Peanuts NeXT Software Archives

home *** CD-ROM | disk | FTP | other *** search

/ Peanuts NeXT Software Archives / Peanuts-2.iso / Developer / hardware / dsp / drbub / dr.bub.hier < prev next >

Wrap

Text File | 1991-09-07 | 42.6 KB | 863 lines

This was taken off of Motorola's Dr. BuB DSP support BBS. Details about how to log on are located in file 'dr.bub'. 1) Free DSP Software 1) DSP56000/1 - General Purpose DSP 1) Introduction to the DSP56000 library (READ FIRST) LIBRARY ORGANIZATION: The software libraries are grouped into menus by topic. For example all FFT routines are grouped with their support routines into one menu. Dr. Bub will be adding programs to existing menus as well as adding more menu topics as application programs are tested and documented. FILE ORGANIZATION: In each topic menu application programs are grouped together with their help files and test files. The file naming convention is: xxxxxxx.asm The application program or assembler MACRO. The file name xxxxxxx indicates the type of program such as fftr2e.asm for a radix 2 fft application. xxxxxxx.hlp A text help file for the application program. The .hlp files explain the application, give performance numbers, application notes, tips on usage, limitations, etc. Press RETURN to continue or q RETURN to exit to menu: xxxxxxxt.asm Most of the application programs include a test program that demonstrates the usage of the application program. Test programs have the same name as the application with a 't' appended. xxxxxxxt.hlp Some test programs also have help files USAGE NOTES: 1. The DSP56000 software can be downloaded using: KERMIT, XMODEM, YMODEM, or ASCII. Dr. Bub allows the use of a "*" wildcard when downloading. 2. Periodically check the version numbers. Bug fixes and improvements are noted by a new version number and last update. If you find a problem, please leave us a message on Dr. BuB. 3. The test programs use a set of standard equates which are contained in the ioequ.asm file provided in the STANDARD I/O EQUATE FILES menu. Press RETURN to continue or q RETURN to exit to menu: 4. Many test programs use INCLUDE assembler directives which assume that the application program and the ioequ.asm file are in a "dsplib" subdirectory on your system. 5. The INCLUDE assembler directives used in many test program files may need to be modified to reference the desired files and to match your computer's file path naming convention. 2) Standard I/O Equate Files 3) FFTs 4) FIR/IIR Filters --- FIR and IIR FILTERS --- The following files contain FIR and IIR filter programs, macros, help files, and test routines for the DSP56000/1. +-------------------------------------------------------------------+ | You will need to remember the document id for further reference. | +-------------------------------------------------------------------+ DOCUMENT ID VERSION SYNOPSIS SIZE UPDATED ---------- ------- -------- ---- ------- fir.asm 1.0 Direct Form FIR Filter 545 7/15/87 fir.hlp Help for fir.asm 2161 7/15/87 firt.asm 1.0 Test program for fir.asm 1164 7/16/87 iir1.asm 1.0 Direct Form Second Order All Pole 656 7/15/87 IIR Filter iir1.hlp Help for iir1.asm 1786 7/15/87 iir1t.asm 1.0 Test program for iir1.asm 1157 7/16/87 iir2.asm 1.0 Direct Form Second Order All Pole 801 7/15/87 IIR Filter with Scaling iir2.hlp Help for iir2.asm 2286 7/15/87 iir2t.asm 1.0 Test program for iir2.asm 1311 7/16/87 iir3.asm 1.0 Direct Form arbitrary Order All 776 7/15/87 Pole IIR Filter iir3.hlp Help for iir3.asm 2605 7/15/87 iir3t.asm 1.0 Test program for iir3.asm 1309 7/16/87 iir4.asm 1.0 Second Order Direct Canonic IIR Filter 713 7/15/87 (Biquad IIR Filter) iir4.hlp Help for iir4.asm 2255 7/15/87 iir4t.asm 1.0 Test program for iir4.asm 1202 7/16/87 iir5.asm 1.0 Second Order Direct Canonic IIR Filter 842 7/15/87 with Scaling (Biquad IIR Filter) iir5.hlp Help for iir5.asm 2803 7/15/87 iir5t.asm 1.0 Test program for iir5.asm 1289 7/16/87 iir6.asm 1.0 Arbitrary Order Direct Canonic IIR 923 7/15/87 Filter iir6.hlp Help for iir6.asm 3020 7/15/87 iir6t.asm 1.0 Test program for iir6.asm 1377 7/16/87 iir7.asm 1.0 Cascaded Biquad IIR Filters 900 7/15/87 iir7.hlp Help for iir7.asm 3947 7/15/87 iir7t.asm 