IAR Embedded Workbenches for Renesas

Transcription

1 IAR Embedded Workbenches for Renesas Shawn A. Prestridge, Senior Field Applications Engineer IAR Systems, Inc. Class ID: 3C19I Renesas Electronics America Inc.

2 Shawn A. Prestridge Senior Field Applications Engineer Responsible for Embedded Workbench training Develops applications for many different boards Develops applications for middleware such as: RTOS GUI Projects as examples and for customers PREVIOUS EXPERIENCE: Embedded hardware/software engineer for Texas instruments. Contractual-based development for Ministry of Software Degree work from Southern Methodist University: BSEE, BS Mathematics, MSEE, MS Software Engineering, PhD in EE 2

3 Renesas Technology & Solution Portfolio 3

4 Agenda Tour of the Embedded Workbench Renesas MCU support Hardware debugger support Example Projects C-SPY debugger advanced features visualstate overview Eclipse support Customer support Writing more efficient code Live demonstration Summary 4

5 Introduction The Embedded Workbench enables developers to write the tightest and fastest code possible for their embedded designs. This session will explore the features of the toolchain and show you how to use them effectively. 5

6 IAR Embedded Workbench 6

7 Overview IAR Embedded Workbench IDE IDE tools Build tools IAR C-SPY Debugger Plug-in to Eclipse IDE Custom plug-ins (editors, source code control systems, etc.) Editor Project manager Library builder IAR C/C++ Compiler Assembler Simulator driver Hardware system drivers Power debugging Segger J-Link Renesas E1 Renesas E20 3 rd party probes IAR visualstate Librarian Linker RTOS plug-ins Integrated RTOS partners 7

8 Renesas MCU Support 32-bit RX SuperH V850 R32C RH bit RL78 78K0R M32C/M16C R8C H8S H8/300H 8-bit 78K0 78K0S 8

9 IAR Embedded Workbench for Renesas RH850 Beta version planned for Autumn 2012 Based on the current EWV850 with: New instructions added Improved optimization including re-written arithmetic libraries C99 support New editor (source browser, auto completion, code folding, block selection, etc.) Improved inline assembler (GNU style) Improved stack analysis (Link-time analysis of max stack depth) Elf/Dwarf as object format Renesas ABI (Application Binary Interface) compliant Link compatibility between toolchains Ability to port existing object code into EWRH850 C-SPY support for Renesas emulators 9

10 IAR Embedded Workbench for Renesas The only toolchain that supports virtually ALL Renesas devices! One IDE to access all your Renesas projects An Embedded Workbench workspace can have multiple projects, each one targeting a different Renesas architecture 10

11 Hardware Debugging Support for Renesas Renesas E(x) Emulators (OCD) Download to flash Debug in real time on hardware Support for hardware breakpoints Renesas IECUBE (in-circuit) Adds support for instruction and function trace, event breakpoints, timers, etc 11

12 Example Projects Ready-made example projects demonstrating the different peripheral s of the MCU Most popular boards supported Kickstart version of EW included in Renesas Starterkits 12

13 IAR Embedded Workbench for RX v Coremarks/MHz! Coremark Board: RDKRX62N, 96Mhz Result: 2.76 Coremarks/Mhz CoreMark Size : 666 Total ticks : Total time (secs): Iterations/Sec : Iterations : 3000 Compiler version : Compiler flags : Memory location : STACK seedcrc : 0xe9f5 [0]crclist : 0xe714 [0]crcmatrix : 0x1fd7 [0]crcstate : 0x8e3a [0]crcfinal : 0xcc42 Correct operation validated. See readme.txt for run and reporting rules. CoreMark 1.0 : / / STACK 13

14 EEMBC Benchmarks Published by Renesas 14

15 IAR Embedded Workbench for RX v2.40 RX size benchmark This is a code size test on real customer applications where the size of linked code + constants has been compared. The test was performed with the following tools: IAR EWRX v Options: -e -Ohz --double=32 --data_model=f --align_func=1 --endian l --core RX600 mfc Renesas RX v Options: -cpu=rx600 -fpu -optimize=max -size -goptimize Linker: -noprelink optimize GNURX v Os -fno-function-cse -funit-at-a-time -falign-jumps -fdata-sections -ffunctionsections -g3 -g -flto -mlittle-endian-data -mcpu=rx600 -c -x c Linker: --gc-sections -no-keep-memory -e _PowerON_Reset -loptm -loptc - lnosys flto GNURX HEWRX EWRX Customer application Name GNURX HEWRX EWRX EWRX vs GNURX EWRX vs HEWRX Blowfish algorithm blowfish ,2% 76,3% Functions to descramble CSS encrypted DVD content. decrypt ,1% 79,8% Reed-Solomon Reed_Solom ,7% 73,1% encoding/decoding on Microwave sensor microwave ,4% 64,6% application Bluetooth stack bt_stack ,3% 84,7% Car navigation system car_navig ,8% 84,5% Turbogenerator technology, engine_contr ,7% 93,4% "turbocompounding" Remote monitoring and operation for generators and engines. generator ,2% 96,3% Sum: ,2% 89,7% GNURX HEWRX EWRX 15

