition" of the function to be expanded inline,
the CA850 outputs a warning message and ignores the specification.
(4)
The types of return value and argument are the same between "call" and "definition" of the function to be
expanded inline.
If the return value type and argument type are different between "call" and "definition" of the function to be
expanded inline, the CA850 outputs a warning message and ignores the specification. If the type can be
converted, however, it is converted as follows and the function is expanded inline.
-
The return value type is the type of the "calling side".
-
The argument type is the type of the "function definition".
If the "-ansi" option is specified, however, the type is not converted and an error is output.
User’s Manual U18513EJ1V0UM
97
CHAPTER 3 C LANGUAGE EXPANSION
(5)
The size of the function to be expanded inline and the stack size are not too large.
If the size of the function to be expanded inline and the stack size are too large, neither an error nor warning
is output, and the inline expansion specification is ignored. This "size" means the size in the intermediate
language and is different from the size of the actual object. The upper limit of the size can be changed in the
CA850. The function size in the intermediate language can be changed by this option.
The function size in the intermediate language can be changed by this option.
-Wp,-Nnum
The stack size used by the function in the intermediate language can be changed by this option.
-Wp,-Gnum
In addition, the size of each function and stack size used in the intermediate language can be checked by
using this option.
-Wp,-l
This option can be used to determine the size for specification.
(6)
The number of arguments of the function to be expanded inline is not variable.
If inline expansion is specified for a function with a variable number of arguments, the CA850 outputs neither
an error nor warning message and ignores the specification.
(7)
Recursive function is not specified to be expanded inline.
If a recursive function that calls itself is specified for inline expansion, the CA850 outputs neither an error nor
warning message and ignores the specification. If two or more function calls are nested and if a code that calls
itself exists, however, inline expansion may be executed.
(8)
An interrupt handler is not specified to be expanded inline.
A function specified by the #pragma interrupt, __ interrupt, or t__multi_interrupt directive is recognized as
an interrupt handler. If inline expansion is specified for this function, the CA850 outputs a warning message
and ignores the specification.
(9)
A task of a real-time OS is not specified to be expanded inline.
A function specified by the #pragma rtos_task directive is recognized as a task of a real-time OS. If inline
expansion is specified for this function, the CA850 outputs a warning message and ignores the specification.
(10) Interrupts are not disabled in a function by the #pragma block_interrupt directive.
If inline expansion is specified for a function in which interrupts are declared by the #pragma block_interrupt
directive to be disabled, the CA850 outputs a warning message and ignores the specification.
98
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.8.3
Controlling inline expansion via options
Inline expansion can be controlled using options when inline expansion by the compiler should be suppressed.
The cases in which inline expansion can be controlled and the options are as follows. If execution speed priority
optimization (-Ot) is specified, however, refer to "3.8.4
Execution speed priority optimization and inline
expansion".
(1)
To expand inline all static functions that are referenced only once
-Wp,-S
If this option is specified, a static function that is referenced only once is expanded inline, regardless of
optimization specification and the presence or absence of a #pragma inline specification. If optimization other
than the size priority optimization (-Os) is specified, however, inline expansion may not be executed even if the
-Wp,-S option is specified, because of the internal processing of the CA850.
(2)
To suppress inline expansion of all functions
-Wp,-no_inline
In this case, inline expansion is suppressed even if the -Wp,-S option or the #pragma inline directive is
specified.
User’s Manual U18513EJ1V0UM
99
CHAPTER 3 C LANGUAGE EXPANSION
3.8.4
Execution speed priority optimization and inline expansion
If the "execution speed priority optimization (-Ot)" option of the CA850 is specified, the CA850 uses inline
expansion as one of the means of optimization. If the -Ot option is specified, the CA850 selects an appropriate
function and expands it inline as long as the inline expansion conditions in "3.8.2 Conditions of inline expansion"
are satisfied, even if the function is not specified for inline expansion by the #pragma inline directive. Inline
expansion can be controlled using options when inline expansion by the compiler should be suppressed. The
items that can be controlled and the options are as follows.
(1)
To suppress inline expansion of all functions even though the -Ot option is specified
-Wp,-no_inline
In this case, inline expansion is suppressed even if the -Wp,-S option or the #pragma inline directive is
specified.
(2)
To expand inline only the function specified by the #pragma inline directive even though the -Ot option is
specified
-Wp,-inline
In this case, the function for which inline expansion is specified must meet the conditions explained in "3.8.2
Conditions of inline expansion".
100
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.8.5
Examples of differences in inline expansion operation depending on
option specification
Here are examples of differences in inline expansion operation depending on whether the #pragma inline
directive or an option is specified.
When -Os (size priority optimization) is specified (other than -Ot)
#pragma inline func0
/* expanded if inline expansion conditions are satisfied because
#pragma inline is specified #pragma inline */
void func0(){...}
void func1(){...} /* not expanded */
void func2(){...} /* not expanded */
When -Ot (execution speed priority optimization) is specified
#pragma inline func0
/* expanded if inline expansion conditions are satisfied because -Ot is
specified */
void func0(){...}
/* expanded if inline expansion conditions are satisfied because -Ot is
specified */
void func1(){...}
/* expanded if inline expansion conditions are satisfied because -Ot is
specified */
void func2(){...}
When -Ot (execution speed priority optimization)
+ -Wp,-inline (inline expansion of only function specified by #pragma inline) are specified
#pragma inline func0
/* expanded if inline expansion conditions are satisfied because
#pragma inline is specified */
void func0(){...}
/* not expanded because -Wp,-inline is specified but #pragma inline is not
specified */
void func1(){...}
/* not expanded because -Wp,-inline is specified but #pragma inline is not
specified */
void func2(){...}
Remarks 1 The CA850 does not treat a function specified for inline expansion by the #pragma inline directive
as a static function. To use such a function as a static function, static must be explicitly specified.
2 When executing debugging, a breakpoint cannot be specified for a function specified for
inlineexpansion in the C language source.
User’s Manual U18513EJ1V0UM
101
CHAPTER 3 C LANGUAGE EXPANSION
3.9
Real-Time OS Support Function
The CA850 has functions to improve programming description and to reduce the number of codes, making
allowances for organizing a system using the V850 microcontrollers real-time OS RX850 or RX850 Pro.
3.9.1
Description of task
An application using a real-time OS performs processing in task units. The real-time OS schedules a task
using a system call issued in that task or interrupt servicing. Register contents are saved and restored by the
real-time OS when the task is switched (when the context is switched). Therefore, prologue and epilogue
processing are different from those of an ordinary function. In other words, the prologue and epilogue processing
generated by the CA850 when a function is called are not executed by a task.
To use a function described as a task, the code can be reduced by deleting the prologue and epilogue
processing that are executed when a function is called. However, ordinary functions and tasks are not
distinguished according to the description method of C language. Therefore, the CA850 has the following
#pragma directive so that a function can be recognized as a task of a real-time OS.
#pragma rtos_task [function-name]
Consequently, the function specified by "function-name" can be recognized as a task of a real-time OS. A
function name described in C is specified as "function-name". The following description is made, for example, to
use the function "void func1(int inicode){}" as a task.
#pragma rtos_task func1
"function-name" can also be omitted. If omitted, the function following the #pragma rtos_task directive in that
file is recognized as a task.
Specifying the #pragma rtos_task directive has the following effect.
(1)
The prologue/epilogue processing output by an ordinary function is not performed. Specifically, the
following codes are not output.
102
(a)
Saving/restoring of register contents for register variables
(b)
Saving/restoring of link pointer (lp)
(c)
Jump to return address
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
(2)
The system call "ext_tsk" can be used as a defined function.
This system call can be used even if a prototype declaration is not made in the application. Functions other
than the one specified as a task can be called in the same manner as long as they are described after the
#pragma rtos_task directive. When this system call is called, a code using the jr instruction is output to reduce
the code size. If the main body of system call "ext_tsk" is not in the range in which the jr instruction can branch
execution, the linker (ld850) outputs an error. In this case, take the following actions.
(a)
Change the memory allocation by the link directive.
(b)
Replace the jr instruction with the jmp instruction in the assembly language source.
(c)
Specify far jump
Note the following points when the #pragma rtos_task directive is specified.
-
A task cannot be called in the same manner as calling a function. A task called from a separate file is not
checked. A task cannot be expanded inline because it cannot be called as a function. That is, even if the
#pragma inline directive is specified for a function specified by the #pragma rtos_task directive, the
#pragma inline specification is ignored.
-
An error occurs if "#pragma rtos_task function-name" is described after the function definition in the same
file. If the function is not defined after "#pragma rtos_task function-name" is described in the file, the
#pragma directive for that function is ignored.
-
A function specified by the #pragma rtos_task directive cannot be specified as an ordinary interrupt/
exception handler (refer to "3.7 Interrupt/Exception Processing Handler").
Refer to the User's Manual of each real-time OS for the real-time OS functions.
User’s Manual U18513EJ1V0UM
103
CHAPTER 3 C LANGUAGE EXPANSION
3.10
Embedded Functions
In the CA850 some of the instructions can be described in C language source as embedded functions.
Table 3 - 12 shows the instructions that can be described as functions.
Table 3 - 12 Embedded Functions
Assembler
Instruction
Function
Description
di
ei
Interrupt control (DI/EI)
__DI()
__EI()
nop
nop
__nop()
halt
halt
__halt()
satadd
Saturated addition (satadd)
long a, b;
long __satadd(a, b)
satsub
Saturated subtraction (satsub)
long a, b;
long __satsub(a, b)
bsh
Halfword data byte swap (bsh) [V850E]
long a;
long __bsh(a)
bsw
Word data byte swap (bsw) [V850E]
long a;
long __bsw(a)
hsw
Word data halfword swap (hsw) [V850E]
long a;
long __hsw(a)
sxb
Byte data sign extension (sxb) [V850E]
char a;
long __sxb(a)
sxh
Halfword data sign extension (sxh) [V850E]
short a;
long __sxh(a)
mul
Instruction that assigns higher 32 bits of multiplication
result to variable using mul instruction [V850E]
long a; long b;
long __mul32(a, b)
mulu
Instruction that assigns higher 32 bits of unsigned
multiplication result to variable using mulu instruction
[V850E]
unsigned long a, b;
unsigned long _ _mul32u(a, b)
sasf
Flag condition setting with logical left shift (sasf)
[V850E]
long a;
unsigned int b;
long __sasf(a, b)
Remark [V850E] mark indicates that only V850Ex core is available.
Caution Even if a function is defined with the same name as an embedded function, it cannot be used.
If an attempt is made to call such a function, processing for the embedded function provided by the
compiler takes precedence.
104
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.10.1
Interrupt control (DI/EI)
An example of describing the interrupt control (DI/EI) instruction is shown below.
Example
void func(void)
{
:
_ _ DI(); /* Describe the processing to be executed while interrupts are disabled. */
__EI();
:
}
Compiler output of above example
_func:
- - Prologue code
di
-- Describe the processing to be executed while interrupts are disabled.
ei
:
- - Epilogue code
jmp [lp]
3.10.2
nop
An example of describing the nop instruction is shown below.
Example
void func(void)
{
:
_ _ nop();
:
}
Compiler output of above example
_func:
:
nop
:
User’s Manual U18513EJ1V0UM
105
CHAPTER 3 C LANGUAGE EXPANSION
3.10.3 halt
An example of describing the halt instruction is shown below.
Example
void func(void)
{
:
_ _halt();
}
Compiler output of above example
_func:
:
halt
3.10.4
Saturated addition (satadd)
An example of describing the saturated addition instruction is shown below.
Example
void func(void)
{
long
a, b, c;
:
c = _ _ satadd(a, b);
/* The result of the saturated operation of a */
/* and b is stored in c. */
:
}
Compiler output of above example
_func:
:
ld.w
ld.w
satadd
st.w
- - The
:
106
-4 +.A2[sp], r10
--8 +.A2[sp], r11
-r11, r10
-r10, -12 +.A2[sp]
result of the saturated
Load variable a
Load variable b
Saturated subtraction (a + b)
operation is stored in variable c.
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.10.5
Saturated subtraction (satsub)
An example of describing the saturated subtraction instruction is shown below
Example
void func(void)
{
long
a, b, c;
:
c = _ _ satsub(a, b);
/* The result of the saturated operation of a */
/* and b is stored in c (c = a - b). */
:
}
Compiler output of above example
_func:
:
-4 +.A2[sp], r10
-8 +.A2[sp], r11
ld.w
ld.w
-- Load variable a
-- Load variable b
satsub r11, r10
-- Saturated subtraction (a - b)
st.w
r10, -12 +.A2[sp]
- - The result of the saturated operation is stored in variable c.
:
3.10.6
Halfword data byte swap (bsh) [V850E]
An example of describing the halfword data byte swap (bsh) instruction is shown below.
Example
void func(void)
{
long
a, b;
:
b = _ _ bsh(a);
/* Halfword data of a is byte-swapped */
/* and the result is stored in b. */
:
}
Compiler output of above example
_func:
:
ld.w
bsh
st.w
-4+.A2[sp], r10
r10, r10
r10, -8+.A2[sp]
-- Load variable a
-- Halfword data byte swap
- - The result of halfword data byte swap is stored in variable b.
:
User’s Manual U18513EJ1V0UM
107
CHAPTER 3 C LANGUAGE EXPANSION
3.10.7
Word data byte swap (bsw) [V850E]
An example of describing the word data byte swap (bsw) instruction is shown below.
Example
void func(void)
{
long
a, b;
:
b = _ _ bsw(a);
:
}
/* Word data of a is byte-swapped and the result is stored in b.*/
Compiler output of above example
_func:
:
ld.w
bsw
st.w
:
3.10.8
-8+.A2[sp], r10
r10, r10
r10, -12+.A2[sp]
-- Load variable a
-- Word data byte swap
-- Stored in variable b
Word data halfword swap (hsw) [V850E]
An example of describing the word data halfword swap (hsw) instruction is shown below.
Example
void func(void)
{
long
a, b;
:
b = _ _ hsw(a);
:
}
/* Word data of a is halfword-swapped and the result is stored in b. */
Compiler output of above example
_func:
:
ld.w
hsw
st.w
:
108
-8+.A2[sp], r10
r10, r10
r10, -12+.A2[sp]
-- Load variable a
-- Word data halfword swap
-- Stored in variable b
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.10.9
Byte data sign extension (sxb) [V850E]
An example of describing the byte data sign extension (sxb) instruction is shown below.
Example
void func(void)
{
char
a;
long
b;
:
b = _ _ sxb(a);
/* Sign extension of the byte data of a is performed */
/* and the result is
stored in b. */
:
}
Compiler output of above example
_func:
:
ld.w
sxb
st.w
-8+.A2[sp], r10
r10, r10
r10, -12+.A2[sp]
-- Load variable a
-- Sign extension of byte data
-- Stored in variable b
:
3.10.10 Halfword data sign extension (sxh) [V850E]
An example of describing the halfword data sign extension (sxh) instruction is shown below.
Example
void func(void)
{
short
a;
long
b;
:
b = _ _ sxh(a);
/* Sign extension of the halfword data of a is */
/* performed and the result is stored in b. */
:
}
Compiler output of above example
_func:
:
ld.w
sxh
st.w
:
-8+.A2[sp], r10
r10
r10, -12+.A2[sp]
-- Load variable a
-- Sign extension of halfword data
-- Stored in variable b
User’s Manual U18513EJ1V0UM
109
CHAPTER 3 C LANGUAGE EXPANSION
3.10.11 Instruction that assigns higher 32 bits of multiplication result to
variable using mul instruction [V850E]
An example of describing the instruction that assigns variable for the higher 32 bits of the multiplication result
using the mul instruction is shown below.
Example
void func(void)
{
long
a, b, c;
:
c = _ _ mul32(a, b); /* The higher 32 bits of the result of a * b are stored in c. */
:
}
Compiler output of above example
_func:
:
ld.w
ld.w
mul
st.w
:
110
-4+.A2 [sp], r10
-8+.A2 [sp], r11
r11, r10, r12
r12, -12+.A2 [sp]
-----
Load variable a
Load variable b
a * b
Stored in variable c
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.10.12 Instruction that assigns higher 32 bits of unsigned multiplication result
to variable using mulu instruction [V850E]
An example of describing the instruction that assigns the higher 32 bits of the unsigned multiplication result to
variable using mulu instruction is shown below.
Example
void func(void)
{
unsigned long a, b, c;
:
c = _ _ mul32u(a, b); /* The higher 32 bits of the result of a * b are stored in c. */
:
}
Compiler output of above example
_func:
:
ld.w
ld.w
mulu
st.w
:
-4+.A2 [sp], r10
-8+.A2 [sp], r11
r11, r10, r12
r12, -12+.A2 [sp]
-----
Load variable a
Load variable b
a * b
Stored in variable c
User’s Manual U18513EJ1V0UM
111
CHAPTER 3 C LANGUAGE EXPANSION
3.10.13 Flag condition setting with logical left shift (sasf) [V850E]
An example of describing the flag condition setting instruction with logical left shift when a conditional
expression is written in the second argument is shown in Example 1.
An example of describing the flag condition setting instruction with logical left shift when a variable is written in
the second argument is shown in Example 2.
Example 1
/* When a conditional expression is written in the second argument */
void func(void)
{
unsigned long a, b, c;
:
c = _ _ sasf(c, a == b); /* If a == b is true, c is shifted left logically by */
/* 1 bit and 1 is added. */
/* If a == b is not true, c is shifted left logically */
/* by 1 bit. The result is stored in c. */
:
}
Compiler output of above Example 1
_func:
:
ld.w
-4+.A2 [sp], r10
-- Load variable a
ld.w
-8+.A2 [sp], r11
-- Load variable b
cmp
r11, r10
-- Compare variable a and b.
ld.w
-12+.A6 [sp], r12 -- Load variable c
sasf
0x2, r12
- - If a == b is true, c is shifted left logically by 1 bit and 1 is added.
- - If a == b is not true, c is shifted left logically by 1 bit.
st.w
r12, -12+.A2 [sp] -- Stored in variable c.
:
Example 2
/* When a variable is written in the second argument */
void func(void)
{
unsigned long a, b;
:
b = _ _ sasf(b, a);
/* If a is not 0, b is shifted left logically by 1 bit */
/*
/*
/*
/*
and 1 is added. */
If a is other than 0, b is shifted left logically */
by 1 bit. */
The result is stored in b. */
:
}
112
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
Compiler output of above Example 2
_func:
:
ld.w
-4+.A2 [sp], r10
-- Load variable a
cmp
r0, r10
-- Compare variable a and 0.
ld.w
-8+.A2 [sp], r11
-- Load variable b
sasf
0xa, r11
- - If a is not 0, b is shifted left logically by 1 bit and 1 is added.
- - If a is 0, b is shifted left logically by 1 bit.
st.w
r11, -8+.A2 [sp]
-- Stored in variable b
:
User’s Manual U18513EJ1V0UM
113
CHAPTER 3 C LANGUAGE EXPANSION
3.11
Structure Packing Function
In the CA850, the alignment of structure members can be specified at the C language level. This function is
equivalent to the -Xpack option, however, the structure-packing directive can be used to specify the alignment
value in any location in the C language source.
Note
The data area can be reduced by packing a structure, but the program size increases and the
execution speed is degraded.
3.11.1
Structure packing specified
The structure packing function is specified in the following format.
#pragma pack([1248])
#pragma pack changes to an alignment value of the structure member upon the occurrence of this directive.
The alignment value is called the packing value and the specifiable numeric values are 1, 2, 4, and 8. When the
packing value is not specified, the default alignment (1)Note is specified. Since this directive becomes valid upon
occurrence, several directives can be described in the C language source.
Example
/* Structure member aligned using 1-byte alignment */
#pragma pack(1)
struct TAG {
char
c;
int
i;
short
s;
};
Note
114
Alignment values "4" and "8" are treated as the same in Ver. 2.70.
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.11.2
Rules of structure packing
The structure members are aligned in a form that satisfies the condition whereby members are aligned
according to whichever is the smaller value: the structure packing value or the member’s alignment value.
For example, if the structure packing value is 2 and member type is int type, the structure members are
aligned in 2-byte alignment.
Example
struct S {
char
c;
int
i;
};
#pragma pack(1)
struct S1 {
char
c;
int
i;
};
#pragma pack(2)
struct S2 {
char
c;
int
i;
};
struct S
sobj;
struct S1
s1obj;
struct S2
s2obj;
/* Satisfies 1-byte alignment condition */
/* Satisfies 4-byte alignment condition */
/* Satisfies 1-byte alignment condition */
/* Satisfies 1-byte alignment condition */
/* Satisfies 1-byte alignment condition */
/* Satisfies 2-byte alignment condition */
/* Size of 8 bytes */
/* Size of 5 bytes */
/* Size of 6 bytes */
sobj
c
0
i
7 8
31 32
63
s1obj
c
0
i
7 8
39
s2obj
c
0
i
7 8
15 16
47
User’s Manual U18513EJ1V0UM
115
CHAPTER 3 C LANGUAGE EXPANSION
3.11.3
Union
A union is treated as subject to packing and is handled in the same manner as structure packing.
Example 1
union U {
struct S {
char c;
int i;
} sobj;
};
#pragma pack(1)
union U1 {
struct S1 {
char c;
int i;
} s1obj;
};
#pragma pack(2)
union U2 {
struct S2 {
char c;
int i;
} s2obj;
};
union
U
uobj;
union
U1 u1obj;
union
U2 u2obj;
/* Size of 8 bytes */
/* Size of 5 bytes */
/* Size of 6 bytes */
Example 2
union U {
int 7:i;
};
#pragma pack(1)
union U1 {
int 7:i;
};
#pragma pack(2)
union U2 {
int 7:i;
};
union
U
uobj;
union
U1 u1obj;
union
U2 u2obj;
116
/* Size of 4 bytes */
/* Size of 1 byte */
/* Size of 2 bytes */
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.11.4
Bit field
Data is allocated to the area of the bit field element as follows.
(1) When the structure packing value is equal to or larger than the alignment condition value of the
member type
Data is allocated in the same manner as when the structure packing function is not used. That is, if the data is
allocated consecutively and the resulting area exceeds the boundary that satisfies the alignment condition of the
element type, data is allocated from the area satisfying the alignment condition.
(2) When the structure packing value is smaller than the alignment condition value of the element type
(a)
If data is allocated consecutively and results in the number of bytes including the area becoming larger
than the element type
The data is allocated in a form that satisfies the alignment condition of the structure packing value.
(b)
Other conditions
The data is allocated consecutively
Example
struct S {
short
short
short
short
a
b
c
d
:
:
:
:
7;
7;
7;
7;
/*
/*
/*
/*
} sobj;
#pragma pack (1)
struct S1 {
short
a : 7;
short
b : 7;
short
short
} s1obj;
0 to 6th bit */
7 to 13th bit */
16 to 22nd bit (aligned to 2-byte boundary) */
32 to 46th bit (aligned to 2-byte boundary) */
/* 0 to 6th bit */
/* 7 to 13th bit */
c : 7;
d : 15;
/* 14 to 20th bit */
/* 24 to 38th bit (aligned to byte boundary) */
sobj
a
0
b
6 7
c
13
16
d
22 23
31 32
46 47
63
s1obj
b
a
0
6 7
c
13 14
d
20 21 23 24
38 39 40
User’s Manual U18513EJ1V0UM
117
CHAPTER 3 C LANGUAGE EXPANSION
3.11.5
Alignment condition of top structure object
The alignment condition of the top structure object is the same as when the structure packing function is not
used.
3.11.6
Size of structure objects
Perform padding so that the size of structure objects becomes a multiple value of whichever is the smaller
value: the structure alignment condition value or the structure packing value. The alignment condition of the top
structure object is the same as when the structure packing function is not used.
Example 1
struct S {
int
i;
char
c;
};
#pragma pack(1)
struct S1 {
int
i;
char
c;
};
struct S
sobj;
struct S1
s1obj;
struct S2
s2obj;
/* Size of 8 bytes */
/* Size of 5 bytes */
/* Size of 6 bytes */
sobj
i
0
c
31 32
39 40
63
s1obj
i
0
c
31 32
39
s2obj
i
0
118
c
31 32
39 40
User’s Manual U18513EJ1V0UM
47
CHAPTER 3 C LANGUAGE EXPANSION
Example 2
struct S {
int
i;
char
c;
};
struct T {
char
c;
struct S s;
};
#pragma pack(1)
struct S1 {
int
i;
char
c;
};
struct T1 {
char
c;
struct S1 s1;
};
#pragma pack(2)
struct S2 {
int
i;
char
c;
};
struct T2 {
char
c;
struct S2 s2;
};
struct T
tobj;
struct T1
t1obj;
struct T2
t2obj;
/* Size of 12 bytes */
/* Size of 6 bytes */
/* Size of 8 bytes */
sobj
c
0
s.i
s.c
31 32
7 8
63 64
71 72
95
s1obj
c1
0
s1.i
s1.c
7 8
39 40
47
s2obj
c2
0
s2.i
7 8
15 16
s2.c
47 48
55 56
User’s Manual U18513EJ1V0UM
63
119
CHAPTER 3 C LANGUAGE EXPANSION
3.11.7
Size of structure array
The size of the structure object array is a value that is the sum of the number of elements added to the size of
structure object.
Example
struct S {
int
i;
char
c;
};
#pragma pack(1)
struct S1 {
int
i;
char
c;
};
#pragma pack(2)
struct S2 {
int
i;
char
c;
};
struct S
sobj[2];
struct S1
s1obj[2];
struct S2
s2obj[2];
/* Size of 16 bytes */
/* Size of 10 bytes */
/* Size of 12 bytes */
sobj
i
c
c
i
31 32 39 40
0
95 96 103 104
63 64
s1obj
i
0
c
i
c
31 32 39 40
71 72
79
s2obj
i
0
120
c
i
31 32 39 40 47 48
c
79 80 87 88
User’s Manual U18513EJ1V0UM
95
127
CHAPTER 3 C LANGUAGE EXPANSION
3.11.8
Area between objects
For example, sobj.c, sobj.i, and cobj may be allocated consecutively without a gap in the following source
program (the allocation order of sobj and cobj is not guaranteed).
Example
#pragma pack(1)
struct S {
char
c;
int
i;
} sobj;
char
cobj;
sobj,cobj
c
0
3.11.9
i
7 8
cobj
39 40
47
Notes concerning structure packing function
(1) Specification of -Xpack option and #pragma pack directive at the same time
If the -Xpack option is specified when structure packing is specified with the #pragma directive in the C language source, the specified option value is applied to all the structures until the first #pragma pack directive
appears. After this, the value of the #pragma directive is applied.
Even after the #pragma directive appears, however, the specified option value is applied to the area specified
as default
Example
/* When specify -Xpack = 2 */
struct S2 { ... };
/* Packing value is specified as 2 in option */
/* Option -Xpack = 2 is valid: packing value is 2 */
#pragma pack(1)
/* Packing is specified as 1 in #pragma directive */
struct S1 { ... };
/* Pragma pack (1) is valid: packing value is 1 */
#pragma pack( )
/* Packing value is specified as default in #pragma directive */
struct S2_2 { ... };/* Option -Xpack = 2 is valid: packing value is 2 */
User’s Manual U18513EJ1V0UM
121
CHAPTER 3 C LANGUAGE EXPANSION
(2) Restrictions
When using the V850 microcontrollers and a CPU that is set to disable misalign access for V850Ex products,
the following restrictions apply.
(a)
Access using the structure member address cannot be executed correctly.
As shown in the following example, the structure member address is acquired, and the access to that
address is then performed with the address masked in accordance with the data alignment of the device.
Therefore, some data may disappear or be rounded off.
Example
struct test {
char
c;
int
i;
} test;
/* offset 0 */
/* offset 1-4 */
int *ip ,i;
void func(void)
{
i = *ip;
}
void func2(void)
{
ip = &(test.i);
}
(b)
/* Accessed with address masked */
/* Acquire structure member address */
In bit field access, an area with no data to be read using the member’s type is also accessed.
If the width of the bit field is smaller than the member’s type as shown in the following example, access
occurs outside the object because reading is performed using the member’s type.
Generally, there is no problem with the function, but if I/O are mapped, an illegal access may occur.
Example
struct S {
int x : 21;
} sobj;
/* 3byte */
sobj.x = 1
122
User’s Manual U18513EJ1V0UM
CHAPTER 3 C LANGUAGE EXPANSION
3.12
Binary Constants
The CA850 can handle integer constants in binary form.
A binary constant is a string that consists of "0b" or "0B" followed by one or more "0" or "1".
Example
0b00010110111101010111111010010111
Note
Binary constants cannot be used when the -ansi option is specified.
User’s Manual U18513EJ1V0UM
123
CHAPTER 4 CALLING PROGRAM
CHAPTER 4 CALLING PROGRAM
This chapter explains how to handle arguments when a program is called by the CA850.
4.1
-
Calling Between C Functions
Normal function call
→ jarl instruction
-
Function call using a pointer indicating a function (and returning from function call)
→ jmp instruction (dispose instruction [V850E])
When a C function is called from another C function, a 4-word argument is stored in the argument registers (r6
to r9). An argument in excess of 4 words is stored in the stack frame of the calling function. Control is then
transferred (jumps) to the called function and the value in the argument registers stored when the function was
called is stored in the stack frame of the calling function.
The stack frame is generated when the prologue code of the function, i.e., the code that is executed before the
code of the main body of the function is called (processing (4) to (7) in Figure 4 - 3 and Figure 4 - 5 is the
prologue code), is executed and the stack pointer (sp) is shifted by the necessary size. The stack frame disappears when the epilogue code of the function, i.e., the code that is executed after the code of the main body of
the function is executed and until control returns to the calling function, is executed and the stack pointer (sp) is
returned.
124
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
4.1.1
Stack frame/function call
This section explains the stack frame format and how the stack frame is generated and disappears when a
function is called.
(1) Stack frame format
The CA850 allocates the argument register area to either the beginning of the stack or center of the stack in
the stack frame, according to the argument condition. The argument conditions are as follows.
(a)
When the argument register area is allocated to the beginning of the stack
The argument register area is allocated to the beginning of the stack when the area is accessed
successively, exceeding the area for the 4-word argument, in the following two cases.
-
If the number of arguments is variable
-
If the argument is the entity of a structure and its area extends over a 4-word area
(b)
When the argument register area is allocated to the center of the stack
The argument register area is allocated to the center of the stack under conditions other than (a).
Figure 4 - 1 shows the stack frame when the argument register area is at the beginning of the stack, and
Figure 4 - 2 shows the stack frame when the argument register area is at the center of the stack.
Figure 4 - 1 Stack Frame (When Argument Register Area Is Located at Center of Stack)
old sp
r20
Register area for r21
...
register variables r28
r29
lp
.S=.F
Argument register area
(4-word argument area)
.X
.R
Work register area
Automatic variable
area
new sp
Argument area for
argument of more
than 4 words
.A
.T
User’s Manual U18513EJ1V0UM
125
CHAPTER 4 CALLING PROGRAM
Figure 4 - 2 Stack Frame (When Argument Register Area Is Located at Beginning of Stack)
old sp
Argument register area
(4-word argument area)
.S
r20
Register area for r21
...
register variables r28
r29
lp
Work register area
.R=.X
Automatic variable
area
.A
Argument area for
argument of more
than 4 words
new sp
.F
.T
".S, .F, .X, .R, .A, and .T" in the figure are macros for functions output by the compiler internally. These
macros are used for a specific purpose, as shown in Table 4 - 1.
Table 4 - 1 Meanings of Macros for Functions
Macro Name
.S
Stack size
.F
Stack size - Size of argument register area (if at the beginning of the stack)
.X
Size of argument register area (if at the center of the stack) + .R
.R
Size of work register area + .A + .T
.A
Size of automatic variable area + .T
.T
Size of area for arguments of function to be called in excess of 4 words
.P
Always 0 (macro for code generation)Note
Note
126
Meaning
.P is not shown in Figure 4 - 1 and Figure 4 - 2 because it is always 0.
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
These macros are used to access the stack area. Table 4 - 2 shows specific access methods (access codes
to be output).
Table 4 - 2 Method of Accessing Stack Area
Stack Area
Access Method (Displacement [sp])
Register area for register variables (including lp)
-offset+.Fxx[sp]
Work register area
-offset+.Rxx[sp]
Automatic variable area
-offset+.Axx[sp]
Area for arguments in excess of 4 words
offset+.Pxx[sp]
Argument register area (if at the beginning of the stack)
offset+.Fxx[sp]
Argument register area (if at the center of the stack)
offset+.Rxx[sp]
"offset" in this table is a positive integer and means the offset in each area. "xx" after a macro is a positive
integer and indicates the frame number of the function.
(2) Generation/disappearance of stack frame when function is called (when argument register area is at
center of stack)
The following explains the generation and disappearance of the stack frame when a function is called if the
argument register area is at the center of the stack.
This case applies to most function calls. Figure 4 - 3 shows an example of the generation/disappearance of
the stack frame when the function "func2()" is called from the function "func1()" and then execution returns to
"func1()".
User’s Manual U18513EJ1V0UM
127
CHAPTER 4 CALLING PROGRAM
Figure 4 - 3 Generation/Disappearance of Stack Frame
(When Argument Register Area Is Located at Center of Stack)
Processing on func1() side when func2() is called
(1)
The arguments are stored in the argument
registers.
The arguments of func2 to be called are stored
in r6 to r9.
The arguments in excess of 4 words are stored
in the stack.
The excess arguments that cannot be stored in
r6 to r9 are stored in the stack. This processing
is performed if the number of arguments is five
or more.
Execution branches to func2() by the jarl
instruction.
Higher address
(2)
Area for automatic
variables
Area for arguments
in excess of 4 words
(3)
Area for saving contents of
registers for register variables
Processing on func2() side when called by func1
lp saving area
Stack frame
for func1
(4)
Argument register area
(4 words)
(5)
Work register area
(6)
Area for automatic
variables
(7)
Area for arguments
in excess of 4 words
Area for saving contents of
registers for register variables
lp saving area
Stack frame
for func2
(2)
sp of func1
(iii)
(6) (i)
[Note]Since the V850Ex can perform processing (4),
(5), and (6) with the prepare instruction, the
CA850 outputs the prepare instruction.