1.0 Test program for iir7.asm 1432 7/16/87 lms.hlp 1.0 LMS Adaptive Filter Algorithm 5818 4/08/88 transiir.asm 1.0 Implements the transposed IIR filter 1981 8/19/88 transiir.hlp 1.0 Help file for transiir.asm 974 8/19/88 FIR & IIR filters: fir.asm fir.hlp firt.asm iir1.asm iir1.hlp iir1t.asm iir2.asm iir2.hlp iir2t.asm iir3.asm iir3.hlp iir3t.asm iir4.asm iir4.hlp iir4t.asm iir5.asm iir5.hlp iir5t.asm iir6.asm iir6.hlp iir6t.asm iir7.asm iir7.hlp iir7t.asm lms.hlp transiir.asm transiir.hlp 5) Lattice Filters 6) Matrix Operations 7) Functions 8) Companding Routines --- Companded CODEC to Linear PCM Data Conversion --- The following files contain function routines and macros for the DSP56000/1 to perform conversions between companded CODEC and Linear PCM data. +----------------------------------------------------------------+ |You will need to remember the document id for further reference.| +----------------------------------------------------------------+ DOCUMENT ID VERSION SYNOPSIS SIZE UPDATED ---------- ------- -------- ---- ------- loglin.asm 1.0 Companded CODEC to linear PCM data 4572 4/20/87 conversion loglin.hlp Help for loglin.asm 1479 4/20/87 loglint.asm 1.0 Test program for loglin.asm 2184 4/20/87 loglint.hlp Help for loglint.asm 1993 4/20/87 linlog.asm 1.1 Linear PCM to companded CODEC data 4847 8/17/88 conversion linlog.hlp Help for linlog.asm 1714 4/20/87 CODEC: linlog.asm linlog.hlp loglin.asm loglin.hlp loglint.asm loglint.hlp 9) Speech --- Speech --- The following files contain function routines and macros for the DSP56000/1 to perform speech processing and analysis. +-------------------------------------------------------------------+ | You will need to remember the document id for further reference. | +-------------------------------------------------------------------+ DOCUMENT ID VERSION SYNOPSIS SIZE UPDATED ---------- ------- -------- ---- ------- lgsol1.asm 2.0 Leroux-Gueguen solution for PARCOR 4861 6/02/87 (LPC) coefficients lgsol1.hlp Help for lgsol1.asm 3971 6/02/87 ---------- durbin1.asm 1.2 Durbin Solution for PARCOR 6360 6/05/87 (LPC) coefficients durbin1.hlp Help for durbin1.asm 3616 6/05/87 Speech: durbin1.asm durbin1.hlp lgsol1.asm lgsol1.hlp a) Sorting & Searching b) Floating Point Library c) Error Coding Routines 2) DSP56200 - Digital Adaptive Filter 3) Software Tools and Utilities --- Tools and Utilities --- +---------------------------------------------------------------------+ |NOTE:You will need to remember the document id for further reference.| +---------------------------------------------------------------------+ DOCUMENT ID VERSION SYNOPSIS SIZE UPDATED --------- ------- -------- ---- ------- srec.c 1.7 Utility to convert DSP56000 OMF format 37038 7/07/88 to SREC. srec.h 1.0 Include file for srec.c 3599 4/18/88 srec.exe 1.7 Srec executable for IBM PC 19547 7/07/88 sloader.asm 1.1 Serial loader from the SCI port for the 3986 5/24/88 DSP56001 sloader.hlp 1.1 Help for sloader.asm 2598 5/24/88 sloader.p 1.1 Serial loader s-record file for download 736 5/24/88 to EPROM parity.asm 1.0 Parity calculation of a 24-bit number in 1641 5/24/88 accumulator A parity.hlp 1.0 Help for parity.asm 936 5/24/88 parityt.asm 1.0 Test program for parity.asm 685 5/24/88 parityt.hlp 1.0 Help for parityt.asm 259 5/24/88 dspbug Ordering information for free debug 799 5/26/88 monitor for DSP56000/1 parity.asm parity.hlp parityt.asm parityt.hlp sloader.asm sloader.hlp sloader.p srec.c srec.h srec.exe dspbug 2) Motorola DSP Documentation 3) Answers to Frequently Asked Questions ANSWER ID VERSION SYNOPSIS SIZE UPDATED ---------- ------- -------- ---- ------- a1 1.1 Parallel Data Moves 3898 5/22/87 a2 1.0 Operating Temperature 450 4/09/86 a3 1.0 Data Shifting 8662 5/17/86 a4 1.0 Stop/Wait Mode 478 8/19/86 a5 1.1 Samples 773 11/03/87 a6 1.0 Bootstrap Program 732 2/23/87 a7 1.1 DSP56000 Programs 878 5/18/87 a8 1.0 P,X,Y Address Spaces 2339 1/07/87 a9 1.0 DSP56000 Simulator 1253 1/07/87 a10 1.1 DSP56001 ROM Contents 1445 5/18/87 a11 1.0 VAX, SUN Assembler/Simulator 1183 1/07/87 a12 1.0 DSP56000 C Compiler 562 1/07/87 a13 1.1 1024-point FFT 7743 5/18/87 a14 1.1 DSP56001 Development System 977 5/18/87 a15 1.0 What is Convergent Rounding? 