16 Embedded Workbenches Available Now! See for more information 16

17 IAR C-SPY Debugger Advanced Features 17

18 IAR C-SPY Debugger Performance Analysis Dedicated breakpoint type Capable of measuring these execution aspects: total time total number of cycles number of cycles spent in interrupts and other exceptions the number of executed instructions the number of accepted interrupts and other exceptions. 18

19 IAR C-SPY Debugger Complex Breakpoints Allows break at address condition and/or data condition Useful to detect access to a certain memory area and within a certain data range Can help you find where data gets clobbered in code Can help you see unauthorized access to memory areas 19

20 IAR C-SPY Debugger Profiling/Code Coverage Profiling helps you to find the functions in your source code where the most time is spent during execution. Lets you know where to focus optimization efforts Can also help you see when you should inline a function call Code coverage helps you to verify whether all parts of the code have been executed. 20

21 IAR C-SPY Debugger Trace Functionality Trace lets you inspect the program flow to a specific state. Information based on E1/E20 and J-Link trace capabilities. Choose between non-intrusive mode and richer trace information modes 21

22 IAR C-SPY Debugger During Execution Features Breakpoints can be set/removed Live watch displays live data in the Watch window Live memory displays live data in the memory window The RAM monitor enables display of data read, written or both, with different colors in the memory window. 22

23 IAR C-SPY Debugger Hot Attach Feature Possibility to let the debugger attach to a running application at its current location, without resetting the target system. When a board in the field is in the weeds, you can attach to it without resetting in order to determine what went wrong. 23

24 IAR C-SPY Debugger Power Debugging Feature Ability to sample power consumption and correlate it to the program's instruction sequence and hence with the source code. C-SPY visualizes power consumption data in different views. Provides a view of the power profile of an application. Allows a developer to try different approaches to the application to minimize power footprint. 24

25 IAR C-SPY Debugger RTOS Awareness RTOS kernel awareness Displays RTOS properties like: task lists, mailboxes, timers, semaphores, queues, and resources Built in support is currently available for: -ThreadX - Micrium uc/os-ii / III - Segger embos - OSEK ORTI -FreeRTOS Task List RTOS Overview 25

26 visualstate C-SPY Debug Integration Direct graphical feedback in C-SPY for state machines on various levels of detail, like current state vector, executed actions, received events etc. 26

27 IAR C-SPY Debugger Advanced Features Be sure to attend our lab section to see advanced debug features in action! 27

28 Learning Renesas Tools Renesas provides a wide variety of starter, evaluation and development kits to fit any price point. All of these kits are supported by IAR Systems and many have a bevy of example applications to get your development effort kick-started! 28

29 Eclipse Support Eclipse build chain plug-ins available for RX, V850, RL78, R32C and 78K Supports any Eclipse IDE distribution with CDT v3.5 (Galileo) or later Plug-ins included in Renesas e2studio 29

30 Customer Support Support and Update Agreement (SUA) Free software updates First class world-wide technical support by telephone, and fax VIP Support contracts available My Pages at Read about new product releases Download product updates Transfer licenses to co-workers Manage your contact information Different architectures, one solution! We have compilers to support over 30 different architectures We are your one-stop compiler supplier 30

31 Writing Efficient Code 31

32 Controlling Optimization Global setting Over entire project Per-file settings When functions in a file must run as fast as possible. Per-function settings Use #pragma Compiler version dependent, so check your manual #pragma optimize=s 0 void delay(int time) {... } 32

33 Structuring an Application To be efficient and portable, an application should: Isolate device-dependent code Leave most of the code undisturbed Use tuned code where needed Generic Program Files Tuned Program Files Device Driver Files Hardware 33