(5) (ii)
Argument register area
(4 words)
(7)
Processing on func2() side when execution
returns from func2() to func1()
Work register area
(i)
Area for automatic
variables
(ii)
Area for arguments
in excess of 4 words
Area for saving contents of
registers for register variables
lp saving area
sp is shifted.
The stack pointer moves to the stack to be
used by func2.
lp is saved.
The return address of func1() is stored.
Register variable registers are saved.
These registers are saved because the register
values used by func1 must be retained when
func2 also uses the register variable registers.
Arguments in argument register area are
stored.
The values of r6 to r9 are stored. The current
argument values are stored in the stack
because when another function is called from
func2, the arguments at that time are stored in
registers r6 to r9.
(iii)
sp of func2
(4)
(iv)
The contents of the registers for register
variables are restored.
The values of the register variable registers of
func1() is restored to registers.
lp is restored.
The return address of func1() is restored.
sp is returned.
The stack pointer moves back to the stack to be
used by func1().
Execution is returned by the jmp [lp] instruction.
[Note]Since the V850Ex can perform processing (i),
(ii), (iii), and (iv) with the dispose instruction,
the CA850 outputs the dispose instruction.
Lower address
128
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
The items saved to the stack frame and the stack frame to be used are summarized below.
(a)
(b)
-
Calling side - func1() The values of the excess arguments are called if the arguments of func2() to be called exceed 4 words
Called side - func2() Passing the arguments stored in the argument registers
(The calling side (func1()) stores the argument in the register.)
-
Saving the link pointer (lp) (= return address of func1()) of the calling side (func1())
-
Saving the contents of the register variable registers
-
The register variable registers are allocated as follows.
In 22-register mode: r25, r26, r27, r28, r29
In 26-register mode: r23, r24, r25, r26, r27, r28, r29
In 32-register mode: r20, r21, r22, r23, r24, r25, r26, r27, r28, r29
Of these registers, those that are used are saved.
-
Area for automatic variables
-
Allocating an area used for operation if a very complicated expression is used in a function
Although this area is not shown in Figure 4 - 3, it is allocated at the lower address of the area for automatic
variables if it is necessary.
If the function has a return value, that value is stored in r10.
The location of each area of the stack frame and the image of the stack growth direction of each area are
illustrated below (it is assumed that func2() to be called has five arguments).
User’s Manual U18513EJ1V0UM
129
CHAPTER 4 CALLING PROGRAM
Figure 4 - 4 Stack Growth Direction of Each Area of Stack Frame
Growth direction
of each area
Stores 5th argument
sp for func1
Area for saving contents of registers
for register variables
Area for saving link pointer (lp)
Stores 4th argument
Stores 3rd argument
Stores 2nd argument
Stores 1st argument
Area for automatic variables
Area for complicated operations
Area for arguments of function to be called
from func2 in excess of 4 words
sp for func2
An example of a source calling a C function from a C function and an assembly source when that source is
compiled is shown below.
Example
void func1(void)
{
int
a, b, c, d, e;
func2(a, b, c, d, e);
:
}
int func2(int a, int b, int c, int d, int e)
{
register int
i;
:
return i;
}
130
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
Assembler instructions generated when func2 is called in example above
[V850]
_func1:
jbr .L3
.L4:
ld.w
-8+.A3 [sp], r6
ld.w
-12+.A3 [sp], r7
ld.w
-16+.A3 [sp], r8 --- (1)
ld.w
-20+.A3[sp], r9
ld.w
-24+.A3[sp], r10
st.w
r10, [sp]
- - - (2)
jarl
_func2, lp - - - (3)
:
- - epilogue for main
- - processing from (ii) to (iv)
.L3:
- - prologue for main
- - processing from(4) and (5)
:
jbr .L4
[V850E]
_func1:
jbr .L3
.L4:
ld.w
-8+.A3[sp], r6
ld.w
-12+.A3[sp], r7
ld.w
-16+.A3[sp], r8 --- (1)
ld.w
-20+.A3[sp], r9
ld.w
-24+.A3[sp], r10
st.w
r10, [sp]
--- (2)
jarl
_func2, lp --- (3)
:
-- epilogue for main
-- processing from (ii) to (iv)
.L3:
-- prologue for main
-- processing from(4) and (5)
:
jbr .L4
_func2:
jbr .L5
.L6:
st.w
r6, .R2[sp]
st.w
r7, 4+.R2[sp]
st.w
r8, 8+.R2[sp] - - - (7)
st.w
r9, 12+.R2[sp]
st.w
r29, -4+.A2[sp]
:
jbr
.L2
.L2:
ld.w
-4+.A2[sp], r10
ld.w
-4+.F2[sp], r29 - -- (i)
ld.w
-8+.F2[sp], lp - -- (ii)
add
.F2, sp
- -- (iii)
jmp [lp]
- - - (iv)
_func2:
jbr .L5
.L6:
st.w
r6, .R2[sp]
st.w
r7, 4+.R2[sp]
st.w
r8, 8+.R2[sp] --- (7)
st.w
r9, 12+.R2[sp]
st.w
r29, -4+.A2[sp]
:
jbr
.L2
.L2:
ld.w
-4+.A2[sp], r10
dispose .X2, 0x3, [lp]
- -(i), (ii), (iii), (iv)
.L5:
add
st.w
st.w
jbr
.L5:
-.F2, sp
- - - (4)
lp, -8+.F2[sp] - - - (5)
r29, -4+.F2[sp] - -- (6)
.L6
prepare 0x3, .X2
- -(4), (5), (6)
jbr
.L6
(3) Generation/disappearance of stack frame when function is called (when argument register area is at
beginning of stack)
The following explains the generation and disappearance of the stack frame when a function is called if the
argument register area is at the beginning of the stack.
Figure 4 - 5 shows an example of the generation/disappearance of the stack frame when the function "func2()"
is called from the function "func1()" and then execution returns to "func1()".
User’s Manual U18513EJ1V0UM
131
CHAPTER 4 CALLING PROGRAM
Figure 4 - 5 Generation/Disappearance of Stack Frame
(When Argument Register Area Is Located at Beginning of Stack)
Processing on func1() side when func2() is called
Higher address
(1)
Area for automatic
variables
(2)
Area for arguments
in excess of 4 words
(3)
Argument register area
(4 words)
Processing on func2() side when called by func1
lp saving area
(4)
Stack frame
for func1
Area for saving contents of
registers for register variables
(5)
Work register area
(6)
Area for automatic
variables
Area for arguments
in excess of 4 words
(2)
sp of func1
(iii)
Argument register area
(4 words)
Area for saving contents of
registers for register variables
(7)
Area for automatic
variables
Argument register area
(4 words)
132
[Note]Since the V850Ex can perform processing (4),
(5), and (6) with the prepare instruction, the
CA850 outputs the prepare instruction.
(i)
Area for arguments
in excess of 4 words
Lower address
sp is shifted.
The stack pointer moves to the stack to be
used by func2.
lp is saved.
The return address of func1() is stored.
Register variable registers are saved.
These registers are saved because the
register values used by func1 must be
retained when func2 also uses the register
variable registers.
Arguments in argument register area are
stored.
The values of r6 to r9 are stored. The current
argument values are stored in the stack
because when another function is called from
func2, the arguments at that time are stored in
registers r6 to r9.
Processing on func2() side when execution
returns from func2() to func1()
Work register area
lp saving area
(7)
(6) (i)
(5) (ii)
lp saving area
Stack frame
for func2
The arguments are stored in the argument
registers.
The arguments of func2 to be called are
stored in r6 to r9.
The arguments in excess of 4 words are
stored in the stack.
The excess arguments that cannot be stored
in r6 to r9 are stored in the stack. This
processing is performed if the number of
arguments is five or more.
Execution branches to func2() by the jarl
instruction.
(ii)
sp of func2
(4)
(iii)
(iv)
The contents of the registers for register
variables are restored.
The values of the register variable registers of
func1() is restored to registers.
lp is restored.
The return address of func1() is restored.
sp is returned.
The stack pointer moves back to the stack to
be used by func1().
Execution is returned by the jmp [lp]
instruction.
[Note]Since the V850Ex can perform processing (i),
(ii), (iii), and (iv) with the dispose instruction,
the CA850 outputs the dispose instruction.
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
The items that are saved to the stack frame and the stack frame to be used are summarized below.
(a)
(b)
-
Calling side - func1() The values of the excess arguments are called if the arguments of func2() to be called exceed 4 words
Called side - func2() Passing arguments stored in argument registers
(The calling side (func1()) stores the arguments in the registers.)
-
Saving the link pointer (lp) (= return address of func1()) of the calling side (func1())
-
Saving contents of register variable registers
-
Area for automatic variables
-
Allocating an area used for operation if a very complicated expression is used in a function
Although this area is not shown in Figure 4 - 3, it is allocated at the lower address of the area for automatic
variables if it is necessary.
If the function has a return value, it is stored in r10.
The location of each area of the stack frame and the image of the stack growth direction of each area are
illustrated below (it is assumed that func2() to be called has five arguments).
User’s Manual U18513EJ1V0UM
133
CHAPTER 4 CALLING PROGRAM
Figure 4 - 6 Stack Growth Direction of Each Area of Stack Frame
Growth direction
of each area
Stores 5th argument
sp for func1
Stores 4th argument
Stores 3rd argument
Stores 2nd argument
Stores 1st argument
Area for saving contents of registers
for register variables
Area for saving link pointer (lp)
Area for automatic variables
Area for complicated operations
Area for arguments of function to be called
from func2 in excess of 4 words
sp for func2
An example of a source calling a C function from a C function and an assembly source when that source is
compiled is shown below.
Example
void func1(void)
{
int a, b, c, d, e;
func2(a, b, c, d, e);
:
}
int func2(int a, int b, int c, int d, int e)
{
register int
i;
:
return i;
}
134
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
Assembler instructions generated when func2 is called in example above
[V850]
_func1:
jbr .L3
.L4:
ld.w
-8+.A3[sp], r6
ld.w
-12+.A3[sp], r7
ld.w
-16+.A3[sp], r8 - -- (1)
ld.w
-20+.A3[sp], r9
ld.w
-24+.A3[sp], r10
st.w
r10, [sp]
- -- (2)
jarl
_func2, lp
- -- (3)
:
- - epilogue for main
- - processing from (ii) to (iv)
.L3:
- - prologue for main
- - processing (4) and (5)
:
jbr .L4
[V850E]
_func1:
jbr .L3
.L4:
ld.w
-8+.A3[sp], r6
ld.w
-12+.A3[sp], r7
ld.w
-16+.A3[sp], r8 --- (1)
ld.w
-20+.A3[sp], r9
ld.w
-24+.A3[sp], r10
st.w
r10, [sp]
--- (2)
jarl
_func2, lp
--- (3)
:
-- epilogue for main
-- processing from (ii) to (iv)
.L3:
-- prologue for main
-- processing (4) and (5)
:
jbr .L4
_func2:
jbr .L5
.L6:
st.w
st.w
st.w
st.w
:
st.w
jbr .L2
.L2:
ld.w
ld.w
ld.w
add
jmp
.L5:
sub
st.w
st.w
jbr
_func2:
jbr .L5
.L6:
st.w
r6, .F2[sp]
st.w
r7, 4+.F2[sp]
st.w
r8, 8+.F2[sp] --- (7)
st.w
r9, 12+.F2[sp]
:
st.w
r29, -4+.A2[sp]
jbr .L2
.L2:
ld.w
-4+.A2[sp], r10
dispose .X2, 0x3
- - (i), (ii), (iii)
add
.S2-.F2, sp
--- (iii)
jmp
[lp]
--- (iv)
.L5:
add
.F2 -.S2, sp
--- (4)
prepare 0x3, .X2
- -- (4), (5), (6)
jbr .L6
r6,
r7,
r8,
r9,
.F2[sp]
4+.F2[sp]
8+.F2[sp] - - - (7)
12+.F2[sp]
r29, -4+.A2[sp]
-4+.A2[sp], r10
-4+.F2[sp], r29 - -- (i)
-8+.F2[sp], lp - - - (ii)
.S2, sp
- -- (iii)
[lp]
- -- (iv)
-.S2, sp
- -- (4)
lp, -8+.F2[sp] - - - (5)
r29, -4+.F2[sp] - -- (6)
.L6
User’s Manual U18513EJ1V0UM
135
CHAPTER 4 CALLING PROGRAM
4.2
Calling Between C Function and Assembler Function
This section explains the points to be note when an assembler function is called by a C function or vice versa.
4.2.1
Calling assembler function from C function
Note the following points when calling an assembler function from a C function.
(1) Identifier
If external names, such as functions and external variables, are described in the C language source by the
CA850, they are prefixed with "_" (underscore) when they are output to the assembler.
Table 4 - 3 Identifier
C
Assembler
func1( )
_func1
Prefix "_" to the identifier when defining functions and external variables with the assembler. Remove "_" when
referencing them from a C function.
(2) Stack frame
The CA850 outputs codes on the assumption that the stack pointer (sp) always indicates the lowest address of
the stack frame. Therefore, the address area lower than the address indicated by sp can be freely used in the
assembler function after branching from a C language source to an assembler function. Conversely, if the
contents of the higher address area are changed, the area used by a C function may be lost and the subsequent
operation cannot be guaranteed. To avoid this, change sp at the beginning of the assembler function before
using the stack. At this time, however, make sure that the value of sp is retained before and after calling. When
using a register variable register in an assembler function, make sure that the register value is retained before
and after the assembler function is called. In other words, save the value of the register variable register before
calling the assembler function, and restore the value after calling.
The register variable registers that can be used differ depending on the register mode.
Table 4 - 4 Registers for Register Variables
Register Mode
136
Registers for Register Variables
22-register mode
r25, r26, r27, r28, r29
26-register mode
r23, r24, r25, r26, r27, r28, r29
32-register mode
r20, r21, r22, r23, r24, r25, r26, r27, r28, r29
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
(3) Argument passed to assembler function
The CA850 stores 4-word arguments in argument registers r6 to r9 and arguments in excess of 4 words in the
stack frame on the calling side (refer to "4.1.1 Stack frame/function call" for details). Reference each stored
value when using an argument value in an assembler function. An argument value in a C function is the value
itself that is specified as an argument. The operation of the C function is not affected even if this value is
changed in an assembler function.
(4) Return value returned from assembler function
The CA850 generates codes on the assumption that the return value of a function is stored in the r10 register.
Store the value returned from an assembler function in r10. If the function returns a structure, the return value,
i.e., the structure, is stored in the stack frame of the calling function.
(5) Return address passed to C function
The CA850 generates codes on the assumption that the return address of a function is stored in link pointer lp
(r31). When execution branches to an assembler function, the return address of the function is stored in lp.
Execute the jmp [lp] instruction to return to a C function.
User’s Manual U18513EJ1V0UM
137
CHAPTER 4 CALLING PROGRAM
4.2.2
Calling C function from assembler function
Note the following points when calling a C function from an assembler function.
(1) Stack frame
The CA850 generates codes on the assumption that the stack pointer (sp) always indicates the lowest address
of the stack frame. Therefore, set sp so that it indicates the higher address of an unused area of the stack area
when execution branches from an assembler function to a C function. This is because the stack frame is
allocated toward the lower addresses.
(2) Work register
The CA850 retains the values of the register variable registers before and after a C function is called but does
not retain the values of the work registers. Therefore, do not leave a value that must be retained assigned to a
work register.
The register variable registers and work registers that can be used differ depending on the register mode.
Table 4 - 5 Registers for Register Variables
Register Mode
Registers for Register Variables
22-register mode
r25, r26, r27, r28, r29
26-register mode
r23, r24, r25, r26, r27, r28, r29
32-register mode
r20, r21, r22, r23, r24, r25, r26, r27, r28, r29
Table 4 - 6 Work Registers
Register Mode
Work Registers
22-register mode
r10, r11, r12, r13, r14
26-register mode
r10, r11, r12, r13, r14, r15, r16
32-register mode
r10, r11, r12, r13, r14, r15, r16, r17, r18, r19
(3) Argument passed to C function
The CA850 stores 4-word arguments in argument registers r6 to r9 and arguments in excess of 4 words in the
stack frame of the calling function (refer to "4.1.1 Stack frame/function call" for details). Store the arguments in
excess of 4 words upward from the address indicated by sp.
(4) Return value returned from C function
The CA850 generates codes on the assumption that the return value of a function is stored in the r10 register.
Reference the r10 register when using the value returned from a C function. If the function returns a structure, a
value is stored in an area for the return value of the calling function, and a code that passes the address of that
area as an argument is output. An area for the return value must be allocated in advance on the calling side.
138
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
(5) Return address returned to assembler function
The CA850 generates codes on the assumption that the return address of a function is stored in link pointer lp
(r31). When execution branches to a C function, therefore, the return address of the function must be stored in
lp. Execution is generally branched to a C function using the jarl instruction.
User’s Manual U18513EJ1V0UM
139
CHAPTER 4 CALLING PROGRAM
4.3
Prologue/Epilogue Processing of Function
The CA850 can reduce the object size in part of the prologue/epilogue processing of a function by calling a
runtime library. Because the prologue/epilogue processing of a function is predetermined, it is prepared as
runtime library functions and these functions are called when a function is called or execution returns to a
function.
An example of the assembler code of the prologue/epilogue processing of a function is shown below. Numbers
in parentheses in this example correspond to those in Figure 4 - 3.
Example
int func(int a, int b, int c, int d, int e)
{
register int
i;
:
return i;
}
Assembler instruction in prologue/epilogue processing of function "func" in above example
140
[Code when runtime library function is not used]
_func :
[Code when runtime library function is used]
_func :
jbr .L5
.L6 :
st.w
st.w
st.w
st.w
:
st.w
jbr
.L2 :
ld.w
ld.w
ld.w
add
jmp
.L5:
add
st.w
st.w
jbr
jbr .L5
.L6 :
st.w
st.w
st.w
st.w
:
st.w
jbr
.L2 :
ld.w
add
jarl
- - (i),
.L5 :
jarl
- - (4),
add
jbr
r6,
r7,
r8,
r9,
.R2[sp]
4+.R2[sp]
8+.R2[sp] - - - (7)
12+.R2[sp]
r29, -4+.A2[sp]
.L2
-4+.A2[sp], r10
-4+.F2[sp], r29 - -- (i)
-8+.F2[sp], lp - -- (ii)
.F2, sp
- -- (iii)
[lp]
- -- (iv)
-.F2, sp
- -- (4)
lp, -8+.F2[sp] - -- (5)
r29, -4+.F2[sp] - -- (6)
.L6
User’s Manual U18513EJ1V0UM
r6,
r7,
r8,
r9,
.R2[sp]
4+.R2[sp]
8+.R2[sp] --- (7)
12+.R2[sp]
r29, -4+.A2[sp]
.L2
-4+ .A2[sp], r10
.R2, sp
--- (iii)
___pop2904, lp
(ii), (iii), (iv)
___push2904, r10
(5), (6)
-.R2, sp
--- (4)
.L6
CHAPTER 4 CALLING PROGRAM
4.3.1
Specifying use of runtime library function for prologue/epilogue of
function
Specify the compiler option "-Xpro_epi_runtime=on" to call the runtime library for prologue/epilogue. Specify
"-Xpro_epi_runtime=off" if the runtime library is not called. When an optimization option other than "-Ot
(execution speed priority optimization)" is specified, however, the runtime library is automatically called for the
prologue/epilogue of a function. That is, the compiler option "-Xpro_epi_runtime=on" is automatically specified.
If an option other than "-Ot" is specified and if a runtime library should not be called, specify the
"-Xpro_epi_runtime=off" option. The "-Xpro_epi_runtime" option can be specified in each source file, so a file for
which the runtime library is called and a file for which the runtime library is not called can be used together.
When a runtime library is called for the prologue/epilogue of a function by specifying the "Xpro_epi_runtime=on" option, a dedicated section ".pro_epi_runtime" is necessary. Consequently, the following
definition must be described by a link directive.
.pro_epi_runtime = $PROGBITS ?AX .pro_epi_runtime;
Table information of the prologue/epilogue runtime function is allocated to this section.
User’s Manual U18513EJ1V0UM
141
CHAPTER 4 CALLING PROGRAM
4.3.2
Calling runtime library for prologue/epilogue of function in V850Ex
When the V850Ex is used, the following instruction is used to call the prologue/epilogue runtime function of a
function.
CALLT instruction
The CALLT instruction is a 2-byte instruction. The code size can be reduced by using this instruction for calling
a function. The CALLT instruction requires a pointer that indicates that the table of the function subject to the
CALLT instruction is set to the CTBP (Callt Base Pointer) register. If processing of the setting is missing from the
program, the following error message is output during linking.
F4414: CallTBasePointer(CTBP) is not set. CTBP must be set when compileroption
"-Ot" (or "-Xpro_epi_runtime=off") is not specified.
Since setting a value to the CTBP register is performed by an assembler instruction, it should be performed in
the startup routine. Add the following instruction to the startup routine.
mov
ldsr
#_ _ _ PROLOG_TABLE, r12 -- three underscores "_" before "PROLOG"
r12, 20
At this time, _ __PROLOG_TABLE is the first symbol of the function table of the runtime function of the
prologue/epilogue of a function, and the function table itself is allocated to the ".pro_epi_runtime" section. The
r12 register is used in the above example, but it is not always necessary to use r12. If the CALLT instruction
provided in the CA850 is used for any purpose other than calling a runtime library for the prologue/epilogue of a
function, the CTBP register contents must be saved/restored. If the CALLT instruction is used by another object,
such as middleware or a user-created library, and if a code that saves/restores the CTBP register contents is
missing or cannot be inserted in that object, a runtime library for the prologue/epilogue of a function cannot be
called. In this case, suppress calling the runtime library by specifying the "-Xpro_epi_runtime=off" option.
Refer to the Relevant Device’s Architecture User’s Manual of each device for details of the CALLT instruction
and the CTBP register.
142
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
4.3.3
Notes on calling runtime library for prologue/epilogue of function
Note the following points when calling a runtime library for the prologue/epilogue of a function.
(1)
Calling a runtime library for the prologue/epilogue of a function degrades the execution speed because a
function is called. Specify the "-Xpro_epi_runtime=off" option to avoid this. Specifying this option in file units
is effective.
(2)
In the case of a program in which few functions are called, the code size may not be reduced even if a
runtime library is called for the prologue/epilogue. If no real effect can be expected, specify the
"-Xpro_epi_runtime=off" option.
(3)
A runtime library is not called for the prologue/epilogue of an interrupt function.
However, a function called from an interrupt function is subject to runtime library calling.
User’s Manual U18513EJ1V0UM
143
CHAPTER 4 CALLING PROGRAM
4.4
Far Jump Function
The CA850 outputs a code using the jarl instruction when a function is called.
jarl
_func1, lp
The architecture allows only a sign-extended value of up to 22 bits (22-bit displacement) to be specified as the
first operand of the jarl instruction. This means that, if the branch destination is not within a 1 MB range from the
branch point, branching cannot take place and the linker outputs the following error message.
F4161:symbol " function-name"(output section : section-name) is too far from
output section " section-name".(value : disp-value, file : main.o, input section : .text, offset: offset-value, type : R_V850_PC22 ).
This can be solved easily by allocating the branch destination within a 1 MB range from the branch point.
However, the branch destination may not be able to be located within this range depending on target system.
The "far jump" function solves this.
When the far jump function is used, a code that uses the jmp instruction is output when a function is called. As
a result, execution can branch to the entire 32-bit space of the V850. Function calling using the far jump function
is called "far jump calling".
4.4.1
Specifying far jump
When calling a function using the far jump function, prepare a file in which functions to be called by the far
jump function are enumerated (file listing functions to be called by the far jump function), and use the compiler
option "-Xfar_jump".
-Xfar_jump file listing functions to be called by far jump function
The "-Xfar_jump" option can also be used with "=" as follows.
-Xfar_jump=file listing functions to be called by far jump function
Refer to the next section for the format of the file listing the functions to be called by the far jump function.
144
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
4.4.2
File listing functions to be called by far jump function
This section explains the format of the file that enumerates the functions to be called by using the far jump
function. Describe one function to which the far jump function is applied in one line. Describe a C function name
with "_" (underscore) prefixed.
Sample of file listing functions to be called by far jump
_func_led
_func_beep
_func_motor
:
:
_func_switch
If the following description is made instead of "_function-name", all the functions are called using the far jump
function.
{all_function}
If {all_function} is specified, all the functions are called by the far jump function, even if "_function-name" is
specified.
The far jump function can also be applied to the following functions, as well as to user functions.
-
Standard library functions
-
Runtime library functions
-
Prologue/epilogue runtime function of function
-
System calls of real-time OS
Refer to "4.4.3 Examples of using far jump function" for examples of specification.
Note the following points when describing the file listing the functions to be called by the far jump function.
-
Only ASCII characters can be used.
-
Comments must not be inserted.
-
Describe only one function in one line.
-
A blank and tab may be inserted before and after a function name.
-
Up to 1,023 characters can be described in one line. A blank or tab is also counted as one character.
-
Describe a C function name with "_" (underscore) prefixed to the function name.
-
The far jump function cannot be used together with the re-link function of the flash memory/external ROM.
User’s Manual U18513EJ1V0UM
145
CHAPTER 4 CALLING PROGRAM
4.4.3
Examples of using far jump function
Examples of using the far jump function are shown below.
(1) User function (same applies to standard functions)
[C language source file]
extern void func3(void);
void func(void)
{
func3();
}
[file listing functions to be called by far jump]
_func3
[Normal calling code]
#@CALL_ARG
jarl
_func3, lp
146
[far jump calling code]
#@CALL_ARG
movea
#_func3, tp, r10
movea
.L18, tp, lp
jmp
[r10]
.L18:
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
(2) Runtime function (when calling a macro)
[file listing functions to be called by far jump]
___mul
[Normal calling code]
.macro mul arg1, arg2
add
st.w
st.w
mov
mov
jarl
ld.w
mov
ld.w
add
.endm
-8, sp
r6, [sp]
r7, 4[sp]
arg1, r6
arg2, r7
_ _ _ mul, lp
4[sp], r7
r6, arg2
[sp], r6
8, sp
[far jump calling code]
.macro mul arg1, arg2
.local macro_ret
add
-8, sp
st.w
r6, [sp]
st.w
r7, 4[sp]
mov
arg1, r6
mov
arg2, r7
movea
macro_ret, tp, r31
.option nowarning
movea
#___mul, tp, r1
jmp
[r1]
.option warning
macro_ret:
ld.w
mov
ld.w
add
.endm
4[sp], r7
r6, arg2
[sp], r6
8, sp
(3) Runtime function (direct calling)
[file listing functions to be called by far jump]
___mul
[Normal calling code]
mov
r12, r6
mov
r13, r7
#@CALL_ARG r6, r7
#@CALL_USE r6, r7
jarl
_ _ _ mul, lp
mov
r6, r13
[far jump calling code]
mov
r12, r6
mov
r13, r7
#@CALL_ARG r6, r7
#@CALL_USE r6, r7
movea
#___mul, tp, r14
movea
.L13, tp, lp
jmp
[r14]
.L13 :
mov
r6, r13
The compiler automatically selects whether a runtime macro is called or a runtime function is directly called by
judging the register efficiency in the process of optimization.
User’s Manual U18513EJ1V0UM
147
CHAPTER 4 CALLING PROGRAM
(4) System calls of real-time OS
[file listing functions to be called by far jump]
_ext_tsk
[Normal calling code]
#@B_EPILOGUE
#@BEGIN_NO_OPT
add
.S4, sp
jr
_ext_tsk
- - C NR
#@END_NO_OPT
#@E_EPILOGUE
[far jump calling code]
#@B_EPILOGUE
#@BEGIN_NO_OPT
add
.S4, sp
movea
#_ext_tsk, tp, r10
jmp
[r10]
#@END_NO_OPT
#@E_EPILOGUE
-- C NR
(5) Prologue/epilogue runtime function
[file listing functions to be called by far jump]
_ _ _pop2900
_ _ _push2900
[Normal calling code]
#@B_EPILOGUE
jarl
_ _ _ pop2900, lp - - 1
#@E_EPILOGUE
.L3 :
jarl
_ _ _ push2900, r10
#@E_PROLOGUE
148
[far jump calling code]
#@B_EPILOGUE
movea
#___pop2900, tp, r11
jmp
[r11] --1
#@E_EPILOGUE
.L3 :
movea
#___push2900, tp, r11
movea
.L5, tp, r10
jmp
[r11]
.L5 :
#@E_PROLOGUE
User’s Manual U18513EJ1V0UM
CHAPTER 4 CALLING PROGRAM
Table 4 - 7 shows the prologue/epilogue function names that can be specified by the far jump function. Before
specifying a prologue/epilogue runtime function, confirm the functions used in the assembly source output after
compilation.
Table 4 - 7 List of Prologue/Epilogue Runtime Functions
_ _ _pop2000,
_ _ _pop2100,
_ _ _pop2200,
_ _ _pop2300,
_ _ _pop2400,
_ _ _pop2500,
_ _ _pop2600,
_ _ _pop2700,
_ _ _pop2800,
_ _ _pop2900,
_ _ _poplp00,
___push2000,
___push2100,
_ _ _ pop2001, _ _ _ pop2002, ___ pop2003, _ _ _ pop2004, ___ pop2040,
_ _ _ pop2101, _ _ _ pop2102, ___ pop2103, _ _ _ pop2104, ___ pop2140,
_ _ _ pop2201, _ _ _ pop2202, ___ pop2203, _ _ _ pop2204, ___ pop2240,
_ _ _ pop2301, _ _ _ pop2302, ___ pop2303, _ _ _ pop2304, ___ pop2340,
_ _ _ pop2401, _ _ _ pop2402, ___ pop2403, _ _ _ pop2404, ___ pop2440,
_ _ _ pop2501, _ _ _ pop2502, ___ pop2503, _ _ _ pop2504, ___ pop2540,
_ _ _ pop2601, _ _ _ pop2602, ___ pop2603, _ _ _ pop2604, ___ pop2640,
_ _ _ pop2701, _ _ _ pop2702, ___ pop2703, _ _ _ pop2704, ___ pop2740,
_ _ _ pop2801, _ _ _ pop2802, ___ pop2803, _ _ _ pop2804, ___ pop2840,
_ _ _ pop2901, _ _ _ pop2902, ___ pop2903, _ _ _ pop2904, ___ pop2940,
_ _ _ poplp01, _ _ _ poplp02, ___ poplp03, _ _ _ poplp04, ___ poplp40,
___push2001, ___push2002, ___push2003, ___push2004, ___push2040,
___push2101, ___push2102, ___push2103, ___push2104, ___push2140,
___push2200,
___push2300,
___push2400,
___push2500,
___push2600,
___push2700,
___push2800,
___push2900,
___pushlp00,
___push2201,
___push2301,
___push2401,
___push2501,
___push2601,
___push2701,
___push2801,
___push2901,
___pushlp01,
___push2202,
___push2302,
___push2402,
___push2502,
___push2602,
___push2702,
___push2802,
___push2902,
___pushlp02,
___push2203,
___push2303,
___push2403,
___push2503,
___push2603,
___push2703,
___push2803,
___push2903,
___pushlp03,
___push2204,
___push2304,
___push2404,
___push2504,
___push2604,
___push2704,
___push2804,
___push2904,
___pushlp04,
___push2240,
___push2340,
___push2440,
___push2540,
___push2640,
___push2740,
___push2840,
___push2940,
___pushlp40
Refer to "4.3 Prologue/Epilogue Processing of Function" for details of the prologue/epilogue runtime library of
functions.
User’s Manual U18513EJ1V0UM
149
CHAPTER 5 STARTUP ROUTINE
CHAPTER 5 STARTUP ROUTINE
This chapter explains the startup routine.
5.1
Operation of Startup Routine
A startup routine is a routine that is executed after the V850 is reset and before the main function is executed.
Basically, it performs initialization as follows after the system is reset. The specific operations are shown below.
-
Setting RESET handler when a reset is input
-
Setting register mode of startup routine
-
Securing stack area and setting stack pointer
-
Securing argument area of main function
-
Setting text pointer (tp)
-
Setting global pointer (gp)
-
Setting element pointer (ep)
-
Setting mask value to mask registers r20 and r21
-
Initializing peripheral I/O registers that must be initialized before execution of main function
-
Initializing user target that must be initialized before execution of main function
-
Clearing sbss area to 0
-
Clearing bss area to 0
-
Clearing sebss area to 0
-
Clearing tibss.byte area to 0
-
Clearing tibss.word area to 0
-
Clearing sibss area to 0
-
Setting CTBP value for prologue/epilogue runtime library of functions [V850E]
-
Setting programmable peripheral I/O register values [V850E]
-
Setting r6 and r7 as argument of main function
-
Branching to main function (when real-time OS is not used)
-
Branching to initialization routine of real-time OS (when real-time OS is used)
Depending on the system, some of these operations may not be necessary and can be omitted. In addition to
the above, the user can describe necessary processing. The above operations must be described basically with
assembler instructions. How to describe each operation is explained below. Startup routine samples are provided
in the CA850.
150
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
The samples are stored in the location shown in the following table.
Table 5 - 1 Startup Routine Samples
Storage Location
Install Folder\lib850\r22
Install Folder\lib850\r26
Install Folder\lib850\r32
File Name
Meaning
crtN.s
Startup routine sample for V850 core
and for 22-register mode
crtE.s
Startup routine sample for V850Ex core
and for 22-register mode
crtN.s
Startup routine sample for V850 core
and for 26-register mode
crtE.s
Startup routine sample for V850Ex core
and for 26-register mode
crtN.s
Startup routine sample for V850 core
and for 32-register mode
crtE.s
Startup routine sample for V850Ex core
and for 32-register mode
If the startup routine is not added to the project, the CA850 automatically links a default startup routine
(object). The file to be linked is as follows.
-
Project for V850 core: crtN.o
-
Project for V850Ex core: crtE.o
These files result from compiling (assembling) sample startup routines "crtN.s" and "crtE.s". These objects are
assembled with the assembler options "-cn" and "-cnv850e" and can be used commonly in the V850
microcontrollers.