2414 3/24/87 a16 1.0 Rev. 1.1 Simulator/Assembler Updates? 830 2/24/87 a17 1.0 DSP56000 Assembler Rev.1.1 Release Notes 2177 3/06/87 a18 1.0 DSP56000 Host Port Register Use in DMA 975 3/04/87 Mode a19 1.0 "Free" FIR Output w/DSP56200 Adaptive 3571 3/05/87 Mode a20 1.0 Using Decimal-Fractional #s With DSP5620 5851 3/05/87 a21 1.0 DSP56000 Simulator Timing Questions 4558 3/10/87 a22 1.2 Common Benchmarks for the DSP56000/1 2677 6/02/87 a23 1.0 Multi-Channel Adaptive Filter w/DSP56200 7406 4/16/87 a24 1.0 XMODEM File Transfer Problems w/Dr. BuB? 591 4/16/87 a25 1.0 DSP56000/1 compatible CODECs 985 5/29/87 a26 1.0 DSP56000 Simulator Session Example 9645 6/02/87 a27 1.0 DSP56000 Assembler Rev.2.0 Release Notes 1641 11/02/87 a28 1.0 Radix 4 and Radix 8 complex FFT software 538 11/05/87 a29 1.0 Complex FFT with Real Data 918 11/05/87 a30 1.0 Connecting SCSI to the DSP56000/1 3481 3/03/88 a31 1.0 Other Motorola Bulletin Boards 811 3/22/88 a32 1.0 Interrupt processing and the instruction 3569 3/22/88 pipeline a33 1.0 Sockets available for DSP56000/1 1820 4/07/88 a34 1.0 Separate RD and WR signals on DSP56000/1 2451 10/06/88 a35 1.0 Information about military products 593 12/21/88 a36 1.0 How Dr. BuB answers questions 1264 1/05/89 Enter answer id: a26 +-------------------------------------------------------------------------+ | | | Can you give me an example of how to use the DSP56000/1 Simulator ? | | | +-------------------------------------------------------------------------+ Example DSP56000/1 Simulator Session using Input and Output Data Files The example DSP56000/1 Simulator session described below was developed for an IBM PC environment. The example session assumes the following file directory structure: +----------+ | DSP56001 | +----------+ | | ------------------------- | | | | +-----+ +-----+ | ASM | | SIM | +-----+ +-----+ Files: asm56000.exe sim56000.exe lgsol1.asm lgin.io < lgsol1.lod > -------- < lgsol1.lod > < lgsol1.lst > < lgout.io > Note: Files that are created by the DSP56000/1 Assembler or Simulator are shown above in < brackets >. The user does not need to create these files prior to working through the session described below. All other files should exist in their respective directories as shown above. The prompt from the ASM directory will be shown as "DSP56001\ASM>", the prompt from the SIM directory will be shown as "DSP56001\SIM>" and the prompt from within the DSP56000/1 Simulator will be shown as ">". Steps: 1) Obtain DSP56000/1 assembly language source file. The assembly language program to be simulated can be created using an editor or downloaded from Dr. BuB. For this example, the assembly language source file "lgsol1.asm" should be downloaded from Dr. BuB. Remember to copy the downloaded file "lgsol1.asm" into the ASM directory as shown above. 2) Assemble the assembly language source file "lgsol1.asm" using the DSP56000/1 Assembler. DSP56001\ASM>asm56000 -b -l lgsol1.asm This command assembles the DSP56000/1 assembly language source file "lgsol1.asm" which was downloaded from Dr. BuB. The assembler output listing is placed in the file "lgsol1.lst" and the assembler output load file is placed in the file "lgsol1.lod". 3) Create the input data file to be used by the DSP56000/1 Simulator. DSP560001\ASM>cd ..\SIM DSP560001\SIM> (Use the editor program of your choice to create the input data file "lgin.io".) DSP560001\SIM>type lgin.io The first command changes the current directory to the SIM directory. The second command will vary depending upon the editor used to create the simulator's input data file "lgin.io". Note that this file should be located in the SIM directory as shown above. The third command is used to verify that the contents of the newly created input data file are correct. Note: The contents of the DSP56000/1 Simulator input file "lgin.io" are shown later in this file for reference. 4) Simulate the assembly language program using the DSP56000/1 Simulator. DSP56001\SIM>copy ..