34 Use Correct Data Sizes Different architectures have different natural data size Different available memories, sizes, etc. Using Unnatural data size might cost A 32-bit MCU might need to shift, mask and signextend operations to use smaller types A smaller MCU will need to store 32-bit data in multiple registers to hold all its contents and perform operations in RAM It is therefore better to use a natural size unless there is a compelling reason not to do so, e.g.: Perhaps you are doing I/O and you need a precise number of bits Bigger types like a char array might take up too much room 34

35 Use Signedness Appropriately Think about signedness Division by a negative number is treated differently than that for a positive number (by the rules of the C language) Signed Negative values possible Arithmetic operations performed Using a signed number that never will be negative, incurs an extra test-and-jump condition Unsigned Negative values impossible Bit operations performed + - * / % << >> & ^ ~ 35

36 Avoid Floating Point Arithmetic if Possible Floating point very expensive Brings in large library (from C runtime) Use only when really needed Can be done inadvertently: Example code: #define Other 20 #define ImportantRatio (1.95 * Other) #define ImportantRatioBetter int i=a + b * ImportantRatio; 36

37 Avoid Floating Point Arithmetic if Possible IEEE 754: float Wide range: Float to 10 38, Double to Good precision: Float 10-7, Double Designed for giving small error in complex computations Expensive in size and speed...unless there s hardware support Real-world data usually have: Fixed range Fixed precision Fixed-point arithmetic: Implemented using integers Can give significant savings (size and speed)...do the math before writing the code Use relaxed floating-point semantics 37

38 Avoid Becoming a Castaway C has implicit and explicit casts Casting is (usually) not free: Sign-extension and zero-extension Complex conversions to and from floating-point numbers Extending pointers Casting to/from pointers is bad Can lose information (smaller size) Inefficient code (larger size) Avoid unless really necessary Don t do explicit casts Don t mix types in expressions 38

39 Structure your Structures Carefully Structures are padded when the CPU requires alignment or gains from it struct foo { uint8_t byte; uint32_t word; }; byte 1 byte padding to align word word 4 bytes thus, foo 8 bytes 39

40 Structure your Structures Carefully Waste of memory to pad? Packed structures require more code Simple guideline reduces need of padding Guideline: Order fields by size Largest object first, smallest object last struct record { long id; /* 4 bytes */ char* name; /* 2 bytes */ char* title; /* 2 bytes */ char tag; /* 1 byte */ }; 40

41 Use Global Variables Locally Copy value into a local variable Local variable probably placed in register, so operations on it is fast No memory reads Write back the value if needed. Gives more info to compiler No function call will change the variable 41

42 Use Parameters and Local Variables Parameters and local variables: Give compiler more freedom Usually placed in registers but can be placed in memory Parameter passing: First parameters usually in registers Registers to use = calling convention Register allocation targets: Simple values (integers, pointers, floating-point numbers) Hard to allocate: Arrays indexed addressing the norm Structs large, often pointed to Struct as parameter: Value copied, and a pointer passed Pass pointer directly for efficiency! 42

43 Use Parameters and Local Variables void foo(char a) { char b,i,j; char h[10]; b=a<<5; for(i=0; ) OUTPUT(b); live ranges a b i j Register usage: Variables can share a register Only present in register while live Don t worry about extra variables Optimize: Minimize lifetime Minimize overlap } for(j=0; ) h[j]=a; 43

44 Don t Write Clever Code! Clever code: Fewer source characters = better code Using dark corners of C semantics Straightforward code: Easy to read = easy to optimize Easy to maintain Use common constructions: Likely to be better optimized Clever code unsigned long int a; unsigned char b; b =!!(a<<11); Better code unsigned long int a; unsigned char b; if((a & 0x1FFFFF)!= 0) b = 0x01; 44

45 Avoid Tight Timing Loops Empty delay loop does not work Code with no effect Gets removed Delay is always zero To really delay: Access volatile variable Use OS services Use CPU timer void delay(int time) { int i; for (i=0;i<time;i++) continue; return; } void InitHW(void) { OUT_SIGNAL (0x20); delay(120); OUT_SIGNAL (0x21); delay(121); OUT_SIGNAL (0x19); } 45

46 Live Demonstration 46

47 Questions? 47

48 Summary Tour of the Embedded Workbench Renesas MCU support Hardware debugger support Example Projects C-SPY debugger advanced features visualstate overview Eclipse support Customer support Writing more efficient code Live demonstration Summary 48

49 Please Provide Your Feedback Please utilize the Guidebook application to leave feedback or Ask me for the paper feedback form for you to use 49

50 Renesas Electronics America Inc.