User’s Manual U18513EJ1V0UM
151
CHAPTER 5 STARTUP ROUTINE
5.1.1
Setting RESET handler when reset is input
Describe the processing to be performed when a reset (reset interrupt) is input. Execution branches to the
handler address 0x0 when a reset is input in the V850. Therefore, allocate an instruction that branches to the
beginning of the startup routine to address 0x0. As explained in "3.7.4 Notes on describing interrupt/exception
handler", describe a reset interrupt with assembler instructions because it cannot be described in C language by
specifying the #pragma interrupt directive. The description is as follows.
.section "RESET", text
jr _ _ start
_ _ start:
Use the .section quasi directive to allocate an instruction to the handler address. If the above description is
made, the "jr _ _start" instruction is allocated to the handler address. If the jr instruction cannot reach the
destination, i.e., if "_ _start" is not within +1Mbytes from address 0x0, use the jmp instruction as follows.
.section "RESET", text
mov #_ _ start, lp
jmp [lp]
_ _ start:
In this case, one register is used. The lp (r31) register is used in the above example. Any general-purpose
register whose contents can be lost at this point can be used. The lp (r31) register in which the return address
from a function is stored is not used when a reset is input. Therefore, it is safe to use the lp (r31) register.
The description of the .section quasi directive does not always have to be in the startup routine. In the above
example, the symbol of the startup routine is "__start". This may be another name.
152
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.2
Setting register mode of startup routine
Describe the setting of the register mode in the startup routine described with assembler instructions.
However, this setting is necessary only when the 22-register mode or 26-register mode is used for the overall
system. It is not necessary to describe this setting when the 32-register mode is used.
(1)
22-register mode
.option reg_mode 5 5
(2)
26-register mode
.option reg_mode 7 7
If this setting is not described, the linker outputs the following warning message.
W4608: input files have different register modes. use -rc option for more
information.
User’s Manual U18513EJ1V0UM
153
CHAPTER 5 STARTUP ROUTINE
5.1.3
Securing stack area and setting stack pointer (sp)
Secure the stack area used by the system and set the stack pointer (sp = r3) at the beginning of this area.
When a real-time OS is used, however, the stack specified here is used until execution branches to the
initialization routine of the real-time OS. In other words, it is hardly used or not used at all. If a large stack area is
secured, therefore, the RAM area is wasted. Check if the stack is used before execution branches to the
initialization routine of the real-time OS. Interrupts must be especially noted. It seems, however, that the startup
routine is mostly executed with interrupts disabled.
The stack area is secured as follows.
.set STACKSIZE, 0x200
.bss
.lcomm _ _ stack, STACKSIZE, 4
mov #_ _ stack+STACKSIZE, sp
This is an example of securing a 0x200-byte stack in the .bss area. The contents of the stack are allocated to
a bss attribute area because they do not have an initial value. Of course, they can be allocated to the sbss area,
but the size of the stack that can be allocated to the sbss area is limited because the sbss area is accessed with
a single gp-relative instruction. It is recommended to allocate the stack contents to the bss area if the stack size
is great, as it may be better to allocate other variables to the sbss area. Change the value written to the .set
instruction to change the stack size to be secured.
The CA850 generates codes on the assumption that the sp is at a 4-byte boundary when it references the
memory relatively with the stack pointer (sp). Therefore, be sure to allocate the stack pointer at a 4-byte
boundary. If necessary, use the quasi directive ".align 4".
The stack has a serious effect on the operation of the system. If the stack area runs short, the stack size
exceeds the secured area and the stack contents are lost, which may cause a system hang-up. Estimate the
stack size to be used by functions using stack850 included with the CA850, and secure a sufficient stack size.
154
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.4
Securing argument area for main function
The ANSI C specification defines the format of the main function as follows without dummy argument:
int main(void) { /* ... */ }
or, as the following function with two dummy arguments.
int main(int argc, char *argv[]) { /* ... */ }
argc of the function having two dummy arguments is a value that is not negative and indicates the total number
of dummy arguments. argv indicates an array of pointers to argument character strings. argv[argc] is null (vacant
pointer). If argc is 1 or more, argv[0] to argv[argc-1] are pointers to character strings.
Secure the areas for argc and argv in the startup routine, as shown below.
.data
.size _ _ argc, 4
.align 4
__argc:
.word 0
.size _ _ argv, 4
__argv:
.word #.L16
.L16:
.byte 0
.byte 0
.byte 0
.byte 0
The above area is not necessary if the main function is defined in this format.
int main(void) { /* ... */ }
The used RAM area can be reduced by deleting the above area.
Actually, processing that sets arguments (r6 and r7) of the main function is performed immediately before the
main function. If r6 and r7 are not used in the startup routine, the processing can be executed immediately after
the above program. Refer to "5.1.19 Setting r6 and r7 as argument of main function" for the processing to be set.
User’s Manual U18513EJ1V0UM
155
CHAPTER 5 STARTUP ROUTINE
5.1.5
Setting text pointer (tp)
The text pointer (tp) is a pointer prepared to implement referencing (PIC: Position Independent Code)
independent of the position at which the text area of an application, i.e., program code is allocated when the
program code is referenced. For example, if it is necessary to reference a specific location in the code during
program execution, the CA850 outputs the code to be accessed in tp-relative mode. Since the code is output on
the assumption that tp is correctly set, tp must be correctly set in the startup routine.
The text pointer value is determined during linking, and is in a symbol defined by a symbol directive that is
described in the link directive file. For example, suppose that the symbol directive of the text pointer is described
as follows.
_ _ tp_TEXT @ %TP_SYMBOL {TEXT};
The text pointer value is the beginning of the TEXT segment, and is in "__ tp_TEXT".
Describe as follows to set tp in the startup routine.
.extern _ _ tp_TEXT, 4
mov #_ _ tp_TEXT, tp
Refer to CA850 for Link Directive User’s Manual for details of symbol directives and link directives.
156
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.6
Setting global pointer (gp)
External variables or data defined in an application are allocated to the memory. The global pointer (gp) is a
pointer prepared to implement referencing independent of location position (PID: Position Independent Data)
when the variables or data allocated to the memory are referenced. The CA850 outputs a code for the section
that is to be accessed in gp-relative mode. Since the code is output on the assumption that gp is correctly set, gp
must be correctly set in the startup routine.
The global pointer value is determined during linking, and is in a symbol defined by a symbol directive that is
described in the link directive file. For example, suppose that the symbol directive of the global pointer is
described as follows.
_ _ gp_DATA @ %GP_SYMBOL {DATA};
The gp symbol value can be defined the beginning of "data segment" of the DATA segment as shown above,
or offset from a text symbol. A gp symbol can be specified not only by specifying the start address of a data
segment (such as the DATA segment), but also by using an offset value from the text symbol as its address.
Using the second method, the gp symbol value is determined by adding an offset value from tp to tp. In other
words, a code that is independent of location can be generated. To copy a program code and data used by that
code to the RAM area simultaneously and execute them, the value of gp can be acquired immediately if the start
address of the copy destination is known. In this case, the symbol directive is described as follows.
_ _ tp_TEXT @ %TP_SYMBOL {TEXT};
_ _ gp_DATA @ %GP_SYMBOL&_ _ tp_TEXT {DATA};
The global pointer value is "__tp_TEXT to which the value of __gp_DATA is added", and the value to be
added, i.e., offset value, is stored in "__gp_DATA". Therefore, describe as follows to set gp in the startup routine.
.extern _ _ tp_TEXT, 4
.extern _ _ gp_DATA, 4
mov #_ _ tp_TEXT, tp
mov #_ _ gp_DATA, gp
add tp, gp
This sets the correct value of the global pointer to gp.
Refer to CA850 for Link Directive User’s Manual for details of symbol directives and link directives.
User’s Manual U18513EJ1V0UM
157
CHAPTER 5 STARTUP ROUTINE
5.1.7
Setting element pointer (ep)
Of the external variables or data defined in an application, those that are allocated to the following sections are
accessed from the element pointer (ep) in relative mode.
-
sedata/sebss attribute section
-
sidata/sibss attribute section
-
tidata.byte/tibss.byte section
-
tidata.word/tibss.word section
If these sections exist, the CA850 outputs a code to access these areas in ep-relative mode. Since the code is
output on the assumption that ep is correctly set, ep must be correctly set in the startup routine. The element
pointer value is determined during linking, and is in a symbol defined by a symbol directive that is described in
the link directive file. For example, suppose that the symbol directive of the element pointer is described as
follows.
The element pointer value is the beginning of the SIDATA segment by default, and its value is in
"__ ep_DATA". Therefore, describe as follows to set ep in the startup routine.
_ _ [email protected]%EP_SYMBOL;
The element pointer value is the beginning of the SIDATA segment by default, and its value is in
"__ ep_DATA".
Therefore, describe as follows to set ep in the startup routine.
.extern _ _ ep_DATA, 4
mov #_ _ ep_DATA, ep
Reference the absolute address of _ _ep_DATA and set that value to ep.
Refer to CA850 for Link Directive User’s Manual for details of symbol directives and link directives.
158
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.8
Setting mask value to mask registers (r20 and r21)
When using mask registers, set them in the startup routine. The mask registers are "r20" and "r21". Set these
registers to the following values.
-
r20 to 8-bit mask value "0xff"
-
r21 to 16-bit mask value "0xffff"
Set these values as follows.
.option nowarning
mov 0xff, r20
mov 0xffff, r21
.option warning
The portion between ".option nowarning" and ".option warning" is the quasi directive that suppresses output of
warning messages during assembling. If the assembler option "-m" (use of mask option) is set, codes in which
mask values are set are output to r20 and r21. If the user intentionally attempts to substitute values in r20 and
r21, therefore, the following warning message is output.
W3013: mask register r20 or r21 used as destination register.
Refer to "2.5 Mask Register" for details of the mask registers.
User’s Manual U18513EJ1V0UM
159
CHAPTER 5 STARTUP ROUTINE
5.1.9
Initializing peripheral I/O registers that must be initialized before
execution of main function
When the external RAM is initialized by the startup routine, the external memory must first be set to the
peripheral I/O; otherwise the memory area cannot be accessed and initialized. In addition, initialize the
peripheral I/O registers that must be set for executing the startup routine.
Register setting can be described with assembler instructions, or execution may once branch from the startup
routine to a C function and register setting can be described in this function. If it is described in C, reading and
substitution in the peripheral I/O can be described in a visually simple way. For example, when creating the C
function "void reset(void)" and calling it from the startup routine, describe the following instruction in the startup
routine.
jarl
_reset, lp
Differences between assembler instruction description and C description are shown below using the following
examples. An instruction that substitutes "1" in P0 (port 0) is described as an assembly language source and as
a C language source is as follows.
Assembly language source
mov
st.b
1, r10
r10, P0
r10 is used in this example.
C language source
#pragma ioreg
P0 = 1;
The external memory setting differs depending on the device. Refer to the Relevant Device’s Hardware User’s
Manual of each device.
With a clock generation function, the "internal system clock" that is supplied to each unit built in the V850
needs to be generated. In this case, the clock needs to be multiplied by a PLL (Phase locked loop) synthesizer
before use. In other words, the clock must be correctly set to the frequency used; otherwise the clock operates
slower or faster than the assumed operation speed.
Regarding the default value of the PLL, usually, the multiplication value is small and the operation frequency is
low. These also apply to the startup routine. If the clearing of the memory area that is explained in "5.1.11
Clearing sbss area to 0" and later sections is executed while the operating frequency is low, it takes a lot of time
to complete the execution. Therefore, it is recommended that the PLL be set during the early stages of the
startup routine.
160
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
- - Setting
mov
st.b
st.b
nop
nop
nop
nop
nop
set1
5 MHz to the value multiplied by four (20 MHz) in V850ES/SG
0x80,r10
r10,PRCMD
r10,PCC
- - fcpu = fxx
0, PLLCTL
- - PLLON = 1
Aside from the above settings, set the following settings: the "system wait control register (VSWC)", the
"command register (PRCMD)", and, if necessary, the "watch dog timer (WDT)". For the correct settings, refer to
the Relevant Device’s Hardware User’s Manual.
User’s Manual U18513EJ1V0UM
161
CHAPTER 5 STARTUP ROUTINE
5.1.10
Initializing user target that must be initialized before execution of main
function
Describe the necessary initialization processing for the user target, if any, in the startup routine.
The processing can be described with assembler instructions, or execution may once branch from the startup
routine to a C function and the processing can be described in this function.
162
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.11
Clearing sbss area to 0
Initialize the sbss area, one of the bss attribute areas that do not have an initial value. Since the memory
contents are undefined after the V850 is reset, it is recommended to clear the sbss area to zero. This processing
is not necessary if the sbss attribute section has not been created or if it is not necessary to clear the sbss area
to zero.
Use symbols "__ssbss" and "__esbss" reserved for the CA850 to clear the sbss area. The meaning of each
symbol is as follows.
Table 5 - 2 Symbols of sbss Area
Symbol Name
Meaning
__ssbss
Symbol indicating start of sbss area
__esbss
Symbol indicating end of sbss area
The values (addresses) of these symbols are determined during linking. The program that clears the sbss area
using these symbols is as follows.
.extern
.extern
mov
mov
cmp
jnl
.L12:
st.w
add
cmp
jl
.L11:
_ _ ssbss, 4
_ _ esbss, 4
#_ _ ssbss, r13
#_ _ esbss, r12
r12, r13
.L11
r0, [r13]
4, r13
r12, r13
.L12
This program clears the sbss area to zero in 4-byte units.
User’s Manual U18513EJ1V0UM
163
CHAPTER 5 STARTUP ROUTINE
5.1.12
Clearing bss area to 0
Initialize the bss area, one of the bss attribute areas that do not have an initial value. Since the memory
contents are undefined after the V850 is reset, it is recommended to clear the bss area to zero. This processing
is not necessary if the bss attribute section has not been created or if it is not necessary to clear the bss area to
zero.
Use symbols "_ _sbss" and "__ebss" reserved for the CA850 to clear the bss area. The meaning of each
symbol is as follows.
Table 5 - 3 Symbols of bss Area
Symbol Name
Meaning
__sbss
Symbol indicating start of bss area
__ebss
Symbol indicating end of bss area
The values (addresses) of these symbols are determined during linking. The program that clears the bss area
using these symbols is as follows.
.extern
.extern
mov
mov
cmp
jnl
.L15:
st.w
add
cmp
jl
.L14:
_ _ sbss, 4
_ _ ebss, 4
#_ _ sbss, r13
#_ _ ebss, r12
r12, r13
.L14
r0, [r13]
4, r13
r12, r13
.L15
This program clears the bss area to zero in 4-byte units.
164
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.13
Clearing sebss area to 0
Initialize the "sebss area", one of the bss attribute areas that do not have an initial value.
Since the memory contents are undefined after the V850 is reset, it is recommended to clear the sebss area to
zero. This processing is not necessary if the sebss attribute section has not been created or if it is not necessary
to clear the sebss area to zero.
Use symbols "__ ssebss" and "__esebss" reserved for the CA850 to clear the sebss area. The meaning of
each symbol is as follows.
Table 5 - 4 Symbols of sebss Area
Symbol Name
Meaning
__ssebss
Symbol indicating start of sebss area
__esebss
Symbol indicating end of sebss area
The values (addresses) of these symbols are determined during linking. The program that clears the sebss
area using these symbols is as follows.
.extern
.extern
mov
mov
cmp
jnl
.L18:
st.w
add
cmp
jl
.L17:
_ _ ssebss, 4
_ _ esebss, 4
#_ _ ssebss, r13
#_ _ esebss, r12
r12, r13
.L17
r0, [r13]
4, r13
r12, r13
.L18
This program clears the sebss area to zero in 4-byte units.
User’s Manual U18513EJ1V0UM
165
CHAPTER 5 STARTUP ROUTINE
5.1.14
Clearing tibss.byte area to 0
Initialize the tibss.byte area, one of the bss attribute areas that do not have an initial value.
Since the memory contents are undefined after the V850 is reset, it is recommended to clear the tibss.byte
area to zero. This processing is not necessary if the tibss.byte section has not been created or if it is not
necessary to clear the tibss.byte area to zero.
Use symbols "_ _stibss.byte" and "__ etibss.byte" reserved for the CA850 to clear the tibss.byte area. The
meaning of each symbol is as follows.
Table 5 - 5 Symbols of tibss.byte Area
Symbol Name
Meaning
__stibss.byte
Symbol indicating start of tibss.byte area
__etibss.byte
Symbol indicating end of tibss.byte area
The values (addresses) of these symbols are determined during linking. The program that clears the tibss.byte
area using these symbols is as follows.
.extern
.extern
mov
mov
cmp
jnl
.L21:
st.w
add
cmp
jl
.L20:
_ _ stibss.byte, 4
_ _ etibss.byte, 4
#_ _ stibss.byte, r13
#_ _ etibss.byte, r12
r12, r13
.L20
r0, [r13]
4, r13
r12, r13
.L21
This program clears the tibss.byte area to zero in 4-byte units.
166
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.15
Clearing tibss.word area to 0
Initialize the tibss.word area, one of the bss attribute areas that do not have an initial value.
Since the memory contents are undefined after the V850 is reset, it is recommended to clear the tibss.word
area to zero. This processing is not necessary if the tibss.word section has not been created or if it is not
necessary to clear the tibss.word area to zero.
Use symbols "_ _stibss.word" and "__ etibss.word" reserved for the CA850 to clear the tibss.word area. The
meaning of each symbol is as follows.
Table 5 - 6 Symbols of tibss.word Area
Symbol Name
Meaning
__stibss.word
Symbol indicating start of tibss.word area
__etibss.word
Symbol indicating end of tibss.word area
The values (addresses) of these symbols are determined during linking. The program that clears the
tibss.word area using these symbols is as follows.
.extern
.extern
mov
mov
cmp
jnl
.L24:
st.w
add
cmp
jl
.L23:
_ _ stibss.word, 4
_ _ etibss.word, 4
#_ _ stibss.word, r13
#_ _ etibss.word, r12
r12, r13
.L23
r0, [r13]
4, r13
r12, r13
.L24
This program clears the tibss.word area to zero in 4-byte units.
User’s Manual U18513EJ1V0UM
167
CHAPTER 5 STARTUP ROUTINE
5.1.16
Clearing sibss area to 0
Initialize the sibss area, one of the bss attribute areas that do not have an initial value.
Since the memory contents are undefined after the V850 is reset, it is recommended to clear the sibss area to
zero. This processing is not necessary if the sibss attribute section has not been created or if it is not necessary
to clear the sibss area to zero.
Use symbols "__ ssibss" and "__esibss" reserved for the CA850 to clear the sibss area. The meaning of each
symbol is as follows.
Table 5 - 7 Symbols of sibss Area
Symbol Name
Meaning
__ssibss
Symbol indicating start of sibss area
__esibss
Symbol indicating end of sibss area
The values (addresses) of these symbols are determined during linking. The program that clears the sibss
area using these symbols is as follows.
.extern
.extern
mov
mov
cmp
jnl
.L25:
st.w
add
cmp
jl
.L26:
_ _ ssibss, 4
_ _ esibss, 4
#_ _ ssibss, r13
#_ _ esibss, r12
r12, r13
.L26
r0, [r13]
4, r13
r12, r13
.L25
This program clears the sibss area to zero in 4-byte units.
168
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.17
Setting CTBP value for prologue/epilogue runtime library of functions
This setting is necessary when the V850Ex core is used and when the prologue/epilogue runtime library is
used.
Since the CALLT instruction is used when the prologue/epilogue runtime library of functions is called by the
V850Ex core, the value of CTBP necessary for the CALLT instruction must be set at the beginning of the function
table of the prologue/epilogue runtime library of functions.
The prologue/epilogue runtime library is used in the following case.
-
Compiler option "-Xpro_epi_runtime=on" is set
If a compiler option other than "-Ot" is specified for optimization, "-Xpro_epi_runtime=on" is automatically
specified.
-
___ PROLOG_TABLE
Describe the following code using this symbol.
mov
ldsr
#_ _ _ PROLOG_TABLE, r12
r12, 20
CTBP is system register 20. Set a value to it using the ldsr instruction.
User’s Manual U18513EJ1V0UM
169
CHAPTER 5 STARTUP ROUTINE
5.1.18
Setting BPC value of programmable peripheral I/O register
The peripheral area select control register (BPC) must be set when using a V850 microcontrollers product in
which programmable peripheral I/O registers are provided and using a programmable peripheral I/O register.
For example, the peripheral area select control register of the V850E/IA1 is configured as follows.
BPC register
15
14
PA15
0
13
12
11
10
9
8
7
6
5
4
3
2
1
0
PA13
PA12
PA11
PA10
PA9
PA8
PA7
PA6
PA5
PA4
PA3
PA2
PA1
PA0
Table 5 - 8 BPC Register
Bit Position
Bit Name
Meaning
15
PA15
Enables or disables use of programmable peripheral I/O area.
0: Use of programmable peripheral I/O area disabled.
1: Use of programmable peripheral I/O area enabled.
13 - 0
PA13 - PA0
Set address of programmable peripheral I/O area.
When using a programmable peripheral I/O register, a value must be set to the programmable peripheral I/O
register using the compiler option "-Xbps". As a result, the CA850 outputs a code to access the programmable
peripheral I/O register. However, this option does not set a value to BPC. To set a value to BPC, processing to
write a value to the BPC register must be described in the startup routine.
In the case of the V850E/IA1, PA15 is set to 1, and a programmable peripheral I/O area address is set to PA13
to PA0. Set the BPC register, for example, to set the address of the programmable peripheral I/O area to 0x1234
as follows.
mov
st.h
0x9234, r13
r13, BPC
Because PA15 must be set to 1, set BPC to the logical sum (OR) of 0x1234 and 0x8000. The value set by the
compiler option "-Xbps" is 0x1234, and the value set to BPC is 0x9234. Therefore, care must be exercised that
no contradiction occurs.
Refer to the Relevant Device’s Architecture User’s Manual of each device for details of the programmable
peripheral I/O registers.
170
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.19
Setting r6 and r7 as argument of main function
If the main function is defined to have two dummy arguments as follows,
int main(int argc, char *argv[]) { /* ... */ }
processing that sets a value to the arguments (r6 and r7) must be performed before execution branches to the
main function. Refer to "5.1.4 Securing argument area for main function" for how to secure an area. This
processing is not necessary for an application using a real-time OS because the main function is not created.
Processing to set a value to r6 and r7 is as follows.
ld.w
movea
$_ _ argc, r6
$_ _ argv, gp, r7
The argument area of the main function is allocated to the .sdata section, so describe an access code in gprelative mode.
User’s Manual U18513EJ1V0UM
171
CHAPTER 5 STARTUP ROUTINE
5.1.20
Branching to main function
Then the processing necessary for the startup routine has been completed, execute an instruction that
branches to the main function. However, this processing is not necessary for an application using a real-time OS
because the main function is not created. Instead, an instruction that branches to the initialization routine of the
OS is necessary. Refer to "5.1.21 Branching to initialization routine of real-time OS" for details.
Describe the following code to branch to the main function.
jarl
_main, lp
When the main function has been executed, execution returns to the 4 bytes subsequent to this branch
instruction. The following instruction can also be used if it is known that execution does not return.
jr
_main
mov
jmp
#_main, lp
[lp]
The entire 32-bit space can be accessed using the jmp instruction. When the "jarl_main, lp" instruction is used,
execution returns after the main function is executed. It is recommended to take appropriate action to prevent
deadlock from occurring when execution returns.
172
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
5.1.21
Branching to initialization routine of real-time OS
In an application using a real-time OS, execution branches to the initialization routine when the processing that
must be performed by the startup routine has been completed. In an application not using a real-time OS,
execution branches to the main function. Refer to "5.1.20 Branching to main function".
Branching to the initialization routine is performed differently depending on whether NEC Electronics' real-time
OS RX850 or RX850 Pro is used.
In RX850
.extern _ _ urx_start
jr
_ _ urx_start
In RX850 Pro
.extern
mov
.extern
mov
jmp
_sit
#_sit, r10
_ _ rx_start
#_ _ rx_start, lp
[lp]
Note that the start symbol of the initialization routine of each OS differs. Refer to the User's Manual of each
real-time OS for details.
User’s Manual U18513EJ1V0UM
173
CHAPTER 5 STARTUP ROUTINE
5.2
Example of Startup Routine
This section shows an example of the startup routine.
Figure 5 - 1 Example of Startup Routine
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # external label declaration reserved for CA850 (1)
# for tp, gp, ep
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ tp_TEXT, 4
.extern _ _ gp_DATA, 4
.extern _ _ ep_DATA, 4
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # external label declaration reserved for CA850 (2)
# for initializing bss attribute section deleted if there is
# a section not used If the section to be used is not determined,
# write all sections and suppress the assemble error of the startup
# routine that occurs due to addition/deletion of sections.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ ssbss, 4
.extern _ _ esbss, 4
.extern _ _ sbss, 4
.extern _ _ ebss, 4
.extern _ _ ssebss, 4
.extern _ _ esebss, 4
.extern _ _ stibss.byte, 4
.extern _ _ etibss.byte, 4
.extern _ _ stibss.word, 4
.extern _ _ etibss.word, 4
.extern _ _ ssibss, 4
.extern _ _ esibss, 4
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # external label declaration of symbol reserved for CA850
# Declare start address of function table as external label
# when using prologue/epilogue runtime library in V850Ex.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ _ PROLOG_TABLE
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # external label declaration of main function
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _main
174
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
Figure 5 - 1 Example of Startup Routine
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # external label declaration of main function
# unnecessary if void main(void) type is used
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .data
.size _ _ argc, 4
.align 4
__argc:
.word 0
.size _ _ argv, 4
__argv:
.word #.L16
.L16:
.byte 0
.byte 0
.byte 0
.byte 0
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # The following is dummy data for section generation.
# This dummy data is used to clear the bss attribute section
# that appears later to zero.
#
# The start symbol and end symbol are generated if data exists
# in the corresponding section during linking.
# If the section to be used has not yet been determined, however,
# an assemble error of the startup routine occurs each time a section
# is added or deleted. To avoid this, generate the start and
# end symbols of a section by allocating dummy data to the section.
# The bss attribute section is not described because data is allocated
# by a stack generation code and dummy data does not have to be created
# in that section.
#
# If the section to be used is determined, delete this dummy data and
# the zero clear routine except the necessary part of the routine.
# This can eliminate waste and enhance the code efficiency.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .sbss
.lcomm _ _ sbss_dummy, 0, 0
.sebss
.lcomm _ _ sebss_dummy, 0, 0
.tibss.byte
.lcomm _ _ tibss_byte, 0, 0
.tibss.word
.lcomm _ _ tibss_word, 0, 0
.sibss
.lcomm _ _ sibss_dummy, 0, 0
User’s Manual U18513EJ1V0UM
175
CHAPTER 5 STARTUP ROUTINE
Figure 5 - 1 Example of Startup Routine
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # securing stack
# securing 0x200 bytes in bss area
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .set STACKSIZE, 0x200
.bss
.lcomm _ _ stack, STACKSIZE, 4
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # reset handler
# Describe instructions to be allocated to the reset handler.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .section "RESET", text
jr _ _ start
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # startup routine entity
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .text
.align 4
.globl _ _ start
.globl _ _ exit
.globl _ _ startend
_ _ start:
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # It is assumed that _ _ gp_DATA is set by a symbol directive
# that uses a relative value from tp.
# Therefore, gp adds the value of __gp_DATA to tp.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - mov
#_ _ tp_TEXT, tp
mov
#_ _ gp_DATA, gp
add
tp, gp
mov
#_ _ stack+STACKSIZE, sp
mov
#_ _ ep_DATA, ep
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # mask register setting
# Delete this description to reduce the code if a mask register is
# not used. There is no problem even if it is not deleted in operation
# because it is overwritten in the program.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .option nowarning
mov
0xff, r20
mov
0xffff, r21
.option warning
.L11:
176
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
Figure 5 - 1 Example of Startup Routine
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # clearing sbss attribute section to zero
# delete this description to reduce the code if the sbss attribute
# section is not used
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ ssbss, 4
.extern _ _ esbss, 4
mov
#_ _ ssbss, r13
mov
#_ _ esbss, r12
cmp
r12, r13
jnl
.L11
.L12:
st.w
r0, [r13]
add
4, r13
cmp
r12, r13
jl
.L12
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # clearing bss attribute section to zero
# delete this description to reduce the code if the sbss attribute
# section is not used
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ sbss, 4
.extern _ _ ebss, 4
mov
#_ _ sbss, r13
mov
#_ _ ebss, r12
cmp
r12, r13
jnl
.L14
.L15:
st.w
r0, [r13]
add
4, r13
cmp
r12, r13
jl
.L15
.L14:
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # clearing sebss attribute section to zero
# delete this description to reduce the code if the sbss attribute
# section is not used
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ ssebss, 4
.extern _ _ esebss, 4
mov
#_ _ ssebss, r13
mov
#_ _ esebss, r12
cmp
r12, r13
jnl
.L17
User’s Manual U18513EJ1V0UM
177
CHAPTER 5 STARTUP ROUTINE
Figure 5 - 1 Example of Startup Routine
.L18:
st.w
r0, [r13]
add
4, r13
cmp
r12, r13
jl
.L18
.L17:
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # clearing tibss.byte attribute section to zero
# delete this description to reduce the code if the sbss attribute
# section is not used
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ stibss.byte, 4
.extern _ _ etibss.byte, 4
mov
#_ _ stibss.byte, r13
mov
#_ _ etibss.byte, r12
cmp
r12, r13
jnl
.L20
.L21:
st.w
r0, [r13]
add
4, r13
cmp
r12, r13
jl
.L21
.L20:
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # clearing tibss.word attribute section to zero
# delete this description to reduce the code if the sbss attribute
# section is not used
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ stibss.word, 4
.extern _ _ etibss.word, 4
mov
#_ _ stibss.word, r13
mov
#_ _ etibss.word, r12
cmp
r12, r13
jnl
.L23
.L24:
st.w
r0, [r13]
add
4, r13
cmp
r12, r13
jl
.L24
.L23:
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # clearing sibss attribute section to zero
# delete this description to reduce the code if the sbss attribute
# section is not used
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - .extern _ _ ssibss, 4
.extern _ _ esibss, 4
178
User’s Manual U18513EJ1V0UM
CHAPTER 5 STARTUP ROUTINE
Figure 5 - 1 Example of Startup Routine
mov
mov
cmp
jnl
#_ _ ssibss, r13
#_ _ esibss, r12
r12, r13
.L26
.L25:
st.w
r0, [r13]
add
4, r13
cmp
r12, r13
jl
.L25
.L26:
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # setting of prologue/epilogue runtime library of functions
# The start address of the library function table is set to
# CTBP (system register #20). Delete this description when a core
# other than the V850Ex is used.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - mov
#_ _ _ PROLOG_TABLE, r12
ldsr
r12, 20
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # programmable peripheral I/O register setting
# Delete this description if a V850 not having programmable
# peripheral I/O registers is used.
# Shown below is an example where the BPC register value
# (set address) is 0x1234. The logical sum of 0x1234 (address) and
# 0x8000 (use of programmable peripheral I/O) is set to BPC.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - mov
0x9234, r13
st.h
r13, BPC
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # setting argument of main function to r6 and r7
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - ld.w
$_ _ argc, r6
movea
$_ _ argv, gp, r7
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # branching to main function
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - jarl
_main, lp
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - # processing when main function returns
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - - - -- - - - - - - -- - - - __exit:
halt
_ _ startend:
User’s Manual U18513EJ1V0UM
179
CHAPTER 6 LIBRARY FUNCTION
CHAPTER 6 LIBRARY FUNCTION
This chapter explains the library functions provided in the CA850.
6.1
Supplied Libraries
The CA850 provides the following libraries.
Table 6 - 1 Supplied Libraries
Library Type
Library Name
Function
Standard library
libc.a
Definition of Function with Variable Number of Arguments
Management of Character String and Memory
Character Type Macros and Functions
Standard Input/Output
Standard Utility Functions
Runtime Library
Prologue/epilogue runtime library of functions
Mathematical library
libm.a
Mathematical Functions
ROMization library
libr.a
ROMization copy functions
When the standard library or mathematical library is used in an application, include the related header files to
use the library function. Reference these libraries using the linker option (-l). However, it is not necessary to
reference the libraries if only "definition of a function with a variable number of arguments" and "character type
macro/character type function" are used.
When PM+ is used, these libraries are referenced by default. Since the mathematical library internally
references the standard library, the standard library is required when the mathematical library is used. The
runtime library is a part of the standard library.
It is a routine that is automatically called by the CA850 when a floating-point operation or integer operation
(such as 32-bit integer multiplication, division, or remainder calculation) is performed. Unlike the other library
functions, therefore, the runtime library and prologue/epilogue runtime library of functions is not described in the
C language source or assembly language source.
When the mask register function is used in the 32-register mode, use the standard library stored in the mask
register folder (Install Folder\lib850\r32msk).
The linker automatically references the standard library in the above folder in the following cases.
-
When 32-bit register mode is specified.
-
When the mask register function is used with the compiler option "-Xmask_reg".
The ROMization library is referenced by the linker when the compiler option "-Xr" is specified. This library
stores functions "_rcopy", "_rcopy1", "_rcopy2", and "_rcopy4", which are used to copy packed data.
180
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.1.1
Standard library
The functions contained in the standard library are listed below. These functions are described in the "libc.a"
file. The prologue/epilogue runtime library of functions is explained in 6.1.5. The meaning of each element in the
list is as follows.
Function name
Name of function
Outline
Functional outline of function
Header file
Header file that must be included in the C language source when this function is
used. Include this file using the #include directive. "errno.h" must also be included if
errno is used when an exception occurs.
ANSI
Indicates whether or not the function is stipulated by the ANSI standard. If it is
stipulated, "O" is shown in this column; if not, "- - -" is shown.
Use of sdata
Indicates whether or not this function uses the memory area "sdata". In other words,
whether or not data for which the function has an initial value is allocated to RAM is
indicated. Because the section name must be ".sdata", generate the .sdata section
even when this area is not used by the user application. If the .sdata section is used,
"O" is shown in this column; if not, "- --" is shown. If "O" is shown, data with an initial
value is necessary, so the initial value must be copied to RAM before program
execution. In other words, ROMization processing must be performed using the
_rcopy function. Refer to CA850 for Operation User’s Manual for details of this
processing.