\ASM\lgsol1.lod DSP56001\SIM>sim56000 The first command copies the assembler output load file "lgsol1.lod" from the ASM directory to the SIM directory. The second command invokes the DSP56000/1 Simulator. Once the DSP56000/1 Simulator has been invoked, the simulator's prompt ">" should appear as shown below. The following simulator commands should then be entered, one at a time, as shown. >reset s This command sets the simulator to the reset state. >change bcr 0 This command changes the Port A Bus Control Register (bcr) contents to 0. This means that all external memory and external I/O accesses will require zero wait states for purposes of simulation. >display on x:0..$a x:$20..$29 This command enables the display of two banks of X memory in the DSP56000/1. >load lgsol1 This command loads the file "lgsol1.lod" into the simulator. The simulator should initialize the contents of the Program Counter (pc) register to the top of the lgsol1 program at p:$100. >input y:$ffff lgin.io This command defines the file "lgin.io" as an input file to the DSP56000/1 Simulator. Thus, all read operations from the DSP56000/1 external I/O memory location y:$ffff will get their data value from the file "lgin.io". >output y:$fffe lgout.io This command defines the file "lgout.io" as an output file to the DSP56000/1 Simulator. Thus, all write operations to the DSP56000/1 external I/O memory location y:$fffe will put their data value into the file "lgout.io". >break pc>$139 This command defines a breakpoint for any program memory location greater than p:$139. If the Program Counter (pc) register should ever exceed p:$139, program simulation will halt and display the currently enabled registers and memory locations. >display This command displays all of the currently enabled registers and memory locations. This command is used here to verify that all changes have been made and that everything is in order prior to actual simulation. >go This command enables program simulation beginning at the current value of the Program Counter (pc). Program simulation will continue until a previously defined breakpoint is encountered. >quit This command exits the DSP56000/1 Simulator and returns the "DSP56001\SIM>" prompt. Note that the DSP56000/1 Simulator output file "lgout.io" will not be available for inspection until the simulator has been terminated using the "quit" command. 5) Verify that the contents of the DSP56000/1 Simulator output file "lgout.io" are correct. DSP560001\SIM>type lgout.io Note: The contents of the DSP56000/1 Simulator output file "lgout.io" are shown later in this file for reference. DSP56000/1 Simulator Input Data File: "lgin.io" +------------------------------------------------------------+ | 1 | | 400000 200000 200000 200000 200000 200000 | | 200000 200000 200000 200000 200000 | +------------------------------------------------------------+ The first value in the input data file represents the number of speech data frames to be processed by the "lgsol1" program. This program performs Linear Predictive Coding (LPC) analysis on speech data using the Leroux - Gueguen (LG) algorithm. The following eleven (11) values in the input data file represent the normalized autocorrelation coefficients required to perform a tenth order LPC analysis. Note: All DSP56000/1 input data files must be composed of standard ASCII characters. DSP56000/1 Simulator Output Data File: "lgout.io" +------------------------------------------------------------+ | C00000 | | D55556 | | E00000 | | E66667 | | EAAAAC | | EDB6DD | | F00001 | | F1C71D | | F33333 | | F45D18 | +------------------------------------------------------------+ The ten values in the output data file shown above are the output of the LG LPC algorithm. These ten (10) values represent the desired reflection coefficents (K's) produced by the tenth order LPC analysis. Note: All DSP56000/1 output data files are composed of standard ASCII characters. 4) Pose a Question or Make a Comment 1) Read editor documentation To use the Motorola DSP bulletin board note editor you only need to know the following: 0. Backspace is <ctrl-h>. 1. There are two main modes to this editor, INSERT and COMMAND. 2. When you enter the editor, you're in INSERT mode. 3. To exit INSERT mode and enter COMMAND mode, you need to enter a blank line. Doing this will bring up a list of valid editor commands. 