Use of sbss
Indicates whether or not this function uses the memory area "sbss". In other words,
whether or not the function uses RAM as a temporary area is indicated. Because the
section name must be ".sbss", generate the .sbss section even when this area is not
used by the user application. If the .sbss section is used, "O" is shown in this
column; if not, "-- -" is shown. If "O" is shown, data without an initial value is
allocated, so unlike when .sdata is used, it is not necessary to perform ROMization
processing.
Re-entrancy
Indicates whether or not the function is re-entrant. If it is re-entrant, "O" is shown; if
not, "-- -" is shown. "Re-entrant" means that the function can "re-enter". A re-entrant
function can be correctly executed even if an attempt is made in another process to
execute that function while the function is being executed. In an application using a
real-time OS, for example, this function is correctly executed even if dispatching to
another task is triggered by an interrupt while a certain task is executing this
function, and even if the function is executed in that task. A function that must use
RAM as a temporary area may not necessarily be re-entrant.
(1) Definition of Function with Variable Number of Arguments
Table 6 - 2 Definition of Functions with Variable Number of Arguments
Function
Name
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
va_start
Initialization of variable for scanning
argument list
stdarg.h
---
---
---
---
va_arg
Moving variable for scanning
argument list
stdarg.h
---
---
---
---
va_end
End of scanning argument list
stdarg.h
---
---
---
---
User’s Manual U18513EJ1V0UM
181
CHAPTER 6 LIBRARY FUNCTION
(2) Management of Character String and Memory
Table 6 - 3 Character String Functions
Function
Name
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
bcmp
Memory comparison
(char argument version of memcmp)
string.h
---
---
- --
O
bcopy
Memory copy
(char argument version of memcpy)
string.h
---
---
- --
O
memchr
Memory search
string.h
O
---
- --
O
memcmp
Memory comparison
string.h
O
---
- --
O
memcpy
Memory copy
string.h
O
---
- --
O
memmove
Memory move
string.h
O
---
- --
O
memset
Memory set
string.h
O
---
- --
O
ANSI
Use of
sdata
Use of
sbss
Reentrancy
Table 6 - 4 Memory Management Functions
Function
Name
182
Header
File
Outline
index
Character string search (start position)
string.h
---
---
- --
O
rindex
Character string search (end position)
string.h
---
---
- --
O
strcat
Character string concatenation
string.h
O
---
- --
O
strchr
Character string search
(start position of specified character)
string.h
O
---
- --
O
strcmp
Character string comparison
string.h
O
---
- --
O
strcpy
Character string copy
string.h
O
---
- --
O
strcspn
Character string search (maximum
length not including specified
character)
string.h
O
---
- --
O
strerror
Character string conversion
of error number
string.h
O
O
- --
- --
strlen
Length of character string
string.h
O
---
- --
O
strncat
Character string concatenation
(with number of characters specified)
string.h
O
---
- --
O
strncmp
Character string comparison
(with number of characters specified)
string.h
O
---
- --
O
strncpy
Character string copy
(with number of characters specified)
string.h
O
---
- --
O
strpbrk
Character string search (start position)
string.h
O
---
- --
O
strrchr
Character string search (end position)
string.h
O
---
- --
O
strspn
Character string search (maximum
length including specified character)
string.h
O
---
- --
O
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 4 Memory Management Functions
Function
Name
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
strstr
Character string search
(start position of specified character)
string.h
O
---
- --
O
strtok
Token division
string.h
O
---
O
- --
ANSI
Use of
sdata
Use of
sbss
Reentrancy
(3) Character Type Macros and Functions
Table 6 - 5 Conversion of Character
Function
Name
Header
File
Outline
_tolower
Conversion from uppercase to
lowercase (correctly converted only if
argument is in uppercase)
ctype.h
- --
---
---
O
_toupper
Conversion from lowercase to
uppercase (correctly converted only if
argument is in lowercase)
ctype.h
- --
---
---
O
toascii
Conversion from integer to ASCII
character
ctype.h
- --
---
---
O
tolower
Conversion from uppercase to
lowercase (not converted if argument
is not in uppercase)
ctype.h
O
O
---
O
toupper
Conversion from lowercase to
uppercase (not converted if argument
is not in lowercase)
ctype.h
O
O
---
O
ANSI
Use of
sdata
Use of
sbss
Reentrancy
Table 6 - 6 Classification of Characters
Function
Name
Header
File
Outline
isalnum
Identification of ASCII letter or
numeral
ctype.h
O
O
- --
O
isalpha
Identification of ASCII letter
ctype.h
O
O
- --
O
isascii
Identification of ASCII code
ctype.h
- --
---
- --
O
iscntrl
Identification of control character
ctype.h
O
O
- --
O
isdigit
Identification of decimal number
ctype.h
O
O
- --
O
isgraph
Identification of display character
other than space
ctype.h
O
O
- --
O
islower
Identification of lowercase character
ctype.h
O
O
- --
O
isprint
Identification of display character
ctype.h
O
O
- --
O
ispunct
Identification of delimiter character
ctype.h
O
O
- --
O
isspace
Identification of space/tab/carriage
return/line feed/vertical tab/page feed
ctype.h
O
O
- --
O
User’s Manual U18513EJ1V0UM
183
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 6 Classification of Characters
Function
Name
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
isupper
Identification of uppercase character
ctype.h
O
O
- --
O
isxdigit
Identification of hexadecimal number
ctype.h
O
O
- --
O
ANSI
Use of
sdata
Use of
sbss
Reentrancy
(4) Standard Input/Output
Table 6 - 7 Standard I/O Functions
Function
Name
184
Header
File
Outline
perror
Error processing
stdio.h
O
O
-- -
Note
fread
Read from stream
stdio.h
O
O
-- -
---
fwrite
Write to stream
stdio.h
O
O
-- -
---
fgetc
Read one character from stream
(same as getc)
stdio.h
O
O
-- -
---
fgets
Read one line from stream
stdio.h
O
O
-- -
---
getc
Read one character from stream
(same as fgetc)
stdio.h
O
O
-- -
---
getchar
Read one character from standard
input
stdio.h
O
O
-- -
---
gets
Read character string from standard
input
stdio.h
O
O
-- -
---
ungetc
Push one character back to input
stream
stdio.h
O
O
-- -
---
rewind
Reset file position indicator
stdio.h
O
O
-- -
---
fputc
Write character to stream
stdio.h
O
O
-- -
---
fputs
Output character string to stream
stdio.h
O
O
-- -
---
putc
Write character to stream
stdio.h
O
O
-- -
---
putchar
Write character to standard output
stream
stdio.h
O
O
-- -
---
puts
Output character string to standard
output stream
stdio.h
O
O
-- -
---
sprintf
Output with format
stdio.h
O
O
O
---
fprintf
Output text in specified format to
stream
stdio.h
O
O
O
---
printf
Output text in specified format to
standard output stream
stdio.h
O
O
O
---
vfprintf
Write text in specified format to stream
stdio.h
O
O
O
---
vprintf
Write text in specified format to
standard output stream
stdio.h
O
O
O
---
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 7 Standard I/O Functions
Function
Name
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
vsprintf
Write text in specified format to
character string
stdio.h
O
O
O
---
sscanf
Input with format
stdio.h
O
O
O
---
fscanf
Read and interpret data from stream
stdio.h
O
O
O
---
scanf
Read and interpret text from standard
output stream
stdio.h
O
O
O
---
Note
stderr is not re-entrant.
(5) Standard Utility Functions
Table 6 - 8 Standard Utility Functions
Function
Name
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
abs
Output absolute value (int type)
stdlib.h
O
---
---
O
labs
Output absolute value (long type)
stdlib.h
O
---
---
O
bsearch
Binary search
stdlib.h
O
---
---
O
qsort
Align
stdlib.h
O
---
---
O
div
Division (int type)
stdlib.h
O
---
---
O
ldiv
Division (long type)
stdlib.h
O
---
---
O
ecvtf
Conversion of floating-point value to
numeric character string (with total
number of characters specified)
stdlib.h
---
O
O
- --
fcvtf
Conversion of floating-point value to
numeric character string (with number
of digits below decimal point specified)
stdlib.h
---
O
O
- --
gcvtf
Conversion of floating-point value to
numeric character string
(in specified format)
stdlib.h
---
O
O
- --
itoa
Conversion of integer (int type) to
character string
stdlib.h
---
---
---
O
ltoa
Conversion of integer (long type) to
character string
stdlib.h
---
---
---
O
ultoa
Conversion of integer (unsigned long
type) to character string
stdlib.h
---
---
---
O
calloc
Memory allocation (initialized to zero)
stdlib.h
O
O
O
Note 1
free
Memory release
stdlib.h
O
O
O
Note 1
malloc
Memory allocation
(not initialized to zero)
stdlib.h
O
O
O
Note 1
realloc
Memory re-allocation
stdlib.h
O
O
O
Note 1
User’s Manual U18513EJ1V0UM
185
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 8 Standard Utility Functions
Function
Name
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
rand
Pseudorandom number sequence
generation
stdlib.h
O
O
---
- --
srand
Setting of type of pseudorandom
number sequence
stdlib.h
O
O
---
- --
atoff
Conversion of character string to
floating-point number
stdlib.h
O
O
---
Note 2
strtodf
Conversion of character string to
floating-point number (storing pointer
in last character string)
stdlib.h
O
O
O
Note 2
atoi
Conversion of character string to
integer (int type)
stdlib.h
O
O
---
Note 2
atol
Conversion of character string to
integer (long type)
stdlib.h
O
O
---
Note 2
strtol
Conversion of character string to
integer (long type) and storing pointer
in last character string
stdlib.h
O
O
O
Note 2
strtoul
Conversion of character string to
integer (unsigned long type) and
storing pointer in last character string
stdlib.h
O
O
O
Note 2
Notes 1 A function that can be called recursively.
2 A function is not re-entrant if errno is updated when an exception occurs.
Remark errno.h must be included if errno is used when an exception occurs.
(6) Non-Local Jump Functions
Table 6 - 9 Non-Local Jump Functions
Function
Name
186
Outline
Header
File
ANSI
Use of
sdata
Use of
sbss
Reentrancy
setjmp
Set destination of non-local jump
setjmp.h
O
---
- --
O
longjmp
Non-local jump
setjmp.h
O
---
- --
O
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.1.2
Mathematical library
The functions contained in the mathematical library are listed below. These functions are described in the
"libm.a" file. The meaning of each element in the list is as follows.
Function name
Name of function
Outline
Functional outline of function
Header file
Header file that must be included in a C language source when this function is used.
Include this file using the #include directive. "errno.h" must also be included if errno
is used when an exception occurs, "limits.h" if limit values of general integer type
shown in "1.1.10 Quantitative limit" are used as a macro name, and "float.h" if limit
values of floating-point type are used.Header file that must be included in the C
language source when this function is used. Include this file using the #include
directive.
ANSI
Indicates whether or not the function is stipulated by the ANSI standard. If it is
stipulated, "O" is shown in this column; if not, "- - -" is shown.
Use of sdata
Indicates whether or not this function uses the memory area "sdata". In other words,
whether or not data for which the function has an initial value is allocated to RAM is
indicated. Because the section name must be ".sdata", generate the .sdata section
even when this area is not used by the user application. If the .sdata section is used,
"O" is shown in this column; if not, "- --" is shown. If "O" is shown, data with an initial
value is necessary, so the initial value must be copied to RAM before program
execution. In other words, ROMization processing must be performed using the
_rcopy function. Refer to CA850 for Operation User’s Manual for details of this
processing.
Use of sbss
Indicates whether or not this function uses the memory area "sbss". In other words,
whether or not the function uses RAM as a temporary area is indicated. Because the
section name must be ".sbss", generate the .sbss section even when this area is not
used by the user application. If the .sbss section is used, "O" is shown in this
column; if not, "-- -" is shown. If "O" is shown, data without an initial value is
allocated, so unlike when .sdata is used, it is not necessary to perform ROMization
processing.
Re-entrancy
Indicates whether or not the function is re-entrant. If it is re-entrant, "O" is shown; if
not, "-- -" is shown. "Re-entrant" means that the function can "re-enter". A re-entrant
function can be correctly executed even if an attempt is made in another process to
execute that function while the function is being executed. In an application using a
real-time OS, for example, this function is correctly executed even if dispatching to
another task is triggered by an interrupt while a certain task is executing this
function, and even if the function is executed in that task. A function that must use
RAM as a temporary area may not necessarily be re-entrant.
User’s Manual U18513EJ1V0UM
187
CHAPTER 6 LIBRARY FUNCTION
(1) Mathematical Functions
Table 6 - 10 Mathematical Functions
Function
Name
188
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
j0f
Bessel function of first kind (0 order)
math.h
- --
O
O
Note
j1f
Bessel function of first kind (first order)
math.h
- --
O
O
Note
jnf
Bessel function of first kind (n order)
math.h
- --
O
O
Note
y0f
Bessel function of second kind (0
order)
math.h
- --
O
O
Note
y1f
Bessel function of second kind (first
order)
math.h
- --
O
O
Note
ynf
Bessel function of second kind (n
order)
math.h
- --
O
O
Note
erff
Error function (approximate value)
math.h
- --
O
O
Note
erfcf
Error function (complementary
probability)
math.h
- --
O
O
Note
expf
Exponent function
math.h
O
O
O
Note
logf
Logarithmic function (natural
logarithm)
math.h
O
O
O
Note
log2f
Logarithmic function (base = 2)
math.h
O
O
O
Note
log10f
Logarithmic function (base = 10)
math.h
O
O
O
Note
powf
Power function
math.h
O
O
O
Note
cbrtf
Cubic root function
math.h
- --
---
- --
O
sqrtf
Square root function
math.h
O
O
O
Note
ceilf
ceiling function
math.h
O
---
- --
O
fabsf
Absolute value function
math.h
O
---
- --
O
floorf
floor function
math.h
O
---
- --
O
fmodf
Remainder function
math.h
O
O
O
Note
frexpf
Divide floating-point number into
mantissa and power
math.h
O
O
O
Note
ldexpf
Convert floating-point number to
power
math.h
O
O
O
Note
modff
Divide floating-point number into
integer and decimal
math.h
O
---
- --
O
gammaf
Logarithmic gamma function
math.h
- --
O
O
Note
hypotf
Euclidean distance function
math.h
- --
O
O
Note
matherr
Error processing function
math.h
- --
---
- --
O
acoshf
Inverse hyperbolic cosine
math.h
- --
O
O
Note
asinhf
Inverse hyperbolic sine
math.h
- --
O
O
Note
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 10 Mathematical Functions
Function
Name
Header
File
Outline
ANSI
Use of
sdata
Use of
sbss
Reentrancy
atanhf
Inverse hyperbolic tangent
math.h
- --
O
O
Note
coshf
Hyperbolic cosine
math.h
O
O
O
Note
sinhf
Hyperbolic sine
math.h
O
O
O
Note
tanhf
Hyperbolic tangent
math.h
O
O
O
Note
acosf
Inverse cosine
math.h
O
O
O
Note
asinf
Inverse sine
math.h
O
O
O
Note
atanf
Inverse tangent
math.h
O
O
O
Note
atan2f
Inverse tangent (y/x)
math.h
O
O
O
Note
cosf
Cosine
math.h
O
O
O
Note
sinf
Sine
math.h
O
O
O
Note
tanf
Tangent
math.h
O
O
O
Note
Note
These functions are not re-entrant only when errno is updated when an exception occurs and when
matherr is called.
Remark errno.h" must also be included if errno is used when an exception occurs, "limits.h" if limit values of
general integer type shown in "1.1.10 Quantitative limit" are used as a macro name, and "float.h" if
limit values of floating-point type are used.
User’s Manual U18513EJ1V0UM
189
CHAPTER 6 LIBRARY FUNCTION
6.1.3
Runtime library
The functions contained in the runtime library are listed below. These functions are described in the "libc.a"
file. The functions of the runtime library are automatically called by the CA850 when a floating-point operation or
integer operation (32-bit integer multiplication/division or remainder calculation) is executed in a C-source
program, so they are not described in the C language source or assembly-language source, like the prologue/
epilogue runtime library of functions. The meaning of each element in the list is as follows.
190
Function name
Name of function
Outline
Functional outline of function
Use of sdata
Indicates whether or not this function uses the memory area "sdata". In other words,
whether or not data for which the function has an initial value is allocated to RAM is
indicated. Because the section name must be ".sdata", generate the .sdata section
even when this area is not used by the user application. If the .sdata section is used,
"O" is shown in this column; if not, "- --" is shown. If "O" is shown, data with an initial
value is necessary, so the initial value must be copied to RAM before program
execution. In other words, ROMization processing must be performed using the
_rcopy function. Refer to CA850 for Operation User’s Manual for details of this
processing.
Use of sbss
Indicates whether or not this function uses the memory area "sbss". In other words,
whether or not the function uses RAM as a temporary area is indicated. Because the
section name must be ".sbss", generate the .sbss section even when this area is not
used by the user application. If the .sbss section is used, "O" is shown in this
column; if not, "-- -" is shown. If "O" is shown, data without an initial value is
allocated, so unlike when .sdata is used, it is not necessary to perform ROMization
processing.
Re-entrancy
Indicates whether or not the function is re-entrant. If it is re-entrant, "O" is shown; if
not, "-- -" is shown. "Re-entrant" means that the function can "re-enter". A re-entrant
function can be correctly executed even if an attempt is made in another process to
execute that function while the function is being executed. In an application using a
real-time OS, for example, this function is correctly executed even if dispatching to
another task is triggered by an interrupt while a certain task is executing this
function, and even if the function is executed in that task. A function that must use
RAM as a temporary area may not necessarily be re-entrant.
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
(1) Runtime Library
Table 6 - 11 Runtime Library
Function
Name
Outline
Use of
sdata
Use of
sbss
Reentrancy
___ mul
Multiplication of signed 32-bit integer
---
---
O
___ mulu
Multiplication of unsigned 32-bit integer
---
---
O
___ div
Division of signed 32-bit integer
---
---
O
___ divu
Division of unsigned 32-bit integer
---
---
O
___ mod
Remainder of signed 32-bit integer
---
---
O
___ modu
Remainder of unsigned 32-bit integer [V850]
---
---
O
___ addf.s
Addition of single-precision floating-point
O
---
Note
___ subf.s
Subtraction of single-precision floating-point
O
---
Note
___ mulf.s
Multiplication of single-precision floating-point
O
---
Note
___ divf.s
Division of single-precision floating-point
O
---
Note
___ cmpf.s
Comparison of single-precision floating-point and
change of flag
O
---
Note
___ cvt.ws
Conversion from integer to single-precision floatingpoint number
---
---
O
___ trnc.sw
Conversion from single-precision floating-point number
to integer
---
---
O
Note
These functions are not re-entrant only when errno is updated when an exception occurs and when
matherr is called.
Remark "errno.h" must also be included if errno is used when an exception occurs, "limits.h" if limit values of
general integer type shown in "1.1.10 Quantitative limit" are used as a macro name, and "float.h" if
limit values of floating-point type are used.
User’s Manual U18513EJ1V0UM
191
CHAPTER 6 LIBRARY FUNCTION
6.1.4
ROMization library
The functions contained in the ROMization library are listed below. These functions are described in the "libr.a"
file. These functions are routines that copy data and program codes with initial values to RAM. Refer to
Operation User's Manual for the ROMization procedure.
Function name
Name of function
Outline
Functional outline of function
-
A ROMization function itself does not use the sdata and sbss areas but writes data to the sdata area.
-
A ROMization function is usually called only once before the main program is executed, so it is not reentrant.
-
When a load module is downloaded to the in-circuit emulator (ICE), the data with initial values and placed
in the data or sdata area is set as soon as the load module has been downloaded. Therefore, debugging
can be performed without calling the _rcopy function. If a ROMization load module is created and executed
on the actual machine, however, the initial values are not set and the operation is not performed as
expected unless data with an initial value is copied using the _rcopy function. The reason for the trouble is
that an initial value is not set by this _rcopy function in most of the cases. If a routine that clears RAM to
zero is executed during initialization, call the _rcopy function before that routine; otherwise the initial values
will also be cleared to zero.
(1) ROMization copy functions
Table 6 - 12 ROMization Copy Functions
Function
Name
Outline
_rcopy
Copies packed data to RAM, 1 byte at a time (same as _rcopy1).
_rcopy1
Copies packed data to RAM, 1 byte at a time (same as _rcopy).
_rcopy2
Copies packed data to RAM, 2 bytes at a time.
_rcopy4
Copies packed data to RAM, 4 bytes at a time.
Caution _rcopy and _rcopy1 perform the same operation. These functions are provided to maintain
compatibility with the previous version.
When a program code is copied to the internal instruction RAM of a V850 device that has an internal
instruction RAM (such as the V850E/ME2), it must be copied in 4-byte units because of the hardware
specifications. In this case, the program code is copied using the "_rcopy4" function. Any function could be used
were it not for hardware restrictions. When a program code is copied in 2-byte or 4-byte units, the area that must
be copied may be exceeded. If the size of a packed data area is not a multiple of 4, therefore, an area other than
the packed data area is also copied at the same time. Take this into consideration.
192
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.1.5
Prologue/epilogue runtime library of functions
The functions contained in the prologue/epilogue runtime library are listed below. These functions are
described in the "libc.a" file. These functions are routines that are automatically called by the CA850 for
prologue/epilogue processing of functions, so they are not described in the C language source or assemblylanguage source, like the runtime library.
The V850Ex core uses the CALLT instruction to call the prologue/epilogue runtime library of functions. The
code efficiency can be enhanced by calling these functions from the table of the CALLT instruction.
Calling the prologue/epilogue runtime library of functions is valid when:
-
an optimization option other than "-Ot" (execution speed priority optimization) is specified.
-
the compiler option "-Xpro_epi_runtime=on" is specified.
The following functions are used for prologue and epilogue processing of functions.
Table 6 - 13 List of Prologue Runtime Library Functions
Functional Outline
Prologue processing of
functions
Function Name
_ _ _push2000,
_ _ _push2004,
_ _ _push2100,
_ _ _push2104,
_ _ _push2200,
_ _ _push2204,
_ _ _push2300,
_ _ _push2304,
_ _ _push2400,
_ _ _push2404,
_ _ _push2500,
_ _ _push2504,
_ _ _push2600,
_ _ _push2604,
_ _ _push2700,
_ _ _push2704,
_ _ _push2800,
_ _ _push2804,
_ _ _push2900,
_ _ _push2904,
_ _ _pushlp00,
_ _ _pushlp04,
___push2001,
___push2040,
___push2101,
___push2140,
___push2201,
___push2240,
___push2301,
___push2340,
___push2401,
___push2440,
___push2501,
___push2540,
___push2601,
___push2640
___push2701,
___push2740
___push2801,
___push2840,
___push2901,
___push2940,
___pushlp01,
___pushlp40
User’s Manual U18513EJ1V0UM
___ push2002, _ _ _ push2003,
___ push2102, _ _ _ push2103,
___ push2202, _ _ _ push2203,
___ push2302, _ _ _ push2303,
___ push2402, _ _ _ push2403,
___ push2502, _ _ _ push2503,
___ push2602, _ _ _ push2603,
___ push2702, _ _ _ push2703,
___ push2802, _ _ _ push2803,
___ push2902, _ _ _ push2903,
___ pushlp02, _ _ _ pushlp03,
193
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 14 List of Prologue Runtime Library Functions [V850E]
Functional Outline
Prologue processing of
functions
Function Name
_ _ _Epush250,
_ _ _Epush254,
_ _ _Epush260,
_ _ _Epush264,
_ _ _Epush270,
_ _ _Epush274,
_ _ _Epush280,
_ _ _Epush284,
_ _ _Epush290,
_ _ _Epush294,
_ _ _Epushlp0,
_ _ _Epushlp4
___Epush251, ___ Epush252, _ _ _ Epush253,
___Epush261, ___ Epush262, _ _ _ Epush263,
___Epush271, ___ Epush272, _ _ _ Epush273,
___Epush281, ___ Epush282, _ _ _ Epush283,
___Epush291, ___ Epush292, _ _ _ Epush293,
___Epushlp1, ___ Epushlp2, _ _ _ Epushlp3,
Table 6 - 15 List of Epilogue Runtime Library Functions
Functional Outline
Epilogue processing of
functions
194
Function Name
_ _ _pop2000,
_ _ _pop2004,
_ _ _pop2100,
_ _ _pop2104,
_ _ _pop2200,
_ _ _pop2204,
_ _ _pop2300,
_ _ _pop2304,
_ _ _pop2400,
_ _ _pop2404,
_ _ _pop2500,
_ _ _pop2504,
_ _ _pop2600,
_ _ _pop2604,
_ _ _pop2700,
_ _ _pop2704,
_ _ _pop2800,
___pop2001,
___pop2040,
___pop2101,
___pop2140,
___pop2201,
___pop2240,
___pop2301,
___pop2340,
___pop2401,
___pop2440,
___pop2501,
___pop2540,
___pop2601,
___pop2640
___pop2701,
___pop2740
___pop2801,
___pop2002, ___pop2003,
_ _ _pop2804,
_ _ _pop2900,
_ _ _pop2904,
_ _ _poplp00,
_ _ _poplp04,
___pop2840,
___pop2901, ___pop2902, ___pop2903,
___pop2940,
___poplp01, ___poplp02, ___poplp03,
___poplp40
___pop2102, ___pop2103,
___pop2202, ___pop2203,
___pop2302, ___pop2303,
___pop2402, ___pop2403,
___pop2502, ___pop2503,
___pop2602, ___pop2603,
___pop2702, ___pop2703,
___pop2802, ___pop2803,
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 16 List of Epilogue Runtime Library Functions [V850E]
Functional Outline
Epilogue processing of
functions
Function Name
_ _ _Epop250,
_ _ _Epop254,
_ _ _Epop260,
_ _ _Epop264,
_ _ _Epop270,
_ _ _Epop274,
_ _ _Epop280,
_ _ _Epop284,
_ _ _Epop290,
_ _ _Epop294,
_ _ _Epoplp0,
_ _ _Epoplp4
___Epop251, ___Epop252, ___Epop253,
___Epop261, ___Epop262, ___Epop263,
___Epop271, ___Epop272, ___Epop273,
___Epop281, ___Epop282, ___Epop283,
___Epop291, ___Epop292, ___Epop293,
___Epoplp1, ___Epoplp2, ___Epoplp3,
User’s Manual U18513EJ1V0UM
195
CHAPTER 6 LIBRARY FUNCTION
6.2
Header Files
The header files required for using the libraries of the CA850 are listed below. The macro definitions and
function declarations are described in each file.
Table 6 - 17 Header Files
File Name
196
Outline
ctype.h
Header file for character conversion and classification
errno.h
Header file for reporting error condition
float.h
Header file for floating-point representation and floating-point operation
limits.h
Header file for quantitative limiting of integers
math.h
Header file for mathematical calculation
setjmp.h
Header file for non-local jump
stdarg.h
Header file for supporting functions having variable number of arguments
stddef.h
Header file for common definitions
stdio.h
Header file for standard I/O
stdlib.h
Header file for standard utilities
string.h
Header file for memory manipulation and character string manipulation
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.3
Object Names Linked
When a load module is generated by compiling and linking C language sources using libraries, objects stored
in the libraries can be selected and linked as necessary. The names of the objects to be linked can be confirmed
in a link map file that indicates the result of linking. The names of the objects that are linked are almost the same
as the library function names. Routines commonly used for each function are combined and the object names
are different from the library function names. The following objects combine the routines commonly used, and
are automatically linked as necessary.
-
com1f.o
-
com1xf.o
-
com2f.o
-
com3f.o
-
com4f.o
-
com5f.o
-
com6f.o
-
com7f.o
User’s Manual U18513EJ1V0UM
197
CHAPTER 6 LIBRARY FUNCTION
6.4
Explanation of Format
About the library function that CA850 supports, the following format is used for the explanation.
Classification of library function
[Overview]
Outlines the feature of each function.
[Syntax]
Indicates the specification format of each function.
[Description]
Details of features of each function.
[Return value]
Indicates the return value of each function.
[Cautions]
Explains the supplementary points to be noted on each function.
[Example]
Indicates a simple example of each function.
Because the runtime library is written as an assembler, the following items may be added.
[Preprocessing]
Indicates the necessary preprocessing.
[Argument setting register]
Indicates the name of the register used for argument setting.
[Flag]
Indicates the flags that are affected.
198
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.5
Definition of Function with Variable Number of Arguments
This section explains the macros that define functions with a variable number of arguments in a portable form.
The declarations and definitions of these macros are described in the "stdarg.h" file.
Table 6 - 18 Definition of Function with Variable Number of Arguments
Classification
STDARG
Function Name
Outline
va_start
Initializes variable for scanning argument list
va_arg
Moves variable for scanning argument list
va_end
End of scanning argument list
User’s Manual U18513EJ1V0UM
199
CHAPTER 6 LIBRARY FUNCTION
STDARG
[Overview]
Defines a function with a variable number of arguments.
va_start, va_arg, va_end
[Syntax]
#include <stdarg.h>
void
va_start(va_list ap, last-named-argument)
type
va_arg(va_list ap, type)
void
va_end(va_list ap)
[Description]
To define function func having a variable number of arguments in a portable form, the following format is
used.
#include<stdarg.h>
void func(arg-declarations ...)
{
va_list ap;
type
argN;
va_start(ap, last-named-argument);
argN = va_arg(ap, type);
va_end(ap);
}
arg-declarationsis an argument list with the last-named-argument declared at the end. "..." that follows
indicates a list of the variable number of arguments.
va_listis the type of the variable (ap in the above example) used to scan the argument list.
va_start(va_list ap, last-named-argument)
This function initializes variable ap so that it indicates the beginning (argument next to last-namedargument) of the list of the variable number of arguments.
200
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
va_arg(va_list ap, type)
This function returns the argument indicated by variable ap, and advances variable ap to indicate the next
argument. For the type of va_arg, specify the type converted when the argument is passed to the function.
With the CA850, specify the int type for an argument of char and short types, and specify the unsigned int
type for an argument of unsigned char and unsigned short types.
Although a different type can be specified for each argument, stipulate "which type of argument is passed"
according to the conventions between the called function and calling function.
Also stipulate "how many functions are actually passed" according to the conventions between the called
function and calling function.
va_end(va_list ap)
This function indicates the end of scanning the list. By enclosing va_arg ... between va_start and va_end,
scanning the list can be repeated.
[Example]
#include <stdarg.h>
void abc(int first, int second, ...)
{
va_list ap;
int
i;
char
c, *fmt;
va_start(ap, second);
i = va_arg(ap, int);
c = va_arg(ap, int); /* char type is converted into int type. */
fmt = va_arg(ap, char *);
va_end(ap);
}
User’s Manual U18513EJ1V0UM
201
CHAPTER 6 LIBRARY FUNCTION
6.6
Management of Character String and Memory
This section explains the character string processing features and memory area management features.
The declarations and definitions related to these functions are described in the "string.h" file.
Table 6 - 19 Functions for Character String/Memory Management
Classification
STRING
MEMORY
202
Function Name
Outline
index
Character string search (first position)
rindex
Character string search (last position)
strcat
Character string concatenation
strchr
Character string search (first position of specifiedcharacter)
strcmp
Character string comparison
strcpy
Character string copy
strcspn
Character string search (maximum length not including
specified character)
strerror
Character string conversion of error number
strlen
Length of character string
strncat
Character string concatenation (number of characters
specification)
strncmp
Character string comparison (number of characters
specification)
strncpy
Character string copy (number of characters specification)
strpbrk
Character string search (first position)
strrchr
Character string search (last position)
strspn
Character string search (maximum length including specified
character)
strstr
Character string search (first position of specified character
string)
strtok
Token division
bcmp
Memory comparison (char argument)
bcopy
Memory copy (char argument)
memchr
Memory search
memcmp
Memory comparison (void argument)
memcpy
Memory copy (void argument)
memmove
Memory move
memset
Memory set
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
STRING
[Overview]
Manipulates a character string.
index, rindex, strcat, strchr, strcmp, strcpy, strcspn, strerror, strlen, strncat, strncmp, strncpy, strpbrk, strrchr,
strspn, strstr, strtok
[Syntax]
#include <string.h>
char
*index(const char *s, int c)
char
*rindex(const char *s, int c)
char
*strcat(char *dst, const char *src)
char
*strchr(const char *s, int c)
int
strcmp(const char *s1, const char *s2)
char
*strcpy(char *dst, const char *src)
size_t strcspn(const char *s1, const char *s2)
char
*strerror(int errnum)
size_t strlen(const char *s)
char
*strncat(char *dst, const char *src, size_t length)
int
strncmp(const char *s1, const char *s2, size_t length)
char
*strncpy(char *dst, const char *src, size_t length)
char
*strpbrk(const char *s1, const char *s2)
char
*strrchr(const char *s, int c)
size_t strspn(const char *s1, const char *s2)
char
*strstr(const char *s1, const char *s2)
char
*strtok(char *s, const char *delimiters)
User’s Manual U18513EJ1V0UM
203
CHAPTER 6 LIBRARY FUNCTION
[Description]
index(const char *s, int c)
This function is a function having the same feature as strchr.
rindex(const char *s, int c)
This function a function having the same feature as strrchr.
strcat(char *dst, const char *src)
This function concatenates the duplication of the character string indicated by src to the end of the character
string indicated by dst, including the null character (\0). The first character of src overwrites the null character
(\0) at the end of dst.
strchr(const char *s, int c)
This function obtains the position at which a character the same as c converted into char type appears in the
character string indicated by s. The null character (\0) indicating termination is regarded as part of this
character string.
strcmp(const char *s1, const char *s2)
This function compares the character string indicated by s1 with the character string indicated by s2.
strcpy(char *dst, const char *src)
This function copies the character string indicated by src to the array indicated by dst.
strcspn(const char *s1, const char *s2)
This function obtains the length of the maximum and first portion consisting of characters missing from the
character string indicated by s2 (except the null character (\0) at the end) in the character string indicated by
s1.
strerror(int errnum)
This function converts error number errnum into a character string according to the correspondence
relationship of the processing system definition. The value of errnum is usually the duplication of global
variable errno. Do not change the specified array of the application program.
strlen(const char *s)
This function obtains the length of the character string indicated by s.
strncat(char *dst, const char *src, size_t length)
This function concatenates up to length characters (including the null character (\0) of src) to the end of the
character string indicated by dst, starting from the beginning of the character string indicated by src. The null
character (\0) at the end of dst is written over the first character of src. The null character indicating termination
(\0) is always added to this result.
strncmp(const char *s1, const char *s2, size_t length)
This function compares up to length characters of the array indicated by s1 with characters of the array
indicated by s2.