4. To exit COMMAND mode and return to INSERT mode (typing at the end of the note) use the Continue command. BELOW is a description of the editor commands with examples. Abort - Stop note from being posted. Continue - Continue typing a note. Begins on the next line number. Delete - Remove a single line from a note. Edit - Edit any portion of a line. To do this you will need to "mark" off the ends of the portion of the line in question with "^". THE CHANGE will be put AFTER the first ^ and BEFORE the last ^. To insert in a line, put the ^'s next to each other. To delete or replace in a line separate the ^'s by any number of spaces. EVERYTHING between the ^'s will be lost. If you enter a new line, when prompted, everything between the ^'s will be replaced with the new string, otherwise the area between the ^'s is deleted. EXAMPLES: INSERT this is a test ^^ Enter new string: NEW STRING New string is ---> this NEW STRING is a test REPLACE this is a test ^ ^ Enter new string: WAS New string is ---> this WAS a test DELETE this is a test ^ ^ Enter new string: New string is ---> this test Insert - Insert one or more lines of text before an identified line. List - Show the current state of the note Save - Send the message. ? - This message 2) Enter question or comment 5) Motorola Training Courses Information 6) Third Party Information 7) Available Products and Printed Documentation >>>> SUPPORT PRODUCTS <<<< DSP56000 C-Cross Compiler DSP56KCC This package contains a Kernighan and Ritchie C compiler, C preprocessor. Ask for brochure BR541/D or look on Dr. BuB in "Motorola DSP documentation" item d9. 32010 to DSP56001 Translator Software Package DSP320to56001 The DSP320to56001 translator software will convert any 32010 code into code for Motorola's powerful signal processor chip, the DSP56001. The primary features of DSP320to56001 are: * Translation of any 32010 applications software into DSP56001 source code. * Two modes of operation 1) translates to 56001 source code for potential optimization and assembly with the DSP56000SASMA 2) translates and runs 32010 code "as is" directly on the DSP56000ADS. * Runs on IBM-PC, XT or AT* under MS-DOS**. * C source code of DSP320to56001 program is provided. Ask for brochure BR522/D for more information. ----- DSP56000 Development Software Package DSP56000CLASA This package contains a macro cross-assembler, linker/librarian, multi-DSP simulator, disassembler, and user interface for developing DSP56000 and DSP56001 software and applications. Over 800 pages of documentation is provided including user's manuals for the macro cross-assembler and simulator, the technical summary and user's manual for the DSP56000, and the license agreement and software warranty. DSP56000CLASA runs on the IBM-PC, XT, or AT* under MS-DOS** and the price is $495. You may purchase this software package via your local Motorola Semiconductor sales office or Motorola distributor. ----- Application Development System for DSP56000 Family Products DSP56000ADS The DSP56001-based Application Development System (ADS) is a three-component development tool for designing, debugging, and evaluating DSP56000 and DSP56001 target system equipment. The ADS simplifies evaluation of the user's prototype hardware/software product by making all of the essential DSP56000 timing and I/O circuitry easily accessible. This product is described in more detail in brochure BR517. ----- * IBM-PC, XT, AT are trademarks of Intl. Business Machines ** MS-DOS is a trademark of Microsoft Corp. >>>> DOCUMENTATION <<<< 1) BR282 -- DSP56000 Technical Summary 2) BR283 -- DSP56200 Technical Summary 3) BR526 -- DSP56000CLASx Software Summary 4) BR297 -- Dr.BuB Electronic Bulletin Board 5) DSP56001/D -- DSP56001 Technical Summary 6) DSP56000 UM/AD -- DSP56000 User's Manual 7) DSP56200/D -- DSP56200 Advance Information 8) BR517 -- DSP56000ADS Applications Development System 9) BR522/D -- DSP320to56001 Translator Software Summary 10) BR541/D -- DSP56000/1 Family C Language Compiler Software Summary 11) APR1 -- Digital Sine-Wave Synthesis Using the DSP56001 12) APR2/D -- Digital Stereo 10-Band Graphic Equalizer Using the DSP56001 13) DSP56ADC16/D -- DSP56ADC16 16-bit Sigma-Delta Analog-to-Digital Converter, Advance Information >>>> Literature Distribution Centers <<<< USA -- Motorola Literature Distribution P.O. Box 20912 Phoenix, Az 85036 EUROPE -- Motorola Ltd; European Literature Center 88 Tanners Drive Blakelands Milton Keynes MK145BP, England Asia/Pacific -- Motorola Semiconductors H.K. Ltd. P.O. Box 80300 Cheung Sha Wan Post Office Kowloon Hong Kong 8) Motorola Sales Offices 9) Product Updates +------------------------------------------------------------------------+ FILE ID VERSION SYNOPSIS SIZE UPDATED ---------- ------- -------- ---- ------- u1 1.0 DSP56000 Assembler Rev 1.1 Release Notes 2177 6/02/88 u2 1.0 DSP56000 Assembler Rev 2.0 Release Notes 6952 6/02/88 u3 1.0 XSP56001 2B38E/4B38E Silicon Design Memo 3498 2/19/88 ------ u1 - Release Notes -- Motorola DSP56000 Macro Cross Assembler Version 1.10 1) There is a bug in versions 1.00 and 1.10 of the assembler that causes a divide-by-zero error when an integer is used as the divisor in a floating point divide operation (e.g. 3.456/2). The workaround is to write the divisor in floating point format (3.456/2.0). 2) The assembler reference manual contains an error in the description of the logical negate operator on page 3-5. The last sentence of the paragraph reads "If BUF had a value of 1000, !BUF would have a value of 1." Logically negating BUF (!BUF) would actually yield a zero value here. 3) Another error in the manual appears on page 5-4, in the N_R_MUL macro coding example at the top of the page. The operands of the DO instruction are transposed; the instruction should read "DO #N,_END." 4) In the SWAP_REG and SWAP_SYM macros shown on pages 5-6 and 5-7, respectively, the last move instruction in the macros reads "MOVE X0,R\REG1." This instruction should be "MOVE X0,R\REG2." 5) In Appendix C, "Instruction Set Summary," section C.6, "Program Control Instructions," the RTI and RTS instructions DO NOT ALLOW parallel data moves. Therefore none of the program control instructions allow parallel data moves. 6) If an address register is written in an instruction and subsequently used as a pointer in the following instruction, the assembler correctly generates an error flag. This is because due to pipelining the new contents of the register are not available until the next following instruction. However, there is a bug in the assembler such that if a transfer conditional (Tcc) instruction writes to an address register and the following instruction uses that register as a pointer, the assembler does not generate the expected error flag. Here is an example: TGE X0,A R0,R1 ; Write to register R1 MOVE X:(R1),X1 ; R1 used as pointer here; ; Assembler should generate error, ; but does not ------ u2 - Release Notes -- Motorola DSP56000 Macro Cross Assembler Version 2.00 Motorola DSP56000 Cross Linker/Librarian Version 1.00 1) There is a bug in version 2.00 of the assembler which relates to the use of counter modifiers in the ORG directive. When in the assembler relative mode using the ORG directive with either the L or H counter modifiers (e.g. ORG XL:, ORG YH:, etc.) it is possible that the counter information will not make it to the link file, causing incorrect code generation in some cases. The only workaround is to use sections instead of the L and H modifiers whenever a new set of counters is required; the default counters are not affected. This bug affects all implementations and is corrected in 2.01 and subsequent versions of the assembler. 2) In versions 2.00 and 2.01 of the assembler the EQU directive will not correctly handle relative arguments. If the EQU directive is given a relocatable operand (such as '*', the current program counter value) it will not reflect in the link file that this is a relocatable value. This problem frequently crops up when the following programming idiom is used to establish a label value: FOO EQU * In this particular case, the EQU is used simply to assign the current value of the program counter to the symbol FOO, which can also be done by coding FOO on a line by itself. In other instances, however, there is no simple workaround, short of editing the link file directly. This problem is fixed in version 2.02 and subsequent releases of the assembler. 