204
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
strncpy(char *dst, const char *src, size_t length)
This function copies up to length characters (including the null character (\0)) from the array indicated by src
to the array indicated by dst. If the array indicate by src is shorter than length characters, null characters (\0)
are appended to the duplication in the array indicated by dst, until all length characters are written.
strpbrk(const char *s1, const char *s2)
This function obtains the position in the character string indicated by s1 at which any of the characters in the
character string indicated by s2 (except the null character (\0)) appears first.
strrchr(const char *s, int c)
This function obtains the position at which c converted into char type appears last in the character string
indicated by s. The null character (\0) indicating termination is regarded as part of this character string.
strspn(const char *s1, const char *s2)
This function obtains the maximum and first length of the portion consisting of only the characters (except
the null character (\0)) in the character string indicated by s2, in the character string indicated by s1.
strstr(const char *s1, const char *s2)
This function obtains the position of the portion (except the null character (\0)) that first coincides with the
character string indicated by s2, in the character string indicated by s1.
strtok(char *s, const char *delimiters)
This function divides the character string indicated by s into strings of tokens by delimiting the character
string with a character in the character string indicated by delimiters. If this function is called first, s is used as
the first argument. Then, calling with the null pointer as the first argument continues. The delimiting character
string indicated by delimiters can differ on each call.
On the first call, the character string indicated by s is searched for the first character not included in the
delimiting character string indicated by delimiters. If such a character is not found, a token does not exist in the
character string indicated by s, and strtok returns the null pointer. If a character is found, that character is the
beginning of the first token. After that, strtok searches from the position of that character for a character
included in the delimiting character string at that time. If such a character is not found, the token is expanded to
the end of the character string indicated by s, and the subsequent search returns the null pointer. If a character
is found, the subsequent character is overwritten by the null character (\0) indicating the termination of the
token. strtok saves the pointer indicating the subsequent character. If the null pointer is used as the value of
the first argument, a code that is not re-entrant is returned.
This can be avoided by preserving the address of the last delimiting character in the application program,
and passing s as an argument that is not vacant, by using this address.
User’s Manual U18513EJ1V0UM
205
CHAPTER 6 LIBRARY FUNCTION
[Return value]
strcat
Returns the value of dst.
strchr
Returns a pointer indicating the character that has been found. If c does not appear in
this character string, the null pointer is returned.
strcmp
Returns an integer greater than, equal to, or less than 0, depending on whether the
character string indicated by s1 is greater than, equal to, or less than the character
string indicated by s2.
strcpy
Returns the value of dst.
strcspn
Returns the length of the portion that has been found.
strerror
Returns a pointer to the converted character string.
strlen
Returns the number of characters existing before the null character (\0) indicating
termination.
strncat
Returns the value of dst.
strncmp
Returns an integer greater than, equal to, or less than 0, depending on whether the
character string indicated by s1 is greater than, equal to, or less than the character
string indicated by s2.
strncpy
Returns the value of dst.
strpbrk
Returns the pointer indicating this character. If any of the characters from s2 does not
appear in s1, the null pointer is returned.
strrchr
Returns a pointer indicating c that has been found. If c does not appear in this character
string, the null pointer is returned.
strspn
Returns the length of the portion that has been found.
strstr
Returns the pointer indicating the character string that has been found. If character
string s2 is not found, the null pointer is returned. If s2 indicates a character string with a
length of 0, s1 is returned.
strtok
Returns a pointer to a token. If a token does not exist, the null pointer is returned.
[Cautions]
Because the null character (\0) is always appended when strncat is used, if copying is limited by the number
of length arguments, the number of characters appended to dst is n + 1.
[Example]
#include <string.h>
void func(char *str, const char *src)
{
strcpy(str, src);
/* Copies character string indicated by src to */
/*
array indicated by str. */
:
}
206
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
MEMORY
[Overview]
Manipulates the memory contents.
bcmp, bcopy, memchr, memcmp, memcpy, memmove, memset
[Syntax]
#include <string.h>
int
bcmp(const char *s1, const char *s2, size_t n)
void
bcopy(const char *in, char *out, size_t n)
void
*memchr(const void *s, int c, size_t length)
int
memcmp(const void *s1, const void *s2, size_t n)
void
*memcpy(void *out, const void *in, size_t n)
void
*memmove(void *dst, void *src, size_t length)
void
*memset(const void *s, int c, size_t length)
[Description]
bcmp(const char *s1, const char *s2, size_t n)
This function is a function having the same feature as memcmp.
bcopy(const char *in, char *out, size_t n)
This function is a function having the same feature as memcpy.
memchr(const void *s, int c, size_t length)
This function obtains the position at which character c (converted into char type) appears first in the first
length number of characters in an area indicated by s.
memcmp(const void *s1, const void *s2, size_t n)
This function compares the first n characters of an object indicated by s1 with the object indicated by s2.
memcpy(void *out, const void *in, size_t n)
This function copies n bytes from an object indicated by in to an object indicated by out.
memmove(void *dst, void *src, size_t length)
This function moves the length number of characters from a memory area indicated by src to a memory area
indicated by dst. Even if the copy source and copy destination areas overlap, the characters are correctly
copied to the memory area indicated by dst.
memset(const void *s, int c, size_t length)
This function copies the value of c (converted into unsigned char type) to the first length character of an
object indicated by s.
User’s Manual U18513EJ1V0UM
207
CHAPTER 6 LIBRARY FUNCTION
[Return value]
memchr
If c is found, a pointer indicating this character is returned. If c is not found, the null
pointer is returned.
bcmp,
memcmp
An integer greater than, equal to, or less than 0 is returned, depending on whether the
object indicated by s1 is greater than, equal to, or less than the object indicated by s2.
bcopy,
memcpy
Returns the value of out. The operation is undefined if the copy source and copy
destination areas overlap.
memmove
Returns the value of dst at the copy destination.
memset
Returns the value of s.
[Example]
#include <string.h>
int func(const void *s1, const void *s2)
{
int i;
i = memcmp(s1, s2, 5);
/* Compares the first five characters of */
/* the character string indicated by s1 with */
/* the first five characters of the character*/
/* string indicated by s2 */
return i;
}
208
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.7
Character Type Macros and Functions
This section explains the macros that classify characters into several categories (such as letters, numerals,
control characters, and blanks) and macros that perform simple mapping of characters.
These macros can also be used as subroutines.
These macros are defined by "ctype.h".
Table 6 - 20 Character Type Macros
Classification
CONV
CTYPE
Function Name
Outline
_tolower
Conversion from uppercase characters to lowercase characters
_toupper
Conversion from lowercase characters to uppercase characters
toascii
Conversion from integer to ASCII character
tolower
Conversion from uppercase characters to lowercase characters
toupper
Conversion from lowercase characters to uppercase characters
isalnum
Whether ASCII letter or numeral
isalpha
Whether ASCII letter
isascii
Whether ASCII code
iscntrl
Whether control character
isdigit
Whether decimal number
isgraph
Whether display character (other than space)
islower
Whether lowercase character
isprint
Whether display character
ispunct
Whether delimiter
isspace
Whether space, tab, line feed, carriage return, vertical tab, or form
feed
isupper
Whether uppercase character
isxdigit
Whether hexadecimal character
User’s Manual U18513EJ1V0UM
209
CHAPTER 6 LIBRARY FUNCTION
CONV
[Overview]
Converts characters.
_tolower, _toupper, toascii, tolower, toupper
[Syntax]
#include <ctype.h>
int
_tolower(int c)
int
_toupper(int c)
int
toascii(int c)
int
tolower(int c)
int
toupper(int c)
[Description]
_tolower(int c)
This function is a macro that performs the same operation as tolower if the argument is of uppercase
characters. Because the argument is not checked, the correct conversion is performed only if the argument is
of uppercase characters. If otherwise, the operation will be undefined. A compiled subroutine can be used
instead of the macro definition, which is invalidated by using "#undef _tolower".
_toupper(int c)
This function is a macro that performs the same operation as toupper if the argument is of lowercase
characters. Because the argument is not checked, the correct conversion is performed only if the argument is
of lowercase characters. If otherwise, the operation will be undefined. A compiled subroutine can be used
instead of the macro definition, which is invalidated by using "#undef _toupper".
toascii(int c)
This function is a macro that forcibly converts an integer into an ASCII character (0 to 127) by clearing bit 8
and higher of the argument to 0. A compiled subroutine can be used instead of the macro definition, which is
invalidated by using "#undef toascii".
tolower(int c)
This function is a macro that converts uppercase characters into the corresponding lowercase characters
and leaves the other characters unchanged. This macro is defined only when c is an integer in the range of
EOF to 255. A compiled subroutine can be used instead of the macro definition, which is invalidated by using
"#undef tolower".
210
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
toupper(int c)
This function is a macro that converts lowercase characters into the corresponding uppercase characters
and leaves the other characters unchanged. This macro is defined only when c is an integer in the range of
EOF.
[Return value]
toascii
Returns an integer in the range of 0 to 127.
_tolower,
tolower
If isupper is true with respect to c, returns a character that makes islower true in
response; otherwise, returns c.
Also with _tolower, operation can be inconsistent when specifying illegal values for c.
_toupper,
toupper
If islower is true with respect to c, returns a character that makes islower true in
response; otherwise, returns c.
Also with _toupper, operation can be inconsistent when specifying illegal values for c.
[Example]
#include <ctype.h>
int
int
chc = 'a';
ret = func(chc);
int func(int c)
{
int i;
i = toupper(c);
/* Converts lowercase character ’a’ of c into */
/* uppercase character ’A’. */
return i;
}
User’s Manual U18513EJ1V0UM
211
CHAPTER 6 LIBRARY FUNCTION
CTYPE
[Overview]
Classifies characters
isalnum, isalpha, isascii, iscntrl, isdigit, isgraph, islower, isprint, ispunct, isspace, isupper, isxdigit
[Syntax]
#include <ctype.h>
int
isalnum(int c)
int
isalpha(int c)
int
isascii(int c)
int
iscntrl(int c)
int
isdigit(int c)
int
isgraph(int c)
int
islower(int c)
int
isprint(int c)
int
ispunct(int c)
int
isspace(int c)
int
isupper(int c)
int
isxdigit(int c)
[Description]
isalnum(int c)
This function is a macro that checks whether a given character is an ASCII alphabetic character or numeral.
This macro is defined for all integer values. A compiled subroutine can be used instead of the macro definition,
which is invalidated by using "#undef isalnum".
isalpha(int c)
This function is a macro that checks whether a given character is an ASCII alphabetic character. This macro
id defined only when c is made true by isascii or when c is EOF. A compiled subroutine can be used instead of
the macro definition, which is invalidated by using "#undef isalpha".
isascii(int c)
This function is a macro that checks whether a given character is an ASCII code (0x00 to 0x7f). This macro
is defined for all integer values. A compiled subroutine can be used instead of the macro definition, which is
invalidated by using "#undef isascii".
iscntrl(int c)
This function is a macro that checks whether a given character is a control character (0x00 to 0x1F or 0x7F).
This macro is defined only when c is made true by isascii or when c is EOF. A compiled subroutine can be
used instead of the macro definition, which is invalidated by using "#undef iscntrl".
212
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
isdigit(int c)
This function is a macro that checks whether a given character is a decimal number. This macro is defined
only when c is made true by isascii or when c is EOF. A compiled subroutine can be used instead of the macro
definition, which is invalidated by using "#undef isdigit".
isgraph(int c)
This function is a macro that checks whether a given character is a display characterNote (0x20 to 0x7E)
other than space (0x20). This macro is defined only when c is made true by isascii or when c is EOF. A
compiled subroutine can be used instead of the macro definition, which is invalidated by using "#undef
isgraph".
Note
printing character
islower(int c)
This function is a macro that checks whether a given character is a lowercase character (a to z). This macro
is defined only when c is made true by isascii or when c is EOF. A compiled subroutine can be used instead of
the macro definition, which is invalidated by using "#undef islower".
isprint(int c)
This function is a macro that checks whether a given character is a display character (0x20 to 0x7F). This
macro is defined only when c is made true by isascii or when c is EOF. A compiled subroutine can be used
instead of the macro definition, which is invalidated by using "#undef isprint".
ispunct(int c)
This function is a macro that checks whether a given character is a printable delimiter (isgraph(c) &&
!isalnum(c)). This macro is defined only when c is made true by isascii or when c is EOF. A compiled
subroutine can be used instead of the macro definition, which is invalidated by using "#undef ispunct".
isspace(int c)
This function is a macro that checks whether a given character is a space, tap, line feed, carriage return,
vertical tab, or form feed (0x09 to 0x0D, or 0x20). This macro is defined only when c is made true by isascii or
when c is EOF. A compiled subroutine can be used instead of the macro definition, which is invalidated by
using "#undef isspace".
isupper(int c)
This function is a macro that checks whether a given character is an uppercase character (A to Z). This
macro is defined only when c is made true by isascii or when c is EOF. A compiled subroutine can be used
instead of the macro definition, which is invalidated by using "#undef isupper".
isxdigit(int c)
This function is a macro that checks whether a given character is a hexadecimal number (0 to 9, a to f, or A
to F). This macro is defined only when c is made true by isascii or when c is EOF. A compiled subroutine can
be used instead of the macro definition, which is invalidated by using "#undef isxdigit".
User’s Manual U18513EJ1V0UM
213
CHAPTER 6 LIBRARY FUNCTION
[Return value]
These macros return a value other than 0 if the value of argument c matches the respective description (i.e.,
if the result is true). If the result is false, 0 is returned.
[Example]
#include <ctype.h>
void func(void)
{
int
i, j = 0;
char
s[50];
for(i = 50; i <= 99; i++) {
/* Stores characters that can be */
/* displayed in codes 50 through 99 to s. */
if(isprint(i)) {
s[j] = i;
j++;
}
}
}
214
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.8
Standard Input/Output
This section explains the functions that generate and scan character strings in accordance with the
specification of a formatted character string. Definitions and declarations related to these functions are described
in the "stdio.h" file.
Table 6 - 21 Standard Input/Output
Classification
Function Name
Outline
ERROR
perror
Error processing
FILEIO
freadNote
Read from stream
fwriteNote
Write to stream
fgetcNote
Read one character from stream
fgetsNote
Read one line from stream
getcNote
Read one character from stream
getcharNote
Read one character from standard input
getsNote
Read string from standard input
ungetcNote
Push one character back into input stream
rewindNote
Reset file position indicator
fputcNote
Write character to stream
fputsNote
Output string to stream
putcNote
Write character to stream
putcharNote
Write character to standard output stream
putsNote
Output string to standard output stream
sprintf
Formatted output
vsprintf
Write format-specified text to string
fprintfNote
Output format-specified text to stream
printfNote
Output format-specified text to standard output stream
vfprintfNote
Write format-specified text to stream
vprintfNote
Write format-specified text to standard output stream
SSCANF
sscanf
Formatted input
SCANF
fscanfNote
Read and interpret data from stream
scanfNote
Read and interpret text from standard output stream
GETS
PUTS
SPRINTF
PRINTF
Note
These functions are not supported by the NEC Electronics integrated debugger or the system
simulator.
User’s Manual U18513EJ1V0UM
215
CHAPTER 6 LIBRARY FUNCTION
ERROR
[Overview]
Error processing
perror
[Syntax]
#include <stdio.h>
void
perror(const char *s)
[Description]
perror(const char *s)
This function outputs to stderr the error message that corresponds to global variable errno.
The message that is output is as follows.
When s is not NULL
fprintf(stderr, "%s:%s\n", s, s_fix);
When s is NULL
fprintf(stderr, "%s\n", sfix);
s_fix is as follows.
When errno is EDOM
"EDOM error"
When errno is ERANGE
"ERANGE error"
When errno is 0
"no error"
Otherwise
"error xxx"(xxx is abs (errno)%1000)
[Example]
#include <stdio.h>
void func1(double x)
{
double d;
errno = 0;
d = exp(x);
if(errno) perror("func1");
/* If a calculation exception is generated */
/* by exp perror is called */
}
216
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
FILEIO
[Overview]
Direct input/output
fread, fwrite
Caution These functions are not supported by the NEC Electronics integrated debugger or the system
simulator.
[Syntax]
#include <stdio.h>
size_t fread(void *ptr, size_t, size, size_t nmemb, FILE *stream)
size_t fwrite(const void *ptr, size_t size, size_t nmemb, FILE *stream)
[Description]
fread(void *ptr, size_t, size, size_t nmemb, FILE *stream)
This function inputs nmemb elements of size size from the input stream pointed to by stream and stores
them in ptr. Only the standard input/output stdin can be specified for stream.
fwrite(const void *ptr, size_t size, size_t nmemb, FILE *stream)
This function outputs nmemb elements of size size from the array pointed to by ptr to the output stream
pointed to by stream. Only the standard input/output stdout or stderr can be specified for stream
[Return value]
fread
The number of elements that were input (nmemb) is returned.
fwrite
The number of elements that were output (nmemb) is returned.
Error return does not occur for either function.
User’s Manual U18513EJ1V0UM
217
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <stdio.h>
void func(void)
{
struct {
int
c;
double d;
} buf[10];
fread(buf, sizeof(buf[0]), sizeof(buf)/sizeof(buf[0]), stdin);
}
218
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
GETS
[Overview]
Character or string input
fgetc, fgets, getc, getchar, gets, ungetc, rewind
Caution These functions are not supported by the NEC Electronics integrated debugger or the system
simulator.
[Syntax]
#include <stdio.h>
int
fgetc(FILE *stream)
char
*fgets(char *s, int n, FILE *stream)
int
getc(FILE *stream)
int
getchar()
char
*gets(char *s)
int
ungetc(int c, FILE *stream)
void
rewind(FILE *stream)
[Description]
fgetc(FILE *stream)
This function inputs one character from the input stream pointed to by stream. Only the standard input/
output stdin can be specified for stream.
fgets(char *s, int n, FILE *stream)
This function inputs at most n-1 characters from the input stream pointed to by stream and stores them in s.
Character input is also ended by the detection of a new-line character. In this case, the new-line character is
also stored in s. The end-of-string null character is stored at the end in s. Only the standard input/output stdin
can be specified for stream.
getc(FILE *stream)
This function inputs one character from the input stream pointed to by stream. The getc function is
completely equivalent to fgetc.
getchar()
This function inputs one character from the standard input/output stdin.
gets(char *s)
This function inputs characters from the standard input/output stdin until a new-line character is detected
and stores them in s. The new-line character that was input is discarded, and an end-of-string null character is
stored at the end in s.
User’s Manual U18513EJ1V0UM
219
CHAPTER 6 LIBRARY FUNCTION
ungetc(int c, FILE *stream)
This function pushes the character c back into the input stream pointed to by stream. However, if c is EOF,
no pushback is performed. The character c that was pushed back will be input as the first character during the
next character input. Only one character can be pushed back by ungetc. If ungetc is executed continuously,
only the last ungetc will have an effect. Only the standard input/output stdin can be specified for stream.
rewind(FILE *stream)
This function clears the error indicator of the input stream pointed to by stream, and positions the file
position indicator at the beginning of the file. However, only the standard input/output stdin can be specified for
stream. Therefore, rewind only has the effect of discarding the character that was pushed back by ungetc.
[Return value]
fgetc, getc, getchar
The input character is returned.
fgets, gets
s is returned.
ungetc
The character c is returned.
Error return does not occur for any of these functions.
[Example]
#include <stdio.h>
void func(void)
{
int c;
c = fgetc(stdin);
}
220
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
PUTS
[Overview]
Character or string output
fputc, fputs, putc, putchar, puts
Caution These functions are not supported by the NEC Electronics integrated debugger or the system
simulator.
[Syntax]
#include <stdio.h>
int
fputc(int c, FILE *stream)
int
fputs(const char *s, FILE *stream)
int
putc(int c, FILE *stream)
int
putchar(int c)
int
puts(const char *s)
[Description]
fputc(int c, FILE *stream)
This functionoutputs the character c to the output stream pointed to by stream. Only the standard input/
output stdout or stderr can be specified for stream.
fputs(const char *s, FILE *stream)
This function outputs the string s to the output stream pointed to by stream. The end-of-string null character
is not output. Only the standard input/output stdout or stderr can be specified for stream.
putc(int c, FILE *stream)
This function outputs the character c to the output stream pointed to by stream. The putc function is
completely equivalent to fputc.
putchar(int c)
This function outputs the character c to the standard input/output stdout.
puts(const char *s)
This function outputs the string s to the standard input/output stdout. The end-of-string null character is not
output, but a new-line character is output in its place.
User’s Manual U18513EJ1V0UM
221
CHAPTER 6 LIBRARY FUNCTION
[Return value]
fputc, putc, putchar
The character c is returned.
fputs, puts
0 is returned.
Error return does not occur for any of these functions.
[Example]
#include <stdio.h>
void func(void)
{
fputc('a', stdout);
}
222
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
SPRINTF
[Overview]
Formatted output
sprintf, vsprintf
[Syntax]
#include <stdio.h>
int
sprintf(char *s, const char *format [, arg, ...])
int
vsprintf(char *s, const char *format, va_list arg)
[Description]
sprintf(char *s, const char *format [, arg, ...])
This function applies the format specified by the string pointed to by format to the respective arg arguments,
and writes out the formatted data that was output as a result to the array pointed to by s.
If there are not sufficient arguments for the format, the operation is undefined. If the end of the formatted
string is reached, control returns. If there are more arguments that those required by the format, the excess
arguments are ignored. If the area of s overlaps one of the arguments, the operation is undefined.
The argument format specifies "the output to which the subsequent argument is to be converted". The null
character (\0) is appended at the end of written characters (the null character (\0) is not counted in a return
value).
The format consists of the following two types of directives:
Ordinary characters
Characters that are copied directly without conversion (other than "%").
Conversion specifications
Specifications that fetch zero or more arguments and assign a specification.
Each conversion specification begins with character "%" (to insert "%" in the output, specify "%%" in the
format string). The following appear after the "%":
%[flag][field-width][precision][size][type-specification-character]
The meaning of each conversion specification is explained below
User’s Manual U18513EJ1V0UM
223
CHAPTER 6 LIBRARY FUNCTION
(1)
flag
Zero or more flags, which qualify the meaning of the conversion specification, are placed in any order.
The flag characters and their meanings are as follows:
-
The result of the conversion will be left-justified in the field, with the right side filled with
blanks (if this flag is not specified, the result of the conversion is right-justified).
+
The result of a signed conversion will start with a + or - sign (if this flag is not specified, the
result of the conversion starts with a sign only when a negative value has been converted).
Space
If the first character of a signed conversion is not a sign and a signed conversion is not
generated a character, a space (" ") will be appended to the beginning of result of the
conversion. If both the space flag and + flag appear, the space flag is ignored.
#
The result is to be converted to an alternate format. For o conversion, the precision is
increased so that the first digit of the conversion result is 0. For x or X conversion, 0x or 0X
is appended to the beginning of a non-zero conversion result. For e, f, g, E, or G
conversion, a decimal point "." is added to the conversion result even if no digits follow the
decimal pointNote. For g or G conversion, trailing zeros will not be removed from the
conversion result. The operation is undefined for conversions other than the above.
0
For d, e, f, g, i, o, u, x, E, G, or X conversion, zeros are added following the specification of
the sign or base to fill the field width.
If both the 0 flag and - flag are specified, the 0 flag is ignored. For d, i, o, u, x, or X
conversion, when the precision is specified, the zero (0) flag is ignored.
Note that 0 is interpreted as a flag and not as the beginning of the field width.
The operation is undefined for conversion other than the above.
Note
(2)
Normally, a decimal point appears only when a digit follows it.
field width
This is an optional minimum field width. If the converted value is smaller than this field width, the left side
is filled with spaces (if the left justification flag explained above is assigned, the right side will be filled with
spaces). This field width takes the form of "*" or a decimal integer. If "*" is specified, an int type argument is
used as the field width. A negative field width is not supported. If an attempt is made to specify a negative
field width, it is interpreted as a minus (-) flag appended to the beginning of a positive field width.
(3)
precision
For d, i, o, u, x, or X conversion, the value assigned for the precision is the minimum number of digits to
appear. For e, f, or E conversion, it is the number of digits to appear after the decimal point. For g or G
conversion, it is the maximum number of significant digits. The precision takes the form of "*" or "." followed
by a decimal integer. If "*" is specified, an int type argument is used as the precision. If a negative precision
is specified, it is treated as if the precision were omitted. If only "." is specified, the precision is assumed to
be 0. If the precision appears together with a conversion specification other than the above, the operation is
undefined.
224
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
(4)
size
This is an arbitrary optional size character h, l, or L, which changes the default method for interpreting the
data type of the corresponding argument.
When h is specified, a following d, i, o, u, x, or X type specification is forcibly applied to a short or
unsigned short argument.
When l is specified, a following d, i, o, u, x, or X type specification is forcibly applied to a long or unsigned
long argument. l is also causes a following n type specification to be forcibly applied to a pointer to long
argument. If another type specification character is used together with h or l, the operation is undefined.
When L is specified, a following e, E, f, g, or G type specification is forcibly applied to a long double
argument. If another type specification character is used together with L, the operation is undefined.
(5)
type specification character
These are characters that specify the type of conversion that is to be applied.
The characters that specify conversion types and their meanings are as follows.
%
Output the character "%". No argument is converted. The conversion specification is "%%".
c
Convert an int type argument to unsigned char type and output the characters of the
conversion result.
d
Convert an int type argument to a signed decimal number.
e, E
Convert a double type argument to [-]d.dddde+dd format, which has one digit before the
decimal point (not 0 if the argument is not 0) and the number of digits after the decimal point
is equal to the precision. The E conversion specification generates a number in which the
exponent part starts with "E" instead of "e".
f
Convert a double type argument to decimal notation of the form [-]dddd.dddd.
g, G
Convert a double type argument to e (E for a G conversion specification) or f format, with
the number of digits in the mantissa specified for the precision. Trailing zeros of the
conversion result are excluded from the fractional part. The decimal point appears only
when it is followed by a digit.
i
Perform the same conversion as d.
n
Store the number of characters that were output in the same object. A pointer to int type is
used as the argument.
p
Output a pointer in an implementation-defined format. The CA850 handles a pointer as
unsigned long (this is the same as the lu specification).
o, u, x, X
Convert an unsigned int type argument to octal notation (o), unsigned decimal notation (u),
or unsigned hexadecimal notation (x or X) with dddd format. For x conversion, the letters
abcdef are used. For X conversion, the letters ABCDEF are used.
s
The argument must be a pointer pointing to a character type array. Characters from this
array are output up until the null character (\0) indicating termination (the null character (\0)
itself is not included). If the precision is specified, no more than the specified number of
characters will be output. If the precision is not specified or if the precision is greater than
the size of this array, make sure that this array includes the null character (\0).
User’s Manual U18513EJ1V0UM
225
CHAPTER 6 LIBRARY FUNCTION
vsprintf(char *s, const char *format, va_list arg)
This function applies the format specified by the string pointed to by format to the argument string pointed to
by arg, and outputs the formatted data that was output as a result to the array pointed to be s. The vsprintf
function is equivalent to sprintf with the list of a variable number of real arguments replaced by arg. arg must
be initialized by the va_start macro before the vsprintf function is called.
[Return value]
The number of characters that were output (excluding the null character (\0)) is returned.
Error return does not occur.
[Example]
#include <stdio.h>
void func(int val)
{
char
s[20];
sprintf(s,"%-10.51x\n", val);
/* Specifies left-justification, field width 10, precision 5,*/
/*
size long, and hexadecimal notation for the value of val, */
/* and outputs the result with an appended new-line character to */
/* the array pointed to by s. */
}
226
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
PRINTF
[Overview]
Formatted output
fprintf, printf, vfprintf, vprintf
Caution These functions are not supported by the NEC Electronics integrated debugger or the system
simulator.
[Syntax]
#include <stdio.h>
int
fprintf(FILE *stream, const char *format[, arg, ...])
int
printf(const char *format,[, arg, ...])
int
vfprintf(FILE *stream, const char *format, va_list arg)
int
vprintf(const char *format, va_list arg)
[Description]
Stdin (standard input) and stdout (standard error) are specified for the argument streams in each of the
fprintf, printf, vfprintf, and vprintf functions. 1 memory addresses such as an I/O address is allocated for the I/O
destination of stream. To use these streams in combination with a debugger, the initial values of the stream
structure defined in stdio.h must be set. Be sure to set the initial values prior to calling the function.
Definition of stream structure in stdio.h
typedef struct {
int
mode;
unsigend handle;
int
unget_c;
}FILE;
typedef int fpos_t;
/* with error descriptions */
#pragma section sdata begin
extern FILE _ _ struct_stdin;
extern FILE _ _ struct_stdout;
extern FILE _ _ struct_stderr;
#pragma section sdata end
#define stdin(&_ _ struct_stdin)
#define stdout(&_ _ struct_stdout)
#define stderr(&_ _ struct_stderr)
User’s Manual U18513EJ1V0UM
227
CHAPTER 6 LIBRARY FUNCTION
The first structure member, mode, indicates the I/O status and is internally defined as ACCSD_OUT/
ADDSD_IN. The third member, unget_c, indicates the pushed-back character (stdin only) setting and is
internally defined as -1. When the definition is -1, it indicates that there is no pushed-back character. The
second member, handle, indicates the I/O address. Set the value according to the debugger to be used.
Example of I/O address setting
_ _ struct_stdout.handle = 0xfffff000;
_ _ struct_stderr.handle = 0x00fff000;
_ _ struct_stdin.handle = 0xfffff002;
extern FILE _ _ struct_stdout;
extern FILE _ _ struct_stderr;
#pragma section sdata end
#define stdin(&_ _ struct_stdin)
#define stdout(&_ _ struct_stdout)
#define stderr(&_ _ struct_stderr)
fprintf(FILE *stream, const char *format[, arg, ...])
This function applies the format specified by the string pointed to by format to the respective arg arguments,
and outputs the formatted data that was output as a result to stream. Only the standard input/output stdout or
stderr can be specified for stream. The method of specifying format is the same as described for the sprintf
function. However, fprintf differs from sprintf in that no null character (\0) is output at the end.
printf(const char *format,[, arg, ...])
This function applies the format specified by the string pointed to by format to the respective arg arguments,
and outputs the formatted data that was output as a result to the standard input/output stdout. The method of
specifying format is the same as described for the sprintf function. However, printf differs from sprintf in that no
null character (\0) is output at the end.
vfprintf(FILE *stream, const char *format, va_list arg)
This function applies the format specified by the string pointed to by format to argument string pointed to by
arg, and outputs the formatted data that was output as a result to stream. Only the standard input/output stdout
or stderr can be specified for stream. The method of specifying format is the same as described for the sprintf
function. The vfprintf function is equivalent to fprintf with the list of a variable number of real arguments
replaced by arg. arg must be initialized by the va_start macro before the vfprintf function is called.
vprintf(const char *format, va_list arg)
This function applies the format specified by the string pointed to by format to the argument string pointed to
by arg, and outputs the formatted data that was output as a result to the standard input/output stdout. The
method of specifying format is the same as described for the sprintf function. The vprintf function is equivalent
to printf with the list of a variable number of real arguments replaced by arg. arg must be initialized by the
va_start macro before the vprintf function is called.
[Return value]
The number of characters that were output is returned.
228
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <stdio.h>
void func(int val)
{
fprintf(stdout, "%-10.5x\n", val);
}
/* Example using vfprintf in a general error reporting routine */
void error(char *function_name, char *format, ...)
{
va_list arg;
va_start(arg, format);
/* Output function name for which error occurred */
fprintf(stderr, "ERROR in %s:", function_name);
/* Output remaining messages */
vfprintf(stderr, format, arg);
va_end(arg);
}
User’s Manual U18513EJ1V0UM
229
CHAPTER 6 LIBRARY FUNCTION
SSCANF
[Overview]
Formatted input
sscanff
[Syntax]
#include <stdio.h>
int
sscanf(const char *s, const char *format[, arg, ...])
[Description]
sscanf(const char *s, const char *format[, arg, ...])
This function reads the input to be converted according to the format specified by the character string
pointed to by format from the array pointed to by s and treats the arg arguments that follow format as pointers
that point to objects for storing the converted input.
An input string that can be recognized and "the conversion that is to be performed for assignment" are
specified for format. If sufficient arguments do not exist for format, the operation is undefined. If format is used
up even when arguments remain, the remaining arguments are ignored.
The format consists of the following three types of directives:
One or more Space characters
Space ( ), tab (\t), or new-line (\n).
If a space character is found in the string when sscanf is executed, all
consecutive space characters are read until the next non-space
character appears (the space characters are not stored).
Ordinary characters
All ASCII characters other than "%".
If an ordinary character is found in the string when sscanf is executed,
that character is read but not stored. sscanf reads a string from the
input field, converts it into a value of a specific type, and stores it at
the position specified by the argument, according to the conversion
specification. If an explicit match does not occur according to the
conversion specification, no subsequent space character is read.
Conversion specification
Fetches 0 or more arguments and directs the conversion.
Each conversion specification starts with "%". The following appear after the "%":
%[assignment-suppression-character][field-width][size][type-specification-character]
Each conversion specification is explained below.
(1)
Assignment suppression character
The assignment suppression character "*" suppresses the interpretation and assignment of the input field.
230
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
(2)
field width
This is a non-zero decimal integer that defines the maximum field width.
It specifies the maximum number of characters that are read before the input field is converted. If the input
field is smaller than this field width, sscanf reads all the characters in the field and then proceeds to the next
field and its conversion specification.
If a space character or a character that cannot be converted is found before the number of characters
equivalent to the field width is read, the characters up to the white space or the character that cannot be
converted are read and stored. Then, sscanf proceeds to the next conversion specification.
(3)
size
This is an arbitrary optional size character h, l, or L, which changes the default method for interpreting the
data type of the corresponding argument.
When h is specified, a following d, i, n, o, u, or x type specification is forcibly converted to short int type and
stored as short type. Nothing is done for c, e, f, n, p, s, D, I, O, U, or X.
When l is specified, a following d, i, n, o, u, or x type specification is forcibly converted to long int type and
stored as long type. An e, f, or g type specification is forcibly converted to double type and stored as double
type. Nothing is done for c, n, p, s, D, I, O, U, and X.
When L is specified, a following c, i, o, u, or x type specification is forcibly converted to long double type and
stored as long double type. Nothing is done for other type specifications.
In cases other than the above, the operation is undefined.
(4)
type specification character
These are characters that specify the type of conversion that is to be applied.
The characters that specify conversion types and their meanings are as follows.
%
Match the character "%". No conversion or assignment is performed. The conversion
specification is "%%".
c
Scan one character. The corresponding argument should be "char *arg".
d
Read a decimal integer into the corresponding argument. The corresponding argument
should be "int *arg".
e, f, g
Read a floating-point number into the corresponding argument. The corresponding
argument should be "float *arg".
i
Read a decimal, octal, or hexadecimal integer into the corresponding argument. The
corresponding argument should be "int *arg".
n
Store the number of characters that were read in the corresponding argument. The
corresponding argument should be "int *arg".
o
Read an octal integer into the corresponding argument. The corresponding argument must
be "int *arg".
p
Store the pointer that was scanned. This is an implementation definition.
The ca processes %p and %U in exactly the same manner. The corresponding argument
should be "void **arg".
s
Read a string into a given array. The corresponding argument should be "char arg[ ]".
u
Read an unsigned decimal integer into the corresponding argument. The corresponding
argument should be "unsigned int *arg".
User’s Manual U18513EJ1V0UM
231
CHAPTER 6 LIBRARY FUNCTION
x, X
Read a hexadecimal integer into the corresponding argument. The corresponding argument
should be "int *arg".
D
Read a decimal integer into the corresponding argument. The corresponding argument
should be "long *arg".
E, F, G
Read a floating-point number into the corresponding argument. The corresponding
argument should be "double *arg".
I
Read a decimal, octal, or hexadecimal integer into the corresponding argument. The
corresponding argument should be "long *arg".
O
Read an octal integer into the corresponding argument. The corresponding argument
should be "long *arg".
U
Read an unsigned decimal integer into the corresponding argument. The corresponding
argument should be "unsigned long *arg".
[]
Read a non-empty string into the memory area starting with argument arg. This area must
be large enough to accommodate the string and the null character (\0) that is automatically
appended to indicate the end of the string. The corresponding argument should be "char
*arg".
The character pattern enclosed by [ ] can be used in place of the type specification
character s. The character pattern is a character set that defines the search set of the
characters constituting the input field of sscanf. If the first character within [ ] is "^", the
search set is complemented, and all ASCII characters other than the characters within [ ]
are included. In addition, a range specification feature that can be used as a shortcut is also
available. For example, %[0-9] matches all decimal numbers. In this set, "-" cannot be
specified as the first or last character. The character preceding "-" must be less in lexical
sequence than the succeeding character.
Examples
%[abcd]
%[^abcd]
%[A-DW-Z]
%[z-a]
Matches character strings that include only a, b, c, and d.
Matches character strings that include any characters other
than a, b, c, and d.
Matches character strings that include A, B, C, D, W, X, Y, and Z.
Matches z, -, and a (this is not considered a range specification).
Make sure that a floating-point number (type specification characters e, f, g, E, F, and G) corresponds to
thefollowing general format.
[+ | -]ddddd[.]ddd[E | e[+ | -]ddd]
However, the portions enclosed by [ ] in the above format are arbitrarily selected, and ddd indicates a
decimal,octal, or hexadecimal digit.
[Return value]
The number of input fields for which scanning, conversion, and storage were executed normally is returned.
The return value does not include scanned fields that were not stored.
If an attempt is made to read to the end of the file, the return value is EOF.
If no field was stored, the return value is 0.
232
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Cautions]
-
sscanf may stop scanning a specific field before the normal end-of-field character is reached or may stop
completely.
-
sscanf stops scanning and storing a field and moves to the next field under the following conditions.
(a)
The substitution suppression character (*) appears after "%" in the format specification, and the input
field at that point has been scanned but not stored.
(b)
A field width (positive decimal integer) specification character was read.
(c)
The character to be read next cannot be converted according to the conversion specification (for
example, if Z is read when the specification is a decimal number).
(d)
The next character in the input field does not appear in the search set (or appears in the complement
search set).
If sscanf stops scanning the input field at that point because of any of the above reasons, it is assumed that
the next character has not yet been read, and this character is used as the first character of the next field or
the first character for the read operation to be executed after the input.
-
sscanf ends under the following conditions:
(a)
The next character in the input field does not match the corresponding ordinary character in the string to
be converted.
(b)
The next character in the input field is EOF.
(c)
The string to be converted ends.
-
If a list of characters that is not part of the conversion specification is included in the string to be converted,
make sure that the same list of characters does not appear in the input. sscanf scans matching characters
but does not store them. If there was a mismatch, the first character that does not match remains in the
input as if it were not read.
[Example]
#include <stdio.h>
void func(void)
{
int
float
const char
char
i, n;
x;
*s;
name[10];
s = "23 11.1e-1 NAME";
n = sscanf(s,"%d%f%s", &i, &x, name);
/* Stores 23 in i, 1.110000 in x, */
/* and "NAME" in name. */
/* The return value n is 3. */
}
User’s Manual U18513EJ1V0UM
233
CHAPTER 6 LIBRARY FUNCTION
SCANF
[Overview]
Formatted input
fscanf, scanf
Caution These functions are not supported by the NEC Electronics integrated debugger or the system
simulator.
[Syntax]
#include <stdio.h>
int
fscanf(FILE *stream, const char *format[, arg, ...])
int
scanf(const char *format[, arg, ...])
[Description]
fscanf(FILE *stream, const char *format[, arg, ...])
reads the input to be converted according to the format specified by the character string pointed to by format
from stream and treats the arg arguments that follow format as objects for storing the converted input. Only the
standard input/output stdin can be specified for stream. The method of specifying format is the same as
described for the sscanf function.
scanf(const char *format[, arg, ...])
reads the input to be converted according to the format specified by the character string pointed to by format
from the standard input/output stdin and treats the arg arguments that follow format as objects for storing the
converted input. The method of specifying format is the same as described for the sscanf function.
[Return value]
The return value is similar to the one described for sscanf. See the section about sscanf.
234
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <stdio.h>
void func(void)
{
int
i, n;
double x;
char
name[10];
n = scanf("%d%lf%s", &i, &x, name);
/* Perform formatted input of input */
/* from stdin using the format */
/* "23 11.1e-1 NAME" */
}
User’s Manual U18513EJ1V0UM
235
CHAPTER 6 LIBRARY FUNCTION
6.9
Standard Utility Functions
This section explains the utility functions useful for various programs. Definitions and declarations related to
these functions are described in the "stdlib.h" file
Table 6 - 22 Standard Utility Functions
Classification
ABS
BSEARCH
DIV
ECVTF
ITOA
MALLOC
RAND
STRTODF
STRTOL
236
Function Name
Outline
abs
Absolute value (int type)
labs
Absolute value (long type)
bsearch
Binary search
qsort
Sorting
div
Division (int type)
ldiv
Division (long type)
ecvtf
Conversion of floating-point value to numeric character string
(total number of characters specified)
fcvtf
Conversion of floating-point value to numeric character string
(number below decimal point specified)
gcvtf
Conversion of floating-point value to numeric character string
(format specified)
itoa
Conversion of integer (int type) to character string
ltoa
Conversion of integer (long type) to character string
ultoa
Conversion of integer (unsigned long type) to character string
calloc
Dynamic memory allocation
free
Dynamic memory release
malloc
Dynamic memory allocation
realloc
Dynamic memory reallocation
rand
Pseudo random number generation
srand
Setting of pseudo random number seed
atoff
Conversion of character string to floating point
strtodf
Conversion of character string to floating point (stores pointer
in last character string)
atoi
Absolute value (int type)
atol
Absolute value (long type)
strtol
Binary search
strtoul
Sorting
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
ABS
[Overview]
Integer absolute value
abs, labs
[Syntax]
#include <stdlib.h>
int
abs(int j)
long
labs(long j)
[Description]
abs(int j)
This function obtains the absolute value of j (size of j), | j |. If j is a negative number, the result is the reversal
of j. If j is not negative, the result is j.
labs(long j)
This function is the same as abs, but uses long type instead of int type, and the return value is also of long
type.
[Return value]
Returns the absolute value of j (size of j), | j |.
[Example]
#include <stdlib.h>
void func(int i)
{
int val;
val = -15;
i = abs(val);
/* Returns absolute value of val, 15, to 1. */
}
User’s Manual U18513EJ1V0UM
237
CHAPTER 6 LIBRARY FUNCTION
BSEARCH
[Overview]
Binary search
bsearch, qsort
[Syntax]
#include <stdlib.h>
void bsearch(const
void
*key,
const
void
*base,
size_t
nmemb,
size_t
size,
int(*compar)(const void *, const void*))
void qsort(void *base, size_t nmemb, size_t size, int(*compar)(const void*, const
void*))
[Description]
bsearch(const void *key, const void *base, size_t nmemb, size_t size, int(*compar)(const void *, const
void*))
This function searches an element that coincides with key from an array starting with base by means of
binary search. nmemb is the number of elements of the array. size is the size of each element. The array must
be arranged in the ascending order in respect to the compare function indicated by compar (last argument).
Define the compare function indicated by compar to have two arguments. If the first argument is less than the
second, a negative integer must be returned as the result. If the two arguments coincide, zero must be
returned. If the first is greater than the second, a positive integer must be returned.
qsort(void *base, size_t nmemb, size_t size, int(*compar)(const void*, const void*))
This function sorts the array pointed to by base into ascending order in relation to the comparison function
pointed to by compar. nmemb is the number of array elements, and size is the size of each element. The
comparison function pointed to by compar is the same as the one described for bsearch.
[Return value]
A pointer to the element in the array that coincides with key is returned. If there are two or more elements
that coincide with key, the one that has been found first is indicated.
238
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <stdlib.h>
#include <string.h>
int
compar(char **x, char **y);
void func(void)
{
static char *base[] = {"a", "b", "c", "d", "e", "f"};
char
*key = "c";
/* Search key is "c". */
char
**ret;
ret =(char **)bsearch((char *)&key,(char *)base, 6,
sizeof(char *), compar);
/* Pointer to "c" is stored in ret. */
}
int compar(char **x, char **y)
{
return(strcmp(*x, *y));
/* Returns positive, zero, or negative /*
/* integer as result of comparing arguments. */
}
User’s Manual U18513EJ1V0UM
239
CHAPTER 6 LIBRARY FUNCTION
DIV
[Overview]
Division
div, ldiv
[Syntax]
#include <stdlib.h>
div_t
div(int n, int d)
ldiv_t ldiv(long n, long d)
[Description]
div(int n, int d)
This function calculates the quotient and remainder resulting from dividing numerator n by denominator d,
and stores these two integers as the members of the following structure div_t.
typedef struct {
int quot;
int rem;
} div_t
quot the quotient, and rem is the remainder. If d is not zero, and if "r = div(n, d);", n is a value equal to
"r.rem + d * r.quot".
If d is zero, the resultant quot member has a sign the same as n and has the maximum size that can be
expressed. The rem member is 0.
ldiv(long n, long d)
This function is used to divide a value of long type, not a value of int type. The result is stored as the
member of the following structure ldiv_t.
typedef struct {
long
quot;
long
rem;
} ldiv_t
[Return value]
The structure storing the result of the division is returned.
240
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <stdlib.h>
void func(void)
{
div_t r;
r = div(110, 3);
/* 36 is stored in r.quot, and 2 is stored in r.rem. */
}
User’s Manual U18513EJ1V0UM
241
CHAPTER 6 LIBRARY FUNCTION
ECVTF
[Overview]
Conversion from floating point number to character string
ecvtf, fcvtf, gcvtf
[Syntax]
#include <stdlib.h>
char
*ecvtf(float val, int chars, int *decpt, int *sgn)
char
*fcvtf(float val, int decimals, int *decpt, int *sgn)
char
*gcvtf(float val, int prec, char *buf)
[Description]
ecvtf(float val, int chars, int *decpt, int *sgn)
This function generates a character string indicating a numeric value val of float type in number (terminated
with the null character (\0)). The second argument chars specifies the total number of characters to be written
(because only numbers are written, this argument specifies the valid number of numerals in the converted
character string). The digits of the integer of val are always included.
fcvtf(float val, int decimals, int *decpt, int *sgn)
This function is the same as ecvt, except the interpretation of the second argument. The second argument
decimals specify the number of characters to be written after the decimal point.
ecvtf and fcvtf only write a number to an output character string. Therefore, record the position of the
decimal point to *decpt and the sign of the numeric value to *sgn. After the number has been formatted, the
number of digits at the left of the decimal point is stored in *decpt. If the numeric value is positive, 0 is stored in
*sgn; if it is negative, 1 is stored.
gcvtf(float val, int prec, char *buf)
This function converts a numeric value into a character string, and stores it to buffer buf. gcvtf uses the same
rule as the format "%.prec" (sign is appended to the negative number only) of sprintf, and selects an exponent
format or normal decimal point format according to the valid number of digits (specified by prec).
[Return value]
242
ecvtf, fcvtf
Returns a pointer indicating a new character string including the character string
representation of val.
gcvtf
Returns a pointer (same as argument buf) to the formatted character string
representation of val.
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <stdlib.h>
void func(void)
{
float
val;
int
dec, sgn;
val = 111.11;
ecvtf(val, 12, &dec, &sgn);
/* Converts value 111.11 of val to character string of 12 characters. */
/* dec records number of digits, 3, at left of decimal point, */
/* and sgn records sign(0 because numeric value is positive). */
}
User’s Manual U18513EJ1V0UM
243
CHAPTER 6 LIBRARY FUNCTION
ITOA
[Overview]
Conversion from integer to character string
itoa, ltoa, ultoa
[Syntax]
#include <stdlib.h>
char
*itoa(int value, char *string, int radix)
char
*ltoa(long int value, char *string, int radix)
char
*ultoa(unsigned long int value, char *string, int radix)
[Description]
itoa(int value, char *string, int radix)
This function converts an int type numeric value to a character string for a radix-based number and stores it
in the array indicated by string. The terminating null character (\0) always is added at the end of the character
string. Numeric values from 2 to 36 can be specified for radix. If radix is 10, value is handled as a signed
numeric value, and when value < 0, the "-" character is appended at the beginning of the character string.
Otherwise, value is handled as an unsigned numeric value. If radix > 10, the lowercase letters a to z are
assigned for 10 to 35.
ltoa(long int value, char *string, int radix)
This function converts a long int type numeric value to a character string for a radix-based number and
stores it in the array indicated by string. Except for the type of value, this is the same as itoa.
ultoa(unsigned long int value, char *string, int radix)
This function converts an unsigned long int type numeric value to a character string for a radix-based
number and stores it in the array indicated by string. Except for the type of value, this is the same as itoa
[Return value]
itoa, ltoa, ultoa
244
string is returned.
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <stdlib.h>
void func(void)
{
char
buf[128];
itoa(12345, buf, 16); 3/* Converts 12345 to a hexadecimal character string */
}
User’s Manual U18513EJ1V0UM
245
CHAPTER 6 LIBRARY FUNCTION
MALLOC
[Overview]
Memory allocation and management
calloc, free, malloc, realloc
[Syntax]
#include <stdlib.h>
void
*calloc(size_t nmemb, size_t size)
void
free(void *ptr)
void
*malloc(size_t size)
void
*realloc(void *ptr, size_t size)
[Description]
The memory area management functions automatically allocate memory area as necessary from the heap
memory area. Also, since the compiler does not automatically allocate this area, when calloc, malloc, or
realloc is used, the heap memory area must be allocated. The area allocation should be performed first by an
application.
Heap memory setup example
#define SIZEOF_HEAP 0x1000
int
_ _ sysheap[SIZEOF_HEAP>>2];
size_t _ _ sizeof_sysheap = SIZEOF_HEAP;
Remarks 1 The symbol "sysheap" (three underscores "_") of the variable " _sysheap" (two under-scores "_")
points to the starting address of heap memory. This value should be a word integer value.
2 The required heap memory size (bytes) should be set for the variable "_sizeof_sysheap" (two
leading underscores). If assembly language is used for coding, this value should be set for the
symbol " _sizeof_sysheap" (three leading underscores).
calloc(size_t nmemb, size_t size)
This function allocates an area for an array of nmemb elements. The allocated area is initialized to zeros.
free(void *ptr)
This function releases the area pointed to by ptr so that this area is subsequently available for allocation.
The area that was acquired by calloc, malloc, or realloc must be specified for ptr.
malloc(size_t size)
This function allocates an area having a size indicated by size. The area is not initialized.
246
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
realloc(void *ptr, size_t size)
This function changes the size of the area pointed to by ptr to the size indicated by size. The contents of the
area are unchanged up to the smaller of the previous size and the specified size. If the area is expanded, the
contents of the area greater than the previous size are not initialized. When ptr is a null pointer, the operation
is the same as that of malloc (size). Otherwise, the area that was acquired by calloc, malloc, or realloc must be
specified for ptr.
[Return value]
calloc, malloc, realloc
When area allocation succeeds, a pointer to that area is returned. When the area
could not be allocated, a null pointer is returned.
[Example]
#include <stdlib.h>
typedef struct {
double d[3];
int
i[2];
} s_data;
int func(void)
{
sdata *buf;
int
i;
/* Allocate an area for 40 s_data*/
if((buf = calloc(40, sizeof(s_data))) == NULL) return(1);
for(i = 0; i<40; i++)
{
}
/* Release the area */
free(buf);
return(0);
}
User’s Manual U18513EJ1V0UM
247
CHAPTER 6 LIBRARY FUNCTION
[Cautions]
The memory area to be acquired or released by the calloc, free, malloc, and realloc functions is called the
heap area. The area for allocating this heap area and its size must be set in advance. When the heap area is
set via a C language source program, it is written as [Heap memory setup example], however, when specified
using the assembler, the following program should be added to the startup module.
If the heap area is specified by the C language source program and assembler simultaneously, an error
occurs, so specify via one or the other.
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - #
system heap
#----- - - - - - -- - - - - - - -- - - - - - - -- - - - - - - -- - - - - - - - -- - - - - .set
HEAPSIZE, 0x1000
.globl _ _ sysheap
.bss
.lcomm _ _ sysheap, HEAPSIZE, 4
.data
.globl _ _ sizeof_sysheap
_ _ sizeof_sysheap:
.word HEAPSIZE
# In this example, a heap area of 0x1000 bytes is allocated in the .bss area.
248
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
RAND
[Overview]
Pseudo random number generation
rand, srand
[Syntax]
#include <stdlib.h>
int
rand()
void
srand(unsigned int seed)
[Description]
rand()
This function returns a random number that is greater than or equal to zero and less than or equal to
RAND_MAX.
srand(unsigned int seed)
This function assigns seed as the new pseudo random number sequence seed to be used by the rand call
that follows. If srand is called using the same seed value, the same numbers in the same order will appear for
the random numbers that are obtained by rand. If rand is executed without executing srand, the results will be
the same as when srand(1) was first executed.
[Return value]
rand
Random numbers are returned.
[Example]
#include <stdlib.h>
void func(void)
{
if(rand()& 0xf)<4) func1();
}
/* Execute func1 with a probability of 25% */
User’s Manual U18513EJ1V0UM
249
CHAPTER 6 LIBRARY FUNCTION
STRTODF
[Overview]
Conversion from character string to floating-point number
atoff, strtodf
[Syntax]
#include <stdlib.h>
float
atoff(const char *str)
float
strtodf(const char *str, char **p)
[Description]
atoff(const char *str)
This function converts the first portion of the character string indicated by str into a float type representation.
atoff is the same as the following.
strtodf(str, NULL);
strtodf(const char *str, char **p)
This function converts the first part of the character string indicated by str into a long type representation.
The part of the character string to be converted is in the following format and is at the beginning of str with the
maximum length, starting with a normal character that is not a space.
[+|-]digits[.][digits][(e|E)[+|-]digits]
If str is vacant or consists of space characters only, if the first normal character is other than "+", "-", ".", or a
numeral, the partial character string does not include a character. If the partial character string is vacant,
conversion is not executed, and the value of str is stored in the area indicated by ptr. If the partial character
string is not vacant, it is converted, and a pointer to the last character string (including the null character (\0)
indicating at least the end of str) is stored in the area indicated by ptr. This function is not re-entrant.
[Return value]
If the partial character string has been converted, the resultant value is returned. If the character string could
not be converted, 0 is returned. If an overflow occurs (the value is not in the range in which it can be
expressed), HUGE_VAL or -HUGE_VAL is returned, and ERANGE is set to global variable errno. If an
underflow occurs, 0 is returned, and macro ERANGE is set to global variable errno.
250
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <stdlib.h>
#include <stdio.h>
void func(float ret)
{
char
*p, *str, s[30];
str = "+5.32a4e";
ret = strtodf(str, &p);
/* 5.320000 is returned to ret, and pointer */
/* to "a" is stored in area of p. */
sprintf(s, "%lf\t%c", ret, *p);
/* "5.320000 a" is stored in array */
/* indicated by s. */
}
User’s Manual U18513EJ1V0UM
251
CHAPTER 6 LIBRARY FUNCTION
STRTOL
[Overview]
Conversion from character string to integer
atoi, atol, strtol, strtoul
[Syntax]
#include <stdlib.h>
int
atoi(const char *str)
long
atol(const char *str)
long
strtol(const char *str, char **ptr, int base)
unsigned long strtoul(const char *str, char **ptr, int base)
[Description]
atoi(const char *str)
This function converts the first part of the character string indicated by str into an int type representation. atoi
is the same as the following.
(int) strtol(str, NULL, 10);
atol(const char *str)
This function converts the first part of the character string indicated by str into a long int type representation.
atol is the same as the following.
strtol(str, NULL, 10);
252
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
strtol(const char *str, char **ptr, int base)
This function converts the first part of the character string indicated by str into a long type representation.
strol first divides the input characters into the following three parts: the "first blank", "a string represented by
the base number determined by the value of base and is subject to conversion into an integer", and "the last
one or more character string that is not recognized (including the null character (\0))". Then strtol converts the
string into an integer, and returns the result.
(1)
(a)
Specify 0 or 2 to 36 as argument base.
If base is 0, the expected format of the character string subject to conversion is of integer format
having an optional + or - sign and "0x", indicating a hexadecimal number, prefixed.
(b)
If the value of base is 2 to 36, the expected format of the character string is of character string or
numeric string type having an optional + or - sign prefixed and expressing an integer whose base is
specified by base. Characters "a" (or "A") through "z" (or "Z") are assumed to have a value of 10 to 35.
Only characters whose value is less than that of base can be used.
(c)
If the value of base is 16, "0x" is prefixed (suffixed to the sign if a sign exists) to the string of characters
and numerals (this can be omitted).
(2)
The string subject to conversion is defined as the longest partial string at the beginning of the input
character string that starts with the first character other than blank and has an expected format.
(a)
If the input character string is vacant, if it consists of blank only, or if the first character that is not blank
is not a sign or a character or numeral that is permitted, the subject string is vacant.
(b)
If the string subject to conversion has an expected format and if the value of base is 0, the base
number is judged from the input character string. The character string led by 0x is regarded as a
hexadecimal value, and the character string to which 0 is prefixed but x is not is regarded as an octal
number. All the other character strings are regarded as decimal numbers.
(c)
If the value of base is 2 to 36, it is used as the base number for conversion as mentioned above.
(d)
If the string subject to conversion starts with a - sign, the sign of the value resulting from conversion is
reversed.
(3)
The pointer that indicates the first character string
(a)
This is stored in the object indicated by ptr, if ptr is not a null pointer.
(b)
If the string subject conversion is vacant, or if it does not have an expected format, conversion is not
executed.
(c)
The value of str is stored in the object indicated by ptr if ptr is not a null pointer.
This function is not re-entrant.
strtoul(const char *str, char **ptr, int base)
This function is the same as strtol except that the type of the return value is of unsigned long type.
User’s Manual U18513EJ1V0UM
253
CHAPTER 6 LIBRARY FUNCTION
[Return value]
atoi, atol
Returns the converted value if the partial character string could be converted. If it could
not, 0 is returned.
strtol
Returns the converted value if the partial character string could be converted. If it could
not, 0 is returned.
If an overflow occurs (because the converted value is too great), LONG_MAX or
LONG_MIN is returned, and macro ERANGE is set to global variable errno.
strtoul
Returns the converted value if the partial character string could be converted. If it could
not, 0 is returned.
If an overflow occurs, ULONG_MAX is returned, and macro ERANGE is set to global
variable errno.
[Example]
#include <stdlib.h>
void func(long ret)
{
char *p;
ret = strtol("10", &p, 0);
/* 10 is returned to ret. */
ret = strtol("0x10", &p, 0);
/* 16 is returned to ret. */
ret = strtol("10x", &p, 2);
/* 2 is returned to ret, and pointer to "x" */
/* is returned to area of p. */
ret = strtol("2ax3", &p, 16); /* 42 is returned to ret, and pointer to "x" */
/* is returned to area of p. */
:
}
254
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.10
Non-Local Jump Functions
This section describes the non-local jump functions.
Declarations and definitions concerning these functions are described in the setjmp.h file
Table 6 - 23 Non-Local Jump Functions
Classification
SETJMP
Function Name
Outline
setjmp
Sets the destination of the non-local jump
longjmp
Non-local jump
User’s Manual U18513EJ1V0UM
255
CHAPTER 6 LIBRARY FUNCTION
SETJMP
[Overview]
Non-local jumps
setjmp, longjmp
[Syntax]
#include <setjmp.h>
int
setjmp(jmp_buf env)
void
longjmp(jmp_buf env, int val)
[Description]
setjmp(jmp_buf env)
This function sets env as the destination for a non-local jump. In addition, the environment in which setjmp
was run is saved to env.
longjmp(jmp_buf env, int val)
This function performs a non-local jump to the place immediately after setjmp using env saved by setjmp.val
as a return value for setjmp.
[Return value]
setjmp
256
0 is returned if returning from setjmp. The second argument val in longjmp is returned if
a non-local jump is performed by longjmp. However, 1 is returned if val is 0.
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <setjmp.h>
#define ERR_XXX1
#define ERR_XXX2
1
2
jmp_buf jmp_env;
void func(void)
{
for(;;) {
switch(setjmp(jmp_env))
{
case ERR_XXX1 :
/* Termination of error XXX1 */
break;
case ERR_XXX2 :
/* Termination of error XXX2 */
break;
case 0 :
/* No non-local jumps */
default :
break;
}
}
}
void func1(void)
{
longjmp(jmp_env, ERR_XXX1);
/* Non-local jumps are performed upon */
/* generation of error XXX1 */
longjmp(jmp_env, ERR_XXX2);
/* Non-local jumps are performed upon */
/* generation of error XXX2 */
}
User’s Manual U18513EJ1V0UM
257
CHAPTER 6 LIBRARY FUNCTION
6.11
Mathematical Functions
This section explains the mathematical functions.
Definitions and declarations related to these functions are described in the "math.h" file.
Mathematical library libm.a internally references standard library libc.a. When referencing libm.a by starting
ld850 alone, therefore, libc.a must also be referenced. When referencing two or more archive files by starting
ld850 alone, undefined symbols are searched in the sequence in which reference is specified; therefore, specify
the libc.a reference "-lc" at the end of the sequence.
If ld850 is started from the compiler, however, the libc.a file is automatically referenced
Table 6 - 24 Mathematical Functions
Classification
Outline
j0f
Bessel function of first kind (0 degree)
j1f
Bessel function of first kind (first degree)
jnf
Bessel function of first kind (n degree)
y0f
Bessel function of second kind (0 degree)
y1f
Bessel function of second kind (first degree)
ynf
Bessel function of second kind (n degree)
erff
Error function (approximate value)
erfcf
Error function (complementary probability)
expf
Exponential function
logf
Logarithmic function (natural logarithm)
log2f
Logarithmic function (base 2)
log10f
Logarithmic function (base 10)
powf
Power function
cbrtf
Cubic root function
sqrtf
Square root function
ceilf
ceiling function
fabsf
Absolute value function
floorf
floor function
fmodf
Remainder function
frexpf
Divides floating-point number into mantissa and power
ldexpf
Converts floating-point number to power
modff
Divides floating-point number into integer and decimal
GAMMAF
gammaf
Logarithmic gamma function
HYPOTF
hypotf
Euclidean distance function
MATHERR
matherr
Error processing function
BESSEL
ERFF
EXPF
FLOORF
FREXPF
258
Function Name
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 24 Mathematical Functions
Classification
SINHF
TRIG
Function Name
Outline
acoshf
Inverse hyperbolic function, cosine
asinhf
Inverse hyperbolic function, sine
atanhf
Inverse hyperbolic function, tangent
coshf
Hyperbolic function, cosine
sinhf
Hyperbolic function, sine
tanhf
Hyperbolic function, tangent
acosf
Inverse cosine
asinf
Inverse sign
atanf
Inverse tangent
atan2f
Inverse tangent (y/x)
cosf
Cosine
sinf
Sign
tanf
Tangent
User’s Manual U18513EJ1V0UM
259
CHAPTER 6 LIBRARY FUNCTION
BESSEL
[Overview]
Bessel function
j0f, j1f, jnf, y0f, y1f, ynf
[Syntax]
#include <math.h>
float
jnf(int n, float x)
float
j0f(float x)
float
j1f(float x)
float
ynf(int n, float x)
float
y0f(float x)
float
y1f(float x)
[Description]
A Bessel function is a function that is the solution to the following differential equation.
2
2 d y dy
2
2
x --------- + ------ + ( x – p )y = 0
2 dx
dx
jnf(int n, float x)
This function calculates the Bessel function of the first kind of the n degree.
j0f(float x)
This function calculates the Bessel functions of the first kind of the 0 degrees.
j1f(float x)
This function calculates the Bessel functions of the first kind of the first degrees.
ynf(int n, float x)
This function calculates the Bessel function of the second kind of the n degree.
y0f(float x)
This function calculates the Bessel functions of the second kind of the 0 degrees.
y1f(float x)
This function calculates the Bessel functions of the second kind of the first degrees.
260
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Return value]
jnf
Returns the Bessel function of the first kind of the n degree.
j0f
Returns the Bessel function of the first kind of the 0 degree.
j1f
Returns the Bessel function of the first kind of the first degree.
ynf
Returns the Bessel function of the second kind of the n degree.
y0f
Returns the Bessel function of the second kind of the 0 degree.
y1f
Returns the Bessel function of the second kind of the first degree.
[Example]
#include <math.h>
float func(void)
{
float
ret, x;
ret = j1f(x);
/* Calculates Bessel function of first kind and */
/* first decree in response to value of x, */
/* and returns function to ret. */
:
return(ret);
}
User’s Manual U18513EJ1V0UM
261
CHAPTER 6 LIBRARY FUNCTION
ERFF
[Overview]
Error function
erff, erfcf
[Syntax]
#include <math.h>
float
erff(float x)
float
erfcf(float x)
[Description]
erff(float x)
This function calculates the approximate value (numeric value between 0 and 1) of the "error function" that
estimates the probability for which the observed value is in a range of standard deviation x. The expression
that defines the error function is as follows.
2
------- x
π
x
∫0 e
–t
2
dt
erfcf(float x)
This function calculates complementary probability through "1.0-erff(x)". This function is provided to prevent
the accuracy from dropping if erff(x) is called by x with a large value and the result is subtracted from 1.0.
[Return value]
erff
Returns the approximate value (numeric value between 0 and 1) of the "error function".
erfcf
Returns the complementary probability.
[Example]
#include <math.h>
float func(void)
{
float
ret, x;
ret = erff(x);
/* Calculates approximate value of error function in */
/* response to value of x and returns it to ret. */
:
return(ret);
}
262
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
EXPF
[Overview]
Exponent/logarithm/power/cubic root/square root function
expf, logf, log2f, log10f, powf, cbrtf, sqrtf
[Syntax]
#include <math.h>
float
expf(float x)
float
logf(float x)
float
log2f(float x)
float
log10f(float x)
float
powf(float x, float y)
float
cbrtf(float x)
float
sqrtf(float x)
[Description]
expf(float x)
This function calculates the xth power of e (e is the base of a natural logarithm and is about 2.71828).
logf(float x)
This function calculates the natural logarithm of x, i.e., logarithm with base e.
log2f(float x)
This function calculates the logarithm of x with base 2. This is realized by "log(x)/log(2)".
log10f(float x)
This function calculates the logarithm of x with base 10. This is realized by "log(x)/log(10)".
powf(float x, float y)
This function calculates the yth power of x.
cbrtf(float x)
This function calculates the cubic root of x.
sqrtf(float x)
This function calculates the square root of x.
User’s Manual U18513EJ1V0UM
263
CHAPTER 6 LIBRARY FUNCTION
[Return value]
expf
Returns the xth power of e.
expf returns an unnormalized value if an underflow occurs (if x is a negative number that
cannot express the result), and sets macro ERANGE to global variable errno. If an
overflow occurs (if x is too great a number), HUGE_VAL (maximum double type
numerics that can be expressed) is returned, and macro ERANGE is set to global
variable errno.
logf
Returns the natural logarithm of x.
logf returns a non-numeric value and sets macro EDOM to global variable errno if x is
negative. If x is zero, it returns -∞ (0xff800000) and sets macro ERANGE to global
variable errno.
log2f
Returns the logarithm of x with base 2.
log2f returns a non-numeric value and sets macro EDOM to global variable errno if x is
negative. If x is zero, it returns -∞ and sets macro ERANGE to global variable errno.
log10f
Returns the logarithm of x with base 10.
log10f returns a non-numeric value and sets macro EDOM to global variable errno if x is
negative. If x is zero, it returns -∞ and sets macro ERANGE to global variable errno.
powf
Returns the yth power of x.
powf returns a negative solution only if x < 0 and y is an odd integer. If x < 0 and y is a
non-integer or if x = y = 0, powf returns a non-numeric value and sets the macro EDOM
for the global variable errno. If x = 0 and y < 0 or if an overflow occurs, powf returns
+HUGE_VAL and sets the macro ERANGE for errno. If the solution vanished
approaching zero, powf returns +0 and sets the macro ERANGE for errno. If the solution
is a non-normalized number, powf sets the macro ERANGE for errno.
cbrtf
Returns the cubic root of x.
sqrtf
Returns the positive square root of x.
sqrtf returns a non-numeric value and sets macro EDOM to global variable errno if x is a
negative real number.