3) Two local label bugs have been found, one affecting all implementations, the other apparently only a problem on PCs. In the first case the assembler was not switching local label lists during macro expansion on the second pass. The second bug is very sporadic and occurs as a result of improper use of freed dynamic memory space. In both cases the assembler will complain that the local label in question is undefined on pass 2. 4) The description of the assembler message "Phasing error" in the manual is too stringent. Phasing errors IN ISOLATION constitute an internal assembler error. However, cascading phasing errors can result from a number of other error conditions which cause the value of labels and the program counter to get out of synch. If there are other explicable errors in the assembly in conjunction with phasing errors, it is recommended that these errors be investigated and corrected first. The phasing errors might possibly disappear as a result. 5) Under the description of the -B and -L command line options in chapter 1, the assembler manual states that if a pathname (e.g. a directory path) is not supplied along with the filename, the file will be created in the current directory. In version 2.00 this is true only if an explicit filename is supplied. If the filename is omitted, "the assembler will use the basename (filename without extension) of the first filename encountered in the source input file list and append the appropriate file type...to the basename." This means that if the first filename encountered in the source input file list includes a directory path, the object/listing file will be written to that directory. In subsequent releases (2.01 and later) this behavior has been changed such that unless an explicit directory path is specified, the output files will go to the current directory as documented. 6) In chapter 3 of the assembler manual the descriptions of the shift operators in expressions are backwards. The shift right operator causes the left operand to be shifted to the right by the number of bits specified by the right operand, and the sign bit is extended. The shift left operator causes the left operand to be shifted to the left (and zero-filled) by the number of bits specified by the right operand. 7) In relocatable mode the argument to the END directive is not passed correctly as a relative value to the link file, and subsequently to the load file. This can cause loading software to incorrectly set the starting address for program execution. 8) The assembler does not correctly flag a force of I/O short addressing to or from P memory. For example, given the following instruction: MOVEP Y:(R7),P:<<$FFEE The assembler should issue an error, but instead generates an improperly encoded instruction. 9) The assembler may not be able to find macros invoked with the same name from different sections if the macros are defined via the MACLIB directive. A bug in the MACLIB check code causes a macro looked up through MACLIB to be passed over ??lls. 10) If the assembler S or CRE options are used and the same set of symbols are defined both inside and outside of a section a segmentation violation can result on Sun and VAX machines. This is because the count of symbols is not accurate when symbols by the same name are defined privately inside a section and then again in the global name space. 11) The librarian will only accept lowercase command line options. Also, a library name specified in uppercase but without extension will not have the .LIB extension appended correctly. 12) If the assembler PAGE directive is used without arguments inside a macro, when the macro is expanded the PAGE directive will case a page eject (form feed) even if the NOMEX option is given. 13) The assembler will generate an inappropriate error message when two relative terms are subtracted. The subtraction will not be done. 14) It is possible for the assembler to get out of phase if local labels are used in conjunction with structured control statements. The assembler will complain about undefined symbols on pass 2, citing a local label that has been placed between a structured control statement initializer (.FOR, .WHILE, .REPEAT) and its ending statement (.ENDF, .ENDW, .UNTIL). 15) The standalone version of the assembler for the Macintosh (ASM56000SA) does not handle negative fractional values properly. If the assembler sees DC -0.5 it should produce a hex value of C00000, but instead saturates with a positive signed value (7FFFFF). This is not a problem in any of the other assembler implementations, including the MPW version for the Macintosh. 16) The linker will not properly resolve external references to symbols in library modules which contain sections with more than one memory space. When scanning the library for eligible symbols, the linker does not recognize SYMBOL records beyond the first one. If the module is removed from the library it will be linked correctly. ------ u3 - Digital Signal Processor Operation (512)440-2030 SUPERSEDES ANY PREVIOUS ERRATA February 19, 1988 XSP56001 2B38E/4B38E Silicon Design Memo This memo lists and explains the only known discrepancies in the B38E mask set of the XSP56001. These discrepancies will be corrected in future DSP56001 mask sets. 1. If one uses the 8 bit synchronous data format (WDS2, WDS1, WSD0 = 000) of the Serial Communications Interface (SCI) with an external transmitter clock (TCM = 1), the transmitter will transmit two bits when the external clock begins, even when the transmitter is disabled (TE = 0). After transmitting these two bits the device will continue to drive the Transmit Data line (TXD) with the value of the last bit transmitted. The user can avoid this situation by writing all ones ($FF) to the Transmit Data Register (STX) when the SCI is initialized. This will cause the first two transmitted bits to be ones; since there is no start bit in the synchronous mode, the TXD line will remain at a one, i.e., it's idle condition. 2. If one is doing hardware development support for the DSP56000, the following comments will be of interest. If a peripheral interrupt is pending, enabled, and ready to be serviced when the TRACE INTERRUPT is generated, that peripheral interrupt will only be partially serviced by the DSP56001. This condition will not arise if one disables all peripheral interrupts prior to tracing. We have implemented the following solution in the DSP56001ADS Application Development System monitor program software. First, save the Interrupt Priority Register (IPR) value (located at X:$FFFF), then clear the IPR. Set the Trace bit on the stack, then execute the RTI to trace one instruction. After tracing is complete, the IPR register is read. The previously saved value is restored only if the IPR is equal to zero. Also, refer to Page 8-7 in the DSP56000 User's Manual. Note: Tracing a MOVE #0,X:$FFFF instruction is not detectable and will cause the IPR value prior to the trace to be moved into the IPR register after the trace mode interrupt. 3. When Bus Request is asserted, Bus Grant will be negated for one instruction cycle after a read-modify-write instruction, e.g., BSET, is executed from internal program memory on internal X and Y memory locations. 4. Fast interrupt service routines consisting of a single two-word instruction that return to a conditional JSR (condition true) cause both the loop and the trace flags to be cleared and corrupt the interrupt mask bits (I1,I0) in the Status Register (SR) that is stacked. This condition may have a low probability of occurrence in the user's program, but for complete software reliability it is strongly recommended that the user restrict fast interrupt routines to only two single word instructions. 5. When operating in the Host DMA transfer mode (read or write), the user must not switch bit transfer modes prior to completing the current byte/word. A change in the data bit transfer rate prior to completion of the current data bit/word will result in subsequent data being corrupted. The user must therefore complete all byte transfers for the programmed word size. For example, if doing a 24- bit DMA transfer HACK must be asserted three times to transfer all three bytes before the Interrupt Control Register (ICR) bits HM0 and HM1 can be changed successfully. 6. HR/W must be held high during DMA transfers through the Host Port.