The error processing of these functions can be changed by using the matherr function.
[Example]
#include <math.h>
float func(void)
{
float
ret, x, y;
ret = powf(x, y);
:
return(ret);
/* Returns yth power of x to ret. */
}
264
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
FLOORF
[Overview]
ceiling/absolute value/floor/remainder function
ceilf, fabsf, floorf, fmodf
[Syntax]
#include <math.h>
float
ceilf(float x)
float
fabsf(float x)
float
floorf(float x)
float
fmodf(float x, float y)
[Description]
ceilf(float x)
This function calculates the minimum integer value greater than x.
fabsf(float x)
This function fabsf (float x) calculates the absolute value (size) of x by directly manipulating the bit
representation of x.
floorf(float x)
This function calculates the maximum integer value less than x.
fmodf(float x, float y)
This function calculates a floating-point value that is the remainder resulting from dividing x by y. In other
words, it calculates the value "x - i * y" for the minimum integer i that has a sign the same as x and is less than
y, if y is not zero
[Return value]
ceilf
Returns the minimum integer greater than x.
fabsf
Returns the absolute value (size) of x.
floorf
Returns the maximum integer value less than x.
fmodf
Returns a floating-point value that is the remainder resulting from dividing x by y.
fmodf(x, 0) returns x.
The error processing of these functions can be changed by using the matherr function.
User’s Manual U18513EJ1V0UM
265
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <math.h>
float func(void)
{
float
ret, x, y;
ret = fmodf(x, y);
:
return(ret);
/* Returns remainder resulting from dividing x by y to ret. */
}
266
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
FREXPF
[Overview]
Manipulation of each part of floating-point number
frexpf, ldexpf, modff
[Syntax]
#include <math.h>
float
frexpf(float val, int *exp)
float
ldexpf(float val, int exp)
float
modff(float val, float *ipart)
[Description]
All numbers other than zero can be expressed as m x 2p.
frexpf(float val, int *exp)
This function expresses val of float type as mantissa m and the pth power of 2. The resulting mantissa m is
0.5 <= |x| < 1.0, unless val is zero. p is stored in *exp. m and p are calculated so that val = m x 2p.
ldexpf(float val, int exp)
This function calculates val x 2exp.
modff(float val, float *ipart)
This function divides val of float type into integer and decimal parts, and stores the integer part in *ipart.
Rounding is not performed. It is guaranteed that the sum of the integer part and decimal part accurately
coincides with val.
For example, where realpart = modff (val, &intpart), "realpart + intpart" coincides with val.
[Return value]
frexpf
Returns mantissa m.
frexpf sets 0 to *exp and returns 0 if val is 0. Although the value of val can be changed
by using the matherr function, the setting of *exp cannot be changed.
ldexpf
Returns the value calculated by val x 2exp.
lIf an underflow or overflow occurs as a result of executing ldexpf, macro ERANGE is set
to global variable errno. If an underflow occurs, ldexpf returns an unnormalized value. If
an overflow occurs, it returns ∞ (+∞ = 0x7f800000, -∞ = 0xff800000) with the same sign
as HUGE_VAL.
This error processing can be changed by using the matherr function.
modff
Returns a decimal part. The sign of the result is the same as the sign of val.
User’s Manual U18513EJ1V0UM
267
CHAPTER 6 LIBRARY FUNCTION
[Example]
#include <math.h>
float func(void)
{
float
ret, x;
int
exp;
x = 5.28;
ret = frexpf(x, &exp);
/* Resultant mantissa 0.66 is returned to ret, */
/* and 3 is stored in exp */
:
return(ret);
}
268
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
GAMMAF
[Overview]
Logarithmic gamma function
gammaf
[Syntax]
#include <math.h>
float
gammaf(float x)
[Description]
gammaf(float x)
This function calculates In(Γ(x)), i.e., the natural logarithm of the gamma function of x. The gamma function
(expf (gammaf(x)) is a generalized factorial, and has a relational expression of Γ(N) = N x Γ(N - 1). Therefore,
the result of the gamma function itself increases very rapidly. Consequently, gammaf is defined as "In(Γ(x))",
instead of simply "Γ(x)", to expand the valid range of the result that can be expressed.
[Return value]
The natural logarithm of the gamma function of x is returned.
If x is 0 or an overflow occurs, HUGE_VAL is returned, and macro ERANGE is set to global variable errno.
This error processing can be changed by using the matherr function.
[Example]
#include <math.h>
float func(float x)
{
float
ret;
ret = gammaf(x);
:
return(ret);
/* Returns natural logarithm of gamma function of x to ret. */
}
User’s Manual U18513EJ1V0UM
269
CHAPTER 6 LIBRARY FUNCTION
HYPOTF
[Overview]
Euclidean distance function
hypotf
[Syntax]
#include <math.h>
float
hypotf(float x, float y)
[Description]
hypotf(float x, float y)
This function calculates a Euclidean distance
2
2
x + y between the origin (0, 0) and a point indicated by Cartesian coordinates (x, y).
[Return value]
Returns a Euclidean distance
2
2
x + y between the origin (0, 0) and a point indicated by Cartesian coordinates (x, y).
If an overflow occurs, HUGE_VAL is returned, and macro ERANGE is set to global variable errno.
This error processing can be changed by using the matherr function.
[Example]
#include <math.h>
float func(float x)
{
float
ret, y;
ret = hypotf(x, y);
/* Returns Euclidean distance between origin (0, 0) */
/* and coordinates (x, y) to ret. */
:
return(ret);
}
270
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
MATHERR
[Overview]
Error processing function
matherr
[Syntax]
#include <math.h>
int
matherr(struct exception *e)
[Description]
matherr(struct exception *e)
This is a function that is called if an error occurs in a mathematical library function. By preparing a function
named matherr via a user subroutine, therefore, error processing can be customized. Customized matherr
must return 0 if resolution of an error has failed, and a value other than 0 if the error has been resolved. If
matherr returns a value other than 0, the value of global variable errno is not changed.
Error processing can be customized by using the information passed by pointer *e to structure exception.
Structure exception is defined as follows in "math.h".
#if !defined(_ _ cplusplus)
#define _ _ exception exception
#endif
struct exception{
int
type;
char
*name;
double arg1, arg2, retval;
};
The meaning of each member is as follows:
type
Type of mathematical function error that has occurred.
The type of the macro encoding error is also defined in "math.h".
name
Pointer indicating a character string that holds the name of the mathematical library
function in which an error has occurred, and ends with a space character.
arg1, arg2
Arguments responsible for the error.
retval
Error return value that is returned by the calling function.
User’s Manual U18513EJ1V0UM
271
CHAPTER 6 LIBRARY FUNCTION
The types of mathematical library function errors that may occur are as follows.
DOMAIN
The argument is not in the range of the definition area of the function.
Example: logf(-1)
OVERFLOW
Overflow
Example: expf(1000)
UNDERFLOW
Underflow, solutions to non-normalized number.
Solution < 1.1755e-38 and non 0 and precision is lower than the normal value.
Z_DIVISION
Zero division.
Calling matherr when an operation exception occurs and updating global variable errno with a standard
function are not re-entrant.
[Return value]
By changing the value of e ->retval, the result of the function called from the customized matherr can be
changed. This also applies to the function on the calling side.
The matherr returns a value other than 0 if the error has been resolved, and 0 if the error could not be
resolved. If matherr returns 0, set an appropriate value to global variable errono on the calling side.
[Example]
#include <math.h>
#include <stdio.h>
float func(void)
{
float
ret;
ret = logf(-0.1);
:
return(ret);
}
/* 3 is returned to ret. */
int matherr(struct exception *e)
{
char
s[30];
switch(e->type) {
case DOMAIN:
sprintf(s, "%s DOMAIN error %e\n", e->name, e->arg1);
e->retval = 3;
/* Changes error return value to 3. */
break;
default:
sprintf(s, "%s other error %e\n", e->name, e->arg1);
}
return(1);
}
272
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
SINHF
[Overview]
Hyperbolic functions
acoshf, asinhf, atanhf, coshf, sinhf, tanhf
[Syntax]
#include <math.h>
float
acoshf(float x)
float
asinhf(float x)
float
atanhf(float x)
float
coshf(float x)
float
sinhf(float x)
float
tanhf(float x)
[Description]
acoshf(float x)
This function calculates the inverse hyperbolic cosine of x (where x is a numeric value of 1 or greater). The
definition expression is as follows.
2
ln ( x + x – 1 )
asinhf(float x)
This function calculates the inverse hyperbolic sine of x. The definition expression is as follows.
2
sign(x) x ln ( x + 1 + x )
atanhf(float x)
This function calculates the inverse hyperbolic tangent of x.
coshf(float x)
This function calculates the hyperbolic cosine of x. Specify the angle in radian. The definition expression is
as follows.
x
–x
(e + e )
------------------------2
User’s Manual U18513EJ1V0UM
273
CHAPTER 6 LIBRARY FUNCTION
sinhf(float x)
This function calculates the hyperbolic sine of x. Specify the angle in radian. The definition expression is as
follows.
x
–x
(e – e )
------------------------2
tanhf(float x)
This function calculates the hyperbolic tangent of x. Specify the angle in radian. The definition expression is
as follows.
sinh(x) / cosh(x)
[Return value]
acoshf
Returns the inverse hyperbolic cosine of x (x is a numeric number of 1 or greater).
acoshf returns a non-numeric value if x is less than 1. Macro EDOM is set to global
variable errno.
asinhf
Returns the inverse hyperbolic sine of x.
atanhf
Returns the inverse hyperbolic tangent of x.
atanhf returns a non-numeric value and sets macro EDOM to global variable errno if the
absolute value of x is greater than 1.
coshf
Returns the hyperbolic cosine of x.
coshf returns HUGE_VAL and sets macro ERANGE to global variable errno if an
overflow occurs.
sinhf
Returns the hyperbolic sine of x.
sinhf returns HUGE_VAL and sets macro ERANGE to global variable errno if an
overflow occurs.
tanhf
Returns the hyperbolic tangent of x.
The error processing of these functions can be changed by using the matherr function.
[Example]
#include <math.h>
float func(float x)
{
float
ret;
ret = acoshf(x);
:
return(ret);
}
274
/* Returns value of inverse hyperbolic cosine of x to ret. */
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
TRIG
[Overview]
Trigonometric functions
acosf, asinf, atanf, atan2f, cosf, sinf, tanf
[Syntax]
#include <math.h>
float
acosf(float x)
float
asinf(float x)
float
atanf(float x)
float
atan2f(float y, float x)
float
cosf(float x)
float
sinf(float x)
float
tanf(float x)
[Description]
acosf(float x)
This function calculates the inverse cosine (arcosine) of x. Specify x as, -1<= x <= 1.
asinf(float x)
This function calculates the inverse sine (arcsine) of x. Specify x as, -1<= x <= 1.
atanf(float x)
This function calculates the inverse tangent (arctangent) of x.
atan2f(float y, float x)
This function calculates the inverse tangent of y/x. atan2f calculates the correct result even if the angle is in
the vicinity of π/2 or - π/2(if x is close to 0).
cosf(float x)
This function calculates the cosine of x. Specify the angle in radian.
sinf(float x)
This function calculates the sine of x. Specify the angle in radian.
tanf(float x)
This function calculates the cosine of x. Specify the angle in radian.
User’s Manual U18513EJ1V0UM
275
CHAPTER 6 LIBRARY FUNCTION
[Return value]
acosf
Returns the inverse cosine (arccosine) of x. The returned value is in radian and in a
range of 0 to π.
If x is not between -1 and 1, a non-numeric value is returned, and macro EDOM is set to
global variable errno.
asinf
Returns the inverse sine (arcsine) of x. The returned value is in radian and in a range of
-π/2 to π/2.
If x is not between -1 and 1, a non-numeric value is returned, and macro EDOM is set to
global variable errno.
atanf
Returns the inverse tangent (arctangent) of x. The returned value is in radian and in a
range of -π/2 to π/2.
atan2f
Returns the inverse tangent (arctangent) of y/x. The returned value is in radian and in a
range of -π to π.
atan2f returns a non-numeric value and sets macro EDOM to global variable errno if
both x and y are 0.0. If the solution vanished approaching zero, atan2f returns +0 and
sets macro ERANGE to global variable errno. If the solution is a non-normalized
number, atan2f sets macro ERANGE to global variable errno.
cosf
Returns the cosine of x.
sinf
Returns the sine of x.
tanf
Returns the tangent of x.
The error processing of these functions can be changed by using the matherr function.
[Example]
#include <math.h>
float func(float x)
{
float
ret;
ret = atanf(x);
:
return(ret);
/* Returns value of arctangent of x to ret. */
}
276
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
6.12
Runtime Library
This section explains the runtime library.
The microcontroller architecture of the V850 microcontrollers does not have instructions for multiplying or
dividing and performing floating-point operations on 32-bit data. Therefore, to satisfy the language specifications
of the ANSI standards, the CA850 performs multiplication, division, residue calculations, and all floating-point
operations on 32-bit data by calling the runtime library contained in the libc.a file. The runtime library can also be
called when creating a new assembler for the V850 microcontrollers.
However, with the V850Ex, the compiler does not use the runtime library for multiplying, dividing, and residue
calculating 32-bit data. It uses the runtime library for floating-point operations.
The runtime library is a routine automatically used when the compiler executes compiling. This library is
included in the libc.a file along with the standard library. The header file does not need to be included.
When using the runtime library for an application program, libc.a must be referenced by ld850 when an
executable object file is created.
Figure 6 - 1 Image of Using Runtime Library
.s
Floating-point
operation
as850
.o
ld850
a.out
jarl xxx
libc.a
Runtime Library
User’s Manual U18513EJ1V0UM
277
CHAPTER 6 LIBRARY FUNCTION
Table 6 - 25 Runtime Library
Classification
Function Name
Outline
ADDF.S
__ _addf.s
Addition of single-precision floating-point
CMPF.S
__ _cmpf.s
Comparison of single-precision floating-point and change of flag
CVT.WS
__ _cvt.ws
Conversion from integer to single-precision floating-point number
DIV
__ _div
Division of signed 32-bit integer
__ _divu
Division of unsigned 32-bit integer
DIVF.S
__ _divf.s
Division of single-precision floating-point
MOD
__ _mod
Remainder of signed 32-bit integer
__ _modu
Remainder of unsigned 32-bit integer
__ _mul
Multiplication of signed 32-bit integer
__ _mulu
Multiplication of unsigned 32-bit integer
MULF.S
__ _mulf.s
Multiplication of single-precision floating-point
SUBF.S
__ _subf.s
Subtraction of single-precision floating-point
TRNC.SW
__ _trnc.sw
Conversion from single-precision floating-point number to integer
MUL
[Cautions]
(1)
The runtime library is originally used by code generation part (cgen) and is not assumed to be used
alone.Therefore, preprocessing to call the runtime library is necessary when it is used for an assemblylanguage source program.
(2)
The runtime library cannot be used with a C language source program.
(3)
The default processing of the compiler does not use the runtime library’s _ __mul/_ __mulu functions for
multiplication and ___div/___divu functions for division to process integer data of 16 bits or shorter.
Instead, the mulh and divh instructions are used. If the -Xe option is specified with the compiler, the runtime
library is used to process integer data of 16 bits or shorter.
In this case, if the runtime library is used, multiplication/division processing strictly conforming to the ANSI
standards is executed, but the execution speed is slower than when using the mulh and divh instructions.
278
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
ADDF.S
[Overview]
Addition of single-precision floating-point
___ addf.s
[Syntax]
jarl
_ _ _ ad d f . s , lp
[Description]
___ addf.s
This function adds single-precision floating-points.
This function is used by the ca850 as the entity of arithmetic operator "+" of the C language for a singleprecision floating-point number. It is not used for addition of integers.
[Preprocessing]
When using this function for an assembler, general-purpose registers r6 and r7 must be saved, and values
must be substituted into r6 and r7 as arguments.
[Argument setting register]
r6, r7
[Return value]
The result of the addition is set to r6.
[Example]
In the case of "value of reg2 + value of reg1" (reg2 is other than r6 and r7)
add
st.w
st.w
mov
mov
jarl
mov
ld.w
ld.w
add
-8, sp
r6,[sp]
r7, 4[sp]
reg1, r6
reg2, r7
_ _ _ addf.s,lp
r6, reg2
4[sp], r7
[sp], r6
8, sp
- - Saves r6 and r7.
- - Substitutes value as argument.
- - Calls function.
- - Stores result of addition in reg2.
- - Restores r6 and r7.
User’s Manual U18513EJ1V0UM
279
CHAPTER 6 LIBRARY FUNCTION
CMPF.S
[Overview]
Comparison of single-precision floating-point and change of flag
___ cmpf.s
[Syntax]
jarl
_ _ _cmpf.s, lp
[Description]
___ cmpf.s
This function compares single-precision floating-point numbers, and changes flags S and Z according to the
result of the comparison. Changes to the flags are then reflected in the passed PSW, and the PSW is
changed.
[Preprocessing]
When using this function for an assembler, general-purpose registers r6 through r8 must be saved, and
values to be compared as arguments must be substituted into r6 and r7. Moreover, the value of the PSW must
be passed to r8.
This function changes the flags depending on the result of "r7 - r6".
[Argument setting register]
r6, r7, r8
[Return value]
The contents of the PSW are changed depending on the result of the comparison, and the value of the PSW
is set to r6.
[Flag]
280
CY
1 if the result of comparison is negative; otherwise, 0
OV
0
S
1 if the result of comparison is negative; otherwise, 0
Z
1 if the result of comparison is zero; otherwise, 0
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
In the case of "comparing value of reg2 and value of reg1" (reg2 is other than r6 through r8)
add
st.w
st.w
st.w
mov
mov
stsr
jarl
mov
ld.w
ld.w
ld.w
add
-12, sp
r6, [sp]
r7, 4[sp]
r8, 8[sp]
reg1, r6
reg2, r7
5, r8
_ _ _ cmpf.s,lp
r6, reg2
8[sp], r8
4[sp], r7
[sp], r6
12, sp
- - Saves r6 through r8.
- - Substitutes value as argument.
-----
Passes value of PSW to r8.
Calls function
Stores changed PSW value in reg2.
Restores r6 through r8.
User’s Manual U18513EJ1V0UM
281
CHAPTER 6 LIBRARY FUNCTION
CVT.WS
[Overview]
Conversion from integer to single-precision floating-point number
___ cvt.ws
[Syntax]
jarl
_ _ _cvt.ws, lp
[Description]
___ cvt.ws
This function converts an integer to a single-precision floating-point number.
[Preprocessing]
When using this function for an assembler instruction, general-purpose register r6 must be saved and a
value to be converted as an argument must be saved to r6.
[Argument setting register]
r6
[Return value]
The converted value is set to r6.
[Example]
In the case of "converting value of reg1 and storing the result of conversion in reg2" (reg2 is other than r6)
add
st.w
mov
jarl
mov
ld.w
add
282
-4, sp
r6, [sp]
reg1, r6
_ _ _ cvt.ws,lp
r6, reg2
[sp], r6
4, sp
------
Saves r6.
Substitutes function as argument.
Calls function
Stores value resulting from conversion in reg2.
Restores r6.
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
DIV
[Overview]
Division of 32-bit integer
___ div, _ __divu
[Syntax]
jarl
_ _ _div, lp
jarl
_ _ _divu, lp
[Description]
___ div
This function executes division of signed 32-bit integers.
___ divu
This function executes division of unsigned 32-bit integers.
These functions are used by the CA850 as the entities of the arithmetic operator "/" of the C language.
[Preprocessing]
When these functions are used for an assembler instruction, general-purpose registers r6 and r7 must be
saved, and values must be substituted into r6 and r7 as arguments. These functions execute division,
assuming that "r7/r6".
[Argument setting register]
r6, r7
[Return value]
The lower 32 bits of the result of division are set to r6. The remainder is ignored.
User’s Manual U18513EJ1V0UM
283
CHAPTER 6 LIBRARY FUNCTION
[Example]
In the case of "value of reg2/value of reg1" (reg2 is other than r6 and r7
add
st.w
st.w
mov
mov
jarl
mov
ld.w
ld.w
add
284
-8, sp
r6, [sp]
r7, 4[sp]
reg1, r6
reg2, r7
_ _ _ div, lp
r6, reg2
4[sp], r7
[sp], r6
8, sp
- - Saves r6 and r7.
- - Substitutes value as argument.
- - Calls function.
- - Stores result of division in reg2.
- - Restores r6 and r7.
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
DIVF.S
[Overview]
Division of single-precision floating-point
___ divf.s
[Syntax]
jarl
_ _ _divf.s, lp
[Description]
___ divf.s
This function executes division of single-precision floating-points.
This function is used by the CA850 as the entity of the arithmetic operator "/" of the C language in singleprecision floating-point numbers. It is not used for division of integers.
[Preprocessing]
When this function is used for an assembler instruction, general-purpose registers r6 and r7 must be saved,
and values must be substituted into r6 and r7 as arguments. This function executes division, assuming that
"r7/r6".
[Argument setting register]
r6, r7
[Return value]
The result of division is set to r6.
[Example]
In the case of "value of reg2/value of reg1" (reg2 is other than r6 and r7
add
st.w
st.w
mov
mov
jarl
mov
ld.w
ld.w
add
-8, sp
r6, [sp]
r7, 4[sp]
reg1, r6
reg2, r7
_ _ _ divf.s,lp
r6, reg2
4[sp], r7
[sp], r6
8, sp
- -Saves r6 and r7.
- - Substitutes value as argument.
- - Calls function.
- -Stores result of division in reg2.
- - Restores r6 and r7.
User’s Manual U18513EJ1V0UM
285
CHAPTER 6 LIBRARY FUNCTION
MOD
[Overview]
Remainder of 32-bit integer
___ mod, _ __modu
[Syntax]
jarl
_ _ _mod, lp
jarl
_ _ _modu, lp
[Description]
___ mod
This function calculates the remainder resulting from division of signed 32-bit integers.
___ modu
This function calculated the remainder resulting from division of unsigned 32-bit integers.
These functions are used by the CA850 as the entities of the arithmetic operator "%" of the C language.
[Preprocessing]
When these functions are used for an assembler instruction, general-purpose registers r6 and r7 must be
saved, and values must be substituted into r6 and r7 as arguments. These functions execute division,
assuming that "r7%r6".
[Argument setting register]
r6, r7
[Return value]
The remainder resulting from division is set to r6.
286
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
[Example]
In the case of "value of reg2%value of reg1" (reg2 is other than r6 and r7)
add
st.w
st.w
mov
mov
jarl
mov
ld.w
ld.w
add
-8, sp
r6, [sp]
r7, 4[sp]
reg1, r6
reg2, r7
_ _ _ mod, lp
r6, reg2
4[sp], r7
[sp], r6
8, sp
- - Saves r6 and r7.
- - Substitutes value as argument.
- - Calls function.
- - Stores remainder resulting from division in reg2.
- - Restores r6 and r7.
User’s Manual U18513EJ1V0UM
287
CHAPTER 6 LIBRARY FUNCTION
MUL
[Overview]
Multiplication of 32-bit integer
___ mul, _ __mulu
[Syntax]
jarl
_ _ _mul, lp
jarl
_ _ _mulu, lp
[Description]
___ mul
This function executes multiplication of signed 32-bit integers.
___ mulu
This function executes multiplication of unsigned 32-bit integers.
These functions are used by the CA850 as the entities of the arithmetic operator "*" of the C language.
[Preprocessing]
When these functions are used for an assembler struction, general-purpose registers r6 and r7 must be
saved, and values must be substituted into r6 and r7 as arguments.
[Argument setting register]
r6, r7
[Return value]
The lower 32 bits of the result of multiplication are set to r6. The higher 32 bits are invalid.
[Example]
In the case of "value of reg1 * value of reg2" (reg2 is other than r6 and r7)
add
st.w
st.w
mov
mov
jarl
mov
ld.w
ld.w
add
288
-8, sp
r6, [sp]
r7, 4[sp]
reg1, r6
reg2, r7
_ _ _ mul, lp
r6, reg2
4[sp], r7
[sp], r6
8, sp
- - Saves r6 and r7.
- - Substitutes value as argument.
- - Calls function.
- - Stores result of multiplication in reg2.
- - Restores r6 and r7.
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
MULF.S
[Overview]
Multiplication of single-precision floating-point
___ mulf.s
[Syntax]
jarl
_ _ _mulf.s, lp
[Description]
___ mulf.s
This function executes multiplication of single-precision floating-points.
This function is used by the CA850 as the entity of the arithmetic operator "*" of the C language in singleprecision floating-point numbers. It is not used for multiplication of integers.
[Preprocessing]
When this function is used for an assembler instruction, general-purpose registers r6 and r7 must be saved,
and values must be substituted into r6 and r7 as arguments.
[Argument setting register]
r6, r7
[Return value]
The result of multiplication is set to r6.
[Example]
In the case of "value of reg2 * value of reg1" (reg2 is other than r6 and r7)
add
st.w
st.w
mov
mov
jarl
mov
ld.w
ld.w
add
-8, sp
r6, [sp]
r7, 4[sp]
reg1, r6
reg2, r7
_ _ _ mulf.s
r6, reg2
4[sp], r7
[sp], r6
8, sp
-- Saves r6 and r7.
-- Substitutes value as argument.
,lp
-- Calls function.
-- Stores result of multiplication in reg2.
-- Restores r6 and r7.
User’s Manual U18513EJ1V0UM
289
CHAPTER 6 LIBRARY FUNCTION
SUBF.S
[Overview]
Subtraction of single-precision floating-point
___ subf.s
[Syntax]
jarl
_ _ _subf.s, lp
[Description]
___ subf.s
This function executes subtraction of single-precision floating-points.
This function is used by the CA850 as the entity of the arithmetic operator "-" of the C language is singleprecision floating-point numbers. It is not used for subtraction of integers.
[Preprocessing]
When this function is used for an assembler instruction, general-purpose registers r6 and r7 must be saved,
and values must be substituted into r6 and r7 as arguments. This function executes subtraction, assuming that
"r7 - r6".
[Argument setting register]
r6, r7
[Return value]
The result of subtraction is set to r6.
[Example]
In the case of "value of reg2 - value of reg1" (reg2 is other than r6 and r7)
add
st.w
st.w
mov
mov
jarl
mov
ld.w
ld.w
add
290
-8, sp
r6, [sp]
r7, 4[sp]
reg1, r6
reg2, r7
_ _ _ subf.s
r6, reg2
4[sp], r7
[sp], r6
8, sp
-- Saves r6 and r7.
-- Substitutes value as argument.
,lp
-- Calls function.
-- Stores result of subtraction in reg2.
-- Restores r6 and r7.
User’s Manual U18513EJ1V0UM
CHAPTER 6 LIBRARY FUNCTION
TRNC.SW
[Overview]
Conversion from single-precision floating-point number to integer
___ trnc.sw
[Syntax]
jarl
_ _ _trnc.sw, lp
[Description]
___ trnc.sw
This function executes conversion from a single-precision floating-point number to an integer.
The result of conversion is rounded toward 0.
[Preprocessing]
When this function is used for an assembler instruction, general-purpose register r6 must be saved, and a
value must be substituted into r6 as an argument.
[Argument setting register]
r6
[Return value]
The value resulting from conversion is set to r6.
[Example]
To "convert value of reg1 and store result in reg2" (reg2 is other than r6)
add
st.w
mov
jarl
mov
ld.w
add
-4, sp
r6, [sp]
reg1, r6
_ _ _ trnc.sw ,lp
r6, reg2
[sp], r6
4, sp
------
Saves r6.
Substitutes value as argument.
Calls function.
Stores result of conversion in reg2.
Restores r6.
User’s Manual U18513EJ1V0UM
291
CHAPTER 7 FOR EFFICIENT USE
CHAPTER 7 FOR EFFICIENT USE
This chapter explains the programming method and how to use the expansion functions for more efficient use
of the CA850.
7.1
volatile Qualifier
When a variable is declared with the volatile qualifier, the variable is not optimized and is not assigned to
registers. Therefore, keep volatile declaration at the minimum required level.
If it is clear that the value of a variable with volatile declared is not changed externally in a specific section, the
variable can be optimized by assigning the unchanged value to a variable for which volatile not declared and
referencing it, which may increase the execution speed.
292
User’s Manual U18513EJ1V0UM
CHAPTER 7 FOR EFFICIENT USE
7.2
Declaration of Function Without Return Value
If a function without a return value is not declared as void type, an unwanted return processing code is
generated.
Be sure to declare functions without return values as void type
User’s Manual U18513EJ1V0UM
293
CHAPTER 7 FOR EFFICIENT USE
7.3
Pointers and Optimization
The CA850 executes analysis of pointers when the -Og/-O/-Os/-Ot options are specified and optimization is
executed. If the optimization level is lower than that, pointer analysis is not executed. Consequently, if indirect
memory access using a pointer exists, processing is performed on the assumption that all the variables are
accessed by this indirect memory access, and optimization and register allocation cannot be executed efficiently.
Even for size priority optimization or execution speed priority optimization, the same phenomenon may occur
when this indirect memory access using a global pointer or a pointer argument exists. Use the global pointer as
locally as possible.
For example, optimization is not executed in the case of Example 1 below, but optimization is executed
efficiently when a local variable is used as shown in Example 2.
Example 1
int*
int
sp;
s1, s2, s3;
void func(void)
{
int a = 0;
int b = 1;
*sp = s1 / s2 * 100;
if(s1 == 0){
s3 = *sp + a ;
}
else {
s3 = *sp - b;
}
}
294
User’s Manual U18513EJ1V0UM
CHAPTER 7 FOR EFFICIENT USE
Example 2
int*
int
sp;
s1, s2, s3;
void func(void)
{
int a = 0;
int b = 1;
register int tmp = s1 / s2 * 100;
if(s1 == 0) {
s3 = tmp + a; /* a is replaced by 0 */
}
else {
s3 = tmp - b; /* b is replaced by 1 */
}
*sp = tmp;
}
User’s Manual U18513EJ1V0UM
295
CHAPTER 7 FOR EFFICIENT USE
7.4
Assembler Code and Optimization
If descriptions of assembler directives (refer to "3.4
Describing Assembler Instruction") are included,
processing is performed on the assumption that the values of all the variables are used and changed in that
code. Therefore, optimization is not performed beyond the assembler code.
To avoid a drop in the processing efficiency, use functions including assembler code as little as possible. If the
execution speed priority optimization is specified, and if a function including assembler code that defines a label
is used, the same label will be defined at the parts of function definition and inline expansion. In this case, a label
multiple definition error will occur
296
User’s Manual U18513EJ1V0UM
CHAPTER 7 FOR EFFICIENT USE
7.5
7.5.1
Registers
Register specifier
When the debug priority optimization (-Od) option is specified, a variable declared with the register specifier is
assigned to a register variable register, taking precedence over a variable without the register specifier. Even a
variable declared with the register specifier, however, is not assigned to a register if the variable is not referenced
many times. Therefore, the code efficiency will not deteriorate.
Even if a variable is defined with the register specifier and referenced by the debugger, the expected value
may not be obtained. This is due to the number of variables declared with the register specifier or optimization by
the CA850. If the number of register variables is less than the number of variables declared with the register
specifier, the variables are not assigned to registers.
The number of register variable registers in each mode is as follows.
-
22-register mode: 5 (r25 - r29)
-
26-register mode: 7 (r23 - r29)
-
32-register mode: 10 (r20 - r29)
For example, even if six or more variables are declared with the register specifier in the 22-register mode, all
the variables are not stored in registers.
The debug priority optimization (-Od) option gives priority to the variables declared with the register specifier
and assigns these variables to register variable registers. However, a variable for which the register specifier is
specified is not assigned to register if it is referenced only a few times. If an option other than the -Od option is
specified, variables that are relatively frequently referenced are assigned to registers.
Because the variables are assigned to different places in this way, the symbol information may be affected and
therefore the variables cannot be referenced from the debugger. If there is no problem in the result of an
operation that uses variables that cannot be referenced, it is considered that the operation is performed correctly.
7.5.2
Static variables and external variables
When the debug priority optimization (-Od) option or default optimization (-Ob) option is specified, a static
variable or external variable is not assigned to a register when registers are allocated. When these variables are
frequently used in a function and when the values of these variables are not changed by a function call or asm
declaration in that function, the registers are used more frequently and the speed is expected to increase if the
value of that static variable or external variable is substituted at the beginning of the function to an automatic
variable declared with the register specifier and if the value is returned to the variable at the end of the function.
By specifying the following options, even static variables and external variables can be assigned to registers if
they are relatively frequently referenced.
-
Default optimization (-Ob) option
-
Standard optimization (-Og) option
-
Level 1 advanced optimization (-O) option
-
Level 2 advanced optimization (object size) [-Os] option
-
Level 2 advanced optimization (execution speed) [-Ot] option
User’s Manual U18513EJ1V0UM
297
CHAPTER 7 FOR EFFICIENT USE
7.5.3
Argument of function in K&R format
If an argument with a type smaller in size that int type exists in a function definition in the K&R format, the
argument is not allocated to a register, even if object size priority optimization or execution speed priority
optimization is executed, unless a dummy argument is declared with the register specifier. To allocate this
argument to a register, make a declaration with the register specifier. To describe a function definition in the K&R
format, avoid using char, signed char, unsigned char, short, signed short, and unsigned short as the type of the
argument.
7.5.4
Optimum number of local variables to be assigned
Keep the number of local variables (auto variables) to within 10; or preferably to six or seven. Local variables
are assigned to registersNote. The CA850 allows a total of 20 registers, 10 work registers and 10 register variable
registers, to be used for variables (in the 32-bit register mode). It is recommended to use many local variables if
processing in one function takes time. If processing does not take much time, use only the 10 work registers
whenever possible.
The register variable registers require overhead when they are saved or restored. The CA850 automatically
judges whether register variables are to be used or not. Therefore, the efficiency can be enhanced if six to seven
registers are used for local variables and the other three to four registers are used for work by the CA850.
Note
Non-volatile variables that do not use addresses are subject to assignment. Therefore, the local variables that use addresses are secured in the stack area.
7.5.5
Optimum number of arguments to be used for function
Four argument registers, r6 to r9, are available. If the number of arguments is five or more, the stack is used
for the fifth and subsequent arguments. Therefore, keep the number of arguments to within four whenever
possible. If five or more arguments must be used, pass the arguments using the pointer of a structure, in order to
enhance the efficiency.
298
User’s Manual U18513EJ1V0UM
CHAPTER 7 FOR EFFICIENT USE
7.5.6
Other
If a path exists in which a variable may be referenced before a value is set (Example 3), an unwanted transfer
code from memory to register may be generated. Use a variable after setting a value (Example 4).
Example 3
int
s;
void func(int x)
{
int y;
int i;
for(i = x; i < 10; i++) {
if(i == 3) {
y = 10;
}
}
s = y * y * x;
}
Example 4
int
s;
void func(int x)
{
int y = 0;
int i;
for(i = x; i < 10; i++) {
if(i == 3) {
y = 10;
}
}
s = y * y * x;
}
User’s Manual U18513EJ1V0UM
299
CHAPTER 7 FOR EFFICIENT USE
7.6
Stack Size
The compiler allocates one variable to one stack area. Two or more variables cannot be allocated to the same
area. By selecting and using variables for specific purposesNote, the stack size can be reduced.
Note
300
In this case, however, the program may become difficult to read.
User’s Manual U18513EJ1V0UM
CHAPTER 7 FOR EFFICIENT USE
7.7
Aligning Data
Declare data definitions collectively starting from the longest data.
With the V850 microcontrollers, word data such as int type must be aligned at a word boundary, and halfword
data such as short type must be aligned at a halfword boundary.
Consequently, a padding area is generated to enable alignment for the following source.
Higher address
char
short
int
char
int
c
s
i
d
j
=
=
=
=
=
'a';
0;
1;
'b';
2;
j
d
i
s
-
c
Lower address
To prevent the generation of a padding area like this, declare data starting from the longest data.
Higher address
int
int
short
char
i
j
s
c
=
=
=
=
1;
2;
0;
'a';
char
d = 'b';
d
c
s
j
i
Lower address
User’s Manual U18513EJ1V0UM
301
CHAPTER 7 FOR EFFICIENT USE
7.8
Data Type
The V850 microcontrollers sign-extends byte data or halfword data to word length depending on the value of
the most significant bit, when the byte data or halfword data is loaded from memory to a register. Consequently,
a code that masks the higher bits may be generatedNote as a result of operating data of unsigned char or
unsigned short type. Use word data as much as possible. When using byte data or halfword data, use a signed
type.
Note
This mask code is not generated by an operation in which data is already stored in registers.
Caution When the V850Ex is used as the target device with the CA850, mask codes are not created because
the architecture of the V850Ex has unsigned load instructions and type conversion instructions.
In the case of a register variable, a shift instruction is generated to extend the sign because an operation of
signed byte data or signed halfword data integer-expands the operandNote.
When storing the result of the operation in a register variable, a shift instruction is generated in the case of
signed byte data or signed halfword data, or a code that masks the higher bits is generated in the case of
unsigned byte data or unsigned halfword data. To prevent generation of this code, use word data (int, long,
unsigned int, or unsigned long type data) as much as possible when using a register variable.
Note
"Integer-expansion" converts values into int type if all the values of the original type can be expressed
by int type; otherwise, the values are converted into unsigned int type.
Using mask register function
With a program in which word data cannot be used and therefore mask codes are generated, the code
size can be reduced by using the mask register function (refer to "2.5 Mask Register").
302
User’s Manual U18513EJ1V0UM
CHAPTER 7 FOR EFFICIENT USE
Examples of instruction generation in the case of byte data, halfword data, and word data are shown below.
Example (Written in C)
int
i, j, k;
unsigned short s, t, u;
unsigned char c, d, e;
void f(void)
{
register int
ri, rj, rk;
register short rs, rt, ru;
register unsigned char ruc, rud, rue;
c = d + e;
s = t + u;
i = j + k;
rs = rt + ru;
ruc = rud + rue;
ri = rj + rk;
}
User’s Manual U18513EJ1V0UM
303
CHAPTER 7 FOR EFFICIENT USE
(Output instructions)
# Byte data:
ld.b
$_d, r10
andi
0xff, r10, r10 - - Mask code
ld.b
$_e, r11
andi
0xff, r11, r11 - - Mask code
add
r11, r10
st.b
r10, $_c
# Halfword data:
ld.h
$_t, r12
andi
0xffff, r12, r12 -- Mask code
ld.h
$_u, r13
andi
0xffff, r13, r13 -- Mask code
add
r13, r12
st.h
r12, $_s
# Word data:
ld.w
$_j,
ld.w
$_k,
add
r15,
st.w
r14,
r14
r15
r14
$_i
# Signed halfword data
mov
r25,r16
shl
16, r16 - sar
16, r16 - mov
r24, r17
shl
16, r17 - sar
16, r17 - add
r17, r16
shl
16, r16 - sar
16, r16 - -
(register variable):
Shift instruction(integer-expansion)
Shift instruction(integer-expansion)
Shift instruction(integer-expansion)
Shift instruction(integer-expansion)
Shift instruction(sign-expansion of operation result)
Shift instruction(sign-expansion of operation result)
# Unsigned byte data (register variable):
mov
r22, r18
add
r21, r18
addi
0xff, r18, r18 - - Mask code
mov
r18, r23
# Word data
mov
add
mov
st.w
304
(register variable):
r28, r19
r27, r19
r19, r29
r14, $_i
User’s Manual U18513EJ1V0UM
APPENDIX A EXPANDED FUNCTIONS OF CC78Kx
APPENDIX A EXPANDED FUNCTIONS OF CC78Kx
This appendix explains the expanded functions of the CC78Kx.
A.1
#pragma Directive
The following #pragma directive compatible with the CC78Kx can be specified in the CA850.
The [78K-compatible] mark indicates as follows:
Invalid unless -cc78K option is specified
[78K-compatible]
Uppercase and lowercase characters of keywords following #pragma are not
distinguished.
(1) Specifying device type
[78K-compatible]
#pragma pc(device-name)
Specify to reference a device file that defines information dependent on the device to be used.
This directive functions in the same manner as the "#pragma cpu device-name" specification and the device
specification option (-cpu) of the CA850.
(2) Validating peripheral I/O register name
[78K-compatible]
#pragma sfr
A peripheral I/O register of the device is accessed using the specified peripheral I/O register name.
This directive functions in the same manner as the #pragma ioreg directive of the CA850.
(3) Disabling interrupts
[78K-compatible]
#pragma di
The function DI() is treated as the embedded function __ DI().
[78K-compatible]
#pragma ei
The function EI() is treated as the embedded function __EI().
User’s Manual U18513EJ1V0UM
305
APPENDIX A EXPANDED FUNCTIONS OF CC78Kx
(4) Specifying CPU stop function
[78K-compatible]
#pragma halt
The function HALT() is treated as the embedded function __ halt().
(5) Specifying no-operation function
[78K-compatible]
#pragma nop
The function NOP() is treated as the function __ nop().
(6) #pragma directives of CC78Kx
The following directives are not compatible with the 78K.
These directives are treated as the #pragma directive in the CA850.
(a)
Specifying interrupt/exception handler
[78K-compatible]
#pragma interrupt interrupt-request-name function-name [stack selection] ...
#pragma vect interrupt-request-name function-name [stack selection] ...
"#pragma interrupt" and "#pragma vect" of the CC78Kx are treated as "#pragma interrupt interruptrequest-name function-name [allocation-method]" in the CA850.
The following message is output if description is made after "[stack selection]" and if that description cannot be.
W2150: unexected character(s) following directive ’directive’
(b)
Specifying section
[78K-compatible]
#pragma section ...
This directive is treated as "#pragma section section-type ["section-name"] [begin | end]" in the CA850.
The following message is output if it is not recognized by the CA850.
W2162: unrecognized pragma directive ’#pragma directive’, ignored
306
User’s Manual U18513EJ1V0UM
APPENDIX A EXPANDED FUNCTIONS OF CC78Kx
(c)
Specification related to memory manipulation
[78K-compatible]
#pragma inline
The CC78Kx expands memcpy, memset, memchr, and memcmp inline, but the CA850 attempts to
expand the specified function inline, so the following message is output.
W2162: unrecognized pragma directive ’#pragma inline’, ignored
(d)
Specifying module name
[78K-compatible]
#pragma name module-name
The CA850 outputs the following message.
W2162: unrecognized pragma directive ’#pragma name’, ignored
(e)
Specifying data insertion function
[78K-compatible]
#pragma opc
Corresponding embedded function
__ OPC( ) ;
The CA850 outputs the following message and stops compiling.
W2162: unrecognized pragma directive ’#pragma opc’, ignored
E2752: cannot call opc function
(f)
Specifying byte address insertion/generation function
[78K-compatible]
#pragma addraccess
Corresponding embedded function
FP_SEG( ) ;
FP_OFF( ) ;
MK_FP( ) ;
The CA850 outputs the following message and stops compiling.
W2162: unrecognized pragma directive ’#pragma addraccess’, ignored
E2752: cannot call addraccess function
User’s Manual U18513EJ1V0UM
307
APPENDIX A EXPANDED FUNCTIONS OF CC78Kx
(g)
Specifying function directly referencing register
[78K-compatible]
#pragma realregister
Corresponding embedded function
__ absa( ) ;
_ _ ashra( ) ;
_ _ clr1cy( ) ; __ coma( ) ;
__ deca( ) ;
__ getax( ) ;
_ _ getcy( ) ;
_ _ inca( ) ;
__ nega( ) ;
__ not1cy( ) ; _ _ rola( ) ;
_ _ geta( ) ;
__ rolca( ) ;
_ _ rora( ) ;
_ _ rorca( ) ;
__ set1cy( ) ; __ seta( ) ;
__ setcy( ) ;
_ _ shla( ) ;
_ _ shra( ) ;
_ _ setax( ) ;
The CA850 outputs the following message and stops compiling.
W2162: unrecognized pragma directive ’#pragma realregister’, ignored
E2752: cannot call realregister function
(h)
Specifying function directly calling self-writing subroutine of firmware
[78K-compatible]
#pragma hromcall
Corresponding embedded function
__ FlashAreaBlankCheck( ) ; _ _ FlashAreaErase( ) ; __ FlashAreaIVerify( ) ;
__ FlashAreaPreWrite( ) ;
_ _ FlashAreaWriteBack( ) ; __ FlashBlockBlankCheck( ) ;
__ FlashBlockErase( ) ;
_ _ FlashBlockIVerify( ) ;
__ FlashBlockPreWrite( ) ;
__ FlashBlockWriteBack( ) ;_ _ FlashByteRead( ) ;
__ FlashByteWrite( ) ;
__ FlashEnv( ) ;
_ _ FlashGetInfo( ) ;
__ FlashSetEnv( ) ;
__ FlashWordWrite( ) ;
_ _ hromcall( ) ;
__ hromcalla( ) ;
__ setsp( ) ;
The CA850 outputs the following message and stops compiling.
W2162: unrecognized pragma directive ’#pragma hromcall’, ignored
E2752: cannot call hromcall function
308
User’s Manual U18513EJ1V0UM
APPENDIX A EXPANDED FUNCTIONS OF CC78Kx
A.2
Assembler Control Instructions
[78K-compatible]
#asm
assembler instructions
#endasm
This instruction is treated as "#pragma asm" - "#pragma endasm" in the CA850.
The following message is output for each instruction.
W2166: recognized pragma directive ’#pragma asm’
W2166: recognized pragma directive ’#pragma endasm’
A.3
Specifying Interrupt/Exception Handler
An interrupt/exception handler is specified in a C-source program by the following #pragma directive and
qualifier (refer to "3.7 Interrupt/Exception Processing Handler").
[78K-compatible]
#pragma interrupt interrupt-request-name function-nameNote [allocation method]
_ _ interrupt_brk function-definition, or function-declaration
The function qualifier __interrupt_brk is treated as specification of the __interrupt function in the CA850.
Note
A.4
C description
Expanded Functions Not Supported
The CA850 outputs a message if an expanded specification of the CC78Kx that is not supported is specified.
[78K-compatible]
_ _ banked1
_ _ banked6
_ _ banked11
callf
norec
__temp
_ _ banked2
_ _ banked7
_ _ banked12
_ _ callf
_ _ pascal
_ _ banked3
_ _ banked8
_ _ banked13
callt
sreg
__ banked4
__ banked9
__ banked14
__ callt
__ sreg
__banked5
__banked10
__banked15
noauto
__sreg1
The CA850 outputs the following message.
W2761: unrecognized specifier ’specifier’, ignored
User’s Manual U18513EJ1V0UM
309
APPENDIX B CAUTIONS
APPENDIX B CAUTIONS
This chapter explains the points to be noted when using the CA850.
(1) Delimiting Folder Path
Both "\" and "/" are regarded as the delimiters of a folder.
(2) Option Specification Sequence
The CA850 has the following restriction concerning the sequence of an option specified when the driver is
started on the command line:
The actual sequence in which an argument passed to a specific module using the -W option and an argument
of an option recognized by the driver are passed during the module startup is not guaranteedNote.
Note
When ld850 is started from the CA850, -lm -lc is passed to ld850 as the default assumption even if the
-W option is not specified. If ld850 is started from the CA850, startup module crtN.o/crtE.o is passed to
ld850 as the default assumption.
Example
> ca850 -cpu 3201 file.o -Wl,-D,dfile.dir -m
The ld850 passed as follows on starting.
ld850 \Install Folder\lib850\r32\crtN.o -o a.out file.o -lm -lc -D dfile.dir m
However, it is assumed that ld850 has already been placed in Install Folder\bin.
Caution When starting the Id850 directly, allocate "-lc" after "-lm" because the mathematical library references
the standard library (refer to "6.11 Mathematical Functions").
(3) Mixing with K&R Format in Function Declaration/Definition
If the K&R format and ANSI standard format exist together in the declaration and definition of a function, an
error may occur on compilation by the CA850 as a result of argument expansion processing in the K&R format.
For example, a function is declared according to the ANSI standard in the example below, but the function is
defined in the K&R format. Consequently, the types of the arguments do not match, and the CA850 outputs a
"function redeclaration" error.
310
User’s Manual U18513EJ1V0UM
APPENDIX B CAUTIONS
Example of error
void func(int a, int b, float c);
/* Declared in ANSI standard format. */
/* Third argument is declared to be of float type. */
:
void func(a, b, c)
int
a, b;
float
c;
{
/* Defined in K&R format. */
/* Third argument is the expanded default of K&R
and so becomes double type.*/
:
}
In the above example, compilation is performed normally if the K&R format is uniformly used by specifying
"void func();" for the function declaration, or if the ANSI standard format is used by specifying "void func(int a, int
b, float c)" for the function definition.
Note, however, that use of the ANSI standard format is recommended in the CA850.
(4) Output of Other Than Position-Independent Codes
Basically, the CA850 outputs codes not dependent on positions (position-independent codes). However, it
outputs the following codes in response to the "initialization statement with an initial value other than a numeric
value for a pointer type variable other than an automatic variable".
Example
/* Described in C */
char
*ptr = "test\n";
# Output codes
.size
LL20, 6
LL20 :
.str
"test\n\0"
.align 4
.globl _ptr, 4
_ptr :
.word
#LL20
-- Absolute address reference of label
When the -Xd option is specified, the CA850 outputs the following warning message and continues compiling
if an initialization statement with an initial value other than a numeric value for a pointer type variable other than
an automatic variable appears.
W2231:Initialization of non-auto pointer using non-number initializer is not
position independent.
User’s Manual U18513EJ1V0UM
311
APPENDIX B CAUTIONS
(5) Count of Derivative Type Qualification for Type Configuration
The CA850 outputs the following error message and continues compiling if derivative type qualificationNote is
performed 17 times or more for the type configuration
E2260: compiler limit : complicated type modifiers [16]
However, compiling may be stopped depending on the number of times the error has occurred.
Note
*(pointer), [ ] (array), and function declarator included in a declarator.
(6) Length of Identifier and Valid Number of Characters
The CA850 outputs the following error message and continues compiling if an external identifier of 1023
characters or more, or an internal identifier of 1024 characters or more is described.
E2117: compiler limit:too long identifier 'symbol' [1022 / 1023]
However, compiling may be stopped depending on the number of times the error has occurred.
The valid number of characters for an identifier name is 1022 from the beginning of the identifier in the case of
an external identifier and 1023 from the beginning in the case of an internal identifier.
(7) Number of Times of Block Nesting
The CA850 outputs the following message if a pair of "{" and "}" are nested 128 times or more.
F2020: compiler limit : scope level too deep [127]
(8) Number of case Labels in switch Statement
The CA850 outputs the following error message and stops compiling if 1026 or more case labels are described
in one switch statement
F2410: compiler limit : too many case labels [1025]
Depending on the number of nesting switch statements, however, the above message is output and compiling is
stopped even if the number of case labels is less than 1025.
(9) Floating-Point Operation Exception in Operation of Constant Expression
The CA850 outputs the following error message and continues compiling if a floating-point operation exception
occurs during the operation of a constant expression.
E2519: exception has occurred at compile time.
However, compiling may be stopped depending on the number of errors that have occurred.
Moreover, depending on the type of exception, inexact, underflow, overflow, division-by-0, or others is output
for exception.
312
User’s Manual U18513EJ1V0UM
APPENDIX B CAUTIONS
(10) Merging Vast/Large-Quantity File
The CA850 merges intermediate language files according to the optimization level.
At this time, the pre-optimizer (popt850) performs processing on memory to speed up the compiling
processing. To merge a vast or large-quantity intermediate language file, therefore, the following error message
may be output because the memory runs short, and the compiler may be abnormally terminated.
F7009: out of memory
In this case, re-compile on the command line by specifying an option that allows the pre-optimizer to perform
processing to reduce the memory consumption (-Wp, -D).
(11) Optimization of Vast File
If object size priority optimization or execution speed priority optimization is executed, the CA850 analyzes the
data flow in function units inside the global optimization module (opt850) for global optimization.
Because this optimization requires a large amount of the memory, if a source file including a vast function is to
be optimized, the CA850 may output the following error message and be abnormally terminated.
F5104: out of memory
If execution speed priority optimization is performed, inline expansion of a function may result in a function with
a vast size. In such a case, lower the optimization level and execute compilation again.
(12) Library File Search by Specifying Option
The CA850 does not display a message even if a specified library file has not been found as a result of a
library file searchNote initiated by an option (-L or -I). However, if the library file name has been directly specified
on the command line or in the command file, a message is displayed.
Note
If the -L option is not specified, the standard folder (folder \lib850 to which CA850 has been installed,
and each register mode folder below that folder) is searched.
Example
> ca850 -cpu 3201 a.c usr.a
F4002: can not open input file "usr.a".
User’s Manual U18513EJ1V0UM
313
APPENDIX B CAUTIONS
(13) volatile qualifier
When a variable is declared with the volatile qualifier, the variable is not optimized and optimization for
assigning the variable to a register is no longer performed. When a variable with volatile specified is
manipulated, a code that always reads the value of the variable from memory and writes the value to memory
after the variable is manipulated is output. The access width of the variable with volatile specified is not changed.
A variable for which volatile is not specified is assigned to a register as a result of optimization and the code
that loads the variable from the memory may be deleted. When the same value is assigned to variables for which
volatile is not specified, the instruction may be deleted as a result of optimization because it is interpreted as a
redundant instruction. The volatile qualifier must be specified especially for variables that access a peripheral I/O
register, variables whose value is changed by interrupt servicing, or variables whose value is changed by an
external source. When a peripheral I/O register is accessed using the #pragma ioreg directive, however, the
CA850 internally outputs a code for which volatile is specified. Therefore, volatile declaration is not necessary.
The following problem may occur if volatile is not specified where it should.
-
The correct calculation result cannot be obtained.
-
Execution cannot exit from a loop if the variable is used in a for loop.
If it is clear that the value of a variable with volatile specified is changed in a specific section, the variable can
be optimized by assigning the unchanged value to a variable for which volatile not specified and referencing it,
which may increase the execution speed.
[Example of source and output code if volatile is not specified]
If volatile is not specified for "variable a", "variable b", and "variable c", these variables are assigned to
registers and optimized. Even if an interrupt occurs in the meantime and the variable value is changed by the
interrupt, for example, the changed value is not reflected.
int a;
int b;
int c;
void func(void)
{
if (a <= 0) {
b++;
} else {
c++;
}
b++;
c++;
}
314
_func:
#@B_PROLOGUE
#@E_PROLOGUE
ld.w $_a, r12
cmp r0, r12
jgt .L2
ld.w $_b, r11
ld.w $_c, r10
add 1, r11
jbr .L3
.L2:
ld.w $_c, r10
ld.w $_b, r11
add 1, r10
.L3:
addi 1, r11, r13
st.w r13, $_b
addi 1, r10, r14
st.w r14, $_c
#@B_EPILOGUE
jmp [lp]
#@E_EPILOGUE
User’s Manual U18513EJ1V0UM
APPENDIX B CAUTIONS
[Example of source and output code if volatile is specified]
If volatile is specified for "variable a", "variable b", and "variable c", a code that always reads the values of
these variables from memory and writes them to memory after the variables are manipulated is output. Even if
an interrupt occurs in the meantime and the values of the variables are changed by the interrupt, for example,
the result in which the change is reflected can be obtained. (In this case, interrupts may have to be disabled
while the variables are manipulated, depending on the timing of the interrupt.)
When volatile is specified, the code size increases compared with when volatile is not specified because the
memory has to be read and written.
volatile int a;
volatile int b;
volatile int c;
void func(void)
{
if (a <= 0) {
b++;
} else {
c++;
}
b++;
c++;
}
_func:
#@B_PROLOGUE
#@E_PROLOGUE
.option volatile
ld.w $_a, r10
.option novolatile
cmp r0, r10
jgt .L2
.option volatile
ld.w $_b, r11
.option novolatile
add 1, r11
.option volatile
st.w r11, $_b
.option novolatile
jbr .L3
.L2:
.option volatile
ld.w $_c, r12
.option novolatile
add 1, r12
.option volatile
st.w r12, $_c
.option novolatile
.L3:
.option volatile
ld.w $_b, r13
.option novolatile
add 1, r13
.option volatile
st.w r13, $_b
.option novolatile
.option volatile
ld.w $_c, r14
.option novolatile
add 1, r14
.option volatile
st.w r14, $_c
.option novolatile
#@B_EPILOGUE
jmp [lp]
#@E_EPILOGUE
User’s Manual U18513EJ1V0UM
315
APPENDIX B CAUTIONS
(14) Extra Brackets in Function Declaration
If extra brackets "( )" are described in the function declaration, ANSI-C prescribes their handling as shown
below, but the CA850 outputs an error.
Example
typedef int Int;
void f1((Int));
[Prescription in ANSI-C]
In a parameter declaration, a single typedef name in parentheses is taken to be an abstract declarator that
specifies a function with a single parameter, not as redundant parentheses around the identifier for a
declarator.
The above example is therefore interpreted according to ANSI-C.
void f(int (*)(int));
If the code includes extra brackets, delete the unnecessary brackets as shown below.
Example
typedef int Int;
void f1(Int);
316
User’s Manual U18513EJ1V0UM
APPENDIX C INDEX
APPENDIX C INDEX
Symbols
#pragma Directive ... 305
#pragma section Directive ... 58
Example ... 65
Device ... 29
Device File ... 48
Diagnosis Message ... 18
Disabling Interrupt ... 81
___div ... 191, 283
div ... 185, 240
___divf.s ... 191, 285
Division ... 26
___divu ... 191, 283
double Type ... 25
A
abs ... 185, 237
acosf ... 189, 275
acoshf ... 188, 273
___addf.s ... 191, 279
Aligning Data ... 301
Alignment Condition ... 39
ANSI ... 33
Array Type ... 36
asinf ... 189, 275
asinhf ... 188, 273
Assembler Code ... 296
Assembler Control Instruction ... 309
Assembler Function ... 136
Assembler Instruction ... 29, 75
atan2f ... 189, 275
atanf ... 189, 275
atanhf ... 189, 273
atoff ... 186, 250
atoi ... 186, 252
atol ... 186, 252
E
ecvtf ... 185, 242
Embedded Functions ... 104
Enumerate Type ... 36
Enumerate Type Specifier ... 27
erfcf ... 188, 262
erff ... 188, 262
Executing Program ... 19
expf ... 188, 263
F
fabsf ... 188, 265
far jump ... 144
fcvtf ... 185, 242
fgetc ... 184, 219
fgets ... 184, 219
float Type ... 25
Floating-point ... 25
Floating-point Constant ... 23
Floating-point Type ... 35
floorf ... 188, 265
fmodf ... 188, 265
fprintf ... 184, 227
fputc ... 184, 221
fputs ... 184, 221
fread ... 184, 217
free ... 185, 246
Free-standing Environment ... 18
frexpf ... 188, 267
fscanf ... 185, 234
Function Call ... 125
fwrite ... 184, 217
B
Basic Language Specification ... 17
bcmp ... 182, 207
bcopy ... 182, 207
Binary constants ... 123
Bit Field ... 38, 117
BPC ... 170
bsearch ... 185, 238
bsh ... 107
bss ... 164
bsw ... 108
C
C Function ... 136
calloc ... 185, 246
Cast ... 26
cbrtf ... 188, 263
ceilf ... 188, 265
char Type ... 23
Character Indication ... 19
Character Set ... 19
Character String ... 24
___cmpf.s ... 191, 280
Comment ... 24
cosf ... 275
coshf ... 189, 273
CTBP ... 169
___cvt.ws ... 191, 282
D
Data Type ... 18, 302
G
gammaf ... 188, 269
gcvtf ... 185, 242
General Integer ... 25
General-Purpose Register ... 41
getc ... 184, 219
getchar ... 184, 219
gets ... 184, 219
gp ... 157
H
halt ... 106
Header File ... 24
hsw ... 108
User’s Manual U18513EJ1V0UM
317
APPENDIX C INDEX
hypotf ... 188, 270
___mulu ... 191, 288
mulu ... 111
I
Identifier ... 23
index ... 182, 203
Inline Expansion ... 29, 96
Integer Type ... 34
Interrupt Control ... 105
Interrupt Disable ... 30
Interrupt Level ... 78
Interrupt/Exception ... 29
Interrupt/Exception Handler
Example ... 95
Note ... 93
Interrupt/Exception Processing Handler ... 84
isalnum ... 183, 212
isalpha ... 183, 212
isascii ... 183, 212
iscntrl ... 183, 212
isdigit ... 183, 212
isgraph ... 183, 212
islower ... 183, 212
isprint ... 183, 212
ispunct ... 183, 212
isspace ... 183, 212
isupper ... 184, 212
isxdigit ... 184, 212
itoa ... 185, 244
N
nop ... 105
P
Peripheral I/O Register ... 29, 73, 160
perror ... 184, 216
Pointer Type ... 36
Pointers ... 294
powf ... 188, 263
printf ... 184, 227
Prologue/Epilogue ... 140
putc ... 184, 221
putchar ... 184, 221
puts ... 184, 221
Q
qsort ... 185, 238
Quantitative Limit ... 21
R
rand ... 186, 249
_rcopy ... 192
_rcopy1 ... 192
_rcopy2 ... 192
_rcopy4 ... 192
realloc ... 185, 246
Real-Time OS ... 102, 173
Referencing Data ... 42
Register ... 297
Register Mode ... 43, 153
RESET ... 152
rewind ... 184, 219
rindex ... 182, 203
J
j0f ... 188, 260
j1f ... 188, 260
jnf ... 188, 260
L
labs ... 185, 237
ldexpf ... 188, 267
ldiv ... 185, 240
log10f ... 188, 263
log2f ... 188, 263
logf ... 188, 263
longjmp ... 186, 256
ltoa ... 185, 244
M
Macro Name ... 31
main ... 155, 171
malloc ... 185, 246
Mask Register ... 45, 159
matherr ... 188, 271
memchr ... 182, 207
memcmp ... 182, 207
memcpy ... 182, 207
memmove ... 182, 207
Memory Allocation ... 29
memset ... 182, 207
___mod ... 191, 286
modff ... 188, 267
___modu ... 191, 286
___mul ... 191, 288
mul ... 110
___mulf.s ... 191, 289
Multi-byte Character ... 19
318
S
sasf ... 112
satadd ... 106
satsub ... 107
sbss ... 163
scan ... 185
scanf ... 234
sebss ... 165
__set_il ... 78
setjmp ... 186, 256
Shift Operator ... 26
sibss ... 168
sinf ... 275
sinhf ... 189, 273
sizeof ... 25
Software Register Bank ... 43
Special Data Type ... 32
sprintf ... 184, 223
sqrtf ... 188, 263
srand ... 186, 249
sscanf ... 185
sscanff ... 230
Stack Frame ... 125
Stack Pointer ... 154
Stack Size ... 300
Startup Routine ... 150
User’s Manual U18513EJ1V0UM
APPENDIX C INDEX
Example ... 174
Storage Area Class Specifier ... 26
strcat ... 182, 203
strchr ... 182, 203
strcmp ... 182, 203
strcpy ... 182, 203
strcspn ... 182, 203
strerror ... 182, 203
strlen ... 182, 203
strncat ... 182, 203
strncmp ... 182, 203
strncpy ... 182, 203
strpbrk ... 182, 203
strrchr ... 182, 203
strspn ... 182, 203
strstr ... 183, 203
strtodf ... 186, 250
strtok ... 183, 203
strtol ... 186, 252
strtoul ... 186, 252
Structure ... 25
Structure Type ... 37
Structure Type Packing ... 30
___subf.s ... 191, 290
Supplied Library ... 180
sxb ... 109
sxh ... 109
y1f ... 188, 260
ynf ... 188, 260
T
tanf ... 189, 275
tanhf ... 189, 273
Task Specification ... 30
tibss.byte ... 166
tibss.word ... 167
toascii ... 183, 210
_tolower ... 183, 210
tolower ... 183, 210
_toupper ... 183, 210
toupper ... 183, 210
tp ... 156
Translation Limit ... 20
Translation Stage ... 18
___trnc.sw ... 191, 291
Type Qualifier ... 27
U
ultoa ... 185, 244
ungetc ... 184, 219
Union ... 25, 116
Union Type ... 37
User Target ... 162
V
va_arg ... 181, 200
va_end ... 181, 200
va_start ... 181, 200
vfprintf ... 184, 227
volatile Qualifier ... 292
vprintf ... 227
vsprintf ... 184, 185, 223
Y
y0f ... 188, 260
User’s Manual U18513EJ1V0UM
319
For further information,
please contact:
NEC Electronics Corporation
1753, Shimonumabe, Nakahara-ku,
Kawasaki, Kanagawa 211-8668,
Japan
Tel: 044-435-5111
http://www.necel.com/
[America]
[Europe]
[Asia & Oceania]
NEC Electronics America, Inc.
2880 Scott Blvd.
Santa Clara, CA 95050-2554, U.S.A.
Tel: 408-588-6000
800-366-9782
http://www.am.necel.com/
NEC Electronics (Europe) GmbH
Arcadiastrasse 10
40472 Düsseldorf, Germany
Tel: 0211-65030
http://www.eu.necel.com/
NEC Electronics (China) Co., Ltd
7th Floor, Quantum Plaza, No. 27 ZhiChunLu Haidian
District, Beijing 100083, P.R.China
Tel: 010-8235-1155
http://www.cn.necel.com/
Hanover Office
Podbielskistrasse 166 B
30177 Hannover
Tel: 0 511 33 40 2-0
Munich Office
Werner-Eckert-Strasse 9
81829 München
Tel: 0 89 92 10 03-0
Stuttgart Office
Industriestrasse 3
70565 Stuttgart
Tel: 0 711 99 01 0-0
United Kingdom Branch
Cygnus House, Sunrise Parkway
Linford Wood, Milton Keynes
MK14 6NP, U.K.
Tel: 01908-691-133
Succursale Française
9, rue Paul Dautier, B.P. 52
78142 Velizy-Villacoublay Cédex
France
Tel: 01-3067-5800
Sucursal en España
Juan Esplandiu, 15
28007 Madrid, Spain
Tel: 091-504-2787
Tyskland Filial
Täby Centrum
Entrance S (7th floor)
18322 Täby, Sweden
Tel: 08 638 72 00
NEC Electronics Shanghai Ltd.
Room 2511-2512, Bank of China Tower,
200 Yincheng Road Central,
Pudong New Area, Shanghai P.R. China P.C:200120
Tel: 021-5888-5400
http://www.cn.necel.com/
NEC Electronics Hong Kong Ltd.
Unit 1601-1613, 16/F., Tower 2, Grand Century Place,
193 Prince Edward Road West, Mongkok, Kowloon, Hong Kong
Tel: 2886-9318
http://www.hk.necel.com/
NEC Electronics Taiwan Ltd.
7F, No. 363 Fu Shing North Road
Taipei, Taiwan, R. O. C.
Tel: 02-8175-9600
http://www.tw.necel.com/
NEC Electronics Singapore Pte. Ltd.
238A Thomson Road,
#12-08 Novena Square,
Singapore 307684
Tel: 6253-8311
http://www.sg.necel.com/
NEC Electronics Korea Ltd.
11F., Samik Lavied’or Bldg., 720-2,
Yeoksam-Dong, Kangnam-Ku,
Seoul, 135-080, Korea
Tel: 02-558-3737
http://www.kr.necel.com/
Filiale Italiana
Via Fabio Filzi, 25/A
20124 Milano, Italy
Tel: 02-667541
Branch The Netherlands
Steijgerweg 6
5616 HS Eindhoven
The Netherlands
Tel: 040 265 40 10
G07